ADVANCED PROFESSIONAL KNOWLEDGE COURSE

TROPICAL METEOROLOGY &SATELLITE METEOROLOGY

NOTE: - READ THE STARTER KIT WHICH HAS BEEN SENT TO YOU. IN CASE OF DOUBTS YOU MAY CONTACT TRAINING COORDINATION OFFICER (TCO) ON TELE: 2367000/7314

NOTE: - It has been the endeavour of Faculty of Meteorology to update courseware with the latest Policies/ Orders/ Instructions. However the contents of this book are to be used as course material, and not to be quoted as authority at any stage. Officers may refer to the references quoted in the text for official use.

RESTRICTED

INDEX: BOOK – 11/12

Subject Chapter Page No.

SynopticMeteorology

4. EASTERLY WAVE 5 - 20

5. NORTH EAST MONSOON 21 - 30

Tropical Meteorology

4. MEDIUM AND LONG RANGE PREDICTIONOF INDIAN SUMMER MONSOONS 31.- 38

5. TROPICAL STORM 39 – 55

6. PAIR OF CYCLONES 56 - 59

Satellite Meteorology

3. TEMPERATURE RETRIEVAL AND MEASUREMENT OF WINDS AND PRECIPITATION ESTIMATION 61 - 99

4 .SPACE METEOROLOGY 100-124

RESTRICTED

2

RESTRICTED

AMENDMENT RECORDS

Date Amendment Page No.

Authority

RESTRICTED

3

RESTRICTED

SYNOPTIC METEOROLOGY

RESTRICTED

4

RESTRICTED

CHAPTER – 4

EASTERLY WAVES

Definition

1. Easterly waves exist in the tropical region of both the hemispheres, all the year round. However, they are more marked in the northern tropical region during June to September. Also the study has been more for the northern hemisphere than for the southern hemisphere. Hence more information is available for northern hemisphere easterly waves than for southern hemispheric easterly waves.

2. Starting from the west Central Pacific, we see that the easterly waves here begin as weak low-pressure systems, near ITCZ, growing in to tropical depressions and occasionally in to typhoons which affect the Philippines, South China Sea, Vietnam, south-east China and Japan. The easterly waves cross Vietnam, Thailand and Burma and reach North Bay of Bengal. Under favourable circumstances, they may lead to the formation of monsoon depressions over the head Bay of Bengal, which depressions move inland and cause heavy rain over the central parts of India. The easterly waves move across India in to the east Arabian sea and weaken over the west Arabian sea. They can be traced over North Africa, west of longitude 30°E. They gain in intensity as they move westwards, attaining maximum strength near longitude 5°W. Crossing the west African coast, they move in to the east Atlantic, weaken and then move into the central Atlantic and west Atlantic. Here, the easterly waves generally intensify. Some of these waves trigger the development of West Atlantic hurricanes which affect the West Indies, Gulf of Mexico and southeast coast of U.S.A.

3. Some easterly waves cross the West Atlantic Ocean and Central America and move into the eastern Pacific. Here, they generally weaken and do not appear to undergo any significant intensification until they reach the west central Pacific from where we started.

4. During their movement, the easterly waves undergo considerable variations in intensity, wavelength, speed, horizontal tilt, vertical tilt, thermal structure and position of cloudiness and rain relative to the trough line of the wave.

5. Defined as the perturbations in the waves of Easterlies close to ITCZ propagating from east to west normally seen in the lower and middle troposphere. Over Indian region it is seen in the middle & upper troposphere.

Difficulties in the Studies of Easterly Waves

6. (a) Relative feebleness of the migratory easterly waves compared to stationary seasonal waves in the tropics.

(b) Lack of uniform structure of the easterly waves in different regions.

(c) Lack of data in the tropics.

RESTRICTED

5

RESTRICTED

Lack of Uniform Structure

7. Structure depends on structure of zonal current. Hence they have widely different regional and seasonal characteristics. In the typical case where the wave moved slower than the basic current in the lower levels and faster than the basic current in upper levels, the area west of the wave trough was characterized by subsidence and fair wx while areas of convergence and disturbed weather occurred east of the trough.

8. In some regions these have a sinusoidal appearance throughout their depth. In the neighbourhood of ITCZ or near equatorial trough (NET) they often assume the appearance of closed lows in lower levels and near sinusoidal form aloft.

9. In some regions these waves give rise to closed lows and depressions in lower levels. These lows move independently of the easterly waves (of mid/ upper troposphere).

10. In the neighbourhood of ITCZ there are westerlies below and easterlies aloft. Vertical coupling between lower and upper troposphere leads to east to west movement of the perturbations even in the lower tropospheric belt of westerlies.

11. In the upper troposphere over Southern India during Northern summer there are strong easterlies associated with TEJ. Waves in these easterlies are seen as perturbations in the strength of the zonal flow without significant variation in the wind direction.

Synoptic Studies

12. Dunn (1940) from the analysis of sea level pressure pattern observed a series of isallobaric centres moving westward across the tropical Atlantic during the hurricane season. The classical Easterly wave model was developed by Riehl and associates during and after World War II. They found that an isallobaric centre moved westward accompanied by wave like oscillations in the basic Easterly current in the lower troposphere. These waves moved at an average speed of 10 to 15 kt, reached their maximum intensity in the layer from 600 to 500 hPa and sloped eastward with height. The basic Easterly wave model is shown in Figure.

Satellite Studies

13. Merritt (1964) and Frank & Johnson (1968 & 1969) carried out the first ever study of Easterly waves using satellite pictures over the tropical Atlantic. The following are the conclusions:-

(a) The ‘Inverted V’ Formation. Weak Easterly waves that cross the Atlantic between Africa and the Antilles produce a readily recognizable cloud pattern of curved bands, which bulge poleward. The formation resembles an inverted letter 'V'. Frank (1968) proposed this model and adopted the name because the cloud bands were arranged in a herringbone pattern somewhat resembling a nested series of upside down ‘V’s. Figure shows a simplified schematic diagram of this model. The inverted ‘V’ pattern is associated with weak Easterly waves that have no surface vortex. The perturbation in the wind field of these waves, while they are near the African Coast is strongest

RESTRICTED

6

RESTRICTED

between 5,000 & 15,000 feet. The wave has small amplitude in the surface wind field. The cloud bands are aligned to the low-level wind shear and can have more amplitude than the surface flow. The ITCZ cloud band may or may not widen and bulge northward at the cloud axis.

(b) Vortex Formation. Easterly waves often develop a vertical pattern with significant amounts of cloud and weather. If the rotation producing the vertical pattern is weak, it may be only slightly noticeable or completely undetectable in the surface wind field. Figure shows a schematic diagram of tropical vortex and associated cloud pattern.

Northern Hemispheric Summer Easterly Waves

14. Easterly waves exist in both northern and southern tropics all year round.

15. These are seen more prominently in the neighbourhood of ITCZ than elsewhere.

16. Seen all round a latitude circle.

17. In some regions these originate in some regions get rejuvenated, in some regions decay and die.

Summer Easterly Waves over Caribbean Sea

18. Over the Caribbean Sea, the ITCZ runs roughly east – west along longitude 8ºN. The zonal easterly current is strong in lower troposphere (10 m per sec) and decreases to near zero at 400 hPa with westerlies aloft.

19. Tropical Waves and Associated Weather. Riehl (1954) studied the Easterly waves over the Caribbean region and gave the following model with respect to the weather for the classical Easterly wave:-

(a) In Ridge. Trade cumulus of average height, no precipitation.

(b) Ahead of Trough. Cumulus humilis, few build-ups, strong haze and no precipitation.

(c) Close to Trough Line. Cumulus of above-average development, some Cirrus and Altocumulus, improving visibility and scattered showers.

(d) At Trough Line. Large Cumulus and Cumulus congestus, broken to overcast Cirrus, Altocumulus. Frequent showers or rain.

(e) To Rear of Trough Line. Cumulus congestus and Cumulonimbus, with layers of Stratocumulus, Altostratus & Altocumulus and Cirrus. Frequent moderate to heavy showers, with light rain between showers. Thunderstorms.

(f) Eastern Outskirts. Large Cumulus and Cumulus congestus occasionally still Cumulonimbus. Some Stratocumulus, Altocumulus and Cirrus. Moderate showers, decreasing.

RESTRICTED

7

RESTRICTED

20. The weather model as found by Riehl is not always seen in a tropical Easterly wave. Based on the wind and weather pattern, the waves have been classified in the following types in this study:-

(a) Type I. The zonal Easterly wind component of the basic flow is slower in speed than the Easterly wave in the lower levels, with opposite situation in the mid and upper troposphere.

(b) Type II. The low-level winds are westerly, decreasing with height and weak easterlies aloft. Strong easterlies in the mid & upper troposphere.

(c) Type III. The zonal easterly wind component of the basic flow is faster in speed than the speed of the wave in the lower troposphere with opposite situations in the mid and upper troposphere.

21. In Type I & II situations, there is low level divergence, leading to subsidence and relatively clear air to the rear of the trough and cloudy/rainy conditions ahead. However in the case of Type III, there would be cloudiness and rain in the rear of the trough and relatively clear weather ahead of the trough.

Easterly Waves in Indian Region

22. It had been known before the end of the nineteenth century that low pressure waves from the east significantly affect the weather over India and neighbourhood. These waves also helped in the formation of monsoon depressions over the North Bay of Bengal. Mainly due to lack of a dense network of upper air observations in the affected regions, a 3-dimensional model of these easterly disturbances has not been formulated for this region. Situation is as presented below:-

(a) There is a large annual monsoonal oscillation of the zonal winds in Southeast Asia. Figs. 8.4 (15), 8.4 (16), 8.4 (17) and 8.4 (18) based upon the

climatological charts prepared by Ramage & Raman (1972), give the vertical cross-section of zonal winds during winter (January) and summer (July) seasons, along the meridians 73°E and 100°E, which are fairly representative of the southeast Asian region.

Zero line to the north marks the position of the sub-tropical ridge line and the zero line to the south marks the position of near-equatorial trough line.

During the summer season, this trough line is called the ITCZ. The trough is confined to the lower levels while the ridge extends throughout the troposphere. Between the sub-tropical ridge and the near-equatorial trough, we have the lower tropospheric tropical easterlies. In respect of the zonal winds, the following points are noteworthy:-

(i) Over the equator and in the near-equatorial latitudes, in both the seasons, we have westerlies in the lower troposphere and easterlies aloft. This is indeed true for the whole year.

(ii) From winter (January) to summer (July) season, these lower tropospheric westerlies as well as the upper tropospheric easterlies both increase in intensity and also spread northwards.

RESTRICTED

8

RESTRICTED

(iii) From winter to summer season, the lower tropospheric tropical easterlies considerably shrink in latitudinal width.

(b) As a result of this large monsoonal oscillation, the zonal flow in the Indian region cannot be modelled in a simple form which could be assumed to hold for the year as a whole. Correspondingly, it is difficult to construct a model of easterly wave applicable for the whole year.

(c) Throughout the year, the easterly waves are seen through their marked effect on the weather in the neighborhood of the near-equatorial trough line which runs close to 5°N in the winter season and close to 25°N in the summer season. The amplitude of the perturbation is generally largest near this trough line and decreases away from it in the lower troposphere.

(d) Along the trough line, one can always draw a series of quasi-stationary vortices or closed low pressure systems, called seasonal lows. The position and intensity of these seasonal lows is determined largely by orography and the surface temperatures. In general, leeward side of hilly regions and areas of relatively warm temperatures at the bottom of the atmosphere are the preferred regions for the anchoring of the seasonal lows in the lower troposphere.

(e) The arrival of easterly waves is detected by increase of disturbed weather in a region, in which we already have curved or closed isobars and streamlines, and in which the weather is already somewhat disturbed.

(f) In its extremely simple form, the weather development can generally be inferred or interpreted in terms of the following simple rules:-

(i) There is low-level convergence and hence ascending motion in the region of low-level cyclonic vorticity (Ekman pumping). The opposite occurs when the relative vorticity is anticyclonic.

(ii) When a particle enters a region of higher absolute vorticity, it suffers horizontal velocity convergence (simplified form of vorticity equation (dη / dt) + η D = 0).

(iii) In the region of surface pressure fall, the tendency is towards development of more cloudiness and rain (isallobaric convergence).(iv) Diurnal variation of winds as well as local orography have got to be taken into account in computing or estimating low-level horizontal velocity convergence and vorticity motion at a place.

23. Found throughout the year but marked effect: In winter up to 5ºN and in summer up to 25ºN. Maximum amplitude near NET and decreases away from it in lower tropospheric levels.

24. Due to large seasonal changes it is difficult to construct a model of easterly waves applicable for the whole year.

25. Difficult in locating due to lack of data.

RESTRICTED

9

RESTRICTED

26. It is difficult to define structure of easterly waves in this region due to their interaction with quasi stationary eddies and lack of data.

27. These waves were initially identified with the aid of the following:-

(a) P24P24 change charts. (b) Pressure departure from normals.(c) Occurrence of rain over Myanmar

28. For weak easterly waves, the diurnal variation and local topography may lead to convergence and divergence more than synoptic scale convergence and divergence of model easterly wave.

29. The total picture seen on a synoptic chart is the combination of seasonal and the migratory systems. It takes such a variety of forms that we may end up saying “There is nothing like Easterly wave over Indian region”.

Summer Easterly Waves over the Indian Region

30. The frequency and intensity of these waves determine the intensity of monsoon activity over India and neighbourhood and hence determine the occurrence of floods or droughts in these areas.

31. The seasonal quasi-stationary flow pattern over India and neighbourhood itself shows pronounced wave pattern and closed low pressure systems. There is interaction between the strong quasi-stationary eddies and the weak migratory easterly waves. Anomaly in the position and intensity of the quasi-stationary eddies is associated with the arrival and movement of the easterly waves. Due to this interaction, it is more difficult to define the structure of the easterly waves in this region. There is also lack of dense network of upper air observations. Due to these reasons, interpretation of spectral analysis results encounters difficulties. The results given below are based mostly on synoptic analysis and should be considered as tentative at this stage.

32. Figs. 8.4(19), 8.4(20), 8.4(21), 8.4(22) and 8.4(23) give the July streamline pattern at 850, 700, 500, 300 and 200 hPa levels.33. Figs.8.4 (17) & 8.4(18) give meridional cross-section of July zonal winds along longitudes 73°E and 100°E. These are based mainly on the IIOE atlas prepared by Ramage and Raman (1972). The following points are noteworthy:-

(i) There are two troughs, one quite a distance to the north of the equator called ITCZ and the other close to the equator but south of it. In the lower troposphere, ITCZ is seen running across the whole chart but the south equatorial trough is more in the west and the middle of the charts. These troughs exist in the lower and middle troposphere but disappear at 300 hPa level.

(ii) ITCZ generally slopes towards the equator with increasing height.

(iii) There are well-marked seasonal troughs of cyclonic vorticity along ITCZ.

RESTRICTED

10

RESTRICTED

(iv) At 300 and 200 hPa levels, we have mainly easterly flow between 30°N and 10°S.

(v) The easterly zonal wind is stronger at 73°E than at 100°E. The easterly maximum at 200 hPa and aloft is to the south of latitude 17 ½ °N. it is known that the maximum intensity of the upper tropospheric easterlies is within the longitudinal belt of 73°E and 80°E. Thus the region east of 80°E and north of 17 ½ °N becomes the region of right entrance for easterly wind maximum. As such, it is favourable for upper tropospheric horizontal velocity divergence. Right under it, in the lower troposphere, we have the region of ITCZ. Hence the region of north Bay of Bengal is favourable for upward motion and cyclogenesis.

34. It is over this region that the summer easterly waves move and give rise to low pressure areas, depressions etc.

35. Characteristics. There are two types of easterly waves, one in the lower and middle troposphere and the other in the upper troposphere. There is much uncertainty about the structure of the upper tropospheric easterly wave. Its wavelength and speed are just larger than the lower and middle tropospheric wave. The properties of the lower and middle tropospheric wave are given below:-

(a) Wavelength: 2000 km.(b) Period: 5 to 6 days.(c) Slope: The wave trough slopes slightly eastwards as the wave

approaches the Bay of Bengal.(d) Speed: 4.3 m per sec. (e) Tilt: NNE – SSW.(f) Clouding & Precipitation: Mostly to the west of trough line.(g) Diabatic Heating: About 10º K per day with maximum in the upper

troposphere.

36. Keshavamurthy (1971) studied these waves over Indian region during the summer monsoon season and found the following features:-

(a) A wave is seen in the lower troposphere with a maximum amplitude at 850 hPa, for a period of 5 days, having a wavelength of 2000 Km and speed of 4.3 m per sec.

(b) Another Easterly wave with a period of 7-8 days is marked in mid and upper troposphere (maximum amplitude at 400 hPa having wavelength of 3500 Km and speed of 5.7 m per sec).

(c) The waves had a NNE-SSW tilt and sloped eastwards when these waves approached North Bay.

(d) Weather occurred mostly ahead of the wave, to the West of the trough line.

37. Spectral analysis by Murakami (1976) of the wave disturbances over India during the summer monsoon showed that there exists westward propagating

RESTRICTED

11

RESTRICTED

disturbances with a period of 4-5 days and a wavelength of 3000 Kms from the North Bay of Bengal through the monsoon region.

38. Balachandran et al. (1996) studied a case of Easterly wave over the Indian region between 21 Dec 1995 and found the following:-

(a) The remnant of depression, which moved across the South China Sea, could trigger Easterly wave.

(b) Though there is a temporary increase in the intensity of the system between 85°E and 95°E longitudes, the system gets progressively weak as it moved westwards.

39. Results of Spectral Analysis over Indian Region during Monsoon Season. Keshavamurty (1971).

Period(Days)

Level of Max Intensity (hPa)

Vertical Extent (hPa)

Wave Length

(Longitudes)

Speed(deg per day)

5 - 6 850 SLC - 400 20 Long 3.67 - 8 450 700 – 200 35 4.7

(a) Tilt. NNE – SSW.

(b) Slope. Slightly eastward.

(c) Precipitation & Clouding. The rain is mostly ahead to the west of the trough line. An air particle approaches the trough line from west and it experiences an increase of absolute vorticity, horizontal and upward motion.

Winter Easterly Waves over Indian Region

40. Cloud Signatures. The cloud signatures associated with the Tropical Easterly wave are of the following type:-

(a) Convective Cloud Cluster (CCC).

(b) Vortical Appearance (VA).

(c) Latitudinal Band (Lat Band).

(d) Longitudinal Band (Long Band).

(e) Inverted 'V' Type.

41. Over the Bay of Bengal, the cloud signatures associated with the Easterly waves are of the Convective Cloud Cluster (CCC) type, with other patterns observed only at times over South-Central Bay. These cloud signatures can be identified and tracked at least 12 to 24 hours in advance from 100°E.

RESTRICTED

12

RESTRICTED

42. Speed of the Wave. The average speed of the wave based on satellite imagery, between 100°E and 85°E is 250 Km per day and that based on NCMRWF charts is 320 Km per day.

43. Structure. Well-marked waves can be seen between 850 & 700 hPa. They are more prominent at 700 hPa and at times can be marked up to 500 hPa.

44. Slope and Tilt. The wave has an eastward slope with height and a NW-SE tilt at 850 & 700 hPa levels. However at times, a NE-SW oriented tilt is also observed.

45. Dynamical Aspects of the Wave. The regions of occurrence of weather coincides with theory, based on the dynamic aspects of the wave. The waves are observed mainly as Type I (with weather occurring ahead of the trough) and Type III (with weather occurring to the rear of the trough axis). However, a combination of Type I and Type III are at time observed. No wave of Type II was observed during the winter months.

46. Clouds and Precipitation. Associated weather with respect to the position of the trough was found to be consistent with the dynamics of the Easterly trough. However, the Easterly waves that moved across Bay of Bengal, did not show any typical weather pattern. Weather occurred on both sides of the trough axis, mainly ahead of the trough. The passage of the wave gave scattered to fairly widespread precipitation over Bay Islands.

Energetics

47. In the initial stages of development, barotropic instability (Kz→ KE) is a contributory factor, perhaps the main contributory factor.

48. There are certain preferred geographical locations, particularly the lee side of the mountains, where easterly waves are either generated in situ or get intensified when travelling through these regions. Such preferred geographical locations are already regions of considerable relative cyclonic vorticity.

49. Release of latent heat of condensation is a factor which always contributes to the energetics of the easterly waves.

50. In the region of large-scale upward motion, cloudiness and rain, there is found to be warming in the middle and upper troposphere but cooling in the lower troposphere in and near the Planetary Boundary layer (PBL).

Cooling in the lower troposphere is due to the following three factors:-

(a) Cut-off of solar heating by cloudiness.

(b) Evaporational cooling and contact cooling when cold rain drops pass through relatively warm unsaturated air in the lower troposphere.

(c) Cold down-draft from the cloud base. Warming in the middle and upper troposphere is essentially due to the release of heat of condensation inside

the cloud and detrainment and horizontal spreading of the warm cloud air around.

RESTRICTED

13

RESTRICTED

51. In the region of the easterly wave as a whole, there is energy-consuming indirect circulation (rising of cold air and sinking of warm air) in the lower troposphere and energy-producing direct circulation (rising of warm air and sinking of cold air) in the middle and upper troposphere.

52. Production of energy by the direct circulation in the middle and upper troposphere is more than the consumption of energy by the indirect circulation in the lower troposphere in the developing or steady-state condition of the wave. Hence, averaged throughout the depth of the easterly waves there is a net production of energy. This may also be stated as: “Convective condensation generates AE which gets converted to KE”.

53. There is considerable evidence that in the region of an easterly wave, in addition to the above-mentioned warming and cooling associated with cloudiness and clear skies, there is also warming and cooling associated with meridional advection. In general, the vertical shear of zonal wind in the lower and middle troposphere in the tropics is from east to west consistent with the fact that the warm air is to the north and the cold air is to the south. In the field of easterly wave trough, the air moves from north to south advecting warm air in front of the trough; in the rear of the easterly wave trough, the air moves from south to north advecting cold air behind the trough. As a result, the trough bends backwards towards the east with increasing height. Thus Az is converted to AE.

54. AE will be converted to KE if there is upward vertical motion ahead of the trough and downward vertical motion in the rear of the trough.

55. If the low-level wind is westerly or weak easterly such that relative to the moving trough, the air is moving towards the trough ahead of the trough and is moving away from the trough in the rear, then by vorticity equation, the air moves into regions of increasing absolute vorticity ahead of the trough, experiencing low level horizontal velocity convergence and upward vertical motion; the opposite occurs in the rear of the trough. In such a situation, AE gets converted to KE. Such a situation appears to obtain, during northern summer season, in west central Pacific, Southeast Asia, North Bay of Bengal, central and west India and West Africa. This process is qualitatively similar to baroclinic energy conversion in extra-tropical cyclone waves, AZ→AE→KE.

56. It is also conceivable that forcings from sub-tropical ridges on the northern and southern boundaries of tropical region make some contribution to the whole class of easterly wave motions in the tropics.

57. In summary, we may state as follows: “There are extra-tropical forcings on the tropical atmosphere as a whole; these forcings seem to make some contribution to the energetics of the tropical easterly waves. However, inside the region of an easterly wave, barotropic, baroclinic as well as convective condensation are the principal factors providing the kinetic energy of the wave. All these three processes seem to work co-operatively throughout the life history of the easterly wave. There is evidence to suggest that Rayleigh-Kuo type barotropic instability is the principal source of KE in the initial stages of wave-growth. After the initial growth, convective condensation and the baroclinic process (AZ→AE→KE) are the two important processes, the former being the more dominant process”.

RESTRICTED

14

RESTRICTED

58. Convective condensation is the principle component of CISK (Conditional Instability of Second Kind).

Easterly Waves - Possible Sources of Energy

59. (a) Barotropic Instability -Rayleigh-Kuo Type- Lorenz Type (No comments)

R-K Type Barotropic Instability does make a contribution at least at the time of initial developing stage of the waves. After the initial trigger provided by this Barotropic Instability, the energy supply from this source alone is generally far too small to make the wave grow to its full intensity and maintain itself against friction.

(b) Baroclinic Instability with Dry Adiabatic Process. This is unlikely to be a major source of energy, except perhaps is very shallow layers where unusually large vertical shears of wind may develop and persist for some time.

(c) Dynamical Instability in the Presence of Condensation /CISK.CISK is believed to be a very important process for development of

synoptic-scale tropical systems in general, and easterly waves in particular.

(d) Gravitational Wave Energy. Planetary-scale inerto-gravity waves are known to be executed by large scale orographic obstacles and differential heatings. Some of these waves are trapped in the Tropical upper troposphere and lower stratosphere. These waves are considered to provide energy for the mixed R-G waves and the Kelvin waves of the lower stratosphere. Gravitational waves are found to be providing energy for Easterly waves.

(e) Forcing From Sub-Tropical Ridges. Forcing from sub-tropical ridges also make some contribution to the energetics of tropical easterly waves.

Mak’s Hypothesis on Energetics of the Tropical Atmosphere

60. Fig. 8.5 (1) shows Mak’s hypothesis on the energetics of the tropical atmosphere. Its main points are as follows:-

(a) Tropical eddy motions are primarily a response of the dynamically stable tropics to the unstable baroclinic processes in higher latitudes.

(b) There is only one source for KE and that is external, coming from extra-tropical latitudes. The largest sink for KE is its conversion to Kz; the frictional dissipation is also substantial.

(c) There is only one source for AE and that is the conversion from KE by indirect circulation. The losses for AE are partly due to radiation and partly due to conversion to Az. The latter means warming of the already warm latitudes and cooling of the already cool latitudes (refrigeration process). However, amounts of incoming and outgoing energy associated with AE are small.

RESTRICTED

15

RESTRICTED

(d) Budget of Az and Kz cannot be deduced from this limited model.

(e) It is conceivable that since precipitation tends to be concentrated along narrow regions in the tropics, the release of latent heat, by and large, mainly affects the zonally averaged motions and has only minor effects upon the large-scale wave motions over the rest of the tropics.

Tropical Wave

61. A tropical wave is defined as, “a trough, or cyclonic curvature maximum, in the trade-wind easterlies. The wave may reach maximum amplitude in the low or middle troposphere or may be the reflection from the upper troposphere of a cold low or equatorward extension of a mid-latitude trough.” In addition to these mechanisms, tropical waves can be the poleward inflection of low-level cyclones in low latitudes.

62. Since there has always been more data available in the sub-tropical trade-wind zones than in more equatorward latitudes, it has been easier for meteorologists to track the waves than the low-level vortices with which they may have been associated. Similarly, wind data from the upper troposphere have been so sparse that upper cold-lows causing low-level vortices tropical waves could not be identified until the recent availability of wind data from jet aircraft and satellite cloud pictures.

Equatorial Wave

63. The equatorial wave model was postulated by palmer from observations taken in the Marshall Island area of the central pacific. He visualizes these waves as perturbations of the convergent easterly current in the near-equatorial latitudes, having a wave length of roughly 15º of long and an average propagation speed of 10 to 15 Kt toward the west. The waves are most pronounced near the equator and become greatly damped northward & southward in each hemisphere. The equatorial-wave model and associated convergence field is shown in figure.

64. Palmer did not offer a characteristic of the central pacific between 160°E & 150°W. He also asserted that cyclonic vortices frequently develop in these waves and occasionally increase to tropical storms. In fact, palmer postulated that all tropical storms resulted from unstable waves in the easterlies, i.e., equatorial waves or easterly waves, and that mean west winds observed in near-equatorial regions are only a statistical result of many traveling vortices.

65. There has been little further synoptic research on the equatorial-wave model, particularly since weather satellite data have become available. Several theoretical studies of equatorial waves show fair agreement with the empirical model of palmer; however, to obtain theoretical solutions, a large number of simplifying assumptions had to be made. Therefore, at the present time, the equatorial wave model should be regarded as somewhat speculative and requiring much more research and verification before we can recommend it be applied operationally.

RESTRICTED

16

RESTRICTED

RESTRICTED

17

RESTRICTED

RESTRICTED

18

RESTRICTED

RESTRICTED

19

RESTRICTED

RESTRICTED

20

RESTRICTED

CHAPTER – 5

NORTH EAST MONSOON

Chapter objectives.

After reading this chapter, you should be able to

Understand the term “North East Monsoon(NEM’ Understand the process of Establishment of North East Monsoon. Know the areas and distribution of rainfall during this period. Understand the factors influencing NEM

Structure

1. Introduction.2. Regions affected.3. Month wise wind flow pattern over India during the period September-

February4. Onset, withdrawal and duration of northeast monsoon

Introduction

1. The Indian northeast monsoon (NEM), also known as the retreating southwestmonsoon (SWM), is a small scale monsoon confined to parts of southern Indian peninsula with duration October-November-December (OND). It is associated with the seasonal reversal of surface and lower tropospheric winds from southwesterlies (during the SWM season of June–July-August-September (JJAS) to northeasterlies which set in over the Indian region in October (India Meteorological Department (IMD), 1973). In a broader perspective, it is also associated with the northern hemispheric winter circulation dominated by a strong surface high pressure region over Siberia, a primary low over eastern equatorial Pacific region and secondary shallow lows over the north Indian ocean. Though the SWM is the most important weather phenomenon for the country providing it with nearly 75% of the normal annual rainfall of 115 cm (IMD, 2010), the NEM is also an important rainfall season for parts of southern peninsular India. The state and meteorological sub division of Tamil Nadu (TN) is the major beneficiary of the NEM rainfall, which is prodigious over the belt of Coastal Tamil Nadu (CTN). Over CTN, the onset of NEM, preceded by a well defined reversal of low level winds from southwesterlies to northeasterlies, is clearly marked. A synoptic / climatological overview of Indian NEM is given in IMD(1973). In this chapter we present a brief description of a few aspects of Indian NEM viz., rainfall climatology, flow pattern, onset and withdrawal, synoptic scale systems, certain thermodynamic features, relation with global features and seasonal forecasting.

