Extremes of the ENSO phenomenon and Indian summer monsoon rainfall

INTERNATIONAL JOURNAL OF CLIMATOLOGY

Int. J. Climatol. 18: 775–791 (1998)

EXTREMES OF THE ENSO PHENOMENON AND INDIAN SUMMERMONSOON RAINFALL

R.P. KANE*Instituto Nacional de Pesquisas Espaciasis—INPE, Caixa Postal 515-12201-970-Sao Jose dos Campos, Sao Paulo, Brazil

Recei6ed 4 June 1997Re6ised 10 September 1997

Accepted 19 September 1997

ABSTRACT

Characterizing every year during the 120 year interval 1871–1990 as a year of El Nino (EN), or Southern Oscillationminimum (SO), or equatorial eastern Pacific sea-surface temperature (SST) warm (W) or cold (C) episode or none(non-events), the corresponding summer monsoon rainfall departures for all India and for the 29 meteorologicalsubdivisions were examined. The best relationship for droughts was with unambiguous ENSOW (El Nino year withSO and W near the middle of the calendar year) and for floods with C (cold SST). The droughts were generallywidespread, although Assam and Bengal might have had normal rainfall or even floods when other subdivisions haddroughts. In some ENSOW years when all India rainfall was normal, the rainfall in subdivisions was either normalor mixed (droughts in some subdivisions, floods in others). However, droughts and floods occurred during other typesof events also, and ENSOW or C were neither sufficient nor necessary. Some floods and droughts were associatedwith incorrect type (floods during El Nino, etc.), and some occurred during non-events, indicating that factorsunrelated to EN, or SO, or W, or C may be more influential in some years. © 1998 Royal Meteorological Society.

KEY WORDS: rainfall; drought; India; El Nino; Southern Oscillation; ENSO

1. INTRODUCTION

India is a vast country, with rainfall regimes varying greatly from region to region and year to year. Asobserved long ago by Normand (1953), ‘India as a whole is too large to be treated as a single unit. Someareas are negatively correlated with others, for example the monsoon rainfall of Bengal and Assam withBombay and Central India’. The major rainfall occurs during summer (June–September) and can be aslow as 250 mm (West Rajasthan) or as high as 2800 mm (Coastal Karnataka). Also, the year to yearvariability (CV, coefficient of variation, i.e. percentage standard deviation) is quite large, more so for lowaverage rainfall regions (e.g. 39% for West Rajasthan, 18% for Coastal Karnataka).

Regarding the association with warm water episodes and the Southern Oscillation, Pant andParthasarathy (1981) reported a correlation coefficient of 0.59 between June and August area-averagedprecipitation over India and the Southern Oscillation Index (SOI) developed by Wright (1977). Rasmus-son and Carpenter (1983) studied the relationship between equatorial Pacific warm episodes (El Ninos)and the summer monsoon rainfall from 31 Indian subdivisions and reported a strong tendency for a belownormal rainfall during the 25 moderate/strong El Ninos that occurred during 1875–1979.

The El Ninos, the Southern Oscillation (of which Tahiti minus Darwin atmospheric pressure is a usefulindex) minima and the sea-surface temperature (SST) increase in the equatorial eastern Pacific have allbeen shown to be related with Indian summer monsoon rainfall deficits (see e.g. Rasmusson andCarpenter, 1983; Khandekar and Neralla, 1984; Kiladis and Diaz, 1989; Mooley and Paolino, 1989).

* Correspondence to: Instituto Nacional de Pesquisas Espaciasis—INPE, Caixa Postal 515-12201-970-Sao Jose dos Campos, SP,Brazil; e-mail: [email protected]

Contract grant sponsor: FNDCT, Brazil; Contract grant number: FINEP-537/CT

CCC 0899–8418/98/070775–17$17.50© 1998 Royal Meteorological Society

R.P. KANE776

However, these parameters do not always occur simultaneously (Deser and Wallace, 1987). Hence, theevents selected by these workers do not always tally between themselves. Also, not all these selected eventsshow the expected results. For example, from the 25 events chosen by Rasmusson and Carpenter (1983),mainly using the El Nino lists of Quinn et al. (1978), Quinn et al. (1987), nine were associated withdroughts and another 11 were associated with slightly below normal rainfall (total 20) in all India summermonsoon rainfall; but five were associated with above normal rainfall. Similarly, from the 27 eventsselected by Mooley and Paolino (1989), using eastern equatorial Pacific positive sea-surface temperatureanomalies (increases above normal), 24 showed below normal rainfall but three showed above normalrainfall. From the 27 events selected by Kiladis and Diaz (1989), based on minima in the SouthernOscillation Index (Tahiti minus Darwin atmospheric pressure), 23 showed below normal rainfall but fourshowed above normal rainfall. In addition, not all the events in the three lists tallied, although all areexpected to be representative of the ENSO phenomenon. Obviously, something was inappropriate in someof these events. Hence, a finer classification of these events was considered advisable. This was done onthe basis of the presence or absence of El Ninos, Southern Oscillation minima, Pacific SST maxima andthe phases of the latter two. Each of the 120 years (1871–1990) were examined to check as to whether itwas a year of El Nino (EN), or a Southern Oscillation (SO), or a sea-surface temperature anomaly(W=warm, C=cold), or a combination of any or all of these or none, with quantitative criteria asmentioned in the next section. Several categories were observed, namely ENSOW, ENSO, ENW, ENC,EN, SOW, W, SO, SOC, C and non-events. In this paper we examine for each of these categories therainfall characteristics in the 29 subdivisions of India.

2. DATA

Figure 1 shows the latest meteorological subdivisions of contiguous India. The summer monsoon rainfalldata for these were obtained from Parthasarathy et al. (1987, 1992), updated by personal communication.

