2010 - An Analysis of the Fog Distribution in Beijing for the 2001 2005 Period Using NOAA and FY...

Atmospheric Research xxx (2010) xxx–xxx

ATMOS-02125; No of Pages 15

Contents lists available at ScienceDirect

Atmospheric Research

j ourna l homepage: www.e lsev ie r.com/ locate /atmos

ARTICLE IN PRESS

An analysis of the fog distribution in Beijing for the 2001–2005 period usingNOAA and FY data

J.L. Wang a,⁎, S.M. Li b, X.L. Liu c, X.J. Wu b

a Institute of Urban Meteorology, China Meteorological Administration, Beijing 100089, Chinab National Satellite Meteorological Center, Beijing,100081, Chinac Beijing Meteorological Information Center, Beijing, 100089, China

a r t i c l e i n f o

⁎ Corresponding author. Address: No.55 Beiwaxili RBeijing, 100089, China. Tel.: +86 10 6840 0749; fax:

E-mail address: [email protected] (J.L. Wang).

0169-8095/$ – see front matter © 2010 Elsevier B.V.doi:10.1016/j.atmosres.2010.01.007

Please cite this article as: Wang, J.L., et al.,and FY data, Atmos. Res. (2010), doi:10.1

a b s t r a c t

Article history:Received 31 July 2009Received in revised form 24 January 2010Accepted 24 January 2010Available online xxxx

Based on the various fog remote sensing information products abstracted from the polar orbitmeteorological satellite data (NOAA and FY), this paper analyzes the characteristics of thefrequency and distribution that dense fogs have taken place in the Beijing district from 2001 to2005. Among those information products, the statistical graph of the foggy days in Beijing from2001 to 2005 by means of satellite remote-sensing reflects the frequency that dense fog hasoccurred in the different regions in Beijing; and the statistical graph of the foggy days of eachseason by satellite remote-sensing shows the characteristics of the temporal and spatial changeof the fog distribution in Beijing in different seasons; the fog degree index(like pixel-levelspatial measurement) by satellite remote-sensing reflects the fog frequency difference on unitarea in different counties of Beijing. Meanwhile, based on the satellite remote-sensing pictures,data on the main meteorological elements that contribute to the formation of fog, geographicinformation, and DEM data, this paper makes an analysis in 3 areas, the traits of the fogdistribution in the different regions of Beijing in different seasons; meteorological causes fortemporal and spatial changes of the main fog types (advection fog, radiation fog). This paperalso gave a brief introduction of the general principles of meteorological satellite remotesensing fog, fog information extraction method, as well as the process of satellite orbit selectingaccording to the ground visibility, data processing and product generation.

© 2010 Elsevier B.V. All rights reserved.

Keywords:Satellite remote sensingFog distributionMeteorological origin

1. Introduction

As an important weather phenomenon, fog causes a lowsurface visibility, its condensation nuclei usually carry germs,and the thermal inversion phenomenon, which usuallycoincides with it, leads to air pollution. All of these actuallypose a threat to the transport, human health as well as theenvironment. According to incomplete statistics, deaths ofthose traffic accidents which are caused by fog occupies morethan a half of the national unnatural deaths during safeproduction; and many serious photochemical pollutionincidents in the city were actually caused by fog. With the

oad, Haidian District,+86 10 6847 5440.

All rights reserved.

An analysis of the fog d016/j.atmosres.2010.01

improvement of people's living standard, more and moreattention has been paid to fog information. The monitoringand prediction of fog are undoubtedly one of the mosteffective ways to reduce the damage caused by fog disaster.Due to its wide ranging observation and a strong sensitivityto the fog, satellite data plays an important role in the moni-toring of foggy weather.

It has been long since people apply meteorologicalsatellites to fog monitoring and prediction. In the early1970s, satellite data has been used in the instance analysis ofclassic fog. In 1973, American scientists initially used SMS-1visible channel image to study the process of the dissipationof the radiation fog with dynamic monitoring. They madesome experiments to observe and predict the dissipationof the fog in California, which proved to be very successful(Gurka, 1974; Gurka, 1978; Gustafson and Wasserman,

istribution in Beijing for the 2001–2005 period using NOAA.007

mailto:[email protected]

http://dx.doi.org/10.1016/j.atmosres.2010.01.007

http://www.sciencedirect.com/science/journal/01698095


Table 1Table of thresholds of variables for fog detection.