RESTRICTED

21

RESTRICTED

Regions affected by NEM and their rainfall climatology

2. Fig.13.1 presents the geographical locations of five meteorological sub divisions of India viz., Coastal Andhra Pradesh (CAP), Rayalaseema (RYS), South Interior Karnataka (SIK), Kerala (KER) and TN (including the union territory of Pondicherry as well). Here, TN and KER are separate states, SIK is a part of Karnataka state, RYS and CAP are parts of the state of Andhra Pradesh. The boundaries of Bay of Bengal (BOB) which lies east of the southern Indian peninsula and the Arabian Sea (AS) that lies to the west. Both BOB and AS together constitute what is known as the North Indian Ocean.

3. Normal rainfall. The seasonal normal rainfall distribution over the above region for the season OND is depicted in Fig.13.3. Table 13.1a presents the normal monthly, seasonal and annual figures of the above five sub divisions based on the 50 year data of 1951-2000 (IMD, 2010). The rainfall figures expressed as percentage of annual rainfall are also given. In Table 13.1b are presented similar rainfall figures for 14 stations which by and large represent the various regions influenced by NEM. The inferences which could be drawn from the normal monthly and seasonal rainfallfigures are described herein below, subdivision wise:

(a) CAP: This sub division receives 1024 mm of annual normal rainfall out of which 327 mm (32%) is contributed by NEM. The month of October is the major rainy season for CAP. However, there is wide latitudinal variation of rainfall. Visakhapatnam, located on the north CAP (NCAP) at about 17.5°N receives 302 mm during OND, October alone contributing 219 mm. Machilipatnam, located on the central stretch at about 16°N receives 355 mm during OND. Nellore situated on the southern parts at 14°N receives 656 mm with a breakup of 260 mm in October, 295 mm (November) and 104 mm (December) thus experiencing rainfall throughout the season.

(b) RYS: Rainfall over this sub division during OND is 219 mm (31% of annual rainfall of 706 mm). Anantapur representing northwestern dry parts of RYS receives annual rainfall of 567 mm with NEM contribution at only 154 mm. Tirupati located in the southern parts of RYS receives 934 mm of annual rainfall with 467 mm realised during OND.

(d) SIK: SIK receives 1019 mm of annual rainfall and 210 mm during OND (21% of annual). Bangalore located in SIK receives 977 mm of annual rainfall and the NEM seasonal contribution is 245 mm only. Most of this rain is received in October itself (171 mm).

(e) KER: This sub division receives 478 mm of rainfall during OND. But the NEM rainfall is only 16% of the annual rainfall of 2928 mm, wherein, the main contribution comes in during the southwest monsoon season of JJAS (2051 mm, 70%). Kochi, located on the west coast of India in central Kerala at 10°N, receives 497 mm of rainfall during OND, but this becomes insignificant compared to its annual rainfall of 3147 mm. Thiruvananthapuram, situated at 8°N in southern Kerala, receives 528 mm of rainfall during OND, the annual rainfall being 1792 mm.

RESTRICTED

22

RESTRICTED

(e) TN: NEM rainfall of Tamil Nadu (NRT) accounts for 48% (OND, 438 mm) of its annual rainfall of 914 mm making this state / sub division, the major beneficiary of NEM. This is also the lone meteorological sub division of India where rainfall realised during NEM is significantly more than that during the southwest monsoon season (JJAS, 317 mm). There is however wide spatial variation of rainfall within the state. At Salem, located in the northwestern arts of the state, annual rainfall is 1020 mm, JJAS rainfall is 516 mm and OND rainfall, 332 mm only. Coimbatore, located in the southern portion of the northwestern parts is drier than Salem, with OND contributing 309 mm which is nearly 50% of the annual rainfall of 614 mm. Tiruchirapalli, located in the central interior parts receives 872 mm (annual) with NEM contributing to 393 mm. Chennai city located at 13°N receives 857 mm during OND and 1403 mm annually. Vedaranyam, located at 10°N, at the tip of the north-south oriented coastal Tamil Nadu receives prodigious rainfall of 1038 mm during OND making this station the wettest in India during the NEM season. At 347.4 mm, the rainfall realised in the month of December is quite substantial. Tuticorin, located at 8.8°N on the southeast coast of Tamil Nadu abutting Gulf of Mannar, receives 440 mm of its annual rainfall of 631 mm during OND.

Fig.13.1: Geographical locations of the five meteorological sub-divisions of southern peninsularIndia influenced by the Northeast monsoon

RESTRICTED

23

RESTRICTED

Fig.13.3: Normal spatial distribution of rainfall over southern peninsular India during theNortheast monsoon season (OND).

RESTRICTED

24

RESTRICTED

Month wise wind flow pattern over India during the period September-February

4. The IMD has defined four seasons of India viz., winter (January-February (JF)), pre-monsoon or hot weather (March-April-May (MAM)), southwest monsoon (JJAS) and post monsoon or northeast monsoon (OND). During winter for which January is the representative month, the surface pressures which are high over northwestern parts of India decrease towards south. In July however, the pattern reverses with low pressure developing over northwest India and increasing pressures towards south (Fig.13.6a-d) (Rao and Ramamurti, 1983). This reversal is caused by the intense surface heating that takes place over India and neighbouring countries during the MAM season. The southwest monsoon sets in over India by the end of May/ beginning of June and continues up to September / early October. In October the surface pressure pattern changes with the establishment of a dumb bell shaped isobar. The low level winds at 850 hPa during the months of September, October and November are presented in Fig.13.7a-c (Source: Asnani, 2005). As seen, the seasonal trough at 850 hPa is located near 20°N in September but gradually moves southwards to 16-17°N in October and thence to 12-13°N in November. The reversal of winds from southwesterlies to northeasterlies that takes place in October in the peninsula can be clearly observed. This is a prelude to the establishment of northeast monsoon over the southern Indian peninsula. Rao (1976) and Asnani (2005) could be referred which provide excellent description of the wind flow over India during southwest monsoon and as to how the flow changes into northeasterlies after the monsoon’s retreat.

RESTRICTED

25

RESTRICTED

Onset, withdrawal and duration of northeast monsoon

5. The southwest monsoon which is the overriding weather phenomenon duringJune-September starts withdrawing from northwest India in the beginning of September (Rao, 1976). By 15th October, the monsoon withdraws up to 15°N over the peninsula, though the dates may show wide variation in individual years. North of 20°N and in most of the areas north of 17°N, the withdrawal of southwest monsoon is associated with near cessation of rainfall and rise of maximum temperature in October. However south of 15°N, there is a marked rise in rainfall followed by increased clouding and continuation of rainfall in the eastern parts of the peninsula. This marked increase in rainfall occurring after the reversal of winds from southwesterlies to northeasterlies is considered as the northeast monsoon onset. Reference to old issues of Indian Daily Weather Reports (IDWR) published by IMD revealed that northeast monsoon onset was first mentioned in the year 1923 by IMD. Subsequently, the onset was mentioned during some years but not mentioned during several other years. However, since 1977, the setting in of northeast monsoon has been mentioned in the IDWRs every year though phraseologysometimes differed from year to year. There was no well defined and standardcriteria to declare the NEM onset as mentioned in IMD (1973).

Criteria for determination of dates of NEM Onset

6. The 1987 conference of forecasting officers of IMD defined a set of criteria to declare NEM onset over southern peninsula on real time basis (IMD, 1987). These criteria were slightly modified to determine NEM onset over CTN in a diagnostic study (Raj, 1992). The NEM onset dates over CTN were derived for the period 1901-1990 in this study based on daily rainfall of six stations located in CTN viz., Chennai, Cuddalore, Nagapattinam, Vedaranyam, Pamban and Tuticorin

7. The following five rules (R1 to R5) formulated by Raj(1992) were used:

R1: Southwest monsoon should have withdrawn up to coastal Andhra Pradesh

R2: Deep easterlies should have set in over Tamil Nadu or seasonal low should have established in south BOB adjacent to Tamil Nadu coast.

R3: After R1 and R2 are satisfied, the first day of Fairly widespread (FW) rainfall or a higher category over CTN would be the day of northeast monsoon onset.

R4: If the date arrived at by R3 happens to be earlier than 10 October, the winds /surface charts are to be scrutinised to decide as to whether the onset of easterlies are temporary or permanent. If it is permanent, then the date of R3 should be taken as the onset date. If the easterly onset is temporary and if westerlies appear again in the lower troposphere over CTN, the date of permanent onset of easterlies is to be determined and R3 to be applied again.

R5: In case, the date fixed is completely unsatisfactory, a review is to be made and the next date of FW rainfall is to be considered as onset date.

RESTRICTED

26

RESTRICTED

8. IMD Criteria. The criteria for declaring onset of NEM was set by IMD in August 1988 by means of an official circular, which was amended further in August 2006 (IMD, 2008).

The criteria for commencement of NEM rains as per the latest circular are:(a) Withdrawal of SWM up to Latitude 15°N.

(b) Onset of persistent surface easterlies over Tamil Nadu coast.

(c) Depth of easterlies up to 850 hPa over Tamil Nadu coast.

(d) Fairly widespread rainfall over coastal Tamil Nadu, south coastal AndhraPradesh and adjoining areas.

9. As for NEM withdrawal, IMD started announcing withdrawal dates of NEM only from the year 1993 and prior to that no such declaration appears to have been made. A sub-committee constituted by the IMD’s Annual Monsoon Review meet in 2006 submitted a detailed report recommending that real time declaration of NEM rainfall cessation may not be accurate and so may not be resorted to and that the date may be fixed diagnostically by considering rainfall data up to 31st January and a statement to the effect could be released on 31st January (IMD, 2006). This recommendation is still under consideration. IMD has declared cessation of NEM rainfall by considering rainfall and may be a few other parameters such as depth of moisture, temperature etc. in a subjective way but no objective criteria or guidelines to determine NEM have so far been published. In the most recent and extensive study undertaken on this topic, Geetha and Raj, (2011) re-determined the onset and withdrawal dates of NEM for the 100 year period of 1901-2000 based on the daily rainfall data of large number of raingauge stations numbering about 25. Stations of south CAP and CTN which were located within a distance of 100 kms from the coast were considered. The study also freshly determined the dates for the 19 th century 30 year period of 1871- 1900 and the dates for the recent decade of 2001-2010. The methodology used for determination / redetermination of onset and withdrawal dates in the above study are the same as the methodologies described earlier except that instead of 6 stations of CTN, a larger network of 25 raingauge stations of SCAP and CTN have been used.

Synoptic systems that form over Bay of Bengal during NEM season

10. With the shift of the equatorial trough (ET) southwards from September to December, the latitude of formation of low pressure areas also shifts southwards over the BOB. IMD classifies low pressure systems (LPS) based on the maximum wind speed (when they are located over the sea).. Raj(2011) presents a detailed climatology of several aspects of depression (D) / cyclonic storm (CS) / severe cyclonic storm (SCS) which form and move over Indian seas. Most of the statistics have been derived from the data for the 50 year period 1961-2010 which has been extracted from IMD’s Cyclone eAtlas (2008) and its updates. As presented in Raj(2011), during OND, the probability of a BOB depression intensifying into a CS is 61% and that for CS to SCS is 65%, thus exhibiting high probability of intensification. The monthly frequencies of formation of D+CS+SCS, CS+SCS and SCS over BOB during 1961-2010 are 77, 70 and 37 for October, 34, 57 and 22 for November and 19, 42 and 13 for December respectively. The normal area of formation of LPS over BOB is (82.6ºE, 15.1ºN) in October, (86.2ºE, 14.4ºN) in November (84.9ºE, 11.6ºN)

RESTRICTED

27

RESTRICTED

in December, shifting southwards with the advancement of the season . The frequencies of D+CS+SCS, CS+SCS and SCS which crossed the Tamil Nadu coast during OND 1961-2010 are 34, 24 and 19 respectively; for CAP, 46, 29 and 18; for Orissa 14, 10 and 9 and for West Bengal, 10, 7 and 5. For Bangladesh and Arakan (Myanmar) coasts, the frequencies are 36, 25 and 19. Thus large number of BOB LPS move in the northward and northeastward directions. Fig.13.13a presents the tracks of the LPS that formed over BOB during 1961-2010, for the months of October, November and December. In the recent decades, the frequency of LPS crossing Tamil Nadu coast during OND has shown a slight decrease. The frequencies for the decades 1971-80, 1981-90, 1991-2000 and 2001-10 are 6, 4, 11 and 3 respectively. However, the mean rainfall for the decade 2001-2010 for Tamil Nadu is excess by 15% of the LPA thus showing that despite the decrease of frequency of LPS crossing, the NEM sustained an above par activity over the state.

11. Easterly wave (EW) is another transient synoptic scale feature which gives rise to good rainfall during the NEM season. An EW may not always be detected with clarity and precision as normally happens with LPS. However it is possible to track the movement of an EW using vertical time section plots and also by critically studying 24 hr pressure changes (IMD, 1973). Geetha and Raj (2011) studied the EW activity over the southern peninsula for the NEM season of 2010. In this year three EWs were detected and the period of the waves was determined as 4-5 days, the speed of movement as nearly 26 kmph and the amplitude, 2800 km.

Variation of thermodynamic parameters during NEM

12. A study on thermodynamic parameters was undertaken based on 00 UTC (0530 IST) upper air data of Chennai Radio sonde / Radio wind data for the period 1971-80 ( Raj,1996). Some of the results derived in this study are presented below:

(a) During active NEM conditions, an average east to west moisture flux of 21.1 x 108 metric tons per day (mtpd) over an one degree latitudinal wall (approx. 110 km) is transported across CTN.

(b) During an excess NEM year, the mean moisture flux is 16.5 x 108 mtpd whereas the normal flux is 13.3 x 108 mtpd for OND.

(c) The mean liquid water content over CTN for the entire season does not show much interannual variability.

(d) During active NEM conditions, the meridional wind which is northerly in the lower levels veers to southerlies and during dry spells, the meridional winds are northerly at all levels up to mid troposphere.

Relation between NEM and ENSO/Siberian High

13. El Nino and Southern Oscillation (ENSO) are two inter-related global parameters displaying wide and sustained relations with several meteorological features obtained over various parts of the globe. There are numerous studies on the effects of these parameters, reference could be made to Asnani (2005) for a review. The ENSO parameters exhibit considerable influence on Indian SWM which has been extensively studied. Relation between ENSO and the Indian NEM has been

RESTRICTED

28

RESTRICTED

studied recently in a few studies [Sridharan and Muthuchamy (1990), De and Mukhopadhyay (1999), Jayanthi and Govindachari (1999), Khole and De (2003), Raj and Geetha (2008), Geetha and Raj (2011)]. Most of the earlier studies propounded a theory of negative relation between Indian NEM and the southern oscillation index (SOI) and positive relation between the NEM and El Nino, which is opposite to the relation that ENSO exhibits with the Indian monsoon rainfall (IMR). However, in Raj and Geetha (2008) and Geetha and Raj (2011) the relation between SOI and NEM was studied much more critically and it has been found that this relation itself underwent significant intra seasonal variation and also reversed towards the end ofthe season. Table 13.6a presents the CCs between SOI and North east monsoon rainfall of Tamil nadu in antecedent and concurrent modes based on data of a long 104 year period of 1901-02 to 2004-05. As shown, SOI(JJAS) is related to RT(OND) with a CC of -0.38 (significant at 1% level); CC between SOI(OND) and NRT(OND) is -0.33 (1% LS). There are CCs indicating stronger negative relation between SOI and NRT during various periods. However, CC between SOI (Jan) and NRT(Jan) is +0.18 (10% LS). Table 13.6b, which presents the conditional means (CMs) of NRT given SOI, clearly authenticates this reversal of relationship. When SOI (JJAS) values are <-8, -8 to +8 and >+8, the CMs of PDN of NRT (OND) are 16, -1 and -11 respectively. But when SOI (Jan) lies in the above three intervals, the CMs of PDN of NRT (Jan) are -47, 4 and 37 respectively clearly showing how the relation has reversed.

Seasonal forecasting of NEM rainfall

14. The CVs of NRS and NRT are nearly 30% and 25% respectively. The large CV is a manifestation of frequent occurrences of large excess and deficient NEM rainfall during individual years. Failure of NEM affects the state and sub division of Tamil Nadu all the more in view of the substantial dependence of this state on NEM rainfall for agricultural and hydrological sustenance. Seasonal forecasting of NEM rainfall assumes importance for the southeast peninsula especially for Tamil Nadu. The first known attempt on seasonal forecasting of Indian NEM rainfall was made by Doraiswamy Iyer (1941). Raj (1989), carrying out an extensive search of relation between Indian upper air parameters and NRT identified a few predictors belonging to the preceeding southwest monsoon season for predicting the NEM rainfall of CAP, Rayalaseema, Tamil Nadu and Kerala The upper tropospheric (150 hPa) zonal wind

RESTRICTED

29

RESTRICTED

over the region represented by Thiruvananthapuram during August- September, which is the immediate preceding period of the NEM season emerged as a predictor for NRT with a CC value of 0.77 based on a data sample of 17 years and verified over a test sample of 12 years. The interpretation is that a weak and disintegrating Tropical Easterly Jet Stream (TEJ) over Thiruvananthapuram during August- September favours a good NEM season during the succeeding period of October-December whereas a strong and persisting TEJ is unfavourable. Despite the fact that the high value of the CC realised from such a small sample did not get maintained, the relation held true even with a larger dataset and manifested strongly if the concept of CMs was used for analysis.

15. In Raj (1998a), another attempt was made to identify more predictors for prediction of NRT by including parameters of April which is a representative month for pre-monsoon season. Table 13.8 presents the list of predictors indentified and the CCs obtained. The 200 hPa zonal wind over the Indian peninsula exhibited a CC of 0.61 (1% LS) showing that strong westerly zonal winds even in April, favour a good NEM five months later and that weak westerly winds are associated with a weak NEM, based on data for the period 1965-87. The mean temperature at 150 hPa level over Hyderabad and Port Blair during JJAS exhibited a CC value of -0.80 (1% LS) showing that a colder / warmer upper troposphere during the southwest monsoon season over the Indian region favoured good / poor NEM. The zonal wind over Thiruvananthapuram at 150 hPa during August and at 300 hPa level during September displayed significant positive CCs and this is evidently another manifestation of relation between TEJ and NEM. A statistical model combining all the predictors listed in Table 13.8 and tested in an independent sample of 7 years (1988-1994), yielded reasonably correct forecasts with a mean absolute error of 18%.

RESTRICTED

30

RESTRICTED

RESTRICTED

31

RESTRICTED

TROPICAL METEOROLOGY

RESTRICTED

32

RESTRICTED

CHAPTER – 4

MEDIUM AND LONG RANGE PREDICTIONOF INDIAN SUMMER MONSOONS

Medium range forecasting of Southwest Monsoon

1. A Medium Range Weather Forecast (MRF) is defined in the Indian context asa weather forecast with lead-time from 72 hrs to 240 hrs (3 to 10 days). This is the practical definition followed in this discussion. (The ECMWF Charter defines medium range as "the time scale beyond a few days in which the initial conditions are still important"). We consider here only medium range forecast using numerical models that are based on thermo-hydrodynamics of the earth's atmosphere (i.e. using General Circulation Models of the Atmosphere - AGCM). The AGCM is three dimensional in space enclosing the Earth's atmosphere (with a fixed top boundary at around 0.1hPa or less in pressure coordinates). The evolution of the atmosphere is computed forward in time based on the relevant laws of physics and also our scientific understanding of various processes within the atmosphere and its lower and upper boundaries. At the top of the atmosphere the incoming solar radiation (thefuture values of which are known with sufficient accuracy and is independent of atmospheric processes) is prescribed. This radiation along with the exchanges of momentum, heat and water at its lower boundary are the minimum set of external forcings for the model atmosphere. The details of processes as well as their variety increases with the sophistication of the model. The horizontal and vertical resolutions also increase with the sophistication of the model (the current ECMWF MRF model has an approximate horizontal resolution of 25 km and has 91 levels in the vertical). The present day MRF models are seamless in the sense that the same model, sometimes with minor modifications, is used for short range to climate scale simulations. The current trend in MRF is to evolve towards forecast using Earth System Models, i.e., models that include bio-geochemical processes as well as the ocean with two way interactions.

2. Medium range weather forecasting in India got a much-needed thrust with theestablishment of the National Centre for Medium Range Weather Forecasting (NCMRWF) in 1988. The progress of medium range forecasting of South West Monsoon in the country losely followed the developments at NCMRWF. 1 June 1994 may be considered as the date on which usable real-time medium range forecasts for the monsoon season commenced in India, i.e., when the end-to-end forecast-analysis system based on the T80L18 Global spectral Model started running in real time and produced medium range global forecasts. Over the ears a fully operational location specific forecast system for agriculture was successfully developed on the basis of the T80L18 medium range weather forecasts. This location specific forecasting procedures as well as the Agro Advisory Service Units were transferred to the India Meteorological Department in 2008. Recently, the implementation of the high resolution, advanced T254L64 forecast-analysis system in 2007 ushered in a significant improvement in the forecast skill compared to the T80L18 forecast system.

3. The operational numerical weather analysis-forecast system of National Center for Environmental Prediction (NCEP), USA - Global Forecast System (GFS) - was acquired by NCMRWF in the year 2007. The GFS was implemented on CRAYX1E and Param Padma (IBM P5 cluster) computing systems. Observation

RESTRICTED

33

RESTRICTED

preprocessing and the post processing of model output are presently executed in Param Padma whereas the analysis system and Forecast model at T254L64 resolution are implemented in CRAY-X1E. GFS has the capability to assimilate various conventional as well as satellite observations including radiance observations from different polar orbiting and geostationary satellites.

4. The data decoding part runs 48 times in a day, every half hour, as soon as the GTS data files are received at NCMRWF. The data assimilation steps mainly consist of observation processing, data assimilation and model forecast. In the decoding step, all the GTS bulletins are decoded from their native format and encoded into NCEP BUFR format using the various decoders. Global data assimilation system (GDAS) of T254L64 accesses the observational database at a set time each day (i.e., the data cut-off time, presently set as 6 hour), four times a day and performs a time-windowed (± 3 hours) dump of requested observations. Finally the data preprocessing involves the execution of a series of programs designed to assemble observations dumped from decoder databases, encoding of information about the observational error for each data type as well as the background interpolated to each data location, performing of both rudimentary multi- platform quality control and more complex platform-specific quality control. Quality control of satellite radiance data is done within the global analysis scheme.

5. Horizontal resolution of the analysis system is in spectral triangular truncation of 254 (T254). The quadratic T254 Gaussian grid has 768 grid points in the zonal direction and 384 grid points in the meridional direction. The resolution of the quadratic T254 Gaussian grid is approximately 0.5 x 0.5 degree. The analysis is performed directly in the model's vertical coordinate system. This sigma (σ = p / ps) coordinate system extends over 64 levels from the surface (~997.3 hPa) to top of the atmosphere at about 0.27hPa. This domain is divided into 64 layers having enhanced resolution near the bottom and the top, with 15 levels below 800 hPa, and 24 levels above 100 hPa. Meteorological observations from various types of observing platforms that are assimilated in T254L64 global analysis scheme at NCMRWF are shown Table below. The analysis procedure is performed as a series of iterative problems.

Apart from the increased horizontal (150 km in T80L18 to 50 km in T254L64) and vertical (18 levels in T80L18 to 64 levels in T254L64) resolutions and better quality-control packages for conventional observations in T254L64 GDAS, assimilation of direct radiance is one of the major improvements over the T80L18 data assimilation system.

6. The numerical forecast gives the values of a large number of variables ranging from the basic ones like temperature, pressure, horizontal wind components and humidity to precipitation, cloud cover, etc. The forecast could further be processed and diagnosed to obtain physically or practically relevant quantitative or qualitative (e.g. the occurrence of fog; maximum and minimum temperature at a location) information. In meteorological practice it is very useful to convert these numerical values into customary forms like weather charts and also describe the state of the atmosphere in terms that are in consonance with

RESTRICTED

34

RESTRICTED

commonly used conceptual models. (for example, in terms of 'active' monsoon; intensity of 'heat low', etc.) Southwest Monsoon is an exceptionally persistent and stable flow structure with clear-cut morphology and it is normally described (very effectively) in terms of a set of so-called semi-permanent and transient features. We follow this tradition in the Indian meteorological practice, while evaluating medium range weather forecasts during the Southwest Monsoon. In some cases the forecasts are considered like values of variables at grid points and in other cases we describe them more in terms of morphologically distinct features that have considerable geographical extent. Medium range forecast at NCMRWF is carried out for 7 days in to the future. However, the errors beyond 5 days are usually too large and highly variable in the Tropics. The detailed description of errors beyond 5 days has little relevance in the present discussion. We, therefore present error estimates to 5 days (namely, 24hr, 72hr and 120hr forecasts). This is done for the monsoon period of 2008 so as to be representative of the current status of the MRF. There is no single criterion that can satisfactorily measure the error of a forecast. We had used a battery of measures to monitor the errors. However, in this discussion, as a rule, we show the root mean square errors or the changes in mean conditions to keep the size of the discussion within reasonable limits.

7. Fig. below shows the rmse of the magnitude of the wind vector (RMSEV) for the NCMRWF operational model day 03 forecasts against the radiosonde observations over the Indian Region since January 1999, at 850 hPa level. The most notable feature of the error variation is its seasonal cycle with the winter months having least error and the Southwest Monsoon having the largest error. The increase in errors from winter to summer is 2 m/s, which is about one third of the annual mean error. The winter to summer increase of error is a feature seen in forecasts of other centres as well. However the substantial increase in the errors as well as their magnitudes during Southwest Monsoon demonstrate that it is necessary to greatly improve the accuracy of the forecast-analysis system. It is also seen that there was a continuous increase in error from 2003 to 2006. It is also seen that there is a drop in the error before the beginning of Southwest Monsoon. The reasons for these kinds of changes are not clear. One may also note the significant reduction in the forecast errors accompanying the operationalisation of the T254 forecast-analysis system on 1 June 2007. The errors of the T254 model during monsoon 2008 were less than during monsoon 2007. This reduction in error was brought about mainly because of the improvements in the assimilation system and the use of more data. The status of the errors of the forecasts of NCMRWF during Southwest Monsoon (July 2008) vis-a-vis those of NCEP, UKMO and ECMWF were computed and are shown below. Fig.9.9 shows the RMSEV of winds for the T254, NCEP, UKMO and ECMWF model forecasts against the radiosonde observations over the Asian region (25 - 65 N, 60 -145 E) for the Southwest Monsoon of 2008 at 850 hPa. It is seen that the ECMWF forecasts have the least error among the four models. The NCMRWF forecast errors are comparable to that of NCEP. The differences between the errors of the NCEP,UKMO and NCMRWF are small compared to the magnitude of the errors.

RESTRICTED

35

RESTRICTED

Long range forecasting of Southwest Monsoon

8. The 8 and 10 parameter PR models were used for issuing quantitative operational forecast of seasonal rainfall over the country as a whole during the period 2003 to 2006. Table 12.3 shows the predictors used for the development of the PR models. The mathematical form of the power regression model is given below.

Where R is the rainfall, X’s are standardized predictors, and C’s and P’s are constants. N is either 8 or 10. The model is non-linear and the power term, P, in the above equation varies between ± 2. The models were developed using data of 38 years (1958-1995) and independently tested using data of 7 years (1996-2002). Comparison of the model forecasts with IMD’s operational forecasts issued using the 16-parameter power regression model during the independent period is given in

RESTRICTED

36

RESTRICTED

Table 11.4. It is clear that forecasts from the 8- and 10-parameter models were closer to the actual rainfall than the forecasts from the 16-parameter model. The root mean square error of the operational forecasts by 16-parameter model during the period 1996-2002 was 11% of LPA, while that of the new 8-parameter model and 10- parameter models for the same period was 7% and 6 % of LPA respectively. The model errors of the 8- and 10-parameter models were 5% and 4% of LPA respectively which are of the same order as the model error of the 16-parameter model at its inception. For more details of these models can be seen in Rajeevan et al. (2004). However, it may be mentioned that though these models showed better performance in general during the drought years in the hindcast mode, they failed to correctly indicate the large rainfall deficiency during 2002 in the hindcast mode and that during 2004 in real time forecast mode.

9. There were three major changes in the new statistical forecast system used at present (Rajeevan et al. 2007) from that used during 2003 to 2006 which was based on the 8/10 Parameter power regression models. These were:

(a) use of a new smaller predictor data set

( b) use of a new non-linear statistical technique along with conventional multiple regression technique

RESTRICTED

37

RESTRICTED

( c) application of the concept of ensemble averaging.