Figure 2 illustrates the categorization of the various years as having an El Nino (EN) (Quinn et al.,1978, 1987) and/or Southern Oscillation Index minimum (SO) and/or warm (W) or cold (C) events of thesea-surface temperature (SST) of equatorial eastern Pacific, in four panels for 1871–1901, 1901–1931,1931–1961 and 1961–1991. The SOI and SST indices were available as seasonal means. However, theseasonal values had wide fluctuations from season to season (probably due to an annual component).Hence, the values were smoothed by obtaining 12-month running means. These showed smooth, slowchanges and it was possible to locate maxima and minima properly. In each panel, the top plot is for12-month running means of the Southern Oscillation Index (SOI), given by Wright (1977) and obtainedby principle component analysis of seasonal mean atmospheric pressure at eight locations. This index isavailable up to 1974 only. However, Chen (1982) recommended a simpler index, namely Tahiti (T) (18°S,150°W) minus Darwin (D) (12°S, 131°E) mean sea-level pressure difference (T−D), available in Parker(1983) for 1935–1983 and further in Meteorological Data Reports. We found that the 12-monthly meansof (T−D) and Wright’s SOI correlated very well and we obtained a regression equation converting(T−D) into SOI. For 1974 onwards, the SOI shown in panel 4 of Figure 2 represents these convertedvalues. However, for the previous period (1935–1974), we compared the 12-month running means of(T−D) with the Wright SOI and found that the minima of the two tallied almost exactly. We alsoextended the comparison to the earlier period (1876–1935), using the (T−D) values published byRopelewski and Jones (1987). Again, the 12-month running means matched very well with the WrightSOI. As a quantitative limit, SOI minima exceeding −1 (Wright’s dimensionless unit) and SST maximaexceeding +1°C were chosen, but these criteria were superfluous. Because of the 12-month smoothing,only prominent events are seen in Figure 2 and all of these had variations exceeding the above limits.

The next plots in Figure 2 show the SST index of Wright (1984) for the eastern equatorial Pacific(6°N–6°S, 180°N–90°W) and a similar index reported by Angell (Angell (1981), and further personalcommunication) for the same region. The two plots show similar maxima and minima and supplementeach other for missing data. El Ninos are warm water episodes developing along the Peru–Ecuador coast

© 1998 Royal Meteorological Society Int. J. Climatol. 18: 775–791 (1998)

ENSO EXTREMES AND INDIAN MONSOON RAINFALL 777

in some years. Usually, these appear during November–December (hence the term El Nino, meaning, TheChild, referring to the birth of Jesus Christ), develop in the first few months (January and February),spread in the equatorial eastern Pacific and, by the year end, usually disappear. Quinn et al. (1978, 1987)have examined these events and denoted the years of occurrence as those during the January to June ofwhich El Nino was developing or had already developed, and have also denoted the El Ninos strengthsdepending upon the SST anomaly along the Peru–Ecuador coast (\2.9°C is strong, S; 2.0–2.9°C ismoderate, M; B2.0°C is weak, W). Thus, whenever El Nino exists, it is often there at the beginning of

Figure 1. Latest meteorological subdivisions of contiguous India: 2, Arunachal Pradesh; 3, North Assam; 4, South Assam; 5,Sub-Himalayan West Bengal; 6, Gangetic West Bengal; 7, Orissa; 8, Bihar Plateau; 9, Bihar Plains; 10, East Uttar Pradesh; 11, WestUttar Pradesh Plains; 12, West Uttar Pradesh Hills; 13, Harayana; 14, Punjab; 15, Himachal Pradesh; 16, Jammu and Kashmir; 17,West Rajasthan; 18, East Rajasthan; 19, West Madhya Pradesh; 20, East Madhya Pradesh; 21, Gufarat; 22, Saurashtra and Kutch;23, Kokan and Goa; 24, Madhya Maharashtra; 25, Marathawada; 26, Vidarbha; 27, Coastal Andhra Pradesh; 28, Telengana; 29,Rayalseema; 30, Tamilnadu; 31, Coastal Karnataka; 32, North Karnataka; 33, South Karnataka; 34, Kerala. Date for regions 2, 12,

15 and 16 are not considered for analysis


R.P. KANE778

Figure 2. Plots of 12-monthly running means of Southern Oscillation Index (SOI) and sea-surface temperatures in equatorial easternPacific (SST) for 1871–1991 (panels 1–4) and the characterization of each year as having El Nino (EN) and or SOI minima (SO)and/or warm (W) or cold (C) Pacific SST (rectangles). For years having El Nino, the symbols above the rectangles indicate strength

(S, strong; M, moderate; W, weak) of the El Nino. SOI minima and SST maxima are shaded black

the calendar year and could be a preceding warning signal of a few months for locations where majorrainfalls occur later in the year (e.g. June–September for the Indian summer monsoon). Because El Ninoyears usually coincide with SOI minima and equatorial eastern Pacific SST maxima, the SOI minima andSST maxima are shown in black in Figure 2, and SST minima representing cold (C) SST events are shown



as hatched. At the bottom of each panel in Figure 2, the rectangles show the combination (EN, SO, C,etc.) applicable to that year and, for years having an El Nino (EN), the symbols S, M, W at the top ofthe rectangle indicate the El Nino strength. Years without a rectangle are non-events. As expected, mostof the SOI minima and SST maxima (blacks) coincide with El Ninos, but a few do not coincide (Deserand Wallace, 1987).