Variable name Threshold

R670nm N=15%R865nm N=15%R1640nm N=20%NDSI (normalized difference snow index) −0.2

≤NDSI≤0.2Tfog_10.7 μm−Tclearwater_10.7 μm (the difference betweenBrightness Temperature of a fog pixel and that ofa nearest clear-sky water pixel )

N=−4 K

2 J.L. Wang et al. / Atmospheric Research xxx (2010) xxx–xxx

ARTICLE IN PRESS

1976). Mr. Hunt holds that the radiation rate of fog or lowclouds is evidently lower in the mid-infrared (SIR 3.7 μm)channel than that in the thermal infrared LIR (11.0 μm)channel. And this difference will in turn result in a differenceof the brightness temperature on the part of those clouds inthe SIR and LIR channels, whereas in terms of the groundsurface, the difference is not that obvious. The radiationcharacteristics of clouds and fog somehow furnish a theoret-ical basis for the identification of low clouds and fog in thenight. Based on this theory and the spectrum data of NOAAand GOES, in the past 20 years, lots of scientists delve into thephysical trait by calculating the brightness temperaturedifferences between the two channels and using otheridentification methods that is related (Rosenfeld et al.,2004; Anthis and Cracknell, 1999; Ellrod, 2000). On theother hand, a terrific progress was achieved in the field of fogdata collection by using satellite remote sensing with theprosecution of the research into the microphysical traits offog (Bendix, 1998; Greenwald and Christopher, 2000;Nakajima and Nakajima, 1995), the geographical and season-al rules for the fog distribution in the Europe (Bendix, 2001),the model test of the dissipation time of fog with the aid ofgeographical information technology (Bendix et al., 1999;WU et al., 2005).

It's the first time that an analysis of the statisticalcharacteristics of fog in the perspective of climatology hasbeen conducted in China. Statistical analysis of the climaticcharacteristics of the occurrence of fog contributes greatly tothe understanding of seasonal fog distribution characteristic,its deep root as well as the arrangement of the visibilityobservation stations. As an international metropolis, Beijing isthickly populated, and the urban construction has developedrapidly in recent years. The capital has witnessed a sharpincrease in the roads as well as in the number of vehicles. As aresult, the city transportation is heavily burdened. Beijing hasgot more fog, which somehow increasingly aggravates thetransportation problem mentioned above. Focusing on thefog occurrence during the past 5 years (2001–2005) inBeijing, this paper makes a statistical analysis of the fogfrequency and distribution in the Beijing district with the aidof China's FY-1C, FY-1D and NOAA / AVHRR meteorologicalsatellite data (Bendix and Bachmann, 1991; Bendix, 1995a,b;Chen et al., 2002), which conduces to the further research ofthe perennial distribution features of the fog in Beijing and itsmain origins in terms of climatology (Wang et al., 2004;Wang and Liu, 2006).

2. Materials and methods

2.1. Data collection and processing

The meteorological satellite data adopted in this paper isthe round-the-clock data of FY-1C, FY-1D,NOAA-14, andNOAA-16 and the conventional visibility station data of thecorresponding period during 2001 and 2005. After a series ofprocess including calibration, solar zenith angle correctionand geo-location correction and so on, data covering Beijingarea has been projected and a satellite dataset in Beijing hasbeen obtained. The fog frequency analysis of the meteoro-logical satellite data is based on the results of the identifica-tion of the daytime fog, and the fog is an incidental weather

Please cite this article as: Wang, J.L., et al., An analysis of the fog dand FY data, Atmos. Res. (2010), doi:10.1016/j.atmosres.2010.01

phenomenonwith a low frequency and a seasonal occurrencefeature. Therefore, in order to avoid dealing with a largeamount of satellite data, the paper at first sorts out the datawhose visibility value is less than 1 km in light of thecharacteristics that when fog occurs the ground visibilityis very low and the rule that the visibility value is usuallybelow 1km; and then collects the satellite data during thecorresponding period according to station data with lowvisibility that is sorted out.

A comprehensive consideration of the status of eachstation is essential when sorting out the station data with lowvisibility. There are four observation durations per day(2:00 AM, 8:00 AM, 14:00 PM, 20:00 PM). In order to getthe largest fog distribution data during the day, the durationwhen low visibility value occurs most frequently is usuallyselected first. If it is not the low visibility duration but otherduration that is selected first, the low visibility duration isalso taken as one of the selected duration of the satellite data.Since it is much more difficult to identify, check and correctthe night fog detected result in the night, daytime fog data ismore preferred to get fog distribution in Beijing. Usingderived dual channel difference is the only method to detectfog and the low stratus cloud at night, and it is difficult todistinguish stratus cloud from fog. It has some way to detectfog more exactly at daytime, such as using near-infraredchannel data.

A satellite dataset of NOAA/AVHRR, FY-1C and FY-1D from2001 to 2005 covering Beijing and the surrounding areas isselected in the duration order set up in accordance with thelow visibility data, which can be used in the extraction of foginformation and the statistical analysis of thick fog frequency.Satellite data stored in 1b format is collected, and is thenprojected in geographical projection with a spatial resolutionof 0.01° to get local projected data. In the process of theprojection, TOA reflectance in visible, near-infrared andshort-wave infrared channels and brightness temperature inlong infrared channels are obtained by calibration transac-tion, and solar zenith angle correction, is performed. Geo-location correction is also used to decrease the position error.