The new ensemble forecasting system introduced in 2007 used a set of 8 predictors (given in the Table 11.6) that having stable and strong physical linkage with the Indian south-west monsoon rainfall. For the April forecast, first 5 predictors listed in the Table-10.6 were used. For the update forecast issued in June, the last 6 predictors were used that include 3 predictors used for April forecast. In the ensemble forecasting system, the forecast for the seasonal rainfall over the country as a whole was computed as the mean of the two ensemble forecasts prepared from two separate set of models. Multiple linear regression (MR) and projection pursuit regression (PPR) techniques were used to construct two separate sets of models. PPR is a nonlinear regression technique. In each case, models were construed using all possible combination of predictors. Using ‘n’ predictors, it is possible to create (2n-1) combination of the predictors and therefore that many number of models. Thus with 5 (6) predictors it is possible to construct 31 (63) models.

10. For the forecast of July rainfall over the country as a whole, a set of 5 predictors were used. The model was trained using data for the period 1958-2000 and the model was tested for the period 2001 to 2008. PCA analysis was carried over the predictor set using data for the training period and first three PCs explaining about 89% of the total variability of the predictor data set was retained for multiple regression (MR) analysis. Using the PC loadings of the retained PCs, PC scores were calculated for the independent test period and the same were then used for the prediction of July rainfall for the independent test period. A set of six predictors were used for the forecast of August rainfall over the country as a whole. The model was trained using data for the period 1975-2000 and the model was tested for the period 2001 to 2008. PCA analysis was carried over the predictor set using data for the training period and first three PCs explaining about 75% of the total variability of the predictor data set was retained for multiple regression (MR) analysis. The performance of the PCR model for the August rainfall during the independent test period is shown in Table.

RESTRICTED

38

RESTRICTED

11. Since 2005, IMD has been issuing operational forecast for the onset of monsoon over Kerala using an indigenously developed statistical model (Pai & Rajeevan 2009). The model is based on the principal component regression (PCR) method using 6 predictors. Table-11.10 shows the list of the 6 predictors. Sliding fixed wind period of length 22 years was used for deriving the independent forecasts. According to this method, for the prediction of monsoon onset over Kerala each year, data of 22 years just prior to the reference year was first used for PC analysis of the predictor data series. PC scores were calculated for the reference year using the PC loading matrix and predictor values. Only those PCs that having eigen values more than or equal to 1 were then used as the input to the multiple linear regression equation.

RESTRICTED

39

RESTRICTED

CHAPTER – 5

TROPICAL STORM

Introduction

The most vigorous tropical disturbances are the intense cyclonic storm that forms over warm tropical waters. Tropical storm is an intense low pressure area covering large area extending over 6 – 10o latitude / longitude in length and breadth and up to 200 hPa in depth. Wind speed assigned for very severe cyclonic storm and super cyclonic storm are 64 – 119 kts and above and more than 120 kts respectively. Tropical cyclones develop in most tropical oceans at latitudes usually greater than 5o

from the equator. ‘Extremely steep pressure gradients are very frequently attained and hence the surface wind may reach hurricane force. Torrential rain and thunderstorm activity with widespread cloud development accompany the tropical storm. The lowest pressure occurs at centre and is called as Cyclonic Eye. Eye of the storm is only a few km in diameter and is characterised by light winds, no rain and often very little cloud. It is sharply separated from the surrounding strong wind, rain and heavy wall cloud. Tropical storm may move in any direction. However, very often they move west wards and slightly polewards after formation. Strong winds around tropical storm generate high waves. Vertical structure of a tropical storm consists of inflow layer, mid-tropospheric layer and outflow layer. Horizontal structure of a tropical storm comprises of Eye, Wall cloud region and Outer storm area. The cyclonic wind vortex has its maximum diameter at the surface and remains nearly unchanged up to 6.0 km and diminishes rapidly beyond 6.0 km. Anticyclonic flow dominates hurricane circulation above 6.0 km, which reaches its maximum intensity between 12.0 and 16.0 km. Tropical storm has a warm core below 15 km and a cold-core aloft. The warm core is generally 8 – 10o C warmer as compared to the surroundings. In the tropical ocean a 24 hours pressure fall of ≥ 3 hPa and sea surface temperature of the order of 26 – 27oC is indicative of development of cyclonic storm. Under the influence of large amplitude trough in upper tropospheric westerlies, well marked low latitude trough building up in the westerlies to the north of the system and existence of weak trough between two separate subtropical cells leads to the recurvature of the tropical storm. Persistence, climatology, steering current, 24 hours pressure fall and area of major new convection etc indicates the expected direction of movement of the tropical storm. The associated vorticity, divergence and vertical motion fields clearly demonstrate that the strongest activity is concentrated in a narrow vertical tube ahead of the system. This region is characterised by strong cyclonic vorticity, convergence and upward motion. Very moist air prevails to the southwest of the system from its centre. Clouding and precipitation are also found to be maximum in the southwest sector of the storm.

1. Severe atmospheric disturbance in the tropical oceans between latitudes of approximately 5 and 30 in both hemispheres. These storms are characterized by very low atmospheric pressures in the calm, clear centre of a circular structure of rain, cloud, and very high winds. In the western Atlantic and the Caribbean they are called hurricanes; in the western Pacific, typhoons; and in western Australia, willy-

RESTRICTED

40

RESTRICTED

willies (if the surface winds exceed 117 kilometres [73 miles] per hour). Tropical cyclones are cyclonic whirls 80 to 800 kilometres (50 to 500 miles) in diameter; the winds near the centre form an almost circular vortex with a slight inward motion toward the centre near the ocean surface. Because of the Earth's rotation, the vortex circulation is clockwise in the Southern Hemisphere and counterclockwise in the Northern. The areal extent of tropical cyclones is small compared with storms outside the tropics, but the violence of the weather within the disturbed zone is usually far greater. Sustained winds in excess of 160 kilometres (100 miles) per hour are common near the centre, and winds twice that strong have occurred. Very close to the centre of mature cyclones, however, the winds drop abruptly from their extreme maximum to light breezes or even to complete calm. This central circular calm area bears the name eye of the storm and has an average diameter of 24 kilometres (15 miles). The lowest sea-level pressures on Earth occur in or near the eye of a hurricane. All tropical storms develop over water that is warm enough to supply appreciable quantities of vapour to the air. Most storms develop along the equatorial convergence zone, where the trade winds of both hemispheres meet. This zone changes position with the seasons, penetrating to about latitude 15 N between July and October and to latitude 10-15 S from January to March or April. These months represent the principal cyclone seasons of the respective hemispheres. The formation of the storms is most frequent farthest from the Equator, where the Coriolis effect (clockwise or counterclockwise rotation due to the rotation of the Earth) is greater (the effect is zero at the Equator).As the warm water evaporates, moist air is carried aloft, where it condenses, releases latent heat, and is warmed further. This warming strengthens the updraft, and a low-pressure area is created in the lower atmosphere. Surrounding air moves into this low-pressure region and feeds the updraft. The Coriolis effect causes the developing storm to spiral, and it continually draws energy from the condensation of lifted water vapour. After several days of intensification, this process produces a mature traveling cyclone. Dissipation of the storms over the tropical oceans is rare. When removed from their oceanic moisture source, however, they lose intensity; even passage over smaller islands of the tropics can result in great reductions in strength.

Origins

2. A tropical cyclone is likely to occur whenever several of the following prerequisites occur simultaneously: (1) latitude sufficiently high (5-6) for the Coriolis force (see above) to be appreciable; (2) a warm-water surface (at least 27 C) of sufficient area to supply the overlying air with large amounts of vapour; in some cases a cyclone may form over water at 23 or 24 C if much colder air is present at higher altitude; (3) pronounced instability in the air column or relatively low pressure at the surface and often an anticyclone aloft; (4) little or no vertical wind shear (shearing effect produced by the movement of one air mass past another). These conditions are most likely to occur over the oceanic areas where the intertropical convergence zone moves 10 or more away from the Equator. The Coriolis force is proportional to both the latitude and the angular velocity (rotational speed) of the Earth. At these low latitudes, the value of the Coriolis force is minimal; this is why, in large and shallow tropical depressions with very weak winds, the pattern of air flow is indistinct. The general instability and enormous vapour load make the air most susceptible to any triggering factor and especially to convergence due to external wind flow. Convection may rapidly become tumultuous. With a strengthening of convection, the centripetal wind flow gains speed; soon the angular velocity component of the Coriolis force becomes sufficient to impart a definite cyclonic

RESTRICTED

41

RESTRICTED

curvature to the air flow, and a cyclone becomes established. The input of warm, very moist air continues. Large-scale condensation of moisture occurs during the ascent, and enormous amounts of previously latent energy are released. This energy results in stronger winds, which in turn lead to the intake and uplift of larger amounts of humid air, with a further release of energy. The evolution from tranquil tropical depression to violent tropical cyclone takes four to eight days. Strong vertical wind shear (e.g., by a jet stream overhead) would impede convection and prevent the development of the cyclone. Latent heat is the main source of energy in a tropical cyclone. Thus, a rapid inflow of dry air can reduce the cyclone to a much slower tropical depression.

Physical characteristics

3. The shape of the streamlines in a tropical cyclone is not a simple spiral, as it would be if there were no zonal (latitude belt) easterly winds around it. Because of the greater speed around the core, the zonal easterlies are slightly deflected outward before they begin their inward spiraling course toward the centre of cyclonic indraft. Also, downwind of the cyclonic spiral there is a hyperbolic (saddle-shaped) divide between the airstreams that flow into the cyclone and those that rejoin the zonal, easterly flow after having been deflected around the edge of the cyclone. Winds are weakest near this hyperbolic divide: there is no wind at all at the point where the air may be equally likely to flow into the cyclonic indraft or to flow away from it. This is the stagnation point, and it is a necessary element of the cyclonic structure. The corridors of convergence that cross the stagnation point may be marked by a line of clouds, which are only small cumulus on either side of the stagnation point and build up to towering cumulonimbus both eastward and westward where the air streams crowd together. A tropical cyclone is often preceded, at a distance of about 1,000 kilometres, by a short spell of fine weather due to divergence and subsidence in the prevailing (and preceding) easterly airflow. Pressure then begins to fall, but cloudiness does not appear until the cyclone is fewer than 300 to 500 kilometres away.

Distribution

4. Major tracks and frequency of hurricanes and typhoons. More details appear in the Table, which gives the frequency by area or region. Because of the cool surface waters, tropical cyclones do not originate in the eastern South Pacific and South Atlantic oceans (Figure 1). Seasonality and frequency before 1940 many cyclones went undetected. At present every disturbance is detected by satellite photography. There are four main areas of origin of North Atlantic hurricanes: south of the Cape Verde Islands, the open waters east of the Lesser Antilles, the western Caribbean Sea, and the Gulf of Mexico. The early summer hurricanes usually begin in the Gulf of Mexico and western Caribbean, where water temperatures rise early because of the shallows and the nearby heated land. Low pressure farther north often allows the hurricane to enter the Gulf area. As the broad expanses of Atlantic surface waters warm to 28° C or more in late July and August, more hurricanes form much farther east, as far as the Cape Verde Islands. They gather more energy as they travel west before recurving over the West Indies. By mid-September surface waters are cooler, but outbursts of cold air aloft are likely; the area of origin moves

RESTRICTED

42

RESTRICTED

west again to the Caribbean and to the Gulf of Mexico. The high-pressure outflow of continental air often pushes the recurving hurricane along a more easterly path so that it may sweep over Florida or the Atlantic seaboard._

RESTRICTED

43

RESTRICTED

Damage caused by tropical cyclones

RESTRICTED

44

RESTRICTED

5. Tropical cyclones can cause immense damage, both directly (by wind, pressure, and rain) and indirectly (mainly through storm surges and floods). The wind causes damage that generally increases in proportion to the square of its velocity, according to the basic formula P = KV 2, in which P is the pressure exerted by the wind against vertical surfaces, V is the wind velocity, and K is a factor depending on air density and drag. A wind of 74-93 kilometres per hour strips leaves and small branches off trees; at 111-130 kilometres per hour it can topple shallow rooted trees or snap weaker trees outright, blow down thin walls, shift roofing materials, and occasionally lift a whole roof. A wind of this force may blow in large glass windows. At about 130 kilometres, lifting of roofs and snapping of trees is general. Hurricane winds may exert a pressure of more than 400 kilograms per square metre (82 pounds per square foot) on tall structures and can flatten weak buildings at first impact. Gusty winds, combined with a suitable period of vibration of a given structure, have a dynamic effect because they may produce resonance (vibrations in phase), leading to breaking point. Damage also is caused to roofs and windows by the suction produced by strong winds on the downwind side, amounting to 0.5-1 times the windward pressure. As with tornadoes, whirlwinds, and waterspouts (see above), much direct damage is caused by the rapid fall of external pressures. The pressure differential could amount to 300-400 kilograms per square metre for a sealed structure. Damage may be increased by the fact that the strongest gusts, in excess of 185 kilometres per hour, occur immediately after the transit of the eye and blow in the contrary direction to the preceding gusts, thus adding considerable stress to any structures exposed to them. Loose objects lifted by the wind become missiles that shatter glass, batter walls, and flatten roofs. Wind causes injuries and deaths by toppling structures and hurling loose or torn objects about with enormous force. Since the eye may take from a few minutes to an hour or more to pass over a given point, depending on (1) the central or eccentric position of that point, (2) the size of the eye itself, and (3) the travelling speed of the cyclone, many victims are struck after they have left their shelters in the belief that the storm was over. The kinetic energy of the whole cyclonic system is nearly proportional to its power, but the amount of damage and loss of life depends on many other factors, among which lack of warning and insufficient preparedness can be extremely significant. Furthermore, the torrential rain brought by a typhoon may erode the soil, causing landslides in mountain country and making streams and reservoirs overflow. Indirect damages are mostly due to a storm surge. This is a complex surface deformation of the sea induced by the cyclonic winds on coastal waters, which surge as a sudden tide against the coast, flooding the countryside and impeding the flow of rivers. The level of the sea is raised by up to three metres for a period that may last several hours, depending on the characteristics and relative position of the cyclone and of the coastline affected. The level of the sea is raised an additional 0.5-1 metre by the low atmospheric pressure. Extreme tides recorded on the Gulf Coast were all due to hurricanes. Hurricane "Hazel" (1954) brought a 3.5-metre tide to North Carolina. The worst storm surge in Tokyo Bay (October 1917) rose 2.3 metres above normal. Osaka Bay had nine storm surges in half a century, one of them (Sept. 21, 1934) reaching 3.1 metres above normal. Most coastal cities are less than three metres above sea level and may thus be extensively flooded. Industrial plants in coastal areas may be badly damaged by salt spray and seawater. Pounding waves are an additional cause of damage to coastal installations and structures. Coastal erosion may reach catastrophic proportions. Most significant may be the cumulative effect of a close succession of two or three cyclones, as happened in North America in 1954 and 1955. Lakes are affected in the same way as the sea but over shorter periods. Because of their smaller dimensions, they may develop storm-surge seiches

RESTRICTED

45

RESTRICTED

(oscillations) of remarkable amplitude; e.g., the disastrous 5.5-metre swelling of the southern end of Lake Okeechobee in Florida in 1926.

Warning and tracking of tropical cyclones

6. Because of the small diurnal pressure changes common in the low latitudes, it is difficult to single out changes due to approaching cyclones. A local drop in pressure of more than three to four millibars in 24 hours (or over a distance of 500 kilometres or less) may be considered a danger sign. The development of markedly curved isobars moving toward the observer is a much more definite indication of an approaching cyclone, and the roughly circular shape of the isobars soon reveals the position of the cyclone's centre. A rapid increase in the height of any thermal inversion, as revealed by misty or hazy layers aloft, is often a sign of an approaching cyclone, as is reversal of the wind direction in altitude. Humidity is normally high and is no useful indicator, but cloud types and movements--e.g., a sheet of cirrus or bright red sunsets--may provide a belated warning. Clearer advance warning may be given by a long ocean swell with a slower frequency (two to four times the usual interval) than that of normal waves. This swell travels outward from the centre for hundreds of kilometers. In Australia a cyclone warning station was set up on Willis Island, off the Queensland coast, in 1921. Special hurricane forecast centres have been maintained by the United States Weather Bureau since 1935.If a ship reports strong winds or rapidly falling pressure, unusual squall activity, or even a wind flow unusual for the season, special observations at three-hour intervals are requested from all ships in the area, and warnings are issued. Because the shape of the cyclone's path mainly depends on the pressure pattern of the time (the cyclone tends to travel toward any locus of lower pressure and around or away from any locus of high pressure), a synoptic barometric map may allow a gross forecast of a cyclone's track. Pioneer flights into a hurricane took place in 1943. Regular aerial hurricane patrols began in 1945, reporting location, characteristics, and movements of any likely disturbance. Flights may be made into, through, and above the storm. The easiest access to the eye of a cyclone is from the side of the stagnation point, where the hyperbolic divide is located. Because this is essentially an area of very light winds, exact location is very difficult without the use of drift smoke signals. The smoke plumes soon reveal the direction of the air flow. The plume extending toward the lower pressure will follow the indraft-accosting spiral, which leads in the shortest time (but not as the shortest route) to the centre of indraft. A distinct, clear break along a line of growing cumulus to cumulonimbus clouds is likely to be due to the prevailing stillness at the stagnation point and may be taken as an indicator.Radar units provide warning of any storm within range. The characteristic spiral banding of dense clouds and rain makes cyclonic storms easy to identify. Satellites transmit photographs of any part of the Earth and its cloud systems. They give the most reliable and comprehensive coverage of cloud patterns and reveal storm systems from remote areas, where other methods of detection may not always penetrate.

Table 5: Average Frequency of Cyclones by Area

Area Average frequency per year TotalEast and South Asia 29 Northwest Pacific 21Bay of Bengal 6Arabian Sea 2

RESTRICTED

46

RESTRICTED

North and Central America 14Caribbean and northwest Atlantic 8 East Pacific and west Mexico 6Australia and Oceania 9 Southwest Pacific and northeast Australia

6

Timor Sea and west Australia 3Southwest Africa and islands 7Southwest Indian Ocean and Madagascar

7

World 59

Criteria for defining Low Presuure Systems

LOW PRESSURE SYSTEM WIND SPEED IN CIRCULATION (KTS)

LOW < 17 Kts

DEPRESSION 17 – 27 Kts

DEEP-DEPRESSION 28 – 33 Kts

CYCLONIC-STORM 34 – 47 Kts

SEVERE CYCLONIC STORM 48 – 63 KtsVERY SEVERE CYCLONIC STORM

64 – 119 Kts

SUPER CYCLONIC STORM 120 Kts

7. Chara cteristics of Tropical Storm .

(a) The most vigorous tropical disturbances are the intense cyclonic storm that forms over warm tropical waters.

(b) Tropical cyclones develop in most tropical oceans at latitudes usually greater that 5o from the equator. They reach their greatest intensity, while located over warm tropical waters. As soon as they move inland they begin to weaken, but often not before they have caused great destruction / devastation.

(c) The lowest MSL pressure of a tropical cyclone is frequently about 960 hPa, but often if is much less (940 hPa / Bangladesh / Cyclonic Storm / Apr 1991)

(d) Extremely steep pressure gradients are therefore possible and the surface wind may reach hurricane force. Torrential rain & Thunderstorm activity with widespread cloud activity /development accompany the storm.

RESTRICTED

47

RESTRICTED

(e) A typically well developed tropical cyclone consists of a more or less circular area of hurricane winds i.e. over 63 kts, approx in radius of 80 km outside this area the wind force decreases fairly rapidly. At a distance of 150 – 200 kms from centre, the speed may be down to 30 – 40 kts.

(f) The lowest pressure occurs at the centre, which is known as the cyclone eye. This region is sometimes only a few km in diameter and is characterized by light winds, no rain and often very little cloud. It is sharply separated from the surrounding strong wind, rain and heavy cloud mass.

(g) The cyclone eye is surrounded by a spectacular wall of cloud extending in a steep slope to the upper rim at an altitude of 10 km outside the wall cloud, is the maximum speed ring, where the wind may exceed 300 km per hour and towering CB produce torrential rain and thunderstorm. These winds and associated rain are best developed in thee forward quadrant of the cyclone.

(h) Tropical cyclones may move in any direction. However, very often they move west wards & slightly pole wards after formation.

(j) The strong winds around tropical cyclone generate high –waves. These travel outwards in all directions from the storm at an average speed of 1500 km /day. Hence, the arrival of heavy swell is a useful warning of the approach of a tropical-cyclone, while the direction of the swell may provide an indication of the bearing of its centre.

8. Life Cycle of a Cyclonic Storm (CS). The life cycle of a CS may be divided into following stages: -

(a) Formative Stage . Is period from genesis of a cyclonic circulation as a low pressure area, through the stage of depression till it reaches intensity of a SCS. At the end of this stage ‘eye’ and ‘wall-cloud’ are formed. Pressure fall is very slow during the formative stage and the central pressure reaches 10mb or sow below the normal. In the Indian sea area, the normal pressure in the seasonal trough during Post-Monsoon season is about 1010 hPa; consequently, the central pressure in the formative stage of a cyclone is 1000 hPa. Squall and thunder rain occurs over a large area. This stage is unsettled stage.

(b) Immature Stage . During this stage, the Central Pressure (CP) rapidly falls and winds strengthen. The bands of clouds and rain get organized. The area of very strong wind is still small. The CP and the winds reach their maximum limits. The cloud and RRR patterns get organized into narrow bands spiralling inwards.

(c) Mature Stage . The main feature of this stage is that the entire circulation expands while pressure remains relatively constant at the centre. The area covered by winds of hurricane force increases, and area of strong winds extends further from the centre. The mature cyclone exhibits four distinct parts: -

RESTRICTED

48

RESTRICTED

(i) A Calm Central Area or Eye. It is characterized by approx 0 to 20 km in diameter having calm or very light winds with clear to P’Cloudy sky and lowest pressure.

(ii) An Inner Ring of Hurricane Winds. Surrounding the eye of the CS is an inner ring of Hurricane winds ( 64 kts), 50 to 60 km in width within which violent squalls with torrential rain occur under a circular wall of clouds.

(iii) Outer Storm of Area: Wherein, the winds reach to gale force (22 – 47 kts), situated asymmetrically to the pressure centre extending to 400 km or so. Spiral bands of rain clouds and torrential rain are experienced in the area. Winds decrease outwards.

(iv) Outermost Area of Weak CYCIR or Edge: Having comparatively less activity as compared to outer storm area.

(d) Decaying Stage . This is the stage when the system weakens. The weakening may take place on account of the cyclonic storm entering land or moving into regions of cold waters. While those that move inland or over cold water may weaken rapidly, those that re-curve in our latitudes and get into the Westerlies retain their structure and the weather associated with them continues to be severe.

9. The Formative Stage is usually a slow process (upto a week). The Immature Stage when the cyclonic storm rapidly deepens is relatively a quick process. It may occur within 24 hours. Sometimes the intensification is very rapid and the system may develop from a Deep-Depression to a SCS in few hours time.

10. In the formative stage the strong winds in the CS are also confined to a small area. The Mature Stage which marks the outward expansion of the cyclonic circulation may again last for several days. The Dissipating Stage may also be rapid. As Arabian Sea and Bay of Bengal are of limited extent (unlike the Atlantic or Pacific), storms in Indian sea areas may often cross coast and weaken before they develop fully and reach the mature stage.

11. Development Period. Depressions develop into storms on 85% of occasions within 48hours of their formation and 12 – 24 hrs is the most common time interval (40% of cases) cyclonic storms reach severe intensity within 36 hrs on 75% occasions, while 14% of depressions develop into CS within 12 hours and 31% of CS becomes severe within 12 hours.

12. Life Period. The average life period of a storm/depression in the Bay of Bengal is 5 days in October, 4 days in November, and 3 days in December. In the Arabian Sea the corresponding figures are 5, 5 and 4. Thus the average life of storms and depressions in the Indian Sea areas is in general 4 – 5 days.

Horizontal Structure of a Cyclonic Storm

13. Eye. Where winds are nearly calm. As regards ‘Eye’, we have:-

RESTRICTED

49

RESTRICTED

(a) Pressure Eye : Where the sea-level pressure is minimum.

(b) The Wind Eye : Where the wind is calm or light variable.

(c) The Radar Eye : Centre as determined by pattern of rain echoes i.e. maximum echo region and maximum rain patch.

(d) The Satellite Eye : Where an apparent clear spot in the cloud mass is noticed.

14. Wall Cloud Region. It is the region where the strongest winds are noticed. In the wall cloud region torrential weather occurs, which is accompanied by very low stratus cloud and most of the convective cloud mass are confined to this region. Most severe weather occurs in this region.

15. Outer Storm Area. Where winds decrease as we move out in a radial direction, though they may be still strong and rain / shower / thunderstorm also have a decreasing trend.

16. Maximum Wind. Winds associated with CS in Indian Ocean shows that 50% of CS had winds of hurricane force in their circulation while in only 20% of occasions; the winds were less than 50 kts. Winds in excess of 122 kts have also been recorded on many occasions over Indian region.

17. Radial Diameter. 50 – 100 km radius to 2000 km radius. On about 50% occasions the storms have a diameter of about 6 – 10O of latitude or longitude. While the smallest storm has a diameter of 3O, the longest one has 14.5O.

18. Pressure Field. Lowest recorded pressure in Indian Sea: 919 hPa. 55% of storms have a central pressure of 980 hPa and only about 10% have less than 960 hPa. In a wall cloud region the pressure gradient is maximum. Outside the wall cloud, the pressure gradient is considerably weak. Study of 01 – 04 Jun 98/Orrisa Cyclone indicated that pressure gradient of the 26 hPa/Km could also be associated with cyclonic storm.

Pressure Field and Maximum Wind

19. The central pressure and maximum wind in CS are given by Fletcher formula:-

______VMAX = 16 PO - PC

Where,

VMAX = Maximum wind in circulation

PO = Value of outermost closed isobar in hPa

PC = Central Pressure in hPa

RESTRICTED

50

RESTRICTED

By taking PO as 1010 hPa, Fletcher formula yields very good results over India Sea areas. (e.g.: CP = 919 hPa » Max wind = 135 Kt)

20. Vertical Structure of a Cyclonic Storm

RESTRICTED

51

RESTRICTED

(a) Inflow Layer. In this layer there is a pronounced inward radial – component, extending throughout the lower troposphere. The most pronounced inflow is in Planetary Boundary Layer (PBL) i.e 1 Km and below.

(b) Mid-Tropospheric Layer. In this layer there is very little radial motion and it extends up to midtropospheric level.

(c) Outflow Layer. This layer exists in the upper troposphere, which extends up to the top of the storm. The maximum outflow is near 200 hPa.

21. The air that enters the storm circle in the inflow layer raises along the wall-cloud and the rain bands; subsequently, it flows outwards from the storm top and sinks some distance away. The marked cloud free zone around tropical storm in satellite picture is an evidence of sinking–motion (Fig.1).

RESTRICTED

52

RESTRICTED

Fig.1 Schematic Representation of Vertical Structure of Cyclonic Storm

22. Characteristics of a Vertical Structure of a Mature Hurricane Prior To Its Recurvature Into Extra-Tropical Westerlies

(a) Low pressure core of CYCLONIC STORM extends throughout the troposphere and into stratosphere up to 27 km and possibly higher.

(b) The pressure gradient in the core (i.e. maximum winds in the field of CYCLONIC STORM) decreases with height above 6km, but a weak and diffused Low or Trough persists even in the stratosphere. The wind decreases very slowly with height upto 500 hPa and beyond it rapidly.

(c) The cyclonic wind vortex has its maximum diameter at the surface and remains nearly unchanged up to 6km and diminishes rapidly beyond 6km.

RESTRICTED

53

RESTRICTED

(d) Anticyclonic flow dominates hurricane circulation above 6km, which reaches its maximum intensity between 12 and 16km. This flow is usually broken up into two or more anticyclonic vortex which are asymmetrically situated with respect to inner cyclonic vortex.

(e) In the anticyclonic flow, wind speed may reach 60 – 80 kts between 12 km and 15 km in intense storm. These strong winds in the upper troposphere may be inferred from pronounced Ci – flow seen in satellite picture of storms.

(f) The Cyclonic Storm has a warm core below 15 km and a cold core aloft. The warm core is generally 8 – 10O C warmer compared to the surroundings.

23. Pressure Departure and SST. In the tropical oceans a 24 hrs pressure fall of 3 hPa or more is indicative of development of a tropical storm. The sea surface temperature (SST) of 26 – 27 OC is essential for the formation of intense cyclones. Rapid deepening of systems occurs when SST are 28 OC. Higher is the state of fall of pressure, whenever the SST values are in excess of 27oC.

24. Storm – Surge. Is rise in sea level associated with CYCLONIC STORM storm-surges are responsible for the major part of loss of human life and property. Waves of 20 metres are very common in case of Severe Cyclonic Storms.

25. Recurvature. In the Indian Sea storm continues to move from E to W as long as Westerlies are flat or wave in Westerlies have small amplitude and a strong sub-tropical high exists to the North of the system and there is no trough in the Westerlies close to the west of the storm. However, under the influence of the following situations the Storm recurves: -

(a) Large amplitude trough in the upper westerlies to the west of the system.