3. ENSO RELATIONSHIP WITH RAINFALL OF SUBDIVISIONS

3.1. ENSOW years

Among the 120 years (1871–1990), there were 30 ENSOW events. Of these, 16 were unambiguous, witha strong or moderate El Nino (generally in the early part of the calendar year) and SO minima and Win the middle of the calendar year (Indian summer), and 14 were ambiguous, i.e. although there was astrong or moderate El Nino, the SO minima and/or W were in the early or late part of the calendar year.The all India summer monsoon rainfall (I) and the rainfall in the various subdivisions (exceeding onestandard deviation) in the ENSOW years are shown in Figure 3. Positive and negative deviations within91 S.D. (standard deviation) are shown as + or − , whereas deviations exceeding 1 S.D. are shown bytriangles (floods) and circles (droughts).

Figure 3(a) refers to the unambiguous ENSOW events. As can be seen, 11 of the 16 events areassociated with droughts (rainfall exceeding one standard deviation) in all India rainfall (last column),which seem to be fairly widespread, although in most cases, Assam (3 and 4) and West Bengal (5 and 6)seem to have an opposite behaviour (excess rains), confirming the earlier observations of Normand (1953)and many other workers. For the other five events, the all India rainfall was normal, and for theindividual years the subdivisions showed the following.

(i) Year 1896: Assam (3 and 4) and Sub-Himalayan West Bengal (5) showed droughts, accompanied byBihar Plains (9), Uttar Pradesh (10 and 11), Haryana (13), Punjab (14), Telangana (28) andRayalseema (29), but Orissa (7), Bihar Plateau (8), Gujarat (20) and South Karnataka (33) had excessrains.

(ii) Year 1902: Assam (3), West Bengal (5) and coastal Karnataka (31) had excess rains, whereas onlyVidarbha (26) and Telangana 928) had droughts.

(iii) Year 1930: only Sub-Himalayan West Bengal (5), Tamilnadu (3), South Karnataka (33) and Kerala(34) showed mild droughts, whereas elsewhere rainfall was normal (within one standard deviation).

(iv) Year 1957: only mild droughts were seen in parts of Assam (4), Bengal (5), Bihar (8), East MadhyaPradesh (20), Gujarat (21), Tamilnadu (30) and South Karnataka (33), excess rains occurred incoastal Andhra Pradesh (27), and normal rainfall elsewhere.

(v) Year 1976: only Bihar (6 and 7), South Karnataka (33) and Kerala (34) had droughts, whereas partsof Assam (3), Bihar (8), Haryana (13), Punjab (14), West Rajasthan (15) and Gujarat (21) had excessrains.

Thus, for these five unambiguous ENSOW events, rainfall in the various subdivisions was either mostlynormal or mixed (excess in some subdivisions, deficit in others). In Figure 3(a), the pairs of the numbersat the end of each row (each event) show the number of subdivisions (from a total of 29) having positiveand negative divisions. For example, for the first event of 1877, (2/27) means there were two subdivisionsshowing positive deviations (all sizes (+ ) and triangles) and 27 subdivisions showing negative deviations(all sizes (− ) and circles). Thus, for this event, the droughts were widespread. The ratio of (negative tototal) was 27/29=0.93. This ratio can be termed the drought widespreadness index (DWI), the maximumvalue (1.00) implying droughts all over India (29 subdivision). Figure 6(a) top plot (Panel 1) shows thedistribution of this DWI for the 16 unambiguous ENSOW events. There is a broad peak near 0.80,implying that a large number of events (nine out of 16) had droughts spread over more than 80% of theIndian region. However, other events were less widespread, notably 1976, when only 12 out of 29subdivisions had negative deviations.


R.P. KANE780

At the bottom of each column in Figure 3(a), similar ratios are given for the number of events (froma total of 16 in this case) having negative deviations as compared with total events. For example, for thesubdivision 3, there were ten positive and six negative events, and the ratio (negative of total) was6/16=0.38. Figure 6(b) top plot (Panel 1) shows the distribution of this ratio (drought response index,DRI) among the 29 subdivision. As can be seen, there is a preponderance near +0.80, implying that alarge number of subdivisions (17 out of 29) had more than 80% of events (13 or more, out of 16) showingnegative deviations. Here again, some divisions had very poor showing, notably subdivision 3 (DRI=0.38) and 4 (DRI=0.50) in Assam, which had eight or less events (out of 16) showing droughts.

Figure 3. Rainfall distribution in various subdivisions and all India (I) during (a) unambiguous ENSOW; (b) ambiguous ENSOW;and (c) other events involving EN. In the right extreme, the symbols represent years chosen by Rasmusson and Carpenter (RC;

1983), Mooley and Paolino (MP; 1989) and Kiladis and Diaz (KD; 1989) for warm and cold (underlined) episodes



Figure 3(b) refers to 14 ambiguous ENSOW events where there was an El Nino but the SO minima andwarm events (W) did not occur in the middle of the year but occurred earlier or later. For none of theseevents, the all India rainfall showed droughts. Instead, 1878 and 1983 showed floods. The second plots(Panel 2) in Figure 6(a and b) show DRW and DRI distributions for ambiguous ENSOW. These are verybroad, centred around 0.50, and very different from those for unambiguous ENSOW in Panel 1.

Amongst the 30 ENSOW events, there were seven double events, i.e. ENSOW occurring in twosuccessive years, namely 1877–1878, 1918–1919, 1925–1926, 1930–1931, 1940–1941, 1957–1958 and1982–1983. For these, the first years are marked as I and the second years as II in Figure 3(a and b). Forthese seven double events, the values of DWI for year I were 0.93, 0.83, 0.67, 0.76, 0.59, 0.62 and 0.86,giving an average of 0.7590.12 and for year II, 0.24, 0.55, 0.45, 0.38, 0.86, 0.41 and 0.28, giving anaverage of 0.4590.19. Thus, the spread is significantly less in year II. Four other double events1873–1874, 1896–1897, 1899–1900 and 1911–1912 involving EN (but not ENSOW) also showed lessspread in year II. Thus, year I of such double events is likely to be associated with widespread droughts(ca. 70% region coverage), whereas year II is likely to be less widespread. A glaring exception was1940–1941, where 1940 showed a lower DWI (17/29=0.59) compared with that of 1941 (25/29=0.86).