2.2. The extraction information and statistical methods of heavyfog data

As mentioned above, with the aid of satellite data, theanalysis of fog frequency in the paper is based on the resultsof the identification of the daytime fog. In otherwords, the fogidentification with the aid of satellite remote sensing under-lies the identification of the fog frequency. The multi-channelthreshold method (thresholds can be seen in Table 1) which



3J.L. Wang et al. / Atmospheric Research xxx (2010) xxx–xxx

ARTICLE IN PRESS

is commonly used is adapted to fog identification, consideringthe elevation data and land-cover data at the same time. Thepaper combines the threshold method with the closebrightness temperature analysis and boundary analysis.After an automatic identification of the fog based on themulti-channel threshold method and other methods, aninteractive check and revision of the identification results(in case of the errors and missed identification) are needed toensure the accuracy of the fog identification. Dynamicthresholds rather than fixed thresholds are used in the fogdetection algorithm. These make the fog detection moreaccurate. And the identification results generated fog zonebinary raster documents, namely fog zone is 255, while non-fog zone is 0. After automatic identification, an interactivecheck and revision for each image of the fog satellite dataset isperformed to improve fog detection accuracy. Both bycomparing with ground observations and interactive valida-tion by satellite images, the accuracy of the identification offog uncovered by clouds reaches more than 85%, basicallymeeting the demand.

For the satellite data in which fog takes place several timesin one day, the identification is made each time the fog occurs.Further, the composition of the largest fog zone is made basedon the result of each time's identification, and ultimately abinary data of the daily fog distribution is generated.

Fog frequency statistics, namely the statistics of thenumber of the foggy days, refers to the statistics madetowards the number of foggy days of the pixels in a period.Based on the results of the identification of the daytime fog, itdeals with all the identification results of the daytime fog inthe selected period. When the pixels turn out to be fog in oneday, a day is added to the pixel foggy days. Ultimately the fogfrequency statistical result of the pixel in the correspondingperiod is generated.

In this paper, two statistical methods are adopted, one is a5-year statistics of the overall fog frequency and the other isthe seasonal statistics of the fog frequency. The former is todeal with all the identification results of the foggy days duringthe five years, and finally generate an analytic result of theoverall fog frequency during the 5 years, which reflects thegeneral distribution characteristics of the fog frequency inBeijing during the past 5 years. The latter deals with theidentification result of the foggy days in each season: spring(March to May), summer (June-August), autumn (Septemberto November), and winter (December to February) of the5 years, and ultimately generates an analytic result of the fogfrequency in each season, which reflects the distributioncharacteristics of the fog frequency in terms of season.Meanwhile, an interactive check is made between the fogdata attained during the same period and the analytic resultof the fog frequency of the satellite data.

( fcov), the index of the fog frequency distribution intensityis used in this paper to illustrate the intensity of the fogfrequency distribution. ( fcov)is to do areaweighted average tothe foggy days to attain an index which reflect the intensity ofthe fog frequency distribution on per unit area. And thecalculation method is as follows:

fcov =∑n

i=1fdi × fsi

Sð1Þ


In this formula, fdi refers to the foggy pixel days, and fsirefers to the foggy pixel area, S refers to the total area of theregion that is covered in the statistics.

In addition, as far as the seasonal characteristic of the fogdistribution is concerned, this paper makes an analysis of themeteorological cause of the seasonal distribution of the fog,resorting to the data such as surface relative humidity, surfacetemperature and wind data of NCEP (National Centers forEnvironmental Prediction) in the United States during thecorresponding periods. Further, the paper makes an analysisof the impact that the ground elevation and land-cover shedupon the distribution of fog.

3. A comparison between the heavy foggy days in the fourseasons in Beijing from 2001 to 2005

3.1. A comparison between the heavy foggy days in the fourseasons in Beijing from 2001 to 2005

Beijing's climate is characterized by: drought accompa-nied by lots of wind in spring, with fewer foggy days; Morerainfall with a higher humidity in summer, more foggy days;still a high humidity in autumn which is right next tosummer, and fog occurs easily; a lower humidity in winterwith much storm as a major feature, and fewer foggy days(Ackerman, 1987; Gaffen and Ross, 1999) Fig. 1 shows thecomparison between the fog data (the foggy days in the fourseasons of 2001–2005) of Beijing attained through satelliteremote sensing and ground observation. In the northern part,fog is mainly radiation fog which fades away fast and thus ishard to observe (Brown and Roach, 1976; Bott et al., 1990;Bergot and Guedalia, 1994). As a result, the accuracy ofstatistics is somehow lessened. Therefore except for thewinter, Yanqing, Miyun, Huairou stations, three northernregions, are not included in the comparison. The statisticalresult of winter from 2001 to 2005 shows that there are fewerfoggy days in the northern stations, and the comparisonresults are notmuch affected. Fig. 1 shows that there is a goodcorrelation between the two, and the correlation coefficientsare 0.75, 0.64, 0.52, and 0.65. In conclusion, it is feasible to usesatellite remote sensing in the study of the fog frequency ofBeijing.

3.2. A comparison between the overall foggy days from 2001to 2005

Fig. 2 shows the comparison between the overall foggydays attained by satellite remote sensing and groundobservation in Beijing from 2001 to 2005. The general trendsof the two are almost the samewith a correlation coefficient of0.64, which further proves the feasibility of applying satelliteremote sensing to the study of the fog frequency in Beijing.