(b) Well marked low latitude trough building up in the westerlies to the north of the system.

(c) Existence of weak trough between two separate subtropical cells.

26. Movement with respect to 200 hPa ridge line. On 80% of occasion the storm moves in W/NW direction when it is 3o latitude to the south of the ridge line. When, it is with 3o of the ridge it moves north and remains stationery or slows down when close to ridge line.

27. Frequency of Cyclones.

(a) Arabian Sea. Frequency of Tropical storm is almost nil in Feb/Mar. It is least in Jan and highest in Nov. Most of the storm occurs in May, Jun, Oct, Nov and only a few occurs in Jan, Jul, Aug, Sep and Dec.

(b) Bay of Bengal. Frequency is least in Feb and highest in Oct. Steady increase in number of storms from Mar to Nov and then rapid

RESTRICTED

54

RESTRICTED

decrease has been observed statistically. In Jan, Feb and Mar number of storms is reasonably low.

28. Movement of Cyclonic Storm. Indication of movement are given by:

(a) Persistence and steering current give good indication of its movement in next 12 to 24 hours.

(b) A well marked trough in westerlies ahead of a Cyclonic Storm indicates recurvature of the storm.

(c) The 24 hrs pressure falls are seen to be maximum on area of landfall.

(d) Area of formation of major new convection is likely region of Cyclonic Storm movement.

29. Recurvature of Storms:

(a) Storm generally recurves when they are between latitude 16 oN – 18 oN.

(b) Before recurvature the speed of the storm reduces.

(c) Just above the level where circulation ceases to exist, the flow-pattern determines the future track of the storm.

(d) Second degree variation of pressure gives the indication of future movement.

(e) Storms move towards the maximum isallobaric gradient.

(f) Width and strength of westerlies, at 200 hPa at a distance of 10 – 12 O

to the north latitude, is significant for recurvature of a storm. Non-recurving storms do not have westerlies.

30. Formation and Intensification of Tropical Storms.

(a) Central pressure of a tropical storm is generally 8 – 10% below the average sea level pressure.

(b) The situation favourable for the formation of tropical storm are: -

(i) Warm SST > 26 – 27 OC (ii) Easterly waves(iii) ITCZ(iv) Equatorial trough(v) Intensification of E-W trough in lower troposphere in Pre-Monsoon & Post monsoon

31. Details on character / behaviour of Tropical Cyclones.

RESTRICTED

55

RESTRICTED

(a) Well developed tropical cyclone has core of hurricane winds 63 kts in 80 to 100 km radius. At a distance of 150 – 200 km, speed of wind may be down to 30 – 40 kts.

(b) Lowest pressure occurs at the centre, which is known as cyclone eye. Eye is few km in diameter and is characterized by light winds, no rain and often very little clouding. It is sharply separated from the surrounding strong wind, rain and heavy cloud.

(c) Eye is surrounded by a wall of clouding extending in a steep slope to the upper rim at an altitude of 10 km. Outside the wall of the cloud is maximum, where the winds may exceed 300 km/hr and towering CB produce torrential rain and activity of Thunderstorm. These winds and associated rains are best developed in the forward quadrant of the cyclone.

(d) Tropical cyclones may move in any direction. However, very often they move west wards and slightly pole wards after they first form.

(e) The strong winds around tropical cyclones generate high waves. These travel outwards in all direction from the storm at an average speed of 1500 km/day. This may be several times higher than the speed of the storm itself. Hence, the arrival of heavy sell is a useful warning of the approach of a tropical cyclone, while the direction of the swell may provide an indication of the bearing of the centre.

(f) For intensification low-level convergence should be superimposed by upper level divergence. This divergence may be provided either by an upper air anticyclone or an easterly trough.

(g) The associated vorticity, divergence and vertical motion fields clearly demonstrate that the strongest activity is concentrated in a narrow vertical tube ahead of the system. This region is characterized by strong cyclonic voriticity, convergence and upward motion.

(h) Very moist air prevails to the southwest of the system from its centre. Cloudiness and precipitation are also found to be maximum in the southwest sector of the storm. The southwest sector of the storm appears to be the location of the strongest activity. Here winds, temperature, moisture, Vorticity, clouding and precipitation show their maximum strength.

(j) During the course of recurvature, widespread heavy rain occurs to the north and northeast of the centre.

(k) The vortex disappears at 200 hPa and an area of contour High overlies the lower level system at this level.

(l) Upper air CYCIR in case of Low are generally upto 850 hPa, incase of Deep-depression 700/500 hPa and incase of CYCLONIC STORM upto about 200 hPa.

RESTRICTED

56

RESTRICTED

(m) In case of depression tilt is usually towards south. In case of deep depression tilt is less pronounced and in case of SCS tilt is almost nil; rather, the axis is vertical.

32. Details on movement, recurvature and Intensification.

(a) While over sea climatology, persistence and steering level winds give a good indication of its movement in next 12 – 24 hrs.

(b) Average of persistence and climatology also gives a good indication of its movement over sea i.e. (P+C)/2

(c) Surface Ageostrophic Steering Method: First Isobar which is open from the most anticyclonic flow; draw a straight line from the centre of the storm to that point and that will give the direction of the storm (Fig.2).

Fig.2 Ageostrophic Method

(d) P24P24 i.e. 24 hrs pressure falls are seen to be maximum on the area of land fall. However, it has been observed that all storms don’t follow these criteria.

(e) Just above the flow pattern where circulation ceases to exist, the flow pattern determines the future track of the storm.

(f) Second-degree variation of pressure also gives indication of future movement.

(g) For non-recurvature the westerlies should be flat in the upper troposphere. A strong subtropical high to the north of the storm with major trough in the westerlies far to the west indicates non-recurvature.

(h) For intensification, at surface / lower tropospheric level convergence in close vicinity of centre, at 200 hPa, the area of divergence overlies the area of convergence at lower level.

33. Eye of a Tropical Cyclone

(a) Its shape may be circular or elliptical. Generally, the eye shrinks in size during the intensification of a vortex.

(b) As the vortex intensifies, subsidence inside the eye increases and the eye-region becomes more clear of clouds.

RESTRICTED

57

RESTRICTED

(c) In a very intense vortex system, there may be two cloud free regions surrounded by two cloud wells. In such a case, the outer cloud-free region is also called the outer eye of vortex.

(d) The development of an eye also indicates the intensification of vortex.

(e) The eye-wall generally has funnel-type appearance with narrower section near the surface and broader section at the top.

RESTRICTED

58

RESTRICTED

CHAPTER – 6

PAIR OF CYCLONES

Chapter Objectives

After reading this chapter you should be able to assimilate:-

Concept of pair of Cyclones. The Fujiwara Effect

Structure

1. Introduction

2. Pair of Cyclones

3. The Fujiwara effect

4. Binary Interactions

5. Conclusion.

Introduction

1. There are occasions when two cyclones are seen at the same time over oceanic surface. To call them as a pair of cyclones there is a mandatory requirement to fulfilled certain criteria. Movement and intensity prediction of these types of pair of cyclones is a challenging task.

Pairs of Cyclones

2. Errors in prediction of tropical cyclone motion over Western North pacific are due to presence of two spatially proximate storms.

3. Storm pairs subject to binary interaction averaged 1.5 annually over the western North Pacific and 0.33 annually over the Atlantic (1946-81)

4. High frequency over western Pacific is at least partially attributable to the overall higher tropical cyclone frequency over that basis compared to the Atlantic (25:10 annually).

5. Fujiwara (1923, 1931) demonstrated that relative motion was composed to counter clockwise revolution of one vortex about to other and that there was tendency for the approach of circulation with the same sense of rotation. (Fujiwara Effect).

6. Hoover, (1961) conducted an observational study of binary systems over Atlantic and Western pacific. Over Atlantic, the predominant motion was clockwise whereas over Pacific better agreement with Fujiwara expectations of counter

RESTRICTED

59

RESTRICTED

clockwise rotation was noted. Hoover suggested that the Atlantic discrepancies were due to different large scale circulation (steering) pattern between the two basins.

7. Brand (1970) studied 15 years (1953-67) of western Pacific binary cyclone tracks and derived a regression equation for rotation rate Y as a function of cyclone separation distance X,

Y= 119.06 - 0.162 X + 0.000055 X2

X1500where Y is in units of angular degrees of rotation per 12 hours with positive Y

indicating counter clockwise rotation and where X is in km.

8. Statistics. Over 30 years (1947-70), 43 pairs of tropical cyclones (binary) were observed over the western North Pacific Ocean. A binary case is defined as two cyclones which:

(a) Have coexisted for at least 48 hours.(b) Separated at some points by less than 1334 Km. (one unit 111.2 km)(c) Had attained at some points, at least tropical storm (18 mps).

9. Statistics of Rotation.

(a) In 30 (69.8%) of the pairs, Anticlockwise rotation about the pair mid point.

(b) In 05 (11.6%) of the pairs, clockwise rotation.

(c) In remaining 08 (18.6%) of the pairs; indeterminate.

The Fujiwara Effect

10. The interaction between spatially proximate (binary) tropical cyclones is such that the relative rotation in the counter clockwise sense and decreasing separation distance between the storm centres can be expected.

11. A cyclone tends to make other vortices rotate about it in a cyclonic direction, and an anticyclone tends to make them rotate its periphery in an anticyclonic direction.

12. In the same hemisphere, cyclone pairs tend to rotate cyclonically, relative to each other, and anti-cyclonically.

13. Neighbouring cyclones and anti-cyclones exert a “force” upon each other which is directed parallel to the wind flow between them.

14. When the two rotational fields lie in the opposite hemisphere, the clockwise or counter –clockwise direction of rotation is considered, regardless of whether it is called cyclonic or anticyclonic.

15. Thus, two cyclones lying near each other, one on either side of the equator, exert a force on each other towards the east. Usually this effect is much smaller than

RESTRICTED

60

RESTRICTED

that of the poleward anticyclones and appear only as a retardation of the normal westward movement.

16. The apparent effect of one rotational field upon any other varies:-

(a) Directly with its size & rate of rotation.(b) Inversely with the distance between the two centres.

17. Cyclone Pairs. In the same hemisphere, cyclone pairs tend to rotate cyclonically, relative to each other.

18. Anticyclone Pairs. Anticyclone pairs tend to rotate around each other anti-cyclonically.

19. Cyclone - Anticyclone Pair. Neighbouring cyclones and anticyclones exert a force upon each other which is directed parallel to wind flow between them.

Binary Interactions

20. Binary interaction is defined as two co-existing (for at least 48 hours) tropical cyclones, separated at some point less than 1334 km and with associated maximum surface wind of both systems of at least 34 kts.

21. Binary systems are more common over NW Pacific than over North Atlantic.

22. Rotation can be at times erratic (less frequent as compared to the anticlockwise rotation).

23. The possibility of clockwise rotation increases if both cyclones are steered by same 500 hPa ridge, or exist in similar large scale environmental flow.

24. Rotational pattern of counter-clockwise rotating systems is elliptical.

25. Meridionally oriented pairs rotate faster than zonally oriented pairs.

26. Fujiwara effect is highly effective in the case of storms near ITCZ.

27. Exceptions are not ruled out.

28. Environmental currents in which storms are embedded, being significantly effective, are to be filtered before studying Fujiwara effect.

29. Binary systems show both Clockwise (less frequent) and Anti-clockwise rotation about the pair mid-point.

30. Northeast-Southwest systems are likely to approach each other while Northwest-Southeast pairs separate.

31. Separation distance less than 6.5 degree are more suited for Fujiwara forcing. Environmental steering forces also affect if separation distance is larger.

32. Strong pairs (> 64 kt – Hurricanes) show greater Fujiwara effect.

RESTRICTED

61

RESTRICTED

33. Clockwise Rotation.

(a) Three out of five binary pairs exceeded 10 separation distance unit and the steering of each pair was apparently under the influence of the same subtropical high.

(b) Two out of five binary pairs had very less (4.6 units) separation. Both the systems were embedded in a similar environment ; are moved rapidly towards NE ahead of well marked trough in the westerlies and other moved slowly northward in the west of STH.

34. Counter-clockwise Rotation. Out of 30 cases of counter-clockwise rotation of binary rotation, 14 were contained within ITCZ; six were contained with in the ITCZ during their earlier stage and they were influenced by a westerly trough at their later stage while another nine pairs were located on the western side of the subtropical high. A case was located on the southern side of the subtropical high.

Conclusion

35. Various techniques involved in prediction of movement and intensity of pair of cyclones are required to be understood by a operational forecaster for good success in forecasting.

RESTRICTED

62

RESTRICTED

SATELLITE METEOROLOGY

RESTRICTED

63

RESTRICTED

CHAPTER-3

TEMPERATURE RETRIEVAL AND MEASUREMENT OF WINDS AND PRECIPITATION ESTIMATION

Chapter Objectives

After reading this chapter, you should be able to:-

Understand the theory of temperature retrieval and accuracy of the same.

Understand the theory of retrieval of sea surface winds and the utility of the same in weather forecasting.

Understand how precipitation estimation is done using Vis-IR measurements.

Appreciate the limitations of precipitation estimation using Vis-IR measurements.

Structure1. Introduction

2. Principle of Atmospheric Sounding

3. Retrieval Methods

4. Future Developments in Satellite Sounding

5. Atmospheric Motion Vector (wind) from Sequential Satellite Observations

6. Current AMV Derivation Methods

7. Rapid Scan Winds

8. CMV / WVWs Validations and Accuracy

9. Active Microwave Remote Sensing

10. Theory of Radar Sea Return

11. Scatterometer and Model Functions

12. Retrieval Algorithms

13. Current Scatterometer Instruments

14. Visible and IR Techniques for Precipitation Estimation

15. INSAT 3D Rainfall Products

16. Conclusion

RESTRICTED

64

RESTRICTED

TEMPERATURE RETRIEVAL

Introduction

1. Till now we have been considering remote sensing of earth’s surface. The atmosphere which forms an integral part of our planet and important for the very existence of life on earth has not been considered(with the exception of monitoring clouds),instead we have only found fault with its presence for interfering with the remote sensing observation of the earth’s surface. Remote sounding of the atmosphere from the space is usually not covered in books dealing with remote sensing, since they are usually dealt with as a part of meteorological studies. However, for completeness, the principle and techniques involved in remote sounding of the atmosphere are covered briefly in this chapter. Let us revisit the atmospheric composition dealt with in appendix 1.The atmosphere consists of 78% of N2 and 21% O2 and the rest, generally referred to as a minor constituents, consists of H2O,CO2,O3,CH4,CFCs,N2O,etc.All these constituents influence various phenomena taking place on the planet earth.

2. They absorb significant amounts of both solar and terrestrial radiation, thus considerably modifying the radiation field and temperature structure within the atmosphere and the earth. The average earth temperature would have been 18o

C(instead of 15o C) if there were no water vapour and CO2 in the atmosphere, which provide the ‘Greenhouse effect’. Greenhouse gases (GHG) absorb a part of the IR radiation emitted from the earth’s surface and reemit it, thus retaining a part of the IR radiation and thereby warming the surface of the earth and the lower atmosphere. The dominant GHG in the earth’s atmosphere is water vapour, followed by CO2, added to the atmosphere naturally (volcanoes, etc) and anthropogenic ally (fossil burning, decrease in forest cover etc.,). Industrial and agricultural activities have increased the GHGs and added now ones. The increased GHG leads to increase in global average temperature which can lead to sea level rise, weather extremes (floods, droughts ) and their consequence on agriculture, water supply etc. These changed would happen over much shorter periods as compared to the adaptability of the existing ecosystems, thereby leading to the disappearance of many species.

3. Photochemical reaction in the atmosphere is another area of interest. Ozone (O3) plays an important role in the energy balance of the middle atmosphere and also affects the ultraviolet dose on the earth. The species which are involved in the reaction affecting the changes in O3 include the nitrogen family (N2O, NO, NO2, HNO3, etc.,), the chloro-fluro carbons (CFC) etc. Information on the variability of O3

and the predator species with the latitude, longitude, time and season is of importance to understand various photochemical processes and their coupling with the radiation balance in the atmosphere.4. One of the most important parameters required for numerical weather forecasting is the global monitoring of the vertical temperature profile. Measure of the temperature also yields pressure via the hydrostatic equation

p(z) po exp(-z/H)

RESTRICTED

65

RESTRICTED

where p(z) is the pressure at height z, po is the surface pressure,

H is the scale height

KT(z)i.e., H = ---------------, where M is the mean molecular mass

Mg

5. In practice due to gradual change of T with z, an ‘effective’ H has to be used. Therefore, knowledge of T(z) and po gives information on p(z) also. Before the satellite era, temperature data was obtained by the in situ measurements carried out mainly by radiosonde balloons. Thus understanding of the many processes- radiative, chemical and dynamical – that take place in the atmosphere requires global observation of temperature and composition of the atmosphere and their variability. The following section gives the principle behind making such observations by remote sensing and the techniques involved.

Principle of Atmospheric Sounding

6. The basic principle behind the sounding of the atmosphere depends on the fact that the radiance leaving the top of the atmosphere will be a function of the emitting gas and the distribution of temperature throughout the atmosphere (Houghton, 1984). It was first shown by Kaplan, 1959, that the vertical temperature distribution in the atmosphere can be inferred from the measurements of upwelling emission around the absorption band of a relatively abundant gas in the atmosphere, which has a uniform distribution and does not vary with place and time. Conversely, the composition of a varying atmospheric constituent can be estimated if the temperature profile is known. To understand vertical sounding, consider an arbitrary spectral line shown in Fig.1, which gives transmittance (1- absorption) versus wavelength. At λ3, where transmission is very low, most of the radiance received should be from the top layer of the atmosphere, since radiation originating from the lower atmosphere will get absorbed. For observation at λ1, which is close to the transparent region, most of the radiation can come from close to earth surface, while a measurement in λ2 should have most of the radiation from an intermediate height. In other words, by making observation from the top of the atmosphere in a highly absorptive region, the instrument (radiometer) will ‘see’ only a limited distance into the atmosphere, while when observations are made at wavelengths close to the transparent edge of the absorption line, the radiometer gets a deeper ‘look’ into the atmosphere. For observations in wavelengths between the above two, the sensor can see regions in between the top and bottom of the atmosphere. Thus, if measurements are carried out at a number of wavelengths from λ1 to λ3, the relative contributions from the different heights (more correctly, pressure) will vary. Or the radiation received has contribution from throughout the atmosphere but with different ‘weights’. The weighting function is essentially the derivative of the atmospheric transmittance (𝜏) at a wave length, to pressure (to be more precise 𝜕𝜏/𝜕log p). The altitude of the peak of the weighting function depends on the absorption coefficient at the particular wavelength. The weighting function when multiplied by the Planck function gives the upwelling radiance from a given layer of the atmosphere at that pressure. The weighting function is represented as a function of p (in practice as log p) for various wavelengths.

RESTRICTED

66

RESTRICTED

Fig. 1. Arbitrary absorption spectra of an atmospheric constituent transmittance is (1 – absorption)

7. Fig. 2. gives the weighting function of a typical space-borne sounder. Since the weighting function is a convolution of falling density and rising transmittance, it is bell-shaped.

8. The absorption band selected should have the following characteristics to ensure that sounding gives reliable results.

(a) The absorption band selected should not overlap with bands of other atmospheric constituents.

(b) Local thermodynamic equilibrium (LTE) should apply, in which case the emission from the band will be proportional to the Planck function. For this purpose, the probability of excitation by collision should be much larger than the probability of de-excitation by radiation process. As altitude increases, the collision frequency falls and LTE does not hold good.

(c) The wavelength should be long enough so that the scattered solar radiation is insignificant compared to thermal emission.

9. In addition, for temperature sounding, the emitting constituent should be substantially uniformly mixed in the atmosphere so that the emitted radiation can be considered as function of temperature distribution only. 12.For temperature sounding

RESTRICTED

Fig. 2. Weighting function of a filter wheel vertical sounder

0.10 0.30 0.50 0.70 0.90 1.10 1.30

747cm-

1

725cm-

1

708cm-

1

695cm-

1

676cm-

1

668.5cm-

1

Weighting function (dv

/d log p)

0.1

1.0

10.0

100.0

1000.0

Weighting function of a filter wheel vertical sounder

plog

Pressure (mb)

67

RESTRICTED

of the atmosphere both O2 and CO2 satisfy these conditions. CO2 has two absorption bands at 15 𝞵m and 4.3 𝞵m, which can be used up to ~ 80km for 15 𝞵m band and 35 km for 4.3 𝞵m band (Houghton, 1969). O2 absorption band is in the microwave region (5mm) which can be used up to 100 km.

Retrieval Methods

10. The physical problem we would like to solve is this: What temperature and trace gas concentration profiles could have produced a set of observed radiances? This is called the inverse problem or retrieval problem. The opposite problem, called the forward problem, is to calculate outgoing radiances given temperature and trace-gas profiles.

11. Forward Problem. If we know the entire atmospheric temperature profile T(z), then we can uniquely compute the radiances through radiative transfer equation.

12. Inverse Problem. It is the retrieval of the atmospheric temperature profile from a set of measured radiances. But this is ill-posed because the weighting functions are generally broad and have a finite number of chances and an infinite number of different temperature profiles could give the same measured radiances.

13. Clouds cover approximately 50% of the earth; thus about 50% of the soundings are contaminated with clouds. Given clear-column radiances, the forward problem is easy to solve, but the inverse problem (retrieval problem) is difficult because the solution is not unique. When a finite number of wavelengths are observed and the measurements are contaminated with noise, an infinite number of solutions are possible. The retrieval problem becomes one of finding temperature profiles that satisfy the Radiative Transfer Equation and approximate the true profiles as closely as possible. Approaches to the retrieval problem can be classified into three general areas:-

(a) Physical Retrieval

(b) Statistical Retrieval

(c) Hybrid Retrieval

Physical Retrieval14. In physical retrieval schemes, the ease of the forward problem is exploited in an iterative process:

(a) Step I: A guess temperature profile is chosen.

(b) Step II: The weighting functions are calculated.

(c) Step III: The forward problem is solved to yield estimates of the radiance in each channel of the radiometer.

(d) Step IV: If the computed radiances are the observed radiances within the noise level of the radiometer, the current profile is accepted as the solution.

(e) Step V: If the convergence has not been achieved, the current profile is adjusted.

RESTRICTED

68

RESTRICTED

(f) Step III to V (or II to V) are repeated until a solution is found.

15. Chahine (1970) and Smith (1970) provided the two most widely used methods of adjusting the temperature profile.

16. Chahine’s Method. Chahine retrieves temperature for as many levels as there are channels in the radiometer. Suppose the radiometer has J channels, then the scheme retrieves temperature at the J levels located at the peak of the weighting functions. In this scheme there is one-to-one correspondence between channel j of the radiometer and level j, where the weighting function of the channel j peaks. Let Tj

(n) be the nth estimate of the temperature at the jth level and B j(Tj(n) be the resultant

Planck radiance in channel j at the wavelength of channel j. Chahine iterates temperature by iterating the Planck radiance:

Bj(Tj(n+1)) = Bj(Tj

(n))[Lj/Lj(n)]

Where, Tj(n) = nth estimate of temperature at jth level

Bj(Tj(n)) = resultant Planck radiance at level j

Lj(n) = nth estimate of radiance in channel j

Lj = observed radiance in channel j

The iterated temperature at level j is found using the inverse Planck function. This scheme works because if, for example, the calculated radiance in channel is greater than the observed radiance, it is reasonable to adjust downward the Planck radiance (and thus the temperature) at the level where the weighting function for channel j peaks. Since the peak of the weighting function is the highest contributor to the radiance, using the ratio of the observed to the calculated radiance to adjust the Planck radiance is also reasonable.

17. Smith’s Method. Smith’s scheme for adjusting temperature profiles is similar to that of Chahine’s, but he relaxes the requirement that the temperature be retrieved at only J levels. Suppose the temperature is to be retrieved at K levels. Suppose the temperature is to be retrieved at K levels. Let Tk

(n) be the nth estimate of the temperature at level k, and B j (Tk

(n) ) the resulting Planck radiance at the wavelength of channel j. As above, let Lj

(n) and Lj be, respectively, the nth estimate of, and the observed value of, the radiance in channel j. At each level, Smith obtains J estimates of an iterated Planck radiance: Bj(Tjk

(n+1)) = Bj(Tk(n))+[Lj-Lj

(n)] where the J estimates of (Tk

(n+1) ), obtained by the inverse Planck function, are denoted by (Tjk

(n+1) ). Not all of these J estimates are equally good because each channel sees some levels better than others. Smith solves for Tk

(n+1) as a weighted average of the Tjk(n+1) using the weighting function as weights.

Since Smith’s levels are not restricted to be at the peaks of the weighting functions, he cannot iterate the temperature at a level by using a single channel. At each level,

RESTRICTED

69

RESTRICTED

therefore, he obtains a suggested temperature change from each channel and lets the weighting function discriminate among them. Smith’s scheme is more flexible than that of Chahine in that it allows the user to choose the levels at which he will retrieve temperatures (consistent, of course, with the predetermined levels at which the transmittances are calculated). However, from J channels of information, one can calculate independent temperatures at J levels at most. If K is greater than J (as is usually the case) the extra levels are not independent.

18. To retrieve moisture profiles, the temperature profile is assumed,and thus Bl(T), is known. Smith (1970) starts with a first-guess mixing ratio profile q( p) and proceeds in a manner which is formally the same as that for temperature retrievals. The iteration formula is

Where j(n) is a sensitivity factor which estimates the mixing ratio change necessary

to correct for a given radiance imbalance. In equation form,

Where U(n)(p) is the nth estimate of the integrated watervapour above pressure level p. For each level, smih obtains J estimates of the integrated mixing ratio. As above, qk(n+1) is found as a weighted average of the qjk(n+1) using the weighting functions as weights. The logic of this scheme can be demonstrated as follows. Note first that since ∂ /∂U is always less than zero, the sign of j

(n) is opposite that of dT/dp. Consider a channel whose weighting function peaks in the troposphere. Suppose that the calculated radiance in the channel is too high compared with the observed radiance. Since j

(n) is a negative number(because dT/dp >0), qk(n) will be multiplied by a number greater than one, and qk(n+1) will be greater than qk(n). Increasing the moisture will move up in the atmosphere the point where optical depth equals one, and thus the peak of the weighting function for channel will move up. Since troposheric temperature decreases with height, Lj(n+1) will be less than Lj(n) and closer to the observed radiance. Obviously the first guess profile is important. The closer the first guess profile is to the actual profile, the better the solution is going to be. None of the retrieval schemes that we discuss is clearly superior to the others. Users who wish to implement a retrieval scheme must weigh the advantages and disadvantages of each. Based on the experiences, the principal advantages of the physical retrieval method are:-

(a) Physical processes are clearly evident during each stage of the retrieval(b) No large data base of radiosonde data is necessary

19. The dis-advantages are:-(a) The method is computationally intensive(b) It requires accurate knowledge of the transmittances(c) Except for information contained in the first guess, it does not utilise known statistical properties of the atmosphere

Statistical Retrievals

RESTRICTED

70

RESTRICTED

20. In statististical retrievals, the radiative transfer equation is not directly used. These methods assume that the radiometer has been designed so that its channels (weighting functions) will vertically sample the atmosphere. A set of radiosonde soundings that are nearly coincident in time and space with satellite soundings is compiled. This set, called training data set, is used to calculate a statistical relationship between observed radiances and atmospheric temperatures. These relationships are then applied to other observed radiances to retrieve temperatures. These relationships are then applied to other observed radiances to retrieve temperatures. The process can be described as follows.

21. Suppose there are N sounding pairs in the training data set. Let 1 be the J x 1 column vector containing the J radiances observed by the radiometer in one of these soundings. Let t be the K x 1 column vector, paired with 1, which contains the temperatures (and perhaps dew points) at K levels in the atmosphere. Finally, let the symbolx represent the element –by-element average of the vector x. Statistical retrieval methods require finding the K x J matrix C and the vectors 1 and t such that t – t = C(I _ 1 ) Note that once C, t , and 1 are known, retrieval of temperatures from observed radiance is a very simple task which involves only vector subtraction, matrix multiplication, and vector addition. This simplicity makes a statistical method attractive for operational retrieval schemes in which numerous soundings must be processed. There are several ways to find C, t, (I). The simplest is the regression solution. One problem with the regression solution is that no “filtering” of noise from the input temperatures or radiances is done. As a result, the C matrix can be unstable; that is, small radiance errors can produce unacceptably large errors in the retrieved temperatures. Smith and Woolf (1976) developed a technique to filter noise using statistical eigen vectors. Their scheme was used operationally for many years. More recently, Thompson (1992) employed singular value decomposition to improve statistical retrievals.

22. Just as good first-guess profile and an accurate knowledge of the transmittance are crucial to the success of the physical retrieval method, so acquiring a representative training data set is crucial to the success of the statistical method. This is more difficult than it might appear. Some of the requirements for these data sets are the following:-

(a) The data sets must be large to ensure that the retrieval matrix will be stable.(b) Data sets must be collected for ach satellite because of small differences in the radio meter on each.(c) The data sets must be updated frequently because, to site an extreme example, a data set collected in the winter cannot be expected to yield accurate retrievals in the summer. (d) The data set should be divided into latitude zones, and by surface type (land or ocean).