3.2. Other years with EN

Besides the 30 ENSOW events, there were another 16 events that had an El Nino, either alone or withSO (only) or W or C (only). These are shown in Figure 3(c). There were six ENSO events for whichrainfall was often deficient but only in few subdivisions and generally with excess rains in some othersubdivisions. Thus, the combination ENSO is inclined to yield rainfall below average but not necessarilywith overall great deficits.

There was one ENW (1884) which was expected to yield droughts, due to the effect of both EN andW. There were droughts in some subdivisions, but heavy rains in West Uttar Pradesh Plains (11), MadhyaPradesh (19 and 20), Gujarat (21), Saurashtra-Kutch (22) and Vidarbha (26) resulted in overall rainexcess.

There were 4 years involving only EN. Rainfall deficits or excesses occurred in only a few subdivisions,but was normal elsewhere, as also for all India.

There were five ENC, expected to produce a mixed response (droughts due to EN, floods due to C).Except in 1901, when 19 subdivisions showed negative deviations and ten showed positive deviations, theother events showed widespread floods, indicating that the effect of C prevailed.

Figure 6(a and b), Panel 3, shows the DWI and DRI distributions for these events. The distributionsare broad and centred around 0.50, similar to those for ambiguous ENSOW (Panel 2). For the DWI plot(Figure 6(a), Panel 3), ENC events show low ratios, i.e. a larger number of positive deviations, implyingthe effect of C rather than EN.

3.3. Years with SO or W

Figure 4(a) shows events that had no El Nino but only SO and/or W. There were six SOW events,expected to yield droughts due to both SO and W. In general, this was observed, with 1888, 1904, 1913and 1979 yielding widespread droughts in more than 19 subdivisions. However, in the other two events,the rainfall distribution was mixed, with less than 13 divisions showing droughts.

There were three events of type W and all yielded widespread droughts, as expected. There were threeevents of SO type expected to yield droughts but gave mixed results. There were four SOC events,expected to produce mixed results (droughts due to SO, floods due to C). For 1935, 1936, 1946, therainfall distribution was mixed; but for 1949, excess rains occurred, indicating that the effect of Cprevailed.

Figure 6(a and b), Panel 4, shows the DWI and DRI distributions, which are broad and centred around0.50. In Figure 6(a), Panel 5, the four SOC events show low ratios (preponderance of positive deviations)indicating the overpowering influence of C rather than SO. In Figure 6(b), Panel 5, the distribution forSOW, W and SO only (SOC excluded) is more biased toward higher ratios, as compared with the


R.P. KANE782

Figure 4. Same as Figure 3, for (a) events involving SO and W and (b) non-events

distribution of SOW, W, SO and SOC shown in Figure 6(b), Panel 4, thus again indicating that SOCfavours positive deviations (effect of C rather than SO).

3.4. Cold (C) e6ents

Figure 5 depicts 36 cold (C) events, out of which, 14 were associated with floods, fairly widespread over20 or more subdivisions, and 22 had positive deviations, with floods in many subdivisions but accompa-nied with droughts in other subdivisions. Thus, whereas C years are favourable for widespread excessrains, the excess may not be large and droughts in some regions cannot be ruled out. Incidentally, theseare purely cold (C) events. Other events in which C was associated with EN or SO are included in Figure3(c) (ENC) and Figure 4(a) (SOC), and their characteristics show more an effect of C rather than EN orSO.

Figure 6(a and b), Panel 7, shows the DWI and DRI distributions for C events, and reveals a clear biasfor low ratios, i.e. fewer negative deviations and a preponderance of positive deviations.



3.5. Non-e6ents

From the description given above it would seem that although each category has its own expectedcharacteristics, these are not always depicted and sometimes reverse effects are seen, perhaps because ofother causes unrelated to EN, SO or W or C. It would be interesting to check what happens in years whennone of these existed, i.e. during non-events. Figure 4(b) shows the results for 22 non-events. If EN, SOand W or C were the main causes of droughts and floods, the non-events should have shown mostlynormal (+ , − ) rainfall. Instead, Figure 4(b) shows 2 years (1876 and 1901) when droughts werewidespread, and 2 years (1947 and 1961, especially the latter) when floods were widespread. In many othercases, rainfall deficit occurred in some subdivisions and excess occurred in others, so much so that Figure4(b) (non-events) looks similar to Figure 4(a) (SO and W events) and Figure 3(c) (other EN events). Thus,only Figure 3(a) (unambiguous ENSOW) and Figure 5 (cold events) stand out, showing a reasonablygood relationship with droughts and floods, respectively.

Figure 6(a and b), Panel 6, shows the DWI and DRI distributions. These are centred at 0.50 asexpected, but the DWI distribution in Figure 6(a), Panel 6, is very broad. Thus, non-events may showextreme rainfall events due to other causes. In any case, for our purpose of ENSO study, these arenon-events. The distributions of Figure 6(a and b), Panels 1–4, were compared with those of Panel 6. Ax2-test showed Panel 1 distributions were significantly different from those of Panel 6 at better than the

Figure 5. Same as Figure 3, for cold (C) SST events


R.P. KANE784

Figure 6. Frequency distributions of (a) drought widespreadness index (DWI) and (b) drought response index (DRI) forunambiguous ENSOW (Panel 1), ambiguous ENSOW (Panel 2), other EN events (Panel 3), events with SO and/or W (Panels 4 and

5), non-events (Panel 6) and cold (C) SST events (Panel 7)

99% confidence level. Panels 2, 3 and 4, however, were not significantly different from Panel 6. On theother hand distributions of Panel 7 were significantly different from those of Panel 6, indicating that Cevents are certainly related more to floods, with some notable exceptions, such as 1922 and 1928 whennegative deviations prevailed (large DWI in Figure 6(a), Panel 7).