4. The statistical results and analysis of the heavy fogfrequency inBeijingwith the aid of satellite remote sensing

4.1. The distribution characteristics of the heavy fog frequency inBeijing from 2001 to 2005with the aid of satellite remote sensing

Fig. 3 is an overall statistical figure of the fog frequencydistribution in Beijing from 2001 to 2005. In terms of the



Fig. 2. A comparison between the overall fog frequency from 2001 to 2005.

Fig. 1. A comparison of the fog frequency in the four seasons of 2001–2005.


ARTICLE IN PRESS

overall distribution of the fog frequency in the different parts,fog mainly takes place in the southeast and northeast ofBeijing, while rarely in the west part. As far as the countydistribution is concerned, Daxing and Tongzhou are the twocounties that are more frequented by fog than others. Duringthe past five years, both have a total of 50 foggy days or evenmore; in some districts the number reaches 80. The foggy


days are also above 50 in most parts of Shunyi District,eastern Fangshan district, northern Haidian District, south-east of Chaoyang District, the central and eastern parts ofChangping District, south and northeast of Huairou, north-west and southwest of Miyun District. In Mentougou Districtand the central part andwestern part of Fangshan District, fogoccurs less frequently, and the foggy days in the past fiveyears is within 30 days, in some parts even within 20 days. Inother districts, the number is between 30 and 50. In theperspective of the distribution of the main road loop line, inthe urban areas within the fifth-loop line, the foggy days arebetween 30 and 50, and the average annual foggy days arebetween 6 and 10 days.

This paper calculates the intensity index of the fogdistribution in each district and county, and further ordersaccording to the intensity index (see Table 2). The resultshows that among the districts, Daxing and Tongzhou occupythe first two places respectively; while in terms of the season,the index in summer and autumn is higher than that inwinterand spring.

4.2. The seasonal distribution characteristics of the heavy fogfrequency in Beijing during 2001 and 2005 with the aid ofsatellite remote sensing

4.2.1. The analysis of the characteristics of the heavy fogfrequency distribution in the spring

According to the statistical results of the fog frequencydistribution in Beijing in the spring (March to May) of 2001and 2005 (see Fig. 4), the foggy areas of Beijing are mainlylocated in the southeast and northern parts in the spring,while in the southwest and northeast, there are generally lessfoggy zones. Daxing, Tongzhou, and Huairou have a muchhigher fog frequency, and the total number of foggy days isbetween 5 and 15 days, in some areas more than 10 days. Thefog frequency is lower in the southwest (mainly XichengDistrict, most parts of Haidian District, Shijingshan District,MentougouDistrict, andwestern Fangshan District and so on)and in the northeast (eastern Miyun County, north of PingguDistrict and so on). Generally speaking, the foggy days inthese districts are within 3 days, while in the other districts,the number is between 3 and 10 days.

Figs. 5–7 illustrate the causes for the fog distribution in thespring of Beijing in the perspective of the surface–wind–temperature and surface–wind–humidity, which are the



Fig. 3. An overall statistical figure of the fog frequency distribution in Beijing (2001–2005).


ARTICLE IN PRESS

main meteorological elements that form fog. In terms of therelative humidity, there is a relative high humidity center inwhich the surface relative humidity is above 44% in thesoutheast of north China. Beijing is located on the northwestside of the humidity center, and the seasonal average relativehumidity is between 43 and 45%. In terms of the surfacetemperature, Beijing is located in the gradient of the northern

Table 2A statistical table of the fog intensity index in each district and county inBeijing during 2001 and 2005.

District Fog intensity index(2001–2005)

Spring Summer Autumn Winter

Daxing 61.07 9.5 19.42 21.97 10.19Tongzhou 59.64 8.11 22.11 19.74 9.68Shunyi 52.89 5.39 24.56 17.50 5.44Chaoyang 47.94 4.57 22.13 15.74 5.51Changping 47.73 4.81 23.49 15.68 3.75Huairou 47.11 6.72 23.23 14.72 2.44Haidian 46.00 2.96 22.59 15.43 5.03Miyun 44.10 4.35 22.17 14.43 3.15Fengtai 43.31 3.69 18.01 13.98 7.63Dongcheng 41.30 3.63 19.11 12.26 6.30Chongwen 40.11 3.52 17.16 12.63 6.79Shijingshan 40.10 2.57 17.61 14.45 5.47Pinggu 40.03 4.30 17.84 13.18 4.71Xicheng 39.87 2.74 16.13 14.45 6.55Xuanwu 39.48 2.43 18.29 12.67 6.10Yanqing 35.65 5.21 18.07 10.43 1.94Fangshan 33.69 4.44 13.14 10.34 5.78Mentougou 21.53 1.96 11.20 6.59 1.79