23. A clear advantage of the statistical method is that a statistical picture of the structure of the atmosphere is an integral part of the method; that is, the retrieval temperatures cannot be too far from those which have actually been observed in the past. This imposes an ever greater requirement that the training data set be appropriately chosen, however; otherwise the wrong structure will be built into the

RESTRICTED

71

RESTRICTED

retrievals. To summarize, the principal advantages and disadvantages of the statistical method are:-

(a) Advantages

(i) The actual retrievals are computationally easy.(ii) It requires no knowledge of the transmittances or use of the radiative transfer equation.(iii) It extensively utilizes statistical properties of the atmosphere.

(b) Disadvantages

(i) A large training data set of coincident radiosonde and satellite data is necessary.(ii) Physical processes are embedded in the statistics.

Hybrid Retrievals

24. Hybrid retrievals methods are in between physical and statistical retrieval methods. They appear much like purely statistical methods, but they do not require the large training data set. They use weighting functions like physical retrievals, but they do not directly involve integration of radiative transfer equation. Hybrid methods, better known as inverse matrix methods are reviewed by Fritz et al (1972). The main advantage of the Hybrid methods is that they are easier to put in to operation than the statistical or physical methods. They share with the physical retrieval method the disadvantage of depending on knowledge of the transmittances. Most of them share with statististical retrieval method the advantage of including statistical knowledge of the atmospheric structure.

Operational Retrievals

25. The retrieval method is only a small part of the process by which radiances from operational satellites are converted in to temperature and moisture soundings. In the succeeding paragraphs, the processes that are used operationally to produce soundings from NOAA and GOES satellites are discussed.

26. The TIROS N Operational Vertical Sounder. On the NOAA satellites there are three sounders: the HIRS/2, the MSU and SSU. Together they are called the TIROS N Operational Vertical Sounder (TOVS). Fig1. Shows the TOVS weighting functions. The process by which TOVS soundings are retrieved is described by Smith et al(1979a). The reader should be aware, that however, that operational sounding retrieval is not a static process. As new techniques are developed and as problems arise, the system is modified. These modifications can be determined by referring to a periodically updated NESDIS document, The NOAA Polar Orbiter Data User’s Guide. The TOVS system consists of four modules: the Preprocessor, the Atmospheric Radiance Module, the Stratospheric Mapper, and the Retrieval Module.

RESTRICTED

72

RESTRICTED

(a) The TOVS Preprocessor. The Preprocessor processes data received directly from the satellite Operations Control Centre. The digital counts are converted to radiances and each scan spot is Earth- located. The SSU data are further processed by the Stratospheric Mapper. A variety of corrections are applied to the HIRS/2 and the MSU data.

(i) The MSU data are corrected for antenna side-lobes(ii) The limb correction is applied to both the MSU and HIRS/2 data so that the modules can treat the data as if they were obtained at nadir.(iii) An attempt is made to correct the window channels for water vapour absorption.(iv) For day time data, an albedo, estimated using the 3.7µmchannels for reflected sunlight.

Finally, the Preprocessor obtains solar zenith angles, terrain elevations, and initial guess values of skin temperature and surface albedo. All these data are staged to disk to wait further processing.

(b) The TOVS Atmospheric Radiance Module. It’s job is to deliver spatially averaged, clear column radiances to the Retriever Module. Perhaps surprisingly, the Atmospheric Radiance Module consumes most of the computer time in the TOVS system. First the data are divided in to boxes of nine HIRS/2 scan spots and seven spots along track. One sounding is retrieved to represent this group of 63 scan spots. The nominal resolution of the operationally retrieved TOVS soundings is approximately 250Km. Next the 63 scan spots are tested to determine they are contaminated by clouds. Finally the atmospheric Radiation Module produces clear column radiances. If four or more scan spots are determined to be clear the clear column radiances are calculated as weighted average of the observed radiances from the clear spots. If fewer than four clear scan spots are found, the adjacent pair or N technique of Smith and Woolf is attempted. If four or more good adjacent pairs can be found, clear column radiances are estimated. Soundings produced from these radiances are called partly cloudy or second path soundings. If the scene is essentially completely overcast(fewr than four good adjacent pairs), the HIRS/2 channels which sense the troposphere are discarded and the sounding is retrieved using only MSU, SSU and Stratospheric HIRS/2 channels. These soundings are called cloudy or third path soundings. In 1986, a higher resolution scheme was implemented.

(c) The TOVS Stratospheric Mapper. Because the SSU does not scan as far out as the HIRS/2 and the MSU, HIRS/2 scan spots at the edges of the scan have no corresponding SSU data. This problem is solved by the Stratospheric Mapper, which maintains a global map of SSU radiances on a latitude-longitude grid. The map is updated as new observations arrive. The SSU data are corrected for limb effects before mapping.

(d) The TOVS Retrieval Module. The Retrieval Module performs the retrieval using clear-column radiances produces by the Atmospheric Radiance

RESTRICTED

73

RESTRICTED

Module and the Stratospheric Mapper. Until September 1988, temperatures and water vapor below 100 hPa were retrieved using statistical eigenvectors (Smith and Woolf, 1976). Above 100 hPa, temperatures and total ozone were retrieved by regression. As of this writing, temperatures and water vapour are retrieved using the Minimum Variance Simultaneous (MVS) method. The methods is ‘’simultaneous’’ in the sense that temperature and water vapor are retrieved using one matrix.

27. Archived TOVS Data. TOVS sounding data are archived by NESDIS. Users who are interested in ordering historical data, however, should be aware that the archive tapes come in the form of layer-mean virtual temperatures for 15 layers between the surface and 0.4 hPa. Water vapour is archived in the form of layer perceptible water in the three lowest layers (surface-850, 850-700, and 700-500 hPa). The tropopause temperature and pressure, total ozone, cloud-top pressure, cloud amount, and an average value of N*are saved (Kidwell, 1986).

28. The VISSR Atmospheric Sounder. VAS soundings have been retrieved operationally at NESDIS’s VAS Data Utilization Centre (VDUC) since the summer of 1987. The retrieval process, described by Hayden (1988), is similar to that used for TOVS soundings. The retrieval algorithm is a hybrid scheme described by Smith et al. (1986). First-guess temperature and moisture profiles are selected, usually from a model forecast such as the Nested Grid Model (NGM) or the Global Spectral Model. Surface temperature and humidity are obtained from analyzed fields of the corresponding hourly observations. The first-guess estimate of the skin temperature is obtained by regression from the 11.2, 12.7, and (for noncloudy scan spots) 3.9-um brightness temperatures. Temperatures above the 100-hPa level are obtained from the latest analyses or from climatology. Using thfe first-guess profiles, the weighting functions and the brightness temperatures in the VAS channels are calculated. The temperature and moisture profiles are calculated by making a single correction to the first-guess profiles, which is why the scheme is known as the ‘’ simultaneous’’ or ‘’ one-step’’ method.

29. Quality of the Retrieved Soundings. How accurate are soundings from satellite measured radiances? Although a great deal of effort has been expended on this question, the answer is still not simple. The basic difficulty is that to assess accuracy, one needs a standard with which to compare. Radiosonde soundings are the most obvious candidate but two problems occur. Radiosonde soundings are not error free. Satellite soundings are fundamentally different from radiosonde soundings. The problems with satellite soundings are pointed out not to dissuade the reader from using satellite soundings, but to illustrate the point that satellite soundings are not direct substitute for radiosonde soundings. Satellite soundings contain a wealth of information on spatial and temporal scales unattainable with radiosondes. However, those who attempt to use satellite soundings without taking in to account their differences from radiosonde soundings likely will be disappointed.

Future Developments in Satellite SoundingRESTRICTED

74

RESTRICTED

30. Developments are underway to improve sensors to provide satellite soundings at finer resolution in horizontal and vertical. NOAA K, satellites with ATOVS (Advanced TOVS) has been launched this year. ATOVS has twenty channel HIRS - III for sounding in IR region, fifteen channel AMSU - A (Advanced MSU) for temperature sounding and five channel AMSU - B for moisture sounding with improved resolution. Micro-wave soundings are accurate in cloudy areas. Later plans are for ITS (Infrared Thermal Sounder) and IASI (Improved Atmospheric Sounder in Infrared). These will have more channels and better foot-print resolution to give more detailed information of vertical structure of the atmosphere. Indian geosynchronous satellite INSAT –3D would have a multichannel sounder like VAS.

MEASUREMENT OF WINDS

 

RESTRICTED

75

RESTRICTED

31. Wind is a primary variable for the representation of atmospheric state. Accurate estimation of the wind field in areas where no conventional data are available is essential for operational weather forecasting. Measurement of wind from geostationary platforms is important, as it provides near continuous data where conventional observations are lacking, particularly over the data-sparse oceans. Motion of the atmosphere in wide area can be derived by tracing the movement of individual cloud or water vapour patterns in successive geostationary satellite imageries. This product is called Atmospheric Motion Vector (AMV). It includes information on both wind speed and directions and comprises of both CMV (Cloud Motion Vector) and Water Vapour Winds. Studies as early as Bauer (1976) showed that CMV s have a capacity similar in several aspects to that of radiosondes for representing atmospheric flow.

32. Satellite imagery in window channels of different wavelength bands provides instantaneous view of cloud cover. It is therefore an added tool for forecasters. The numerical modellers and forecasters need data from complete vast oceanic data sparse region. Satellite inferred information is intended to be supplementary to conventional observations and not to compete with it. The challenge therefore is to transform information rich imagery into quantitative products. These products will ensure our understating and prediction of meteorological phenomenon beyond existing methods of observing atmosphere. It was therefore imperative that, the measurement of radiance from satellite be done in more channels. Besides window channels, data from channels in absorption bands of some common constituents (CO2, H2O, O3, N2O etc.) can help in inferring other important information, Direct wind observations over oceanic area in tropics are vital. As in low-latitude regions wind field cannot be inferred from mass field. Successive images can provide the information by tracking cloud tracers. Polar orbiting satellites viewing an area once in 6 hours cannot provide the wind derivation capability.

33. Geostationary satellites further enhanced the capability by repetitive view of earth, water cloud system round the clock. In 1966 came first Application Technology Satellite (ATS-1) with spin scan camera. It demonstrated the capability of cloud tracked wind (CMV or CMW) by movie looping. Height was assigned from visual inspection of cloud type and pattern. Latter developments led to establish technique for CMV derivation and height assignment from temperature estimated from infrared band.

Atmospheric Motion Vector (wind) from Sequential Satellite Observations

34. From the early 1970s, wind has been estimated routinely from space using sequential imagery from geostationary satellites. Hubert and Whitney (1971) identified and tracked passive tracers from ATS-1 to determine cloud motion and hence wind velocity. Endlich et al. (1971) referred to the digitisation of ATS data and computer tracking of cloud targets to estimate atmospheric motion velocities. Leese et al. (1971) and later Endlich and Wolf (1981) used cross-correlation and pattern matching methods for identifying targets in different images.

35. Intervals between images, from rapid scan, e.g. 1 minute up to 3 hours have been used to estimate AMVs. In 1974, the Synchronous Meteorological Satellite (SMS-1) was launched and provided images at 5-minute intervals which enabled shorter-lived clouds to be studied. It had a VIS channel with 0.8 km sub-satellite resolution and an IR channel.

RESTRICTED

76

16

16

A (i,j)

RESTRICTED

36. Basic Concept. We know that frequent images of the same area are only possible from the geo-stationary satellites. Therefore, by tracking the movement of the same cloud in successive images can give a good estimate of the winds. As shown in Fig.1(a & b) , a cloud in image at time T is positioned at A( i,j ) and same cloud is positioned at B( i,j) in image at time T + T, then “A vector difference of the location of a cloud in two successive images divided by the time interval between images is an estimate of the horizontal wind at the level of the cloud”

(T) (T)+ T Fig.1 (a). Tracer Image Fig.1 (b). Tracer Image

37. The first Geostationary Operational Environmental Satellite (GOES-1) was launched in 1975. The first CMV system for GOES was similar to that described by Hubert and Whitney (1971). The first European Meteosat and Japanese Geostationary Meteorological Satellite (GMS) satellites were launched in 1977. Routine production of AMV s from GMS began in 1978 with manual target selection and tracking of high-level winds and automatic tracking of low-level winds. Imagery from the water vapour absorption band of MeteoSat-1 became available in 1977. Early work in the generation of CMVs from absorption channel imagery was reported by Allison et al. (1972) and by Steranka et al. (1973). They both examined the application of the Nimbus 4 Temperature and Humidity Infrared Radiometer (THIR) 6.7 µm channel observations to the determination of the moisture distribution and wind flow near the 400 hPa level. Almost a decade later, Kastner et al. (1980) reported using overlapping swaths of consecutive Nimbus 5 orbits to derive wind velocities from water vapour imagery in another pre-Meteosat study. Their work demonstrated the feasibility of using sequential water vapour images for wind determination. Subsequent work by Fischer et al. (1981), Eigenwillig and Fischer (1982), Stewart et al. (1985) using the McIDAS system (Suomi et al., 1983), Hayden and Stewart (1987), Velden et al. (1992) and Le Marshall et al. (1985) described the use of this imagery for deriving mid-tropospheric wind vector by tracking water vapour features. These works further refined the procedures for generating water vapour winds and gave an initial indication of their utility for NWP in particular for tropical cyclone track forecasting where they can provide important information around the 'steering level'.

38. By the 1990s, IR, VIS and WV motion vectors were included in experimental and operational data bases. In 1992, a fully automatic operational cloud drift wind processing system was established in the Bureau of Meteorology, Australia. Later hourly AMV data was routinely generated and high resolution VIS imagery provided high density wind data and was a good source of low-level cloud drift wind data. Inoue and Smith (1994) tracked low-level moisture patterns between the surface and 700 hPa using WV imagery and split-window IR brightness temperature differences.

RESTRICTED

B (i,j)

A (i,j)

77

RESTRICTED

A considerable body of work was produced, in the mid-nineties, on the generation and application of these winds by Velden and associates.

39. Incorrect altitude assignment has always been a major source of error in satellite wind measurement. Szejwach (1982) developed a method for estimating the temperature of cirrus clouds using both the IR window channel and water vapour absorption band data. Menzel et al. (1983) used the VAS sounder to assign cloud heights. In the 1990s, many studies focused on improving altitude assignment. Hayden (1993) described an auto-editor which included an adjustment for speed bias and allowed altitude reassignment when necessary. Research into experimental stereoscopic altitude assignment techniques for cloud drift wind derivation by interactive methods and the use of polar orbiting satellites was conducted by Purdom and Dills (1994).

Current AMV Derivation Methods

40. Agencies currently generating winds for operational use are the European Organisation for the Exploitation of Meteorological Satellites (EUMETSAT), the National Oceanic and Atmospheric Administration/National Environmental Satellite Data and Information Service (NOAA/NESDIS), the Japanese Meteorological Agency, the Indian Meteorological Department (IMD) and the Bureau of Meteorology, Australia.

41. At NOAA/NESDIS IR, WV, VIS and sounder water vapour winds (7.4 µm and 7.0 µm) are produced from the GOES I/M series satellites generally three hourly, with image intervals ranging from 60 minutes for sounder WV winds, through 30 minutes for infrared and WV winds, to 15 or 7.5 minutes for visible cloud drift winds. Low-level height assignment is at cloud base level and uses the method developed at the Bureau of Meteorology, Australia. Upper-level vectors are height assigned using the IR window and WV intercepts method and quality control involves use of an 'auto-editor' and now a Quality Indicator (QI). At EUMETSAT, cloud tracked winds generated from IR, WV and low-resolution VIS imagery are distributed every 90 minutes. Clear sky WV winds from cloud-free areas are also distributed every 90 minutes. High-resolution VIS image based winds are also produced every three hours. Low-level height assignment is to the cloud-base level. Radiance slicing is applied to segments before tracking to enhance the highest cloud layer, and cloud tracking is automatic, using a cross correlation technique, with tracking aided by use of an ECMWF forecast. A semi-transparency correction is applied using IR and WV data. Quality control is fully automatic and a quality indicator (QI) is appended to the vectors.

42. At the Meteorological Satellite Centre (MSC) Japan, cloud and WV winds are produced four times per day with high density low-level visible winds being produced once per day (0400 UTC) in a 20°x20° box around any typhoon. For upper-level vectors, areas containing cumulonimbus are excluded and the IR and WV intercept technique is used in the height assignment for non black-body clouds. Low-level winds are currently assigned to fixed heights. At the IMD cloud drift winds are derived from Kalpana-1 in April 2003 and from INSAT-3A data at 0000, 0600 and 1200 UTC using half hourly image triplets. Satellite winds at 0600 UTC are generated using VIS imagery for cloud tracers while IR imagery is used for cloud tracking at 0000 and 1200 UTC. Cloud top temperatures are used for height assignment. Automatic quality control includes acceleration checks and use of the

RESTRICTED

78

RESTRICTED

forecast model first guess before dissemination of the wind data on the Global Telecommunications System (GTS).

43. The methods used for AMV estimation in the Bureau of Meteorology, Australia are largely covered in Le Marshall et al. (l999, 2000). Three sequential IR, VIS or WV band images (a triplet), usually separated by an hour or half an hour are used for velocity estimation. High density winds are generated continuously at hourly or half hourly intervals. Selected targets are tracked automatically using forecast winds then a lagged correlation technique, which minimises root mean square (RMS) differences in brightness from successive pictures, is used to estimate the vector displacement. Cloud height assignment uses forecast temperature profiles. The cloud height assigned for the low-level winds was that of the cloud base. The benefit of height assignment to the cloud base has been documented in Le Marshall and Pescod. (1994). Height assignment involves fitting Hermite polynomials to smooth raw histograms of brightness temperature enabling estimation of cloud base altitude from cloud base temperature. Upper level AMVs are assigned to the cloud top altitude which is estimated using 11 and 12 µm split window observations.

44. A selection of winds, estimated by the operational Bureau AMV system over the Tasman Sea (30ºS, 160ºE) is shown in Fig. 1 (a). The local IR (11 µm) system alone can provide up to 400 wind vectors around Australia at 0500, 1100, 1700 and 2300 UTC. AMVs around Tropical Cyclones Feng Shen and Fung Wung in the NW Pacific around 0500 UTC 25 July 02 are seen in Fig 1 (b).

Assumptions and Limitations

45. The AMV extraction technique assumes that the clouds will act as rigid bodies. (i.e. no deformation of shape will take place in this duration). The assumption may not hold good sometimes, particularly during the processes of rapid convection, or extremely high winds. These are the limitations of the method.

(a) The tracers chosen may not be passive tracers, as cloud tracers do have vertical extent. In a large cloud system, ideal tracers may not be available. This leads to wind from top layer of the cloud and no low level CMV. Tracers in multilayered clouds produce CMVs which are weighted average of the flow in deeper layer. In regions of strong vertical wind shear, low level cloud movement is rather slow. This causes a negative speed bias in derived CMVs.

(b) Sub pixel sized tracers and cirrus (CI) also lead to error in height assignment to CMVs, as these clouds are transparent to warmer radiation from below.

(c) Errors in navigation and frame to frame registration give additional component to CMVs.

(d) Tracer from finer resolution image data show wind flow better than that derived from coarse resolution IR image. A coarse resolution image triplet produces CMVs that show more zonal flow in computed CMVs. This is not so in visible images, but they are not available round the clock.

RESTRICTED

79

RESTRICTED

(e) Accurate navigation and frame to frame registration from finer resolution imagery can help minimizing the errors. The height assignment is the most challenging problem. Sub pixel sized low cloud tracers do not create a serious problem. Since the IR inferred height are closer to the base rather than top of cloud and these clouds are known to move with flow at the base level.

(f) Large error in height assignment is noticed in case of thin cirrus or sub pixel sized clouds in middle and higher levels. There are different approaches to account for the same.

Methodology used for CMV by I Met D

46. At present Cloud Motion Vectors are being derived by India Meteorological Department both from Infrared and Visible data of Kalpana-1 & Insat-3A Satellites from the triplets of following imagery:-

Time in UTC CMV’s From

(a) 23:30, 00:00, 00:30 IR Kalpana-1(b) 07:00, 07:30, 08:00 IR, VIS Kalpana-1(c) 11:30, 12:00, 12:30 IR Insat-3A

(d) 18:00, 18:30, 19:00 IR Insat-3A

47. CMV computation done interactively by experienced operator can provide good quality CMVs by tracking suitable tracers in tracer and target images and thereby computing the CMVs. In height assignment a developed skill can help in accounting for emissivity and therefore proper height assignment. This interactive process is rather slow. Automotive techniques are now available and these in general have the following five steps in the derivation of Cloud Motion Vectors.

(a) Registration of Triplet

(b) Cloud Tracers Selection

(c) Tracking of Cloud Tracer in Target image

(d) CMV Computation

(e) Height Assignment

48. Registration of Triplet. Registration implies that when two images are displayed in time sequence, the land features should look stationary. In other words, two images are said to be registered if the landmarks in one image exactly overlay on the other image. The registration is achieved by shifting one image with respect to the other till the landmarks are lined up in both the images. In INSAT Meteorological Data Processing System (IMDPS), every reception is navigated automatically as a part of near real time processing. If required, the capabilities (NAV-SHIFT) are applied which exercise the North-South or East-West translational shifts or rotation to achieve the accuracy of Navigation. It is presumed that frame to frame registration is achieved as result of accurate navigation and the computation of Vector Displacement uses latitude/longitude of the initial and final points of the Vector.

RESTRICTED

80

RESTRICTED

49. Cloud Tracers Selection. The steps involved in cloud tracer selection are:-

(a) Tracer Location List Generation

(b) Tracer Threshold Calculation

(c) Tracer Histogram Generation

(d) Tracer Selection

At every selected location, a box of 16 x 16 pixels is considered. Then a four bin histogram is generated for IR imagery.

Total Number of pixel in a box = 256From the Four-Bin Histogram

Total Number of Low Cloud pixels = Ncl

Total Number of Medium Cloud pixels = Ncm

Total Number of High Cloud pixels = Nch

Total Number of Cloudy pixels = Nc

Nc = NCl + NCm + NCh

If the number of clear pixels contained in a box is equal to or greater than 85 % of total pixels, the box is treated as CLEAR

(N - NC) > 85% of N

If the number of cloudy pixels contained in a box is equal to or greater than 90 % of total pixels, the box is treated as CLOUDY

NC > 90% of N

Out of NCl , NCm and NCh , whichever is highest, the tracer is identified to be of that type.

50. Tracking the Cloud Tracers. Manual Tracking and Automatic Tracking are the two basic methods of tracking of clouds.

(a) Manual Tracking. In this technique, an analyst locates the clouds to be tracked on each of two successive images. This is generally done by positioning a cursor at the cloud on a video display device. The advantages are that even low clouds can be tracked through thin cirrus overcast and theoretically, more number of vectors can be obtained. But the disadvantages are that it is tedious and limits the number of tracers

(b) Automatic Tracking. In this method, individual clouds cannot be tracked. A pattern of clouds are tracked. Therefore, the average motion of an

RESTRICTED

81

Reference Window (T)

(p,q)Best Fit

Search Window (S)

16

16

RESTRICTED

area of clouds is calculated. In automatic tracking pattern matching is done primarily using the Cross Correlation method.

(i) Search window size for Low clouds is 36

(ii) Search window size for Medium clouds is 44

(iii) Search window size for High clouds is 52

At each of the lag position (p,q), the cross correlation is calculated by the following formula

2

1

222

11

2

11

2211

,,(),(

),,(),(),(

qpSqjpiSjiT

qpSqjpiSTjiTqpc

rrr

j

r

j

r

j

r

j

rrr

j

r

j

Where

T is the array of pixel intensities from the reference window and is the mean, and S is the array of sub-set of search window at lag position (p,q) and S is the mean at that position

2/1

1 1

2

1 1

2

00

),(2

,2

((),((

)),(2

,2

()(),((),(

r

j

r

j

r

j

r

j

r

j

r

j

qpSrqjrpiSjiT

qpSrqjrpiSTjiTqpc

Where, ),( qpc is cross correlation at lag position (p,q), n is size of tracer reference window, T and S are of the pixel intensities in the tracer reference window and target search window. Bar denotes mean.

The lag position (p,q) where the correlation coefficient is maximum is assumed to be the final position of the vector. Cloud tracking becomes difficult if multi-layer clouds are present in the scene. To overcome this difficulty, IMDPS has a provision of ‘Masking Option’. All pixels in the Reference and the Search window which were not identified as part of a specific cluster are masked out. Before proceeding to vector calculation, target is classified and verified. This is basically a validation step so as to ensure that the same cloud pattern was tracked. For each of the tracked

RESTRICTED

82

RESTRICTED

cloud targets, the four bin histogram is generated as was done for the cloud tracer and the same classification criteria is applied. The cloud target is rejected if it is not of the type of cloud tracer.

51. CMV Computation. The centre of the reference / search window is the initial point of the vector and the location for which absolute maximum peak is obtained as the final position of the vector. From these positions, CMV is calculated. If correlation returns multiple locations with the same maximum value, the first one is accepted.

52. Height Assignment. Height assignment of the CMV’s is being done using Infrared Window (IRW) technique at present. Mean temperature of the 25% coldest IR pixels (John LeMarshal et al. 1993, Merrill R 1989, Nieman S. J et al. 1997) is considered for assigning the height of the CMV. The software of the system is being upgraded and the H2O- IRW Intercept Method will also be tried shortly.

53. Following these steps CMVs can be derived round the clock, (presently are being done at 00, 06, 12 and 18 UTC) using IR tracers and then assigning height to derived CMVs. During daytime visible tracers can be used for cloud tracking and then height assignment using contemporary IR image can give good CMVs. This is because of finer resolution of visible imagery. Some geostationary satellite (GOES, METEOSAT, INSAT) have water vapour imaging in addition to visible and infrared imaging. This provides opportunity to derive water vapour winds (WVWs) even from the areas free from clouds, when no CMVs are available.

Quality Assurance Tests

54. The two sets of CMVs from triplet images are now available and suitable quality assurance tests are necessary to eliminate CMVs of doubtful quality. CMVs sets are subjected to various tests to eliminate vectors not representing flow. Regarding quality control of the generated cloud motion vectors, are tested for the reasonable minimum and maximum limits of CMV, spatial and temporal consistencies of the vectors, comparison of the vectors with forecast field vectors and comparison with climatology in the event of non-availability of forecast field. These are being achieved through the following tests:-

(a) Absolute Threshold Test

(b) Speed Test

(c) Directional Stability Test

(d) Gradient Test

(e) Forecast Test

(f) Climatology Test

55. Absolute Threshold Test. The two vectors (VBA and VBC ) generated from two pairs of images are individually checked for the limits of speeds possible for various cloud types i.e.

RESTRICTED

83

RESTRICTED

|VBA| and |VBc| > Min. Speed, and |VBA| and |VBc| < Max. Speed

56. Speed Test. In order to ensure that the two speeds do not have a large difference, this test is applied.

| |VBA| - | VBc | | < 0.5 V average

where Vaverage = 57. Direction Stability Test. This is basically to ensure temporal consistency.

(a) If Vaverage < 20 Kts and the direction difference between VBA and VBc is more than 60, the CMV is rejected.

(b) If 60 > Vaverage > 20 and the direction difference between VBA and VBc is more than 40 , the CMV is rejected and

(c) If Vaverage > 60 Kts and the direction difference between VBA and VBc

is more than 20, the CMV is rejected.

58. Resultant Vector. At this stage, the average VB of the two vectors viz. VBA

and VBC is computed through U and V components which is then subjected to following tests.

(a) Gradient Test

(b) Forecast Field Test

(c) Climatological Test

59. Gradient Test. Basically, this is to test horizontal consistency.

(a) Speed difference between VB and the surrounding vectors must be less than the gradient speed threshold (20 kts).

(b) Direction difference between VB and surrounding vectors must be less than the gradient direction threshold (45).

(c) Distance from VB and the surrounding vectors must be less than the gradient distance threshold (300 Km.)

60. Forecast Field Test. This is to test CMV’s against the forecast wind field produced by different centres of the world. At present, IMDPS CMV’s are being checked against the forecast field from Limited Area Model, IMD. The vectors those differ the collocated forecast wind vector by beyond the specified limits are flagged. A six or twelve hour forecast field is ideal but may not be operationally available. In that case, longer duration forecast may have to be used. Quality control

RESTRICTED

| VBA | + | VBC |

2

84

RESTRICTED

filters out 50% to 70% of the computed wind vectors. Various centres apply these tests in different order and in combination like forecast test may be applied at the first instance to both the sets rather than final set.

(a) Speed difference between VB and the surrounding field must be less than 20 kts.

(b) Direction difference between VB and the surrounding field must be less than 45

(c) Distance between VB and the surrounding field must be less than 300 Km.

61. Climatological Test. If the forecast field is not available due to any reasons, the CMV’s are checked against climatology. The criteria of the test are same as that of forecast field.

62. Man-Machine Interactive Editing. The CMV’s which have passed objective tests are available for interactive editing by the analyst at the work station. The CMV’s can be displayed on the monitor and the analyst can flag all those vectors which are inconsistent with the general field of motion. However, any deviant CMV can be retained if the cloud pattern suggest the likely existence of some new or rapidly changing circulation system. It is also possible to change the level of CMV.