In Figures 3–5 the symbols + , − , D and O are used, and in Figure 6 the positive deviations (+ ) andthe negative deviations (− ) are used. However, the positive and negative deviations are within 9s andcould be random. More reliable are the D and O symbols, because these exceed the 9s limit. In Figure3(a), the O symbols are frequent and a relationship with droughts is obvious. Nevertheless, to obtain aquantitative estimate, the drought widespreadness index and the drought response index were calculated,using only the symbols O. In case of + and − , the fractions for − were complementary to those for+ , the two adding up to 1.0 (if the fraction of negative deviations was 0.3, the fraction of positivedeviations had to be 0.7, and so on). For D and O, the sum would be any value less than 1.0. Hence, acorresponding flood widespreadness index and flood response index were also calculated. Figure 7 showsthe results, in the right half (c and d) for the response index and in the left half (a and b) for thewidespreadness index. In Figure 7(c), Panel 1 for unambiguous ENSOW shows that most of thesubdivisions show small fractions of positive deviations, D, and in Figure 7(d), Panel 1 shows that mostof the subdivisions show large fractions of negative deviations, O. Thus, unambiguous ENSOW favour Ooverwhelmingly. In Panel 2 for ambiguous ENSOW, Panel 3 for other El Ninos and Panel 4 for SOW,W, SO and SOC, the fractions in (c) and (d) are similar, indicating no preference for positive or negativedeviations. The same is true for Panel 5 for non-events. In contrast, Panel 6 for C events (La Ninas) showspositive deviations (floods) in (c) are larger compared with negative deviations (droughts) in (d). A x2-testshowed that the distributions in Panel 1 and Panel 6 were significantly different from those of Panels 2,



3, 4 and 5, indicating that our classification does indicate more significant relationships with unambiguousENSOW for droughts and C events for floods.

In the case of EN events, could the effect be dependent on the strength of the El Nino? Table I showsthe all India rainfalls for EN events of different strengths. From the 21 events involving strong (S) ElNinos, only seven (all ENSOW) were associated with severe droughts (D), two with mild floods (f), twowith severe floods (F) and ten with normal rainfall in all India. From the 21 events involving moderate(M) El Ninos, only two (both ENSOW) were associated with moderate droughts (d) and only three (allENSOW) with severe droughts (D), one with mild floods (f), one with severe flood (F) and 14 with normalrainfall. From the four events involving weak (W) El Ninos, one (ENSO) was associated with milddroughts and three (all ENSOW) with normal rainfall. Thus, irrespective of whether the El Nino is strongor moderate or weak, more than half (ca. 66%) are associated with normal rainfall or floods. Of theremaining one-third, strong El Ninos seem to be better related to severe droughts (seven) as comparedwith moderate El Ninos (two moderate and three severe droughts), and this difference seems significant,considering the fact that for the 22 non-events, all India rainfall (Figure 4(b)) showed only two severedroughts, one severe and one mild flood, and 18 years of normal rainfall. In any case, the El Ninoseffective for droughts were all in the ENSOW category.

3.6. Years of se6ere droughts

From the 10 years of severe droughts in all India rainfall, 7 years, namely 1877 (−29%), 1899 (−26%),1918 (−24%), 1972 (−23%), 1987 (−19%), 1965 (−17%) and 1905 (−16%) were associated withENSOW, and in these years the droughts were fairly widespread, although occasionally some subdivisions

Figure 7. Frequency distributions of widespreadness indices (a) for floods (FWI) and (b) for droughts (DWI) and of response indices(c) for floods (FRI) and (d) for droughts (DRI) for extreme rainfall deviations only (D and O), for unambiguous ENSOW (Panel1), ambiguous ENSOW (Panel 2), other El Ninos (Panel 3), SOW, W, SO, SOC (Panel 4), non-events (Panel 5) and cold-C events

(Panel 6)


R.P. KANE786

Table I. Indian summer monsoon rainfall (ISMR) percentage deviations for years of strong (S) moderate(M) and weak (W) El Ninos. Symbols d, D, f and F represent mild and severe droughts and floods,

respectively. I and II represent first and second years of double ENSOW events

El Nino strong (S) El Nino moderate (M) El Nino weak (W)

Year Type ISMR (%) Year Type ISMR (%) Year Type ISMR (%)

1871 ENSO −1 1874 ENC +14 F 1873 ENSO −11 d1877 I ENSOW −29 D 1880 ENSO −4 1948 ENSOW +21878 II ENSOW +14 F 1887 ENC +5 1963 ENSOW +11884 ENW +9 f 1889 ENC +9 f 1969 ENSOW −31891 ENSO −7 1896 ENSOW −31899 ENSOW −26 D 1897 EN +41900 ENSO +4 1902 ENSOW −71911 ENSOW −14 D 1905 ENSOW −16 D1912 ENSO −6 1907 ENC −91917 ENC +18 F 1914 ENSOW +61918 I ENSOW −24 D 1919 II ENSOW +41925 I ENSOW −6 1923 ENSOW −41926 II ENSOW +6 1930 I ENSOW −61932 EN −6 1931 II ENSOW +31940 I ENSOW −1 1939 EN −71941 II ENSOW −14 D 1943 EN +21957 I ENSOW −8 1951 ENSOW −13 d1958 II ENSOW +4 1953 ENSOW +81972 ENSOW −23 D 1965 ENSOW −17 D1982 I ENSOW −14 D 1976 ENSOW +11983 II ENSOW +12 f 1987 ENSOW −18 D

showed excess rain. For example, in 1877, 27 subdivisions had deficit rains and two had excess, but in1899, while 23 subdivisions had deficit rains, six had excess rains, mainly in north-east. The same was truefor 1918.