part of the temperature ridge whose center is located in theHuanghuai region, in which the surface temperature isgenerally 8–12 °C. The formation of the temperature gradientis largely due to the frequent cold air activities in the spring inNorth China as well as the altitude difference between theMongolian plateau, Taihang Mountains, and North ChinaPlain. The main wind direction in the spring in the southernpart of Beijing is southwest, while in the central part andnorthern part, the spring wind direction is mainly west. Acomprehensive analysis of the seasonal average field of thesurface wind temperature and humidity mentioned abovebrings it to light that the higher fog frequency zone in thesoutheast of Beijing is largely due to a good humidity fieldcondition and the transportation of the moist air on the partof the southwest air stream near the ground, and that themain cause for the fog formation is the advection fog. In thenorth part where the fog occurs frequently, the west windprevails, and the transportation of the water vapor is verypoor. However, north Beijing is so close to the relative highhumidity center of the east part of north China that the localsurface humidity condition is better. Meanwhile, it is alsolocated in the part that the temperature gradient is thelargest. Comparedwith the central and southern part of southChina plain, the temperature is lower by 5 to 7 °C, whichconduces to the formation of the water steam. According tothe time that the satellite passed the area in the data that isadopted, the formation of the fog in north Beijing is mainlythe result of the local radiation and cooling effect in the earlymorning. As a result, it is classified as the radiation fog. In thesouthwest and the urban areas of Beijing, the relative



Fig. 4. The statistical figure of the of the fog frequency distribution in fog frequency distribution in the spring Beijing (2001–2005).


ARTICLE IN PRESS

humidity condition is poorer. In consequence, the fogfrequency is lower. At the same time, due to the influenceof the urban heat island effect, the temperature in the urban

Fig. 5. The figure of the surface relative humidity


district of Beijing is generally higher by 4–6 °C than that in thesuburbs, which makes it harder for the fog to take place. Inother words, fog occurs less frequently.

in the spring in north China (2001–2005).



Fig. 6. The figure of the average surface temperature distribution in the spring in north China (2001–2005).


ARTICLE IN PRESS

4.2.2. The analysis of the characteristics of the heavy fogfrequency distribution in the summer

According to the fog distribution statistical results (seeFig. 8) in each summer (June to August) from 2001 to 2005,Beijing Summer fog is mainly in the east part of the middle-north and the southern parts of this city, with the fogfrequency much higher than that in the spring. Seen from the

Fig. 7. The figure of the distribution of the wind direc


angle of the urban distribution, in Changping District, Shunyi,Huairou, Miyun County, Chaoyang District, Tongzhou District,Haidian District, Daxing District, and Fengtai District, the localfog frequency is between 20 and 40 days, while in FangshanDistrict, and Mentougou District it is below 15 days; in themost part of Yanqing County, Shijingshan District, part ofFengtai District, Xicheng District, Dongcheng District,

tion in the spring in north China (2001–2005).



Fig. 8. The statistical figure of the fog frequency distribution in the summer in Beijing (2001–2005).


ARTICLE IN PRESS

Xuanwu District, and Chongwen District, the fog frequency isbetween 15 and 20 days. In terms of the distribution of thehighway, the fog frequency within the fifth-loop line is morethan 15 days, with the frequency beyond the North- second-loop-line and the East second-loop-line above 20 days.

Fig. 9. The figure of the surface relative humidity in


As far as the relative humidity is concerned (Fig. 9), in thesummer, the lack of a wet surface center is obvious comparedwith that in the spring. The humidity was in gradient uniformdistribution with the south higher up and the south lowerdown and relative humidity is between 60 and 64% in general.

the summer in north in China (2001–2005).



Fig. 10. The figure of the average surface temperature distribution in the summer in north China (2001–2005).


ARTICLE IN PRESS

The surface temperature was also in the pattern of being highin the northwest stretching to be low in the northwest(Fig. 10), with the surface temperature ranging from 21 to24 °C. Compared with the spring, temperature and humidityhave no obvious high-value center. In the wind field, becausethere is an obvious southeast airstreams on the surface, it ismore favorable than the spring for the Yellow Sea water

Fig. 11. The figure of the distribution of the wind direct


vapor to move north up. Based on the seasonal average fieldof the temperature and humidity of the surface windmentioned above (Fig. 11), the high frequency in thenorthern part of Beijing is due to the fact that the highhumidity conditions (20% higher than that in the spring)make the night long wave radiation cooling effect morenoticeable. The primary cause for the fog is the radiation

ion in the summer in north China (2001–2005).




ARTICLE IN PRESS

cooling effect, so it can be called radiation fog. The higherfrequency of fog in the South is due to good humidityconditions and the transmission of the southeast moistairstreams near the ground. Fog caused by the common effectof stratosphere and radiation cooling is the stratosphereradiation fog. The fog frequency, compared with that in thespring, is on the overall rise, which shows that humidconditions are the most important factors behind the foggeneration. However, the inversion layer formed in thesummer night is very easily damaged due to solar radiationintensity, so the summer fog can only last for a very shortduration. Polar-orbiting satellites in the application ofstatistical methods, limited by transit time, cannot providestatistics on a daily duration of the fog, so in terms of thefrequency, fog appears most frequently in this season.Meanwhile, it is noticeable that there are certain negativecorrelation between the occurrence of fog and urban heatisland effect.