Rapid Scan Winds

63. INSAT Satellites have the capability to scan (Sector-Scan) a smaller area approx. 25 North-South in a span of 7 minutes repetitively three times. The Visible and Infrared images are shown in Fig. 1and 2.

Fig. 1. VIS image.

Fig. 2. IR image

Limitation of CMVs

RESTRICTED

85

RESTRICTED

64. The tracers chosen may not be passive tracers, as cloud tracers do have vertical extent. In a large cloud system, ideal tracers may not be available. This leads to wind from top layer of the cloud and no low level CMV. Tracers in multilayered clouds produce CMVs which are weighted average of the flow in deeper layer. In regions of strong vertical wind shear, low level cloud movement is rather slow. This causes a negative speed bias in derived CMVs. Sub pixel sized tracers and cirrus (Ci) also lead to error in height assignment to CMVs, as these clouds are transparent to warmer radiation from below. Errors in navigation and frame to frame registration give additional component to CMVs.

65.     Tracer from finer resolution image data show wind flow better than that derived from coarse resolution IR image. A coarse resolution image produce CMVs that show more zonal flow in comparison to computed CMVs from finer resolution. This is not so in visible images, but they are not available round the clock. Accurate navigation and frame to frame registration from finer resolution imagery can help minimizing the errors. The height assignment is the most challenging problem. Sub pixel sized low cloud tracers do not create a serious problem. Since the IR inferred height are closer to the base rather than top of cloud and these clouds are known to move with flow at the base level. Large error in height assignment is noticed in case of thin cirrus or sub pixel sized clouds in middle and higher levels. There are different approaches to account for the same.

(a) The height of medium and high cloud can be determined on the basis of mean temperature of coldest 25% of the cloudy pixels in the reference window.

(b) GOES satellite has VAS sounder it has number of channels in CO2

absorption band centre at 15 mm. Channels 3 (14.2 mm), 4 (14.0 mm) and 5 (13.3 mm) are used in radiance slicing (CO2 slicing). Clouds near the band centre represent cloud from top of atmosphere while clouds away from band centre include clouds from lower layers. As one moves closer to 15 mm centre of CO2 band, lower level clouds tend to disappear. Those appearing in channel 5 are low clouds. Similar help could be taken from water vapour channel. Clouds in channel 4 and 5 but not 3 are middle level. These appearing in all channels are high clouds. This slicing of the atmosphere results in closer height assignment (Fig. 7). Another approach of cloud height assignment is a 3D-recursive filter to assign level of best fit to the derived wind.

 CMV / WVWs Validations and Accuracy

66. The quantitative checks of the quality of CMV’s were made by comparing them with the first guess from Limited Area Model (LAM) Forecast field generated operationally by IMD. It is seen that for medium level CMVs, the bias of Kalpana-1 CMVs is 2.0 m/sec or better. Rmse of Kalpana-1 CMVs is 6 m/sec for medium level CMVs which is fairly close to rmse of METEOSAT-5 CMVs. Same is the case with high level CMVs where rmse and biases have about 7 m/sec and 1.5 m/sec respectively.

67. Comparison of Kalpana-1 CMV’s with METEOSAT-5 CMV’s. The Kalpana-1 derived CMVs were compared quantitatively with METEOSAT-5 derived CMVs since

RESTRICTED

86

RESTRICTED

both satellites have fairly large common areas of coverage. It is noted that for all level CMVs, the bias of Kalpana-1 CMVs is around 2.0 m/sec which indicates that the quality of Kalpana-1 CMV’s is close to that of METEOSAT-5. Rmse of Kalpana-1 CMVs of various levels range from 3.4 to 6.5 m/s for the month of May 2004. The comparison does not indicate very high quality of Kalpana-1 CMV’s.

68. AMV Accuracy. The accuracy of AMV s is determined by several factors. The resolution of the imagery used for tracking clearly determines the accuracy in measuring displacement of targets. The time between images is also important in that it needs to be short enough to allow features to be tracked while being long enough that the inherent errors in measuring displacement (due to image resolution) do not lead to large velocity errors. The number of vectors also depends on these factors. A summary of these influences may be seen in Jedlovec (1998) and Le Marshall et al. (2000). The most important factor influencing the accuracy of the AMVs is height assignment. It is well established low-level vectors need to be assigned to the level of the cloud base (Hasler et al., 1976, 1977, Le Marshall and Pescod, 1994). Upper level targets on the other hand travel at the speed of the wind at cloud top and hence the AMVs need be associated with the cloud top altitude. The assignment of low level AMVs to cloud base is first described in Le Marshall et al. (1992) and later in Schmetz et al. (1996). These methods are designed to take into account variations in cloud emissivity in the estimation of cloud top altitude by using multi-pixel, statistical and multi-channel approaches.

69. AMV Quality Control. AMV rejection and expected error determination are usually based on several criteria. These include the correlation between the brightness temperature arrays of the search and target areas, and the difference in meridional and zonal wind components of the two vectors from a tracer tracked in pairs of adjacent images. The difference thresholds for rejection are usually situation dependent (near a jet stream, low level, etc.), and typically require the winds to be within 5 and 7 ms-1 for the zonal and meridional components respectively. The deviation of the calculated wind vectors from the first guess field is also an important consideration. The acceptable deviation is usually situation dependent (near a jet stream, low level, etc.) and zonal and meridional wind thresholds are typically less than around 10 and 7 ms-1 respectively. The weights assigned to AMVs during assimilation are dependent on their expected error. To enhance the use of AMVs in assimilation, the quality control (qc) elements cited above can provide both data selection and be used to estimate expected error, based on previous collocation statistics generated from coincident radiosondes and AMVs. These elements therefore allow selection of vectors with errors appropriate to the assimilation system. It is important to note that in practice the qc system should not be static but changes, for example, with the assimilation system and the accuracy of the background field, i.e. changes, for instance, where the forecast model in the assimilation system is improved. As the accuracy of the background field improves, the ability to distinguish between reliable wind estimates and erroneous data is enhanced.

RESTRICTED

87

RESTRICTED

SEA SURFACE WINDS USING SCATTEROMETER

70. An overview of remote sensing of ocean surface wind vector using spaceborne microwave scatterometer has been presented here. The inadequacies of the theoretical models to predict the observed behaviour of the radar measurements with wind conditions necessitated the use of empirical relationships between them for retrieving wind vector. More so, while extracting the wind vector from the scatterometer measurements, the harmonic nature of the radar backscatter on the wind direction causes multiple solutions which contain one true and several other ambiguous directions. In addition to that, the complicity of the empirical models and the noise in the data makes the inversion of wind vector impossible analytically, for which a numerical method has to be employed. Algorithms for retrieving the wind vector solutions and for removing the directional ambiguity have been presented with a few examples of wind field retrieval using ERS-1 scatterometer data.

71. The ocean surface wind is the main driving force for ocean circulation and for generation of waves and surface currents. It plays an important role in air-sea interaction, upwelling, biogeochemical transport in the ocean and a few other processes. The ocean surface wind vector is an essential input parameter for numerical models of ocean circulation and wave forecast which are used for oceanographic applications and climatic studies. It is also an indispensable parameter, alongwith a few others, for the prediction of storm surges caused by storms formed in the oceans which hit the coasts. These storm surges cause disaster in the coastal regions. Repetitive measurements of surface wind field over the large oceanic regions are thus necessary for the above studies and applications. Wind vectors can be obtained only from a microwave Scatterometer. These scatterometers work on the microwave region of electro-magnetic spectrum at 5-14 GHz. This part of the spectrum has advantage of obtaining data under almost all weather conditions. However, the wind speeds can also be obtained from space-borne microwave radiometers and altimeters also. Besides meteorological applications, ocean surface wind provides forcing parameters for studying the ocean general circulation.

72 Wind vector at the ocean surface is an important component of studies relating to ocean-atmosphere interaction such as exchange of momentum, heat energy across the air-sea interface. Greatest oceanographic gains from scatterometry are likely to be found in conjunction with the numerical modeling of dynamical ocean process. One of the most important boundary conditions of the surface wind-stress field which is direct or indirect forcing of many of the motions occurring in the oceans is from localised upwelling to global ocean circulation. Some research results indicate that scatterometer data can also be useful for ocean wave mapping and wave forecasting, offshore activities, ship routing, fisheries etc. Besides, wind vector from scatterometer, when properly assimilated in an appropriate numerical model, can improve weather forecast.

73. The scatterometer provides global measurements of the synoptic scale ocean surface wind vector at a neutral stability height at 10m. The neutral stability wind is defined as the wind speed that would result from a given frictional velocity u if the atmosphere were neutrally stratified with an adiabatic lapse rate. Thus the neutral stability wind speed is uniquely defined by the frictional wind speed at the sea surface rather than the actual wind speed at the 10m elevation.

RESTRICTED

88

RESTRICTED

74. The physical basis for the measurement technique is the Bragg scattering of microwave energy from centimeter length capillary waves of the oceans created by the action of the surface winds (Wright, 1968; Moore and Fung, 1979). The strength of the radar backscatter is proportional to the capillary wave amplitude, which is assumed to be in equilibrium with the wind friction speed u*. Moreover the backscatter is anisotropic, therefore, wind direction can be derived from radar back scatter. Active microwave remote sensing of ocean surface winds has started since the launch of SEASAT, in 1978. This was the first satellite dedicated to ocean research and was an experimental satellite which demonstrated the retrieval of wind to an accuracy of + 2 m/s, from wide swath scatterometer. ERS-1 satellite, currently in orbit carries a C-band microwave scatterometer and the Japanese Satellite ADEOS shall be in orbit with a Ku-band scatterometer.

Active Microwave Remote Sensing

75. Remote sensing is the science of deriving information about an object from measurements made at a distance without actually coming in contact with it. The observations are synoptic, provide repetitive coverage of large areas and data is quantifiable. An Active microwave sensor transmits a pulsed microwave energy which illuminates the desired area. The return signal is received by the same antenna and is processed. The measurement of radar sea return began during World War II, but the object of early experiments was establishing the size of the radar to determine the speed of the winds at sea using a satellite scatterometer was proposed in 1966. Since that time, many experiments have been conducted to ascertain the way in which the radar signal responds to variations in the surface of the sea caused by variations in local winds and the theory of radar return from the sea has been advanced significantly. The first spaceborne scatterometer was launched onboard Skylab in 1974. The circle flight data of the AAFE RADSCAT program suggested the ways to recover winds from SEASAT SASS measurements. The second spaceborne scatterometer was launched on SEASAT in 1978 which provide data of ocean surface wind vectors over global oceans.

Radar Remote Sensing

76. The average characteristics of radar back scatter from surfaces are described by the differential scattering coefficient, so. the radar backscatter power Wr from a point target is :- Wr = (WtGtArs)/(4PR2)2 , Where

Wt = Transmitter powerGt = Gain of the transmitting antennaAr = Effective aperture of the receiving antennas = Scattering cross section of the targetR = Range from the radar to the target

An area target such as the surface of the sea as the surface of the sea is normally considered to be a collection of point targets contributing returns whose phases are random. The ensemble average of the returns from this scattering complex is described by:-

dAAGWRW orttr s.)4/1( 22

2

0

Where so is the differential scattering cross section. The integration is carried out over the area contributing to the return at any instant. From the measurements of W r

RESTRICTED

89

RESTRICTED

and other quantities, the radar backscatter coefficient so is estimated which is the quantity used for geophysical parameter retrieval form radar measurements.

Theory of Radar Sea Return

77. The nature of radar sea return at microwave frequencies from the sea surfaces was studied in detail before 1951. In the 1970's, scatter theories were developed for studying wind and wave conditions, and the mechanisms of sea scatter. A two-scale surface roughness model was developed for studying these. Later, this was modified to give more accurate results. The success of such a theory to explain so dependence on wind speed, wind direction, and frequency depends almost exclusively upon the adequacy of the sea spectrum of their vertical displacements form the mean. Often the response at the larger angles of incidence appears to rise primarily from resonance in this spectrum, even though the energy in the resonant-frequency range is quite small compared with the total energy. This mechanism has been used to describe scattering from the ocean in terms of resonance with the tiny capillary and short gravity waves, even when these waves have a height measured in tenths of millimeter whereas they ride on waves that are meters high. This phenomenon is described as Bragg resonance. By the well-known interference equation, the dominant capillary wavelength that is sensed equals the E.M. wavelength divided by 2sinq (typically a few cm).

78. In view of the nature of the reported sea spectrum, significant growth of the sea spectrum with wind speed is found only in the capillary region. Thus for so to be sensitive to local wind, incidence angles should be chosen away from nadir (preferably larger than 30o) and the exploring frequency should be high enough to sense the capillary waves but not too high to experience excessive atmospheric attenuation.

80. The derivation of the backscattering coefficient for the sea surface involves two major steps. One is computing the backscattering coefficient for the capillary waves using standard perturbation techniques and other is accounting for the tilting effect due to the large scale waves. The expressions for the polarised scattering

coefficients are given by:- qqs ,8,

24 Wk pppp

where for horizontal polarisation p=h and for vertical polarisation p=v; k is the electromagnetic wave number; q is the incidence angle; is the aspect angle relative to the upwind direction; and w(q,) is the roughness spectrum of the sea surface. The coefficients pp for different polarisation states are defined as

hh = Rhcosqvv = Rvcos2q + (k'2-k2)tv

2sin2q/(2k'2)

Where k' is the wave number in sea water; Rh and Rv are the Fresnel reflection coefficients for horizontal and vertical polarisation, respectively and Tv = 1+Rv.

81. To account for the tilting effect of the large scale waves, it is necessary to identify polarisation changes due to tilting and then to average the resulting expression for the scattering coefficient of the capillary waves over the slope distribution of the large scale waves. Hence when averaging is included, the scattering coefficient for polarised scattering is approximately

RESTRICTED

90

RESTRICTED

yxyxppppo dZdZZZP ),(),(),'(

'''

qqsqs

where spp (q,) is given by (3); q' is the local angle of incidence; p' is the local angle of incidence; Pq(Zx',Zy') is the slope distribution of the large scale waves as viewed at an incidence angle q and is defined in the prime coordinate whose x' -axis is parallel to the wind direction. It is assumed that the plane of incidence is the xz-plane and that the angle between the x-axis and the x'-axis is , so that an upwind observation occurs when =0. The relations between the primed and unprimed co-ordinates

Zx' = Zx cos + Zysin andZy' = Zy cos + Zysin

82. The cross polarised scattering coefficient including polarisation and averaging effects for small tilts is approximately.

yxyxhvhvhvo dZdZZZP ''1

cot

,.,',', q

q

qsqsqs

where .,'sin'/8,

24''1 qqs WZKZZ hhvvyyxhv

The relation between the slope density function Pq(Zx’,Zy’) and the function P(Zx’,Zy’) defined by Cox and Munk (1954) is :- Pq (Zx , Zy) = (1 + Zx tanq) = P(Zx , Zy)

The slope variances given by Cox and Munk (1954) are not restricted to large scale waves. One possible approach to compute these slope variances is to integrate the low frequency portion of the slope of the spectrum of the sea surface wave.The high frequency portion of the sea spectrum, obtained by oceanographers is a semi-empirial result. The general form of the directional sea spectrum is S(K,q) = S(K)(a0 + a1cos2) When higher order terms in are neglected. The quantity determined from measurements is S(K). this sea spectrum shows that the capillary waves grow with the wind and that higher the k value faster is the growth., The parameters a 0 and a1

have been derived by Chan and Fung (1977) and the relation between W(q’, ) and S(K,) is W(q’, ) = [S(K)/2pK]{1+2(1-v)/(1+v)cos2}, Where K = 2Ksinq’ , V = s2

ct /s2

ut, s2ct and s2

ut are the slope variances of the sea surface along the crosswind and the upwind directions. W (q’, ) is restricted to that portion of the sea spectrum which satisfies the assumptions in the perturbation theory. The value of k has a lower bound which is a function of electromagnetic wave length. For angles of incidence above about 20, the predominant mechanism is Bragg scattering. The Bragg resonant wave length is shorter at larger angles of incidence. Hence, the signals from larger angles of incidence correspond to the values of S(K) for higher wave numbers K.

Scatterometer and Model Functions

83. The scatterometer employs three or more single of dual polarization antennae with extreme outer pair separated by 90 azimuthally, with the middle antenna separated by Km wide and the collective data from the multiple beams is available in an area of about 50 Km X 50 Km, with a 25 km-grid, independent backscatter

RESTRICTED

91

RESTRICTED

measurements relating to cell center nodes on a 25 Km-grid, which are obtained with a time delay of 1 to 3 minutes. Calculation of the surface wind vector in terms of speed and direction takes place using these so called ‘triplets’ or ‘quadruplets’ of radar backscatter, wind speed, wind direction and incidence angle of the observation. The specified accuracies are 2 m/s or 10% (whichever is less) for wind speed in the range of 4 to 24 m/s and 20 for direction in 0 to 360 range.

84. The scatterometer measures the radar energy back to scattered by the illuminated sea surface area (Normalised Radar Cross Section, s). The intensity of radar back scatter depends upon the roughness of the sea surface, generated by the action of capillary waves is directly influenced by wind intensity. The basic mechanism at work is the Bragg Resonance, under which the waves satisfying Bragg’s condition produce strong radar back scatter energy. The back scatter signal also depends upon the incidence angle of radar beam. The azimuthal symmetry of sea surface roughness causes azimuthal harmonicity in the radar signal. This causes multiple solutions of wind vectors (especially direction) in which one is true and remaining are termed as the directional ambiguities. Multi-beam scatterometer is used to overcome this problem of ambiguities to a certain extent. The radar back scatter shows a harmonic nature as expressed by (Ulaby et al, 1982).

so = Ao + A1 cos () + A2cos (2) , where, is the angle between radar azimuth and wind direction and the coefficients Ao, A1 and A2 are expressed as

Ao = (su + 2s + sd)/4A1 = (su-sd)/2A2 = (su-2sc+sd)/4 ,

Where, su, sc and sd are specific radar backscatter at upwind, crosswind and downwind directions, respectively. These radar backscatter is related to wind speed as expressed s = a.Wb, where the coefficients a and b are functions of radar wavelength, polarisation p and incidence angle q. As, so far, the observed dependence of radar backscatter on wind conditions has not been well explained theoretically, the empirical relationship are established through simultaneous radar backscatter and insitu measurements at different radar wavelengths. These empirical relationships between radar backscatter and wind vector are known as Model Functions. The model functions available for Ku- and C- bands are known as SASS-I and CMOD4. The functional forms of these model functions are given as follows.

SASS-I Model Functions

85. so(dB) = 10G(q,,p) + 10H(q,,p). log(W) where G and H are model functions given in tabular form which are functions of incidence angle q and relative azimuth angle made with upwind direction with a speed of W in meter per second at 19.5 neutral stability height.CMOD4 Model Functions

86. so(linear) = b0(1+ b1cos() + b3tanh(b2)cos(2))1.6

where b0, b1, b2 and b3 are polynomials of wind speed (10m height) and incidence angle while is relative wind direction.

RESTRICTED

92

RESTRICTED

Retrieval Algorithms

87. The radar backscatter has a harmonic dependence on wind direction which results into multiple solutions of wind vector from a given radar backscatter data. As the radar backscatter depends upon both wind speed and direction, two measurements of radar solutions are not possible due to the harmonicity and the noise in the data. Among these multiple wind vector solutions, only one solution corresponds to the wind vector while others are ambiguities which are called aliases. Although, the wind speed solutions are very close to each other, the direction solutions are quite apart. The solutions are prioritised by an algorithm with certain criterion and the performance of the algorithm is evaluated in terms of percentage of highest priority solutions correctly identifying the true directions out of the total number of data cases processed.The retrieval algorithms yield multiple solutions of wind vector along with their priority in identifying the true wind vector from a given set of radar backscatter measurements. Under noise-free conditions, the highest priority solutions always represents the true wind vector but for noisy or real data, it is always not so. However, the first two highest priority solutions contain the true direction most of the times. The selection of true wind solution is performed by the directional ambiguity removal algorithm.

Wind Vector Solution Retrieval

88. The retrieval algorithm used in the project (Gohil and Pandey, 1994b) is based on the algorithm developed earlier (Gohil and Pandey, 1985) for Seasat scatterometer. The prioritization of wind vector solutions under this algorithm is based n the assumption that the radar backscatter of an area measured in different directions(radar azimuth angles) are associated with the same wind speed conditions. This implies that the wind speeds retrieved from all the radar backscatter data at the given location should be same for an arbitrary wind direction if it represents the true direction of wind. On the other hand, the standard deviation of retrieved wind speeds should be zero or minimum for the true direction of wind.

89. The surface winds are obtained from the so values by making use of empirical relationships relating wind and so. two different algorithms are used for derived unambiguous wind vector. One is retrieval algorithm which provides prioritised wind vector solution. The other algorithm filters out the ambiguities over an area. The wind speeds for an arbitrary wind directions are determined by comparing the measured so data with the simulated so value are equal, is treated as retrieved wind speed at that aspect angle.

wj(x)= wi (when scal (qj, wi, Xj)=sm(qj)), where wj(x)= retrieved wind speed at an aspect angle X for scatterometer antenna sequence j,wi = arbitrary wind speed at ith sequence in the entire range,scal = calculated so, sm = measured so,X = arbitrary wind direction (aspect angle) relative to the antenna azimuth. In the present method, the normalised standard deviation of radar backscatter coefficient (Kp (%)), supplied along with scatterometer data is utilised to obtain the weighted averaged and weighted standard deviation of retrieved seeds as follows.

RESTRICTED

93

RESTRICTED

N

ii

N

iiix wWmwWa

11

/

Wsx = ((Wmi-Wax)2)/N withwi = 1.0/ï Wm+i – Wm-i ï , wherewi = weight for wind speed Wmi for ith so measurements.N = number of so measurements used.Wm+i = obtained from upper limit of so

i data using Kp values.Wm-i = obtained from lower limit of so

i data using Kp values.

90. The average and standard deviation of wind speeds are determined from the entire range of aspect angles. Then the minima of standard deviations of wind speeds are searched along the aspect angles. The aspect angles and the averages of wind speeds corresponding to these minima represent the solutions of direction and speed respectively. The noise weighted standard deviations of wind speeds corresponding to these solutions are used as a measure for prioritising the vector solutions. The solution having minimum standard deviation of wind speed is treated as the highest priority solution and the next minimum as second solution and so on. These prioritised wind vector solutions are required to be corrected for directional ambiguity because the highest priority direction solutions do not always correspond to the true wind directions. However, the first two highest priority solutions contain the true wind vector most of the times.

Fig. 1 shows an example of prioritised wind vector solution retrieval from noise-free simulated so data. For a given wind speed of 10m/s and direction of 45 , the so was generated using the CMOD4 model function.

91. Shown in the figure is the calculated wind speeds, their mean and standard deviation for the arbitrary wind directions ranging from 0 to 360 using the simulated data for the three antennae, forward, Middle and Aft of ERS-1 scatterometer. The incidence angles for Forward and Aft beam are 28.7 and 19.5 for Middle antenna. There are two minima in the standard deviation curve which correspond to the two solutions of wind vector the mean speeds and aspect directions corresponding the minima of standard deviation are the two most probable wind vector solutions, the minimum of the minima corresponds to the true solution. The next minimum value occurs with a difference of about 180 from the true wind direction. In this way, the prioritised wind vector solutions are derived.

RESTRICTED

94

RESTRICTED

Wind Directional Ambiguity Removal

92. The ambiguity removal algorithm utilises the two highest priority vector solutions available from retrieval algorithm for the data locations within the scatterometer data swath and the ambiguity is removed based on the trend or the consensus of the highest priority wind directions solutions over that area. In case of ERS-1 scatterometer antenna configuration, about over half of the highest priority wind directions are in the true wind directional pattern under normal noise conditions while the remaining solutions are in the opposite direction. These true wind directions are randomly located over the scatterometer data swath. A median filtering approach is used to remove these ambiguities. In this approach, a trend or a consensus of highest priority directions is derived and the two highest priority direction are compared with the median or consensus direction. This comparison is performed over a small area called median window and the entire data swath is analysed. For this purpose, the highest priority vectors are first resolved into zonal and meridional components. With a moving window of opimized size (Gohil and Pandey, 1994a), medians of both the scalars at each data location are determined from all the points lying within the window, the median scalars yield median wind direction at each data location which is used to select the direction out of these two highest priority direction which is closer to the median direction. The procedure is followed for entire satellite pass to select the solutions nearer to the median direction. The processing of the solution data of entire pass is repeated in the above mentioned manner till the entire processed wind field is covered. At this state, the resulting wind field mostly resembles the true wind field. In the algorithm (Gohil, 1992), the direction closer to the median direction is retained but not replaced by median direction unlike the earlier algorithm (Schultz, 1985). In case when the scatterometer data is highly noisy over certain parts of the swath, the resulting wind field still may have localised ambiguities which may not always be possible to be removed by the filtering process alone and highly localised ambiguities may still be removed using a supplementary method (Gohil and Pandey, 1994c).

WINDS FROM ERS-1 SCATTEROMETER

93. Summary of ERS-1 satellite, scatterometer system, data analysis, comparison with insitu data and a few results related to retrieval of winds have been mentioned below.

(a) ERS-1 Satellite . ERS-1 is the European Space Agency’s first satellite, devoted entirely to remote sensing from polar orbits. ERS-1 provides global and repetitive observations of the environment using advanced microwave techniques, which enable measurements to irrespective of cloud or sunlight conditions. ERS-1 satellite has a sun-synchronous, near polar quasi-circular orbit with a mean altitude of 785 km and an inclination of 98.5. one of the payloads of ERS-1 satellite is the Active Microwave Instrument (AMI).

(b) ERS-1 Active Microwave Instrument . The AMI comprises two separate radars, one is Synthetic Aperture Radar (SAR) and the other one is Wind Scatterometer both operating at C-band (5.3 Ghz). The SAR operates in two modes. (a) Wave mode: This provides images of 5X5 Km area Separated by several hundred kilometers useful for wave spectrum analysis. The images are transformed into directional spectra providing information about

RESTRICTED

95

RESTRICTED

wavelength and direction of wave systems. (b) Image Mode: It provides high resolution two-dimensional images with a spatial resolution of 26m in range (across track) and between 6m and 30m in azimuth (along track) direction. The imagery obtained is approximately 100 Km in width, 250 Km to the right of sub satellite track. Imagery is built up from the time delay and strength of the return signals, which depend primarily on the roughness and dielectric properties of the surface and its range from the satellite.

(c) ERS-1 Scatterometer. The ERS-1 scatterometer which operates at C-band (5.3 Ghz) provides radar backscatter from ocean surface measured in three different directions by its three vertically polarized antennae azimuthally separated from each other by 45 with the middle antenna pointing normal to the satellite track. As the satellite orbits the earth, each area (measurement cell) is observed in three different directions by three (fore, mid and aft) beams of scatterometer antennae within time intervals of 1 to 3 minutes for near to far ends of the data swath. Doppler filters are employed to get the radar backscatter measurements incidence angles corresponding to the different areas (cells) of the elongated areas on the earth illuminated by these beams. The incidence angles from the fore/aft and the mid beams across the data swath are in the range of 27 to 58 and 18 to 45, respectively. Though, the illumination swath is 500 kms, but the colocated radar backscatter triplets for different Doppler cells of about 50 Km X 50 Km are available for the swath of about 400 Kms. The data grids are separated by 25 Km in the directions across and along the sub-satellite track. The data of incidence and azimuthal angles of radar beams at each grid are also provided along with so triplets required for the retrieval of the wind vectors at each grid.

94. Results of ERS-1 Data Analysis. Ocean surface winds over Indian oceanic region have been extracted from ERS-1 scatterometer data using the above mentioned algorithms. ERS-1 scatterometer data for the period from August 1992 to July 1993 have been analysed to prepare an atlas of monthly and fornightly averaged wind fields (Gohil and Pandey, 1995).

) Fig .2 and 3 depict the mean monthly wind fields during July and January 1993, respectively, which correspond to the southwest and northeast monsoons periods over Indian region. The July 1993 map clearly depicts the strong southwesterly winds near Somalia coast. Also seen in the map is the branching of wind flow near Somalia into Bay of Bengal and Arabian Sea flow near Sri Lanka coast. This feature persists for subsequent two o three months but the intensity weakens during May and wind intensity picks up further during June and south westerly sets in. the Scatterometer data could be very useful in the study of monsoonal winds.

RESTRICTED

96

RESTRICTED

Comparison with Insitu Data

95. The comparison of derived wind velocity has been made with ship wind data collected by NIO under DOD MARSIS Phase-I Projects. The colocated ship and satellite observation within ±6 hrs time interval and within a range of 0.5 degree for about 8 months during August 1992 to July 1993 have been utilized for comparison. Since the satellite data density is high the satellite data nearest to the ship observation has been considered. There are 47 near simultaneous and closely colocated observations available which have been used for comparing the wind speed and direction. It is, however, desirable to have more such data. Comparison of wind speeds for the above mentioned data set shows a good agreement of derived wind speed with ship observed wind speed within the ERS-1 scatterometer instrument error specification of 2 m/s or 10% of actual wind whichever is greater. It has also been observed that the satellite derived wind speeds are slightly overestimated for wind speeds below 5 m/s while these are slightly underestimated for wind speeds above 5m/s. comparison of derived wind directions with ship observed wind directions has also been performed with available limited colocated data points. The rms deviation of derived direction has been estimated using the absolute differences from the ship wind directions (within ± 180). An rms deviation has been found little more as compared to 20 specification which could be due to errors interval and ship measurements.