From the 3 years which did not have an ENSOW, 1979 (−17%) was a SOW and 1920 (−16%) wasa W (already discussed). However, 1901 (−16%) had rain deficits in 27 subdivisions and excess in two,and yet 1901 was neither EN nor SO nor W, but a non-event.

From the 10 years of mild droughts in all India rainfall, 4 years, namely 1941 (−14%), 1911 (−14%),1982 (−14%) and 1951 (−14%) were associated with ENSOW. From the rest, 1986 (−13%) and 1968(−12%) were W, 1974 (−12%) was a SO, 1904 (−12%) was a SOW, 1873 (−12%) was an ENSO and1966 (−14%) was a non-event. Thus, 1901 and 1966 seem to be droughts not related to EN, SO or W.

3.7. Years of se6ere floods

From the 10 years of severe floods in all India rainfall, 6 years, namely 1988 (+16%), 1892 (+16%),1956 (+16%), 1933 (+14%), 1894 (+14%) and 1975 (+13%) were associated with a C (Cold SST). Twoyears 1917 (+18%) and 1874 (+14%) were ENC, indicating that the El Ninos involved (moderate andstrong) could not prevent a flood, and the effect of C prevailed. The flood year 1878 (+14%) was anENSOW. However, it was the second year of the double ENSOW event 1877–1878.

The severest flood, of 1961 (+19%), was a non-event. It occurred in 25 subdivisions, but the north-east(3–5) showed rainfall deficits.

From the 10 years of mild floods, six were associated with C (1942, 1893, 1916, 1970, 1910, 1955). Year1983 was an ENSOW; but it was the second year of the double ENSOW event 1982–1983. Year 1884 wasan ENW and should have been a drought year, both because of a strong El Nino and a warm SST. Itwas, however, an overall mild flood (+9%) year, with 14 subdivisions showing excess rains, whichincluded the very severe floods in West Uttar Pradesh Plains (11), Madhya Pradesh (19 and 20), Gujarat(21), Saurashtra-Kutch (22) and Vidarbha (26), and 15 showing droughts, in north-east and south India.



The year 1959 (+10%) was a SO (which should have resulted in a drought) and the year 1947 (+11%)was a non-event. In both, 20 or more subdivisions showed excess rains and none showed a deficit in 1947,whereas nine showed droughts in 1959.

It would thus seem that although a cold event (C) was favourable for excess rains, floods couldnevertheless occur in any part of the country, and sometimes even in the presence of an El Nino or W orSO minimum.

4. COMPARISON WITH RESULTS OF OTHER WORKERS

Rasmusson and Carpenter (1983) used 25 warm episode years, Mooley and Paolino (1989) used 27 warmphase and 23 cool phase years, and Kiladis and Diaz (1989) gave a list of warm and cold episodes basedon extremes in the Southern Oscillation. Most of their warm episodes tally with the listing of Quinn et al.(1987). These are shown in the last columns of Figures 3–5, as RC, MP and KD. Table II lists theseevents and the corresponding ISMR (%) deviations. The following may be noted.

(i) Rasmusson and Carpenter (1983) chose 14 of the 16 unambiguous ENSOW (Figure 3(a), five of the14 ambiguous ENSOW (Figure 3(b)) and six of the 16 other EN events. They have excluded thesecond years of double events (rightly so, as these give floods instead of droughts). Still, from their25 events, only nine gave mild (one) and severe (eight) droughts. One event (1884) gave mild floodswhereas 15 events were of normal rainfall (11 negative, four positive deviations). Thus, therelationship of El Nino with droughts was loose, similar to what we have depicted.

(ii) Mooley and Paolino (1989) chose 15 of the 16 unambiguous ENSOW (Figure 3(a), five of the 14ambiguous ENSOW (Figure 3(b), five of the other EN, SO or W events and two non-events as their27 warm phase years. These have an association of only 13 mild (four) and severe (nine) droughtswhereas 14 events were of normal rainfall (11 negative, three positive deviations), again depicting aloose relationship between warm phase years and droughts.

(iii) Kiladis and Diaz (1989) chose 13 of the 16 unambiguous ENSOW (Figure 3(a), five of the 14ambiguous ENSOW (Figure 3(b) and nine of the other EN, SO or W events as their 27 warm phaseyears. These are associated with only nine droughts (three mild, six severe). One event (1884) wasassociated with mild floods, whereas 17 events were of normal rainfall (13 negative, four positivedeviations), depicting a loose relationship between warm episodes and droughts.

(iv) Mooley and Paolino (1989) chose 15 of our C events (Figure 5), two ENSOW (1878, 1983, secondyears of the double events 1877–1878, 1982–1983), four EN related events, one SO and one SOC astheir 23 cooling phase years. From these, only nine are associated with mild (6) and severe (3) floods,whereas 14 events were of normal rainfall (13 positive, one negative deviations). Thus, only smallexcess rainfalls are encountered.

(v) Kiladis and Diaz (1989) chose 16 of our C events (Figure 5), one ENSOW (1931, second year of thedouble event 1930–1931), one ENC (1889), one SOC (1949) and one W (1920) as their 20 coldepisodes. The W event (1920) was a wrong choice as it was expected to give a drought and did so(−16%). The year 1928 (C) was also a mild drought year. From the 20 events, only six gave mild(five) and severe (one) floods, 12 events were of normal rainfall (all positive deviations) and two weredroughts.