4.2.3. The analysis of the characteristics of the heavy fogfrequency distribution in the autumn

According to the statistical results of the fog distribution(see Fig. 12) in each autumn (September to December) from2001 to 2005, the higher fog frequency region in Beijing ismainly located in the south–east, while in the east–central–northern part the fog frequency is lower. In terms of thecounty distribution, in Daxing County, Tongzhou District, thenortheast of Huairou District, the foggy days last 20–40 days.In Changping District, Shunyi, east south central part inHuairou District, western Miyun County, west of PingguDistrict, northernHaidian district, northern Chaoyang district,and eastern Fangshan District, the fog frequency in theautumn is between 15 and 20 days. In the western part of

Fig. 12. The statistical figure of the fog frequency dist


both Fangshan and Mentougou District, the foggy days areless than 10 days, in many other districts it is even below15 days. In terms of the distribution of the highway, the fogfrequency within the Sixth-loop-line is more than 10 days.

Seen from the surface humidity (Fig. 13), there is a drycenter in north Hebei in the autumn. The humidity was in agradient uniform distribution with the south higher up andthe south lower down and relative humidity is between 54and 60% in general. The surface temperature was also in thepattern of being high in the northwest stretching to be low inthe northwest (Fig. 14) High surface temperature was alsothe southeast–northwest low pattern (Fig. 14), with thesurface temperature ranging from 8 to 11 °C. Compared withthe spring, temperature and humidity have no obvious high-value center. In the wind field, the prevailing westerlyairstream is not conducive to the southeast water vapor'snorthward transportation. Seen from the seasonal averagefield of the temperature and humidity of the surface windmentioned above (Fig. 15), the occurrence of a high fogsituation in the northern part of Beijing with a low humidityenvironment is the result of surface water. The high incidencelocation is to the northwest of Miyun Reservoir. Based on thebackground analysis, this may be related with the MiyunReservoir vapor, and the special mountainous terrain condi-tions as well as the valley wind direction and cooling effect.The region with a high fog frequency in the north part ofBeijing is caused by the radiation cooling effect in themountainous areas, and it should be classified as radiationfog. The region with a high fog frequency in the southeastBeijing, southern and eastern parts of Hebei should be theresult of the water vapor transportation caused by theprocessing east air stream, and it is mainly advection fog.The high incidence location in the north–central regionwhich

ribution in the autumn in Beijing (2001–2005).



Fig. 13. The figure of the surface relative humidity in the autumn in north China (2001–2005).


ARTICLE IN PRESS

is to the north of Beijing City is the result of a combination ofstratosphere and radiation cooling effect.

4.2.4. The analysis of the characteristics of the heavy fogfrequency distribution in the winter

According to the statistical results of the fog frequencydistribution in Beijing in the winter (December to February the

Fig. 14. The figure of the average surface distribution


next year) from 2001 to 2005 (see Fig. 16), the areas with ahigher fog frequency in Beijing are mainly located in thesoutheast. Contrary to the other 3 seasons, in the north east,there are fewer foggy days, and the number in all the counties isbelow 15. In terms of the specific county distribution, the foggydays in the winter are more than 5 days in Daxing County,Tongzhou District, eastern Fangshan District, Fengtai District,

in the autumn in north China (2001–2005).



Fig. 15. The figure of the distribution temperature of the wind direction in the autumn in north China (2001–2005).


ARTICLE IN PRESS

southern Shijingshan District, southern Chaoyang district,southern Shunyi District, southern Pinggu District, part ofChangpingDistrict, part ofHaidianDistrict, and especially in thesouthern part of Daxing County, the number is above 10. InMentougou District, western Changping District, Yanqing

Fig. 16. The statistical figure of the fog frequency dist


County, most part of Huairou District, most part of MiyunCounty, and the north part of Pinggu District, there are fewerfoggy days, generally below 3; while in other districts thenumber is usually between 3 and 10 days. In terms of thehighway loop line distribution, in the urban areas within the

ribution in the winter in Beijing (2001–2005).



Fig. 17. The figure of the surface relative humidity in the winter in north China (2001–2005).


ARTICLE IN PRESS

sixth-loop-line, the fog frequency is generally within 10 days,and the fog frequency in the south of the central urban area ishigher than that in the north.

In the perspective of the relative humidity (Fig. 17), therelative humidity is higher in the winter in Beijing, and thefigure is about 70%. The surface temperature is generallybetween −5 and 11 °C (Fig. 18). The main wind direction is

Fig. 18. The figure of the average surface temperature dist


northwest (Fig. 19).The three surface elements above areevenly distributed in the gradient field. As a result of the lowhumidity, the fog frequency in Northern Beijing is obviouslylower, and the chance of radiation fog is very slim. In contrast,in the southeast of Beijing, there is a high incidence zone,which is closely linked to the corresponding high incidencezone in the southern part and eastern part of Hebei. The

ribution in the winter in north China (2001–2005).



Fig. 19. The figure of the distribution of the wind direction in the winter in China (2001–2005).


ARTICLE IN PRESS

formation of this high incidence zone is due to the watervapor transportation caused by the east processing airstream,and this type of fog belongs to the advection fog. Meanwhile,although the fog frequency in the winter is lower than that inthe other seasons, due to the weaker solar radiation energyand a thicker inversion layer in the winter, the fog durationusually lasts for a whole day, which impairs the transporta-tion as well as the health of mankind. In this sense, the impactof a fog occurrence in the winter and its impression uponpeople outweighs the total of the impact and impression thatthe fog of the other seasons shed upon people.