Current Scatterometer Instruments

96. The advanced scatterometer (ASCAT) on the European Space Agency's Met Operational (MetOp-A and MetOp-B) platforms are the follow-on to European wind scatterometers. The ISS-RapidScat instrument is a speedy and cost-effective replacement for NASA's QuikScat Earth satellite, which monitored ocean winds to provide essential measurements used in weather predictions, including hurricane monitoring. So essential were QuikScat's measurements that when the satellite stopped collecting wind data in late 2009, NASA was challenged to quickly and cost-effectively conceive of a replacement. The resulting ISS-RapidScat instrument is aboard the International Space Station and will measure Earth's ocean surface wind speed and direction. Rapidscat was launched on 19 Oct 2006. WindSat is a satellite-based polarimetric microwave radiometer developed by the Naval Research Laboratory Remote Sensing Division, the Naval Center for Space Technology, and the National Polar-orbiting Operational Environmental Satellite System (NPOESS) Integrated Program Office (IPO). It was launched in January 2003 aboard the joint DoD/Navy platform Coriolis, with a planned 3-year life. Despite its extended lifespan, it continues to function quite well. WindSat measures the ocean surface wind vector, as well as cloud liquid water, sea surface temperature, total precipitable water, and rain rate (over water only). Derived products include soil moisture and sea ice. Oscat onboard Oceansat 2 developed an anomaly early in Feb 14 which turned out to be not recoverable. After 4.5 years of 5 years design life ISRO formally stopped Oscat operations on 02 Apr 14. The two Ascat instruments on board Metop-A and Metop-B partially mitigate the loss of Oscat data.

Where to Get Scatterometer Data

97. The scatterometer data can be obtained from internet through the following links :-

RESTRICTED

97

RESTRICTED

(a) NRL Monterey http://kauai.nrlmry.navy.mil:80/sat-bin/tc_home

(b) NOAA/NESDIS QuikSCAT http://manati.wwb.noaa.gov/quikscat

(c) http://manati.wwb.noaa.gov/cgi-bin/qscat_storm.pl

(d) Alternative NOAA site, with SSMI wind speeds:

http://polar.wwb.noaa.gov/winds/globdata.html

(e) FNMOC http://www.fnmoc.navy.mil/PUBLIC

(f) http://www.ssmi.com

98. The scatterometer winds are good for detection of circulations and determination of wind speeds over the oceanic regions. Small system could be followed for 3 days. Useful for high seas forecasts over the Southern and Indian Oceans where observations are few and far between. It can be useful in determining the surface location of fronts when IR imagery disguises it. Although there are many problem areas, there is useful information to be gained from most passes. On any individual pass, the problem areas must be identified and worked around. The Wind speeds generally good – useful for areas of gales etc. Use the data if it makes sense. Be aware of low skill areas and different ambiguity removal processes (compare!). Do not use the scatterometer winds in isolation.

VISIBLE INFRARED TECNIQUES FOR ESTIMATION OF PRECIPITATION

100. Today satellites make an important role in the global rainfall detection and estimation. The success of SSM/I type of space borne sensors for global rainfall estimation is widely acknowledged. A further boost to the understanding of the various tropical rain processes, and development of improved rainfall estimation techniques, is expected in the near future with the availability of TRMM (Tropical Rainfall Monitoring Mission) satellite which is already launched later last year.

101. Though microwave measurements from space provide important and more direct measurement of rainfall, they will briefly discussed in this lecture. This lecture deals with mainly visible and infrared (IR) based rainfall estimation techniques.

102. Precipitation estimation techniques may be divided into three categories:-(a) Those that use Visible or IR data.

(b) Those that use passive Microwave data.

(c) Those techniques that will use Radar.

Visible and IR Techniques

103. Visible and IR techniques are grouped together because they share a common characteristic. The radiation does not penetrate through the cloud. Visible and IR techniques estimate precipitation falling from the bottom of the cloud base on the radiation coming from the top and/or the side of the cloud, depending on viewing

RESTRICTED

98

RESTRICTED

geometry. All Visible and IR precipitation-estimation schemes are necessarily indirect; a cloud's brightness or equivalent blackbody temperature may be related to the rain falling from it, but the raindrops themselves are not directly sensed.

104. The Visible and IR techniques are divided into four categories:-

(a) Cloud Indexing Technique.

(b) Life History Technique.

(c) Bi-Spectral Technique.

(d) Cloud Model Technique.

105. Cloud Indexing Technique.Cloud indexing technique assigns a rain rate to each particular type of cloud, which is identified from the satellite image. The rain at a particular location can be written as:

R = ∑i ri fi

Where ri is rain rate assigned to cloud type i and fi is the fraction of time that point is covered with (or fraction of area covered by) cloud type i.

The various studies which basically used cloud indexing technique with some variations are done in the past. Most of the techniques are known with the name of their inventors.

106. Bi-Spectral Techniques. Clouds that are bright in Visible images are more likely to precipitate than dark clouds because brightness is related to optical depth and thus to cloud thickness. Clouds that are cold in IR images are more likely to precipitate than warm clouds because cold clouds have higher tops than warm clouds. There are exceptions to these rules, however, St clouds are bright, but do not rain as much, nor as often, as Cb clouds. Ci clouds are cold but do not produce as much precipitation as some warmer clouds. Bi-spectral methods attempted to combine these rules by saying that clouds which have the best chance of raining are both cold and bright. Lesser amount of precipitation can be expected from cold-but-dark clouds (Ci) and bright-but-warm clouds (St).

107. Life History Techniques. The rain rate of a cloud, particularly a convective cloud, is a function of the stages in its life cycle. Life history techniques take into account a cloud’s life cycle. Geostationary satellite data, with more than one image, is necessary for the algorithms which take into consideration the evolution raining system with time.

108. Cloud Model Techniques. To improve precipitation estimation techniques based on Visible and IR satellite data, it is believed that it is necessary to build the physics of the cloud into retrieval process. One way to this is through the use of cloud models. Several investigations have attempted to use cloud models to relate satellite observations to precipitation. Paramerisation of convection is used to relate fractional cloud cover to rain rate. For example, consider a grid box into which moisture is flowing at the rate I. If Q is the amount of moisture necessary to completely saturate the box, then the rate at which the box is filling with cloud could

RESTRICTED

99

RESTRICTED

be estimated as I/Q. However, convective clouds do not live forever. Suppose that all the cloud which form are convective and have identical lifetime t l. Then the rate at which clouds are dying is approximately c/tl,. where c is the fractional cloud cover. If one makes the further assumption that the rate of influx of moisture I is exactly the same as the rain rate R, then one arrives at the equilibrium relationship R = Qc/t l. Q is calculated from sounding, c from IR satellite data, and R from radar data.

109. About Convective Clouds. Research into precipitation estimation techniques has revealed some basic properties of convective clouds. The most important of these are as follows:-

(a) There is a high correlation between a cloud’s area (as measured in Visible or IR wavelengths) and the total volume of rain per unit time falling from it.

(b) There is little correlation between the visible brightness or IR temperature of a point and the rain rate beneath that point.

 110. Advantages of Rainfall Estimation with Vis/IR Techniques. The satellites provide:-

(a) Uniform and continuous coverage over land and oceans.

(b) Large spatial coverage(c) Uniform resolution.

(d) Continuous observation capability over evolving storm.

(e) Near-real time digital data availability.

(f) Observations unaffected by severe weather

111. Disadvantages with Visible/IR Measurement of Rainfall

(a) Visible/IR methods are indirect as they infer the rain by observing the top of the cloud. They do not penetrate and observe deep inside the cloud.

(b) The convective systems, smaller than the size of the pixel go often unnoticed. For example, IR band of INSAT has a resolution of 8 km, thus any cloud smaller than this will either not be noticed or will provide underestimated rain.

(c) The relationship between CTT (cloud top temperature) and rainfall is not simple. It is often seen that not all cold clouds precipitate, nor rain always come from cold clouds.

(d) The relationship between cloud brightness and rainfall in visible band satellite images is much more complicated than CCT versus rainfall. This is already explained in my previous lecture.

(e) The calibration is another problem. The comparison of areal averaged satellite measurements with point measurements of rain gauges cannot be

RESTRICTED

100

RESTRICTED

expected to match perfectly. However, with no better data available, they are rain gauges are used for calibration.

INSAT 3D Rainfall Products

112. QPE Using INSAT Multispectral Rainfall Algorithm (IMSRA). IMSRA combines variety of techniques (IR and MW) in a single and comprehensive rainfall algorithm. The main components are:-

(a) Identification of the areas for various cloud types from IR-WV ( 11mm-6.7mm), which corresponds well with rainfall.(b) Calibration of Geostationary – IR TB’s from INSAT-3D using Polar orbiting - Microwave Satellite rainfall estimates from TRMM –TMI & PR or SSM/I rainfall.(c) Rainfall estimation using multiple regression approach using IR-MW data base.(d) Generation of total rainfall products on half hourly, daily, weekly, monthly and seasonal scales.

113. Hydro-Estimator (HE).The Hydro-Estimator offers the solution to limitation of IR based R/F estimation using a prior knowledge of the environment from NWP model fields. Initial rain rates are a function of both Tb and its value relative to surrounding mean. The HE considers the brightness temperature relative to the surrounding pixels using the relationship

Where, µ is the mean temperature of the nearby cloudy pixelsσ is the standard deviation of the temperature of the nearby cloudy pixelsPixels colder than their surroundings (positive Z) are assumed to be convective updrafts and hence producing rainfallPixels as warm as or warmer than their surroundings (negative Z) are presumed to be convectively inactive101x101 pixel window, which is adjusted based on Tbmin

If Z < 0; RR =0 (pixel either cirrus or inactive convective)RR1 = [RRc*Z2 + RRn * (1.5 - Z)2] / [Z2 + (1.5 - Z)2]RRc = a exp(-bTb1.2); by function fit with RRc=0.5 mm/h at 240 K and PW dependent RRc at 210 K RRs= (250-Tb) * (RRmax/5) RRmax is again PW dependent rain at 210 K and RRs < min(RRc, 12 mm/h)

RESTRICTED

101

Z=μ−Tbσ

RESTRICTED

Conclusion

114. In this section of the distance learning notes, the very basic principles involved in the retrieval of temperature, AMV, WVW, sea surface winds and measurement of precipitation are covered. Advanced mathematical and statistical techniques are now being used to compute the meteorological parameters from the satellite data. It is important to know the drawbacks or limitations of the technique before using any of the satellite derived product for operational forecasting.

Check Assimilation

Fill in the blanks

1. The basic principle behind the sounding of the atmosphere depends on the fact that the radiance leaving the top of the atmosphere will be a function of __________________________________________________________.

2. In selecting cloud tracer in automatic tracking, ___________________ are tracked.

3. The mean temperature of the ________________________ is considered for assigning the height of the CMV.

4. The physical basis for scatterometer winds is _______________________of microwave energy from centimeter length capillary waves of the oceans.

5. The cloud indexing technique assigns _____________ to each particular type of cloud.

Check Assimilation: The Key

1. The basic principle behind the sounding of the atmosphere depends on the fact that the radiance leaving the top of the atmosphere will be a function of the emitting gas and the distribution of temperature throughout the atmosphere.

2. In selecting cloud tracer in automatic tracking, a pattern of clouds are tracked.

3. Mean temperature of the 25% coldest IR pixels is considered for assigning the height of the CMV.

4. The physical basis for scatterometer winds is the Bragg scattering of microwave energy from centimeter length capillary waves of the oceans.

5. The cloud indexing technique assigns a rain rate to each particular type of cloud.

RESTRICTED

102

RESTRICTED

CHAPTER- 4SPACE METEOROLOGY

Chapter Objectives

After reading this chapter, you should be able to:-

Understand the structure of Sun.

Understand the physics behind sunspots.

Understand about solar flares, hazards associated with it and methods of prediction

Understand about solar winds, coronal mass ejections, coronal loop, and coronal holes.

Appreciate solar energy variations and climatic change.

Understand effects of space weather on aviationStructure

1. Introduction

2. Sunspots

3. Solar flares

4. Solar winds

5. Coronal mass ejections

6. Coronal loop

7. Coronal holes

8. Climate change and space weather

9. Effects of space weather on aviation

INTRODUCTION

1. The term "space weather" refers to the time-variable conditions in the space environment that may affect space borne or ground-based technological systems and, in the worst case, endanger human health or life. The most important social and economic aspects of space weather are related to being aware of and possibly avoiding the consequences of space weather events either by system design or by efficient warning and prediction systems allowing for preventive measures to be taken. On the climatological time scales there are also potential effects on terrestrial climate.

2. During the last few years space weather activities have expanded with an increasing pace world-wide and it has become commonly accepted that improved

RESTRICTED

103

RESTRICTED

space weather services are important and expected to become much more useful in the near future. At the same time the scientific community has strengthened their efforts toward better understanding of the physical foundations of space weather and many scientists have remarked that without strong emphasis on improved scientific understanding the promises for and dreams of improved space weather products may not be fulfilled. This way the scientific Solar-Terrestrial Physics (STP) community and the users of space-sensitive systems are intimately coupled to each other.

3. Space weather has many similarities with atmospheric weather; thus the design of future space weather activities may be able to utilise the experience from meteorological services, at least concerning real-time operations, input data management, and nowcast and forecast distribution. The only operational space weather centres, the Space Environment Center (SEC) of NOAA, and the 55th Space Weather Squadron of the US Air Force, both in Colorado, USA, already operate in this fashion, although in much smaller scale than typical meteorological services.

4. However, there are important differences between the atmospheric and space weather systems:

(a) While many meteorological processes are localised and it is possible to make good limited area weather forecasts, space weather is always global in the planetary scale. Perturbations originating from the Sun disturb the Earth's plasma environment, the magnetosphere, which responds to these disturbances globally.

(b) Space weather events occur over a wide range of time scales: the entire magnetosphere responds to the solar-originated disturbances within only a few minutes, global reconfiguration takes a few tens of minutes, and sometimes extreme conditions may remain for much longer periods. Ground-based magnetometers react immediately when an interplanetary shock hits the magnetosphere whereas enhanced fluxes of energetic particles in radiation belts decay in time scales of days, months, or even years.

(c) Our means to monitor space weather are much more limited than our ability to install weather stations on the Earth's surface: Our prediction schemes must be capable of functioning with input from only a few isolated measurements of the upstream solarwind conditions and magnetospheric parameters. While the present (e.g., magnetic activity indices, interplanetary scintillations) and future (e.g., energetic neutral atom imagery, ionospheric tomography) observations have a global character, they still remain rather far from the present-day rather detailed and continuous coverage of the atmospheric weather. As a consequence, successful space weather activities need to be performed on a global scale, merging a great variety of space-borne and ground-based observational capabilities.

(d) All these aspects put significant requirements and constraints to present and future space weather service systems both in space and on the ground. While the monitoring system cannot be as complete as one would like, it must be extensive enough for reasonable specification of the space environment. This introduces the crucial question of costs versus benefits. On the other hand, data collection and

RESTRICTED

104

RESTRICTED

assimilation, and the service product processes have to be very efficient and fast which calls for well-designed and maintained space and ground segments. Further analyses of all these questions are topics of other parts of the present ESWS study and reported in separate documents.

5. Space weather originates from the Sun. It generally refers to all solar activities such as sunspots and solar flares, and the effects they may have on the Earth. The intensity varies from time to time, sometimes strong and sometimes weak. "Good weather" means a calm period of solar activities, while "bad weather" is a period of frequent and disturbed activities which may affect telecommunications, navigation and power systems on Earth and the operations of satellites or spacecraft. Solar activity is periodic. A solar cycle (also called a sunspot cycle) is an approximately 11year period with increasing and decreasing sunspot numbers. Each cycle starts from the time of minimum activity. The cycle numbering system dates back to the eighteenth century and the current solar cycle is cycle 23.

SUNSPOTS6. Sunspots are temporary features on the surface of the Sun (the photosphere) that appear visibly as dark spots compared to surrounding regions. They are caused by intense magnetic activity, which inhibits convection, forming areas of reduced surface temperature. Although they are at temperatures of roughly 5,000–6,500 K, the contrast with the surrounding material at about 5,800 K leaves them clearly visible as dark spots, as the intensity of a heated black body (closely approximated by the photosphere) is a function of T (temperature) to the fourth power. If a sunspot were isolated from the surrounding photosphere it would be brighter than an electric arc. Sunspots expand and contract as they move across the surface of the sun and can be as large as 80,000 km (50,000 miles) in diameter, making the larger ones visible from Earth without the aid of a telescope.

7. Sunspots, being the manifestation of intense magnetic activity, host secondary phenomena such as coronal loops and reconnection events. Most solar flares and coronal mass ejections originate in magnetically active regions around visible sunspot groupings. Similar phenomena indirectly observed on stars are commonly called starspots and both light and dark spots have been measured.

RESTRICTED

105

RESTRICTED

8. According to George Fischer, a solar astronomer at the University of California, "A sunspot is a dark part of the sun's surface that is cooler than the surrounding area. It turns out it is cooler because of a strong magnetic field there that inhibits the transport of heat via convective motion in the sun. The magnetic field is formed below the sun's surface, and extends out into the sun's corona."

Sunspot Variation9. Sunspot populations quickly rise and more slowly fall on an irregular cycle about every 11 years. Significant variations of the 11-year period are known over longer spans of time. For example, from 1900 to the 1960s the solar maxima trend of sunspot count has been upward; from the 1960s to the present, it has diminished somewhat. Over the last decades the Sun has had a markedly high average level of sunspot activity; it was last similarly active over 8,000 years ago.

10. The number of sunspots correlates with the intensity of solar radiation over the period since 1979, when satellite measurements of absolute radiative flux became available. Since sunspots are darker than the surrounding photosphere it might be expected that more sunspots would lead to less solar radiation and a decreased solar constant. However, the surrounding margins of sunspots are brighter than the average, and so are hotter; overall, more sunspots increase the sun's solar constant or brightness. The variation caused by the sunspot cycle to solar output is relatively small, on the order of 0.1% of the solar constant (a peak-to-trough range of 1.3 W m−2 compared to 1366 W m−2 for the average solar constant). Sunspots were rarely observed during the Maunder Minimum in the second part of the 17th century. This coincides with the middle (and coldest) part of a period of cooling known as the Little Ice Age. There may be a resonant gravitational link between a photospheric tidal force exerted by the planets and the sunspot cycle.

Physics Behind Sunspots

11. Although the details of sunspot generation are still a matter of research, it appears that sunspots are the visible counterparts of magnetic flux tubes in the sun's convective zone that get "wound up" by differential rotation. If the stress on the tubes reaches a certain limit, they curl up like a rubber band and puncture the sun's surface. Convection is inhibited at the puncture points; the energy flux from the sun's interior decreases; and with it surface temperature. The Wilson effect tells us that sunspots are actually depressions on the sun's surface. Observations using the Zeeman effect show that prototypical sunspots come in pairs with opposite magnetic polarity. From cycle to cycle, the polarities of leading and trailing (with respect to the solar rotation) sunspots change from north/south to south/north and back. Sunspots usually appear in groups. The sunspot itself can be divided into two parts:

(a) The central umbra, which is the darkest part, where the magnetic field is approximately vertical (normal to the sun's surface).

(b) The surrounding penumbra, which is lighter, where the magnetic field lines are more inclined.

12. Magnetic field lines would ordinarily repel each other, causing sunspots to disperse rapidly, but sunspot lifetime is about two weeks. Recent observations from the Solar and Heliospheric Observatory (SOHO) using sound waves traveling through the Sun's photosphere to develop a detailed image of the internal structure

RESTRICTED

106

RESTRICTED

below sunspots show that there is a powerful downdraft underneath each sunspot, forming a rotating vortex that concentrates magnetic field lines. Sunspots are self-perpetuating storms, similar in some ways to terrestrial hurricanes.

Butterfly diagram showing paired Sporer’s law behavior

13. Sunspot activity cycles about every eleven years. The point of highest sunspot activity during this cycle is known as Solar Maximum, and the point of lowest activity is Solar Minimum. Early in the cycle, sunspots appear in the higher latitudes and then move towards the equator as the cycle approaches maximum: this is called Spörer's law. Wolf number sunspot index displays various periods, the most prominent of which is at about 11 years in the mean. This period is also observed in most other expressions of solar activity and is deeply linked to a variation in the solar magnetic field that changes polarity with this period, too.

14. The modern understanding of sunspots starts with George Ellery Hale, which links magnetic fields and sunspots. Hale suggested that the sunspot cycle period is 22 years, covering two polar reversals of the solar magnetic dipole field. Horace W. Babcock later proposed a qualitative model for the dynamics of the solar outer layers. The Babcock Model explains that magnetic fields cause the behavior described by Spörer's law, as well as other effects, which are twisted by the Sun's rotation.

Sunspot Observation

15. Sunspots are observed with land-based and Earth-orbiting solar telescopes. These telescopes use filtration and projection techniques for direct observation, in addition to various types of filtered cameras. Specialized tools such as spectroscopes and spectrohelioscopes are used to examine sunspots and sunspot

RESTRICTED

107

RESTRICTED

areas. Artificial eclipses allow viewing of the circumference of the sun as sunspots rotate through the horizon.

Since looking directly at the Sun with the naked eye permanently damages vision, amateur observation of sunspots is generally conducted indirectly using projected images, or directly through protective filters. Small sections of very dark filter glass, such as a #14 welder's glass are effective. A telescope eyepiece can project the image, without filtration, onto a white screen where it can be viewed indirectly, and even traced, to follow sunspot evolution. Special purpose hydrogen-alpha narrow bandpass filters as well as aluminum coated glass attenuation filters (which have the appearance of mirrors due to their extremely high optical density) on the front of a telescope provide safe observation through the eyepiece.

16. Amateur observers should not use filters, no matter how dark, unless they are specifically intended for solar viewing. Other filters may not provide appropriate protection in the non-visible frequency range and eye damage could result. When using a telescope or binoculars for direct viewing with a filter, follow the manufacturers' guidelines closely. These will usually require the filter to be placed at the objective (far) end of the instrument, since the instrument concentrates heat as well as light and may damage an eyepiece filter and cause immediate eye damage. Securely fasten the filter and ensure that attached auxiliary devices such as spotter scopes have hoods in case they concentrate heat dangerously. Direct observation of the sun's surface through an optical device is inherently dangerous, and requires both knowledge and close attention to safety precautions.

Application

17. A large group of sunspots were seen in year 2004. The grey area around the spots can be seen very clearly, as well as the granulation of the sun's surface. Due to their link to other kinds of solar activity, sunspots can predict space weather and the state of the ionosphere. Thus, sunspots can help predict conditions of short-wave radio propagation or satellite communications. Don Easterbrook, a Professor Emeritus of geology at Western Washington University, claims that there is a cause-and-effect relationship between sunspot activity and measured changes in global temperatures on Earth.

Sunspot Numbers18. In 1610, shortly after viewing the sun with his new telescope, Galileo Galilei (or was it Thomas Harriot?) made the first European observations of Sunspots. Continuous daily observations were started at the Zurich Observatory in 1849 and earlier observations have been used to extend the records back to 1610. The sunspot number is calculated by first counting the number of sunspot groups and then the number of individual sunspots.

RESTRICTED

108

RESTRICTED

19. The "sunspot number" is then given by the sum of the number of individual sunspots and ten times the number of groups. Since most sunspot groups have, on average, about ten spots, this formula for counting sunspots gives reliable numbers even when the observing conditions are less than ideal and small spots are hard to see. Monthly averages (updated monthly) of the sunspot numbers (181 kb JPEG image), (307 kb pdf-file), (62 kb text file) show that the number of sunspots visible on the sun waxes and wanes with an approximate 11-year cycle.

SOLAR FLARES

20. A solar flare is a large explosion in the Sun's atmosphere that can release as much as 6 × 1025 joules of energy. The term is also used to refer to similar phenomena in other stars, where the term stellar flare applies.

21. Solar flares affect all layers of the solar atmosphere (photosphere, corona, and chromosphere), heating plasma to tens of millions of kelvins and accelerating electrons, protons, and heavier ions to near the speed of light. They produce radiation across the electromagnetic spectrum at all wavelengths, from radio waves to gamma rays. Most flares occur in active regions around sunspots, where intense magnetic fields penetrate the photosphere to link the corona to the solar interior. Flares are powered by the sudden (timescales of minutes to tens of minutes) release of magnetic energy stored in the corona. If a solar flare is exceptionally powerful, it can cause coronal mass ejections. X-rays and UV radiation emitted by solar flares can affect Earth's ionosphere and disrupt long-range radio communications. Direct radio emission at decimetric wavelengths may disturb operation of radars and other devices operating at these frequencies. Solar flares were first observed on the Sun by Richard Christopher Carrington and independently by Richard Hodgson in 1859 as localized visible brightenings of small areas within a sunspot group. Stellar flares have also been observed on a variety of other stars.

22. The frequency of occurrence of solar flares varies, from several per day when the Sun is particularly "active" to less than one each week when the Sun is "quiet". Large flares are less frequent than smaller ones. Solar activity varies with an 11-year cycle (the solar cycle). At the peak of the cycle there are typically more sunspots on the Sun, and hence more solar flares.

Classification23. Solar flares are classified as A, B, C, or X according to the peak flux (in watts per square meter, W/m2) of 100 to 800 picometer X-rays near Earth, as measured on

RESTRICTED

109

RESTRICTED

the GOES spacecraft. Each class has a peak flux ten times greater than the preceding one, with X class flares having a peak flux of order 10−4 W/m2. Within a class there is a linear scale from 1 to 9, so an X2 flare is twice as powerful as an X1 flare, and is four times more powerful than an M5 flare. The more powerful M and X class flares are often associated with a variety of effects on the near-Earth space environment. Although the GOES classification is commonly used to indicate the size of a flare, it is only one measure. This extended logarithmic classification is necessary because the total energies of flares range over many orders of magnitude, following a uniform distribution with flare frequency roughly proportional to the inverse of the total energy. Stellar flares (and earthquakes) show similar power-law distributions.

24. Soft X-ray light curves showing solar flares of different sizes and durations. The red curve represents the total flux in the band 1 to 8 Angstrom, and the blue curve is the flux in 0.5 to 4 Angstrom. Basically, this means that the curves represent the evolution in time of the X-ray power emitted by the Sun in two energy ranges. Each one of the numerous spikes in the curves represents a temporary increase in the emission due to a solar flare.

Hazards

25. Solar flares strongly influence the local space weather of the Earth. They produce streams of highly energetic particles in the solar wind and the Earth's magnetosphere that can present radiation hazards to spacecraft and astronauts. The soft X-ray flux of X class flares increases the ionisation of the upper atmosphere, which can interfere with short-wave radio communication and can increase the drag on low orbiting satellites, leading to orbital decay. Energetic particles in the magnetosphere contribute to the aurora borealis and aurora australis.

26. Solar flares release a cascade of high energy particles known as a proton storm. Protons can pass through the human body, doing biochemical damage.[2] The proton storms are produced in the solar wind, and hence present a hazard to astronauts during interplanetary travel. Most proton storms take two or more hours from the time of visual detection to reach Earth's orbit. A solar flare on January 20, 2005 released the highest concentration of protons ever directly measured, [3] taking only 15 minutes after observation to reach Earth, indicating a velocity of approximately one-half light speed.

27. The radiation risks posted by solar flares and coronal mass ejections (CMEs) are among the major concerns in discussions of manned missions to Mars, the moon, or any other planets. Some kind of physical or magnetic shielding would be required to protect the astronauts. Originally it was thought that astronauts would have two hours time to get into shelter, but based on the January 20, 2005 event, they may have as little as 15 minutes to do so. Energy in the form of hard x-rays are considered dangerous to spacecraft and are generally the result of large plasma ejection in the upper chromospheres.

RESTRICTED

110

RESTRICTED

Observations

28. The following missions have flares as their main observation target.

(a) Yohkoh - The Yohkoh (originally Solar A) spacecraft observed the Sun with a variety of instruments from its launch in 1991 until its failure in 2001. The observations spanned a period from one solar maximum to the next. Two instruments of particular use for flare observations were the Soft X-ray Telescope (SXT), a glancing incidence low energy X-ray telescope for photon energies of order 1 keV, and the Hard X-ray Telescope (HXT), a collimation counting instrument which produced images in higher energy X-rays (15-92 keV) by image synthesis.

(b) GOES - The GOES spacecraft are satellites in geostationary orbits around the Earth that have measured the soft X-ray flux from the Sun since the mid 1970s, following the use of similar instruments on the SOLRAD satellites. GOES X-ray observations are commonly used to classify flares, with A, B, C, M, and X representing different powers of ten — an X-class flare has a peak 2-8 Å flux above 0.0001 W/m2 .

(c) RHESSI - RHESSI is designed to image solar flares in energetic photons from soft X rays (~3 keV) to gamma rays (up to ~20 MeV) and to provide high resolution spectroscopy up to gamma-ray energies of ~20 MeV. Furthermore, it has the capability to perform spatially resolved spectroscopy with high spectral resolution.