(vi) Besides EN, SO and W or C, other factors have been reported to affect the Indian summer monsoon.Hahn and Shukla (1976) reported an inverse relationship with Eurasian winter snow cover. Duringthe 9-year interval 1967–1975, the snow cover was above average during 1968, 1972, 1973 and 1974,and the rainfall was below normal for 1968, 1972 and 1974, indicating a reverse relationship.However, our classifications for these years were W, ENSOW and SO respectively, all of which werefavourable for droughts. Thus, the role of snow cover is dubious. The year 1973 was a C event(favourable for excess rain) but turned out normal, which could be interpreted as due to contradic-tory effects of C and excess snow cover. During the years 1967, 1969 and 1970, the Eurasian snow


R.P. KANE788

Table II. Indian summer monsoon rainfall (ISMR) percentage deviations for years chosen by Rasmusson andCarpenter (RC) (1983), Mooley and Paolino (MP) (1989) and Kiladis and Diaz (KD) (1989) as warm and coldepisodes. Symbols d, D, f and F represent mild and severe droughts and floods, respectively. I and II represent first

and second years of double ENSOW events

(a) Warm episodes

Kiladis and Diaz (1989)Rasmusson and Carpenter (1983) Mooley and Paolino (1989)

ISMR (%) YearYear Type ISMR (%) Year Type ISMR (%)Type

−29 D 1876 X −10 dS 1877 I S 1877 I ENSOW −29 DENSOWENSO −4 S 1877 I ENSOW −29 D M 1880 ENSO −4M 1880

+10 fS 1884 1883 X −1 S 1884 ENW +10 fENW+5 1888 SOW −5ENC 1888 SOW −5M 1887

ENSOS 1891 −7 M 1896 ENSOW −3 S 1891 ENSO −7−3 S 1899 ENSOW −26 D M 1896M 1896 ENSOW −3ENSOW

−26 D M 1902 ENSOW −7ENSOW S 1899S 1899 ENSOW −26 DENSOWM 1902 −7 1904 SOW −12 d M 1902 ENSOW −7

−16 D M 1905 ENSOW −16 D 1904 SOWM 1905 −12 dENSOW−14 D S 1911 ENSOW −14 DENSOW S 1911S 1911 ENSOW −14 D+6 1913 SOW −8 1913M 1914 SOW −8ENSOW

−24 D S 1918 I ENSOW −24 DENSOW S 1918 I ENSOW −24 DS 1918 IENSOWM 1923 −4 S 1925 I ENSOW −6 M 1923 ENSOW −4

−6 M 1930 I ENSOW −6ENSOW S 1925 IS 1925 I ENSOW −6ENSOWM 1930 I −6 M 1939 EN −7 M 1930 I ENSOW −6ENS 1932 −6 S 1940 I ENSOW −1 S 1932 EN −6

−7 S 1941 II ENSOW −14 DEN M 1939 EN −7M 1939ENSOWS 1941 II −14 D W 1948 ENSOW +2 M 1951 ENSOW −13 d

−13 d M 1951 ENSOW −13 d M 1953M 1951 ENSOW +8ENSOW+8 S 1957 I ENSOW −8ENSOW S 1957 IM 1953 ENSOW −8

ENSOWS 1957 I −8 W 1963 ENSOW +1 W 1963 ENSOW +1−17 D M 1965M 1965 ENSOW −17 D M 1965 ENSOW −17 DENSOW−3 1968 W −11 dENSOW W 1969 ENSOW −3W 1969

−23 D W 1969 ENSOW −3 S 1972S 1972 ENSOW −23 DENSOW+1 S 1972 ENSOW −23 DENSOW M 1976 ENSOW +1M 1976

M 1976 ENSOW +1 S 1982 I ENSOW −14 DS 1982 I ENSOW −14 D 1986 W −13 d

Events 25 27 27

(b) Cold episodes

Mooley and Paolino (1989) Kiladis and Diaz (1989)

Year Type ISMR (%) Year Type ISMR (%)

+7 1886 C1872 +2C+14 F 1889 ENC +10 fENC1874

ENSOW1878 II +14 f 1892 C +16 F−1 1898 C1885 +3SO+9 f 1903 C +1ENC1889

C1892 +16 F 1906 C +4+4 1908 C +51897 EN+3 1916 C +12 fC1898

ENSO1900 +4 1920 W −16 D+1 1924 C +11903 C+4 1928 C −10 dC1906

C1908 +5 1931 II ENSOW +3+12 f 1938 C +71916 C+7 1942 C +12 fC1938

+12 f 19491942 SOC +6C



Table II. (Continued)

(b) Cold episodes

Mooley and Paolino (1989) Kiladis and Diaz (1989)

Year Type ISMR (%) Year Type ISMR (%)

+41946 SOC +6 1954 C+8C1964+4C1954

+10 fC1970+9 fC19551973 C +71964 C +8

+13 f1967 C +1 1975 C1970 C +10 f

+7C1973+12 fENSOW1983 II

20Events 23

cover was below average (favourable for floods) but a rainfall excess occurred only in 1970, whichwas a C year (favourable for excess rain). Thus the contribution of below average snow in this casealso was dubious. The years 1967 and 1969 had normal rainfall and were C and ENSOW. Thus, for1967, C and below average snow both should have led to excess rainfall; but the rainfall turned outto be normal. For 1969, the normal rainfall could be interpreted as the result of contradictory effectsof ENSOW and snow cover below average.

For the 10 year interval 1971–1980, Dey and Bhanu Kumar (1983) mention 1972, 1978 and 1979 asyears of excess Himalayan winter snow cover. The rainfall was below average for 1972 and 1979; but thesewere years of ENSOW and SOW, both favourable for droughts. Thus, the effect of excess snow cover isdubious. The year 1978 was our non-event, and, in spite of an excess Himalayan snow cover, had onlynormal rainfall.