4.2.5. The analysis of the main characteristics of the heavy fogfrequency in terms of the seasonal distribution

The comparative analysis of the fog frequency in thespring, summer, autumn and winter brings to light thefollowing characteristics of the fog frequency in Beijing interms of the seasonal distribution.

Firstly, the southeast of Beijing is the major part in termsof fog distribution. In each of the four seasons, the fogfrequency is either the highest or among the higher ones. Inthe east–central and northern part, fog also occurs veryfrequently in 3 seasons except in winter. The causes of the fogoccurrence in the two parts mentioned above are totallydifferent. The fog in the southeast is mainly caused by thetransportation of the water vapor above the eastern andsouthern sea to the cold ground surface to cool or direct fogtransportation into the stratosphere, and as a result the fog iscalled advection fog. The fog occurrence in the northmountainous part is due to the local radiation cooling effect,and this type of fog is named radiation fog. In Beijing, in thewinter, due to the freezing condition, the poor water vaporcondition lessens the frequency of the fog occurrence.


Secondly, from the angle of seasonal distribution, inBeijing, in the summer and autumn, the fog takes placemost frequently. In contrast, the fog frequency in the winterand spring is the lowest; however, the fog is usually theadvection fog which has a great impact.

Thirdly, except in the summer, the fog frequency is lowerin major urban areas than that in the outskirts.

Lastly, in the west part of Beijing, the fog frequency is thelowest in each of the four seasons, and themain reasons are asfollows:

The advection fog from the east and the south is blockedby mountains, thus it is unable to enter Beijing. The west airstream prevails near the surface of the ground. Moreover, dueto the descending effect of the air currents from the LoessPlateau, a strongwest surfacewindwhich is both cold and dryis easily formed. Due to the strip of mountainous terrain andthe lack of a huge natural water supply, the temperature andhumidity conditions for the radiation fog are hard to form.

5. Discussion

Generally speaking, the use of satellite data in analyzingfog frequency is superior to conventional data (Westcott andIsard, 2001). With wide range coverage and uniformobservation points, the satellite data can better reflect theactual distribution of fog from a macro aspect and thus givesmore representative and practical fog frequency. However, touse satellite data in fog frequency analysis brings about thefollowing questions as well:

Constrained by the satellite transit time every day,although we have selected the most representative satellitetransit orbit, we are still not sure about the duration of the fogprocess and the fog process in the non-transit time.




ARTICLE IN PRESS

Since dense fog is very close to low clouds on the physicalcharacteristics, satellite information can't distinguish all the fogfrom low-cloud clearly. Moreover, due to the sparsity ofconventional meteorological observation stations as well asdifferencesbetweenmeteorological observation timeand transittime and other factors, there is a certain gap between the resultsof satellite remote sensingandconventional observationstationsin fog frequency statistics (Nauss et al., 2005).

When there are other types of high clouds over the fogzone, fog monitoring is impossible.

When the fog zone is located between two tracks, theidentification of fog will be affected to a certain extent, withinformation missing at part of the time.

The speed of fog dissipation is fast, therefore when thesatellite transits, the fog may have begun to disperse and thussatellite observations of the foggy area are not necessarily thelargest, which will have some impact upon the frequencystatistics.

The fog frequency statistics attained by satellite does notnecessarily mean that there is fog; due to the effects ofrainfalls and sandstorms, it is also very likely that the lowground visibility may take place. Therefore, statistics of theprobable fog frequency attained by satellite may be higherthan that attained at the observation stations.

Therefore, the above problems will affect the accuracy of fogfrequency statistics with the use of satellite. Nevertheless, theresults of satellite data and the station observations under lowvisibility areon the same trend; therefore, the results are credible.

6. Conclusion

In summary, according to a total of five years of fogfrequency statistics analysis based on satellite remote sensingfrom 2001 to 2005, fog in Beijing possesses the followingcharacteristics:

Dense fog occurs throughout the year in Beijing. Thedistribution has obvious seasonal characteristics.The fog zone frequented in the southeastern, central andeastern parts (a little closer to the north )of Beijing, and isrelatively rare in the west .Influenced by ground terrain and environmental flow, fogin the southeast region is mainly laminar fog, while fog inthe north and northwest region is mainly radiative fog.Frequency of occurrence of radiation fog is closely relatedto the types of surfaces underneath. Urban heat island hasevident influences upon the fog.

Acknowledgements

This research work was mainly supported by the NationalNatural Science Foundation of China (contract number40675082), the public (Meteorological sector) researchspecial funds (GYHY200806027), and the Beijing MunicipalScience & Technology Commission (contract numbersH020620190091–H020620250230). The authors would liketo thank the participating personnel, as well as expressing our


appreciation to anonymous reviewers for their constructivecomments on the manuscript.

References

Ackerman, B., 1987. Climatology of Chicago area urban rural differences inhumidity. J. Clim. Appl. Meteor. 26, 427–430.