(d) Hinode - A new spacecraft, originally called Solar B, was launched by the Japan Aerospace Exploration Agency in September 2006 to observe solar flares in more precise detail. Its instrumentation, supplied by an international collaboration including Norway, the U.K., and the U.S., focuses on the powerful magnetic fields thought to be the source of solar flares. Such studies shed light on the causes of this activity, possibly helping to forecast future flares and thus minimize their dangerous effects on satellites and astronauts.

29. The most powerful flare of the last 500 years was the first flare to be observed, and occurred in September 1859: it was reported by British astronomer Richard Carrington and left a trace in Greenland ice in the form of nitrates and beryllium-10, which allow its strength to be measured today (New Scientist, 2005).

Prediction

30. Current methods of flare prediction are problematic, and there is no certain indication that an active region on the Sun will produce a flare. However, many properties of sunspots and active regions correlate with flaring. For example, magnetically complex regions (based on line-of-sight magnetic field) called delta spots produce most large flares. A simple scheme of sunspot classification due to

RESTRICTED

111

RESTRICTED

McIntosh is commonly used as a starting point for flare prediction. Predictions are usually stated in terms of probabilities for occurrence of flares above M or X GOES class with 24 or 48 hours. The U.S. National Oceanic and Atmospheric Administration (NOAA) issues forecasts of this kind.

Flare characteristics

31. Solar flares are tremendous explosions on the surface of the Sun. In a matter of just a few minutes they heat material to many millions of degrees and release as much energy as a billion megatons of TNT. They occur near sunspots, usually along the dividing line (neutral line) between areas of oppositely directed magnetic fields.

32. Flares release energy in many forms - electro-magnetic (Gamma rays and X-rays), energetic particles (protons and electrons), and mass flows. Flares are characterized by their brightness in X-rays (X-Ray flux). The biggest flares are X-Class flares. M-Class flares have a tenth the energy and C-Class flares have a tenth of the X-ray flux seen in M-Class flares. The National Oceanic and Atmospheric Administration (NOAA) monitors the X-Ray flux from the Sun with detectors on some of its satellites. Observations for the last few days are available at NOAA's website for Today's Space Weather.

SOLAR WINDS

33. The solar wind is a stream of energized, charged particles, primarily electrons and protons, flowing outward from the Sun, through the solar system at speeds as high as 900 km/s and at a temperature of 1 million degrees (Celsius). It is made of plasma. The solar wind streams off of the Sun in all directions at speeds of about 400 km/s (about 1 million miles per hour). The source of the solar wind is the Sun's hot corona. The temperature of the corona is so high that the Sun's gravity cannot hold on to it. Although we understand why this happens we do not understand the details about how and where the coronal gases are accelerated to these high velocities. This question is related to the question of coronal heating.

Solar Wind Variations34. The solar wind is not uniform. Although it is always directed away from the Sun, it changes speed and carries with it magnetic clouds, interacting regions where high speed wind catches up with slow speed wind, and composition variations. The solar wind

RESTRICTED

112

RESTRICTED

speed is high (800 km/s) over coronal holes and low (300 km/s) over streamers. These high and low speed streams interact with each other and alternately pass by the Earth as the Sun rotates. These wind speed variations buffet the Earth's magnetic field and can produce storms in the Earth's magnetosphere.

1. The Ulysses spacecraft has now completed one orbit through the solar system during which it passed over the Sun's south and north poles. Its measurements of the solar wind speed, magnetic field strength and direction, and composition have provided us with a new view of the solar wind.

2. The Advanced Composition Explorer (ACE) satellite was launched in August of 1997 and placed into an orbit about the L1 point between the Earth and the Sun. The L1 point is one of several points in space where the gravitational attraction of the Sun and Earth are equal and opposite. This particular point is located about 1.5 million km (1 million miles) from the Earth in the direction of the Sun. ACE has a number of instruments that monitor the solar wind and the spacecraft team provides real-time information on solar wind conditions at the spacecraft.

Atmosphereic Effect

35. The solar winds ejected by the sun's corona or center are highlycharged magnetic particles that travel through the atmosphere at 400km per hour. While each planet is protected by a magnetic field thatdeflects these charged volatile solar winds, the earth's convenientposition away form the sun is also a factor that keeps us protectedfrom the ill effects of solar winds. Planets positioned closer to thesun experience considerable degeneration of the magnetic field throughthe power of solar winds.

Outside Interferences

36. We suffer the effects of solar winds on earth today because of thenumber of communication satellites in outer space. The magnetic fieldof solar distorts and even destroys the functioning of communicationsatellites. Astronauts and cosmonauts suffer serious radiation relatedhealth conditions if they are caught in the path of solar winds.Radiation from solar winds is known to cause chromosome damage andcancer, and these conditions may be fatal for humans in outer space.Radio and television communication and satellite based internetservices are disrupted by solar winds. Military satellites are theaffected the worst by solar winds. Geomagnetic storms caused by solarwinds are very strong and can destabilize or destroy power grids. Theyalso affect all navigation and communication systems especially forvessels at sea. Aircraft communications and instruments in theaircraft will be susceptible to faulty functioning during geomagneticstorms.

RESTRICTED

113

RESTRICTED

Effects On The Earth

37. The effects of solar winds on the earth that are visible to naked eyeare the Aurora Borealis (the Northern lights) at the North Pole andthe Aurora Australis (he Southern Lights) at the South Pole. The fierytail seen attached to comets is the effect of solar winds visible tothe naked eye.

Solar Wind Damage

38. Solar winds are highly destructive magnetically charged high energy winds. Satellite communication on earth and the occasional geomagnetic storm near the poles are the main disruptive effects caused by solar winds on planet earth.

CORONAL MASS EJECTIONS

39. Coronal mass ejections (or CMEs) are huge bubbles of gas threaded with magnetic field lines that are ejected from the Sun over the course of several hours. Although the Sun's corona has been observed during total eclipses of the Sun for thousands of years, the existence of coronal mass ejections was unrealized until the space age. The earliest evidence of these dynamical events came from observations made with a coronagraph on the 7th Orbiting Solar Observatory (OSO 7) from 1971 to 1973. A coronagraph produces an artificial eclipse of the Sun by placing an "occulting disk" over the image of the Sun. During a natural eclipse of the Sun the corona is only visible for a few minutes at most, too short a period of time to notice any changes in coronal features. With ground based coronagraphs only the innermost corona is visible above the brightness of the sky. From space the corona is visible out to large distances from the Sun and can be viewed continuously.

40. Coronal Mass Ejections disrupt the flow of the solar wind and produce disturbances that strike the Earth with sometimes catastrophic results. The Large Angle and Spectrometric Coronagraph (LASCO) on the Solar and Heliospheric Observatory (SOHO) has observed a large number of CMEs. Coronal mass ejections are often associated with solar flares and prominence eruptions but they can also occur in the absence of either of these processes. The frequency of CMEs varies with the sunspot cycle. At solar minimum we observe about one CME a week. Near solar maximum we observe an average of 2 to 3 CMEs per day.

41. Coronal mass ejections are explosions in the Sun's corona that spew out solar particles. The CME's typically disrupt helmet streamers in the solar corona. As much as 1x10^13 (10,000,000,000,000) kilograms of material can be ejected into the solar wind. Coronal mass ejections propagate out in the solar wind, where they may

RESTRICTED

114

RESTRICTED

encounter the Earth and influence geomagnetic activity. CME's are believed to be driven by energy release from the solar magnetic field. How this energy release occurs, and the relationship between different types of solar activity, is one of the many puzzles facing solar physicists today.

42. CME's can seriously disrupt the Earth's environment. Intense radiation from the Sun, which arrives only 8 minutes after being released, can alter the Earth's outer atmosphere, disrupting long-distance radio communications and deteriorating satellite orbits. Very energetic particles pushed along by the shock wave of the CME can endanger astronauts or fry satellite electronics. These energetic particles arrive at the Earth (or Moon) about an hour later. The actual coronal mass ejection arrives at the Earth one to four days after the initial eruption, resulting in strong geomagnetic storms, aurora and electrical power blackouts. All of these solar-terrestrial interactions are forecasted and monitored by those who work in the space weather area.

Impact of a CME

43. When the ejection reaches the Earth as an ICME (Interplanetary CME), it may disrupt the Earth's magnetosphere, compressing it on the day side and extending the night-side tail. When the magnetosphere reconnects on the nightside, it creates trillions of watts of power which is directed back toward the Earth's upper atmosphere. This process can cause particularly strong aurora also known as the Northern Lights, or aurora borealis (in the Northern Hemisphere), and the Southern Lights, or aurora australis (in the Southern Hemisphere). CME events, along with solar flares, can disrupt radio transmissions, cause power outages (blackouts), and cause damage to satellites and electrical transmission lines.

Association with other solar phenomena

44. Coronal Mass Ejections are often associated with other forms of solar activity, most notably:

(a) Solar flares

(b) Eruptive prominence and X-ray sigmoids

(c) Coronal dimming (long-term brightness decrease on the solar surface)

(d) EIT and Moreton waves (e) Coronal waves (bright fronts propagating from the location of the eruption) (f) Post-eruptive arcades.

RESTRICTED

115

RESTRICTED

45. The association of a CME with some of those phenomena is common but not fully understood. For example, CMEs and flares were at first thought to be directly connected, with the flare driving the CME. However, only 60% of flares (M-class and stronger) are associated with CMEs.[1] Similarly, many CMEs are not associated with flares. It is now thought that CMEs and associated flares are caused by a common event (the CME peak acceleration and the flare peak radiation often coincide). In general, all of these events (including the CME) are thought to be the result of a large-scale restructuring of the magnetic field.

CORONAL LOOP

46. Coronal loops form the basic structure of the lower corona and transition region of the Sun. These highly structured and elegant loops are a direct consequence of the twisted solar magnetic flux within the solar body. The population of coronal loops can be directly linked with the solar cycle, it is for this reason coronal loops are often found with sunspots at their footpoints. The upwelling magnetic flux pushes through the photosphere, exposing the cooler plasma below. The contrast between the photosphere and solar interior gives the impression of dark spots, or sunspots.

CORONAL HOLES

47. Coronal holes are areas where the Sun's corona is darker, colder, and has lower-density plasma than average. These were found when X-ray telescopes in the Skylab mission were flown above the earth's atmosphere to reveal the structure of the corona. Coronal holes are linked to unipolar concentrations of open magnetic field lines. During solar minimum, coronal holes are mainly found at the Sun's polar regions, but they can be located anywhere on the sun during solar maximum. The fast-moving component of the solar wind is known to travel along open magnetic field lines that pass through coronal holes.

48. A region of the Sun's corona that appears dark in pictures taken with a coronagraph or during a total solar eclipse, and that shows up as a void in X-ray and extreme ultraviolet images. Coronal holes are of very low density (typically 100 times lower than the rest of the corona) and have an open magnetic field structure; in other words, magnetic field lines emerging from the holes extend indefinitely into space rather than looping back into the photosphere. This open structure allows charged particles to escape from the Sun and results in coronal holes being the primary source of the solar wind and the exclusive source of its high-speed component.

49. During the minimum years of the solar cycle, coronal holes are largely confined to the Sun's polar regions (although some exceptions have been observed by SOHO), while at solar maximum they can open up at any latitudes. Coronal holes are regions where the corona is dark. These features were discovered when X-ray telescopes were first flown above the earth's atmosphere to reveal the structure of

RESTRICTED

116

RESTRICTED

the corona across the solar disc. Coronal holes are associated with "open" magnetic field lines and are often found at the Sun's poles. The high-speed solar wind is known to originate in coronal holes.

CLIMATE CHANGE AND SPACE WEATHER

50. Under normal conditions it is the general belief that space weather does not couple significantly to tropospheric weather. Despite many thousands of papers that have been written on the subject, there is very little evidence of any consistent tropospheric phenomenon that can be tied to observed solar variations. One of the main reasons for this is that the bulk of the solar radiation, which lies mainly in the visible and near infrared parts of the EM spectrum, shows very little change over historical time scales. Most of the variation occurs in the lower and upper parts of the solar spectrum; the radio and X-ray bands. At these wavelengths, the solar radiation can vary by many orders of magnitude. These variations are very noticeable in the ionosphere, but since none of the X-ray radiation in particular, penetrates below about 60 km there is no reason to believe any changes would be noticed in the mesosphere "climate" takes place.

51 Particulate radiation from the Sun is also subject to extremely wide fluctuations. However, most of the particles have energies below (and usually well below) 100 MeV. The Earth's magnetic field is very effective in deflecting and/or trapping this radiation with most particles not reaching altitudes below about 50 km. The exception to this may be in the polar regions, although even particles travelling downward along the near-vertical polar magnetic field lines will generally be stopped by the dense layers of atmosphere. Occasionally very high energy particles (some with relativistic energies) do make it to ground level. This occurs just a dozen or so times per solar cycle and is referred to as a Ground Level Event (GLE).

52. Galactic cosmic rays have much higher energies than do solar cosmic rays (protons), and these make it through several atmospheric interactions all the way to the ground, mostly in the form of mu-mesons. This galactic cosmic ray flux is subject to solar influence in the form of heliospheric magnetic fields that can deflect and thus decrease the intensity of the flux. It has been speculated that this variation could cause precipitation variations with cosmic ray interactions providing condensation nuclei ("seeds" for initiating precipitation) in the upper troposphere. Again the evidence is not strong.

53. Recent work has revealed the existence of two areas in which space weather might influence global climate change. The first, of solar origin, relates the small variations in total solar radiation, now well documented by satellite cavity radiometers over many years, to long term climatic effects. The second, entirely

RESTRICTED

117

RESTRICTED

unrelated to solar flux variations, has been the investigation of impacts by kilometre size bodies (asteroids and comets) on the Earth's surface.

Solar Energy Variations and Climatic Change

54. Solar Irradiance Measurements. Full spectrum cavity radiometers on board 7 spacecraft have now monitored the total electromagnetic radiation output from the Sun, called the solar constant, over about 20 years. The average value of the solar constant at the mean distance of the Earth from the Sun (referred to as the Astronomical Unit and equal to 1.496x10^11 m) is about 1370 watt/square metre. This has shown an intrinsic solar variation of just under 0.1% in the last two decades (figure 1). There is also a variation due to the Earth's elliptical orbit, but the measurements in question are standardised to a distance of 1 A.U.] It is uncertain how indicative this recent variation is, measured over just 20 years, of variation in solar radiation.

Figure 1: Solar irradiance variations as recorded by a composite of satellite sensors for the latter part of the 20th Century. The continuous line is an 81 day running average.

55. Recent work has attempted to identify the source of these changes. It is now believed that most solar irradiance variations originate in solar surface magnetic activity. Approximately 80% of the measured variation can be accounted for by the two main visible features on the solar disk; sunspots and faculae. Sunspots are small dark areas on the solar disc whereas faculae are bright areas which form complex networks. Both of these phenomena occur around "active regions" where the magnetic field strengths are particularly high. Around sunspots the magnetic field strength can rise to thousands of gauss, whereas the average field is on the order of 1 gauss. The presence of faculae is highly correlated with the presence of sunspots, and hence with Sunspot Number (SSN). Over long time scales of months or greater it appears that increased irradiance due to the bright faculae exceeds the deficit due to the darker and cooler sunspots (Figure 2).

RESTRICTED

118

RESTRICTED

1. There has been a claim that solar irradiance from one minimum to another an 11 year period], when no spots or faculae are visible, does show a small change, but the general consensus is that any such non-magnetic variation, if it exists, is within the noise level (maybe 0.02%).

2. It is thus thought that sunspot number is a good proxy for solar irradiance changes, and historical variations in this number have led to a belief that the total variation of solar irradiance over time scales of thousands of years is unlikely to exceed 1%. This assumes that no non-magnetic related effects are likely to show up in the longer time scales.

3. The other way to approach the problem is to examine a sample of sun-like stars for irradiance fluctuations. When this is done, the Sun does appear to be abnormally quiet. A recent report suggests that the average G type (sun-like) star shows a 4% irradiance variation over a timescale of decades. Further, this variation may well be due to non-magnetically related activity, and is thus a phenomenon that we do not see at present in our Sun.

Figure 2: Solar satellite irradiance data showing modelled contributions due to sunspots and faculae separately

Solar Driven Global Climatic Change

56. Whilst a 0.1% variation is unlikely to produce measurable climatic change (particularly with a period of just a decade), a 4% variation would certainly be noticeable. Alternatively, a variation with a much longer period and amplitude under 1% could well influence global climate. John Eddy and others have claimed such climatic changes have occurred in the past, one particular association being a colder epoch in Europe during an apparently complete absence of sunspots over a 70 year period in the 17th century. Such claims have generated interest particularly amongst those who challenge the majority belief of enhanced greenhouse warming caused by man's recent activities.

RESTRICTED

119

RESTRICTED

Terrestrial Hypervelocity Impacts and Climate Change

57. Definition and Description. A hypervelocity impact is one where the kinetic energy of the impactor, an asteroid or comet, exceeds the energy that the object would possess if all its mass were conventional explosive. By virtue of the Earth's gravitational potential energy, all bodies on a collision course with the Earth have a velocity in excess of 11 km/sec (and usually less than the solar system escape velocity at Earth's orbit, which is 72 km/sec). Any impact at these speeds is a hypervelocity impact. However, the Earth's atmosphere acts as a good shield against common impactors. Most potential impacts are from very small bodies. Their interactions produce the meteors that we see in the night sky. Small meteoroids are usually turned to dust, while their very high cosmic velocity is reduced to a very low terminal velocity in the atmosphere. The resultant dust particles drift down to the surface over several months or years. Each year our planet accumulates approximately 40,000 tons of extraterrestrial matter from these interactions.

58 A meteoroid needs to be several metres or even tens of metres in diameter to have a chance of making it to the Earth's surface with its space velocity unchanged. If is does so, the energy of impact causes near total vaporisation of the object, as well as excavating an impact crater much larger than the original diameter of the impactor.

59 If the diameter of the impactor is about one kilometre, the amount of material vaporised and thrown into the atmosphere is sufficient to cause global changes. There are two cases to consider, an impact on land and an impact on water. Water is more common because of the greater area of the worlds oceans (70%) compacted to land.

60. Terrestrial Hypervelocity Impacts. Much study has been conducted into the consequences of hypervelocity impacts on land. Detailed computer simulations have been performed to study the entry through the atmosphere, impact, the trajectories of fragments and various consequences. The most significant effect in terms of global climate change comes from the "nuclear winter" resulting from the massive amounts of dust ejected into the stratosphere. This dust would circulate the globe for many months, blocking sunlight essential to plant photosynthesis, causing knock-on effects right up the food chain. It is thought that after a few years, the dust would precipitate out, but not before severe mass extinctions had occurred. Such a scenario has been much discussed in connection with the Cretaceous/Tertiary ("K/T") boundary of around 65 million years ago.

RESTRICTED

120

RESTRICTED

Oceanic Hypervelocity Impacts

61. Hypervelocity impacts into the ocean are only just beginning to receive detailed study. The major effects here relate to the injection of massive quantities of water vapour into the stratosphere and higher atmospheric layers. Water vapour is a greenhouse gas, and the overall effect appears to be a substantial increase in the global temperature, just the opposite of what is expected for a continental impact. A more immediate and perhaps more frightening result of an ocean impact would be the production of massive tsunamis.

Geological Evidence

62. There are currently around 150 impact craters that have been positively identified on the Earth. Over 25 such craters have so far been found in Australia. Seven submarine impact structures are also known. Following the 1980 paper by Alvarez et al, there now appears to be a majority opinion that impact cratering has played a significant role in Earth's recent geological history.

63. A tentative identification of the impact crater that caused the K/T boundary event has been claimed for a large structure off the northern coast of the Yucatan peninsula. Increased values of osmium and iridium, elements associated with extraterrestrial meteorites, have been found in geological strata corresponding to the K/T time period. Ripple-like deposits in Texas and South Dakota have also been association with this event as evidence of the tsunamis resulting from the impact. These claims imply that the tsunami travelled more than 1500 kilometres inland.

Future Probabilities

64. The population of Near Earth Objects (NEOs) over a range of sizes has been estimated by extrapolation from small bodies (whose frequency is well known), and from the rate of return of larger (kilometre size) objects. Figure 3 gives the current estimates for impact rates as a function of impactor size.

65. It is thought that the population of NEOs larger than a kilometre is around 2000 with upper and lower estimates of 4000 and 1000. A kilometre in size is significant because it is believed to be the approximate size of impactor for which major global effects will occur. From Figure 3 we see that the estimated frequency for impacts from kilometre sized bodies is about one every 100,000 years.

RESTRICTED

121

RESTRICTED

Figure 3: Estimated impact frequencies as a function of impactor diameter.

Summary

66. In the current epoch, the Sun is the source of most variations in our near space environment. These variations are called "space weather" and affect a variety of technological systems both ground and space based. Space weather does not appear to cause immediate and direct effects in the troposphere or biosphere of the planet. However, we have reason to expect that space weather might be more significant for the biosphere in the longer term.

67. Small variations in total solar irradiance have been detected from satellite based radiometers over the last 20 years. These variations appear to too small in amplitude and have too short a period to be a factor in recent climate changes. Studies of other sun-type stars indicate that larger variations in irradiance, of the order required to effect climate on earth, do occur on sun-like starts. Some claims of historical climatic effects, resulting from solar irradiance variations, have been made from statistical correlations with various solar indices.

68. Impacts from space debris which travels in Earth crossing orbits can be expected to cause global disturbances to the Earth's climate at intervals of the order of 100,000 years. Much larger disturbances, probably leading to massive biosphere extinctions, appear to have occurred at intervals of the order of 100 million years.

69. Other space weather phenomena from outside the solar system which have the potential to cause global climate changes include galactic cosmic rays, nearby supernovae, interstellar dust clouds, binary star de-orbits, gamma ray bursters and X-ray stars.

RESTRICTED

122

RESTRICTED

EFFECTS OF SPACE WEATHER ON AVIATION

HF and VHF Radio

70. VHF (very high frequency) and UHF (ultra-high frequency) radio channels are used for aircraft communications over populated areas in the low and mid latitudes. Over the remote polar areas (>78°N and >60°S) and the oceans, aviation operations are given primary and secondary HF (high frequency) radio channels for communicating with ground controllers. Radio communications are critical because position reports are required at set intervals, such as every five or ten degrees of longitude along an east-west track, or with any major track or altitude change. If the pilots cannot contact the ground controllers on either the primary or secondary frequency, they must resort to other options, including trying to find an alternate frequency, contacting another aircraft to relay the message, or calling Air Traffic Control (ATC) via satellite communications (SATCOM or Iridium) if the aircraft is equipped and the service is available at the latitudes flown. In general, SATCOM is unreliable above 82 degrees latitude. The Iridium satellite constellation does provide polar coverage, but as of 2012 only a fraction of aircraft were Iridium-equipped.

71. If space weather interferes with HF or VHF/UHF communications, the problems can quickly become serious. No ground-based air traffic control radar is available over the poles or oceans, so without communications, air traffic controllers have no position reports on the plane. This means they cannot verify aircraft separation, which compromises the critical situational awareness needed for flight safety. Furthermore, without a means of contacting the crew, ATC has no way of finding out whether deviations from the flight path have occurred. Ultimately, failure to make contact can result in ATC initiating search and rescue operations.

72. Communications issues related to recent space weather events are reported below:

(a) Thirteen overdue position reports for flights over Central East Pacific and Central West Pacific (1830-1930Z Jan 27, 2012 R3 radio blackout)(b) HF service degraded for over two hours (Oct 19, 2003 R3 radio blackout)(c) Degradation of high latitude comms (Oct 24, 2003 G3 geomagnetic storm)(d) Widespread HF communications problems in Alaska (Jan 22-23, 2012 S3 solar radiation storm)(e) Communications problems in Asia and the Pacific (Jan 22, 2012 daytime R2 radio blackout)(f) Communications problems off U.S. East Coast and West Coast (Jan 27, 2012 daytime radio blackout)

GPS/GNSS

73. The Global Navigation Satellite System (GNSS) has become increasingly used in all aspects of navigation, including in our cars and cell phones. But its use in airline operations has revolutionized the industry by allowing very precise three-dimensional positioning of aircraft. Both the U.S. and Russia have worldwide Global

RESTRICTED

123

RESTRICTED

Positioning System (GPS) coverage available through their respective networks. The high accuracy of the GPS-provided spatial information provides air traffic controllers with extremely precise positioning information. This precise information allows controllers to position planes more closely together in flight patterns; that is, the separation distances are reduced. Additionally, GPS allows a greater number of operations to be undertaken in low visibility environments by supporting approaches and landings in fog or other restricted visibility conditions. Disruptions to GPS are usually one of two types. First, the “lock” on the place might be lost, meaning that controllers no longer have data about the plane’s position. Second, vertical positioning errors can increase as data relays to and from the satellites are affected. In some cases, the satellites themselves could be damaged. GPS disruptions cause the actual vertical separation between airplanes to be unknown. In these situations, approaches or landings relying on GPS guidance will not be possible.

74. If GPS information cannot be relied on, or if controllers lose a plane’s GPS location, greater separation distances between aircraft will be required.

75. If air traffic controllers don't have precise location information for a plane, separation distances will need to be increased. IFR (Instrument Flight Rules) may no longer apply, and planes will need to adhere to larger spacing requirements. (VFR - Visual Flight Rules; VMC - Visual Meteorological Condition; MVFR - Marginal Visual Flight Rules; IFR - Instrument Flight Rules; IMC - Instrument Meteorological Condition.) If GPS is disrupted, the following effects can propagate within a very short time.

(a) Aircraft are suddenly not capable of maintaining actual navigation performance (ANP). This means that the reduced aircraft separation environment relied on through GPS is no longer acceptable.(b) Missed approaches are likely on a required navigation performance (RNP) approach.(c) Air Traffic Control must execute contingency procedures to re-establish traffic within larger separation standards.(d) Reductions in safety can result until larger aircraft separations are achieved.

RESTRICTED

124

RESTRICTED

(e) Reroutes to increase separation or accommodate airspace capacities can lead to delays and diversions.

76. Reroutes and delays can have significant economic impacts on airlines, individuals, and government and other sectors. Rerouting is estimated to cost airlines in the range of $10,000 to $100,000 per flight, and the costs of delays can ripple through multiple sectors, quickly tallying into the millions of dollars. Fortunately for airline operations, in the years since GPS has been used, issues related to space weather events remained relatively uncommon for the time period ending in 2011. Solar activity was very low from 2007 through 2011. But as solar activity increases in advance of the mid-2013 solar maximum, more frequent disruptions are possible.

High Latitude Operations

77. During significant events, space weather effects can be observed over various regions of the globe but the impacts tend to be most pronounced at high latitudes (>50 degrees) and the polar regions (>78 degrees) in both hemispheres. The North Pole region has become particularly well travelled as airlines implement an increasing number of polar flights, particularly between the North America and Asia. In 2011 alone, almost 11,000 flights travelled polar routes.

78. High latitude operations involve not only commercial and military aircraft, but general aviation as well. Because general aviation planes tend to fly shorter routes at lower altitudes, they are in many cases less affected by space weather. However, during severe or extreme events, impacts to general aviation can include navigation errors or outages, particularly on the sunlit side of Earth.

79. Several high latitude routes cross the area of responsibility covered by the Anchorage Center Weather Service Unit (CWSU), collocated with the Air Route Traffic Control Center (ARTCC) in Anchorage, Alaska. The Anchorage Airport is a busy place for international air traffic, as 90% of air cargo sent to and from China goes through Anchorage first. The airport has ranked first in U.S. and third worldwide for cargo poundage (5.9 billion lbs/year). The area of coverage of the Anchorage CWSU is definitely affected by space weather, and impacts to radio communications, GPS positioning, and flight routing can cause significant and expensive disruptions to aviation operations. The blue line surrounds the area of coverage of the Anchorage Center Weather Service Unit in Alaska.

RESTRICTED

125

RESTRICTED

80. As we saw from the reported impacts, disruptions of HF communications and GPS over the polar latitudes mean that planes often must be rerouted. For the airlines, this rerouting means delays, higher fuel consumption, additional crew hours, potential additional stops, and significant expense.

RESTRICTED

126

RESTRICTED

Check Assimilation

Fill in the blanks

1. Sunspots are caused by _________________ which inhibits convection, forming areas of reduced surface temperature.

2. A solar flare is a __________________________ in the Sun's atmosphere.

3. _______________________ is a stream of energized, charged particles, primarily electrons and protons, flowing outward from the Sun, through the solar system.

4. ____________________________ disrupt the flow of the solar wind and produce disturbances that strike the Earth with sometimes catastrophic results.

5. _____________________ are areas where the Sun's corona is darker, colder, and has lower-density plasma than average.

Check Assimilation: The Key

1. Sunspots are caused by intense magnetic activity, which inhibits convection, forming areas of reduced surface temperature.

2. A solar flare is a large explosion in the Sun's atmosphere.

3. The solar wind is a stream of energized, charged particles, primarily electrons and protons, flowing outward from the Sun, through the solar system.

4. Coronal Mass Ejections disrupt the flow of the solar wind and produce disturbances that strike the Earth with sometimes catastrophic results.

5. Coronal holes are areas where the Sun's corona is darker, colder, and has lower-density plasma than average.

RESTRICTED

127