Anthis, A.I., Cracknell, A.P., 1999. Use of satellite images for fog detection(AVHRR) and forecast of fog dissipation (METEOSAT) over lowlandThessalia. Int. J. Remote Sens. 20 (6), 1107–1124.

Bendix, J., 1995a. Ground fog or low level stratus: decision making usingAVHRR data? Proc. of the 1995 Meteorological Satellite Data Users’Conference, Winchester, UK, EUM P17, pp. 385–392.

Bendix, J., 1995b. A case study on the determination of fog optical depth andliquid water path using AVHRR data and relations to fog liquid watercontent and horizontal visibility. Int. J. Remote Sens. 16, 515–530.

Bendix, J., 1998. Observing fog from space: a European example[C]. Proc. 1stInternational Conference on fog and fog collection, pp. 421–424.

Bendix, J., 2001. A 10-years fog climatology of Germany and the Alpine regionbased on satellite data—preliminary results [C]. Proc. 2nd InternationalConference on fog and fog collection, pp. 357–360.

Bendix, J., Bachmann, M., 1991. A method for detection of fog using AVHRR-imagery of NOAA satellites suitable for operational purposes. Meteorol.Rundsch. 43, 169–178.

Bendix, Jorg, Berthmann, Frank, Reudenbach, Christoph, 1999. NOAA-AVHRRand 4D GIS—towards a more realistic view of fog clearance. IEEE 1999International Geoscience and Remote Sensing Symposium (IGARSS'99),Hamburg, Germany, pp. 2235–2237.

Bergot, T., Guedalia, D., 1994. Numerical forecasting of radiation fog. Part I:numerical model and sensitivity tests. Mon.Weather Rev. 122, 1218–1230.

Bott, A., Sievers, U., Zdunkowski, W., 1990. A radiation fog model with adetailed treatment of the interaction between radiative transfer and fogmicrophysics. J. Atmos. Sci. 47, 2153–2166.

Brown, R., Roach, W.T., 1976. The physics of radiation fog II. A numericalstudy. Q.J.R. Meteorol. Soc. 102, 335–354.

Chen, P.Y., Srinivasan, R., Fedosejevs, G., Narasimhan, B., 2002. An automatedcloud detection method for NOAA-14 AVHRR data for Texas, USA. Int. J.Remote Sens. 23, 2939–2950.

Ellrod, G.P., 2000. Proposed improvements to the night time GOES fogproduct to provide ceiling and visibility information. 10th Conference onSatellite Meteorology and Oceanography. Friday, 14 January.

Gaffen, D.J., Ross, R.J., 1999. Climatology and trends of U.S. surface humidityand temperature. J. Climate 12, 811–828.

Greenwald, J.G., Christopher, S.A., 2000. The GOES I-M imagers: new tools forstudying microphysical properties of boundary layer stratiform clouds.Bull. Am. Meteorol. Soc 81, 2607–2619.

Gurka, J.J., 1974. Using satellite data for forecasting fog and stratusdissipation. 5th Conference on Weather Forecasting and Analysis, pp.54–57. Preprints.

Gurka, J.J., 1978. The role of inward mixing in the dissipation of fog andstratus. Mon. Weather Rev. 106, 1633–1635.

Gustafson, A.V., Wasserman, S.E., 1976. Use of satellite information inobserving and forecasting fog dissipation and cloud formation. Mon.Weather Rev. 104, 323–324.

Nakajima, T.Y., Nakajima, T., 1995. Wide-area determination of cloudmicrophysical properties from NOAA AVHRR measurements for FIREand ASTEX regions. J. Atmos. Sci. 52, 4043–4059.

Nauss, T., Kokhanovsky, A.A., Nakajima, T.Y., Reudenbach, C., Bendix, J., 2005.The intercomparison of cloud retrieval algorithms. Atmos. Res. 78,46–78.

Rosenfeld, Daniel, Cattani, Elsa, Melani, Samantha, Levizzani, Vincenzo, 2004.Considerations on daylight operation of 1.6-versus 3.7-µm channel onNOAA and metop satellites. Bull. Am. Meteorol. Soc. 85 (6), 873–881.

Wang, J.L., Liu, X.L., 2006. The discuss on relationship between visibility andmass concentration of pm2.5 in Beijing. Acta Meteorol. Sin. 64 (2),221–228.

Wang, J.L., Xie, Z., Zhang, Y.H., et al., 2004. The research on quantificationalrelationship between mass concentration of fine particles and weatherconditions in Beijing. Acta Meteorol. Sin. 62 (1), 104–110.

Westcott, N., Isard, S., 2001. Dense fog frequency in the North Central UnitedStates. 2nd International Conference on Fog and Fog Collection, St. John's,New Foundland, pp. 387–390. Preprint.

Wu, X.J., Cheng, Y.H., Li, S.M., 2005. Utilizing MODIS Data to Retrieved theVisibility and Microphysical Properties of Fog Happens in The NorthwestChina, vol. 9(6), pp. 688–696.



2010 - An Analysis of the Fog Distribution in Beijing for the 2001 2005 Period Using NOAA and FY...

Documents

Transcript of 2010 - An Analysis of the Fog Distribution in Beijing for the 2001 2005 Period Using NOAA and FY...