The years 1971, 1975, 1976, 1977 and 1980 had Himalayan snow cover below average, favourable forfloods, but only 1975 had excess rains and 1975 was a C event (favourable for floods). Thus the effect ofdeficient snow cover was dubious. The year 1971 was a C and, with deficient Himalayan snow cover,should have yielded excess rainfall but gave only normal rainfall. The years 1976 and 1977 were ENSOWand SOW, both favourable for droughts; but the rainfalls were normal, perhaps due to the contrary effectof Himalayan snow cover deficit.

From the common years 1971, 1972, 1973, 1974 and 1975 for the Hahn and Shukla (1976) and Dey andBhanu Kumar (1983) lists, only 1972 had excess snow cover for both Eurasia and Himalaya. This wasalso an ENSOW year. Thus, everything was favourable for droughts and a severe drought (−23%) wasseen in all India rainfall.

5. CONCLUSIONS AND DISCUSSION

Each year in the 120 year interval 1871–1990 was characterized as having an El Nino (EN), Southernoscillation minimum (SO), eastern equatorial Pacific sea-surface temperature warm (W) or cold (C), orany combination of these, or none (non-events). The rainfall amounts in the 29 meteorological subdivi-sions of India as well as all India rainfall during summer (June–September) were examined for eachcategory of years. The following was noted.

(i) For droughts, the best relationship was with ENSOW type years. There were 30 ENSOW events. Ofthese, 16 were unambiguous, i.e. El Nino years with SO minima and W occurring in the middle ofthe calendar year, and 14 were ambiguous, i.e. El Nino years with SO minima and W occurring earlyor late in the year. From the 16 unambiguous ENSOW, 11 were associated with widespread droughts


R.P. KANE790

over India, except in the north-east (Assam, Bengal), where often excess rain occurred. In theother five events, the rainfall was either almost normal everywhere, or a mixed distributionoccurred, i.e. droughts in some regions and floods in others. From the 14 ambiguous ENSOW,the overall rainfall was normal or there was a mixed distribution (droughts in some regions,floods in others), except in two events (1878, 1983) when there were floods. There were sevendouble events, i.e. ENSOW in two successive years, namely 1877–1878, 1918–1919, 1926–1926,1930–1931, 1940–1941, 1957–1958 and 1982–1983 and, in general, the first year was associatedwith widespread droughts whereas the second year had mixed characteristics (deficit in somesubdivisions, excess in others).

(ii) The other 16 events involving EN (three involving W and 13 involving SO minima) gave mixedresults. Thus, ENSO (six), EN (four), SOW (six) and SOC (four) gave normal rainfall ordroughts. The W (three) events also gave droughts. The ENC (five), ENW (one) and SO (three)gave normal rainfall or floods.

(iii) For floods, the best relationship was with C type years (cold SST). From the 36 C events, 12were associated with widespread floods. In the other 22 events, there were floods only in a fewsubdivisions.

(iv) There were 22 non-events, but for these the rainfall was not always normal. In all India rainfallsevere droughts occurred in 1901 and 1966 and floods in 1947 and 1961. In other years, droughtsoccurred in some regions and floods in others. Thus, in these non-event years, factors other thanEN, SO, or W or C seem to have played important roles.

(v) Considering the 20 years of largest droughts in the all India rainfall, 11 were associated withunambiguous ENSOW (none with ambiguous ENSOW), one with ENSO, two with SOW, onewith SO, three with W and two with non-events, indicating that whereas unambiguous ENSOWwas a powerful combination, it was neither necessary nor sufficient to cause widespread droughts.From the 20 largest floods in the all India rainfall, 12 were associated with C (cold SST) events,two with ENSOW (1878 and 1983, both second years of the double ENSOW events 1877–1878and 1982–1983), 22 with ENC (note, both involved C), two with ENW and SO (opposite toexpectation) and two with non-events. Thus, cold SST was certainly favourable for excess rains.In many drought years, the monsoon seems to set normally over the extreme south of India andAssam and Bengal in the north-east, but fails to extend westward later.

(vi) The drought widespreadness index and the drought response index show that the fractions fornon-events are spread on both sides of 0.5; but for unambiguous ENSOW, the fractions aresignificantly far above 0.5, and for C events they are significantly far below 0.5. Thus, these twocategories have very strong ENSO relationships.

Relationships between large-scale atmospheric anomalies (EN, SO and SST) and Indian monsoonhave been studied for many years. The Walker circulation is the dynamic link between ENSO and theweather and climate over the Indo-Pacific region (Ropelewski and Halpert, 1987; and referencestherein). The present analysis shows that this link is most effective when, in El Nino years, SOminima and SST maxima occur in the middle of the calendar year. However, in some years, someother factors might be playing important roles. The Himalayan and Eurasian snow covers have beenreported to have an inverse relationship with Indian monsoon (Hahn and Shukla, 1976; Dey andBhanu Kumar, 1983; Dickson, 1984) but the years used by these authors had other characteristicsalso. The equatorial stratospheric wind quasi-biennial oscillation (QBO) may be related to the Indianmonsoon (Bhalme et al., 1987; and references therein), as also may the latitudinal location of the axisof the 500 hPa ridge along 75°E (Krishna Kumar et al., 1992; and references therein). Khandekar(1991, 1996) has proposed larger-scale atmospheric flow models incorporating these features. Statisticalmodels using three types of predictors, namely upper air flow over India, heat flow development oversouthern Asia and meridional pressure gradient and cross-equatorial flow over the Indian ocean andthe Southern Oscillation, are being used presently for formulating the Indian summer monsoon rainfall(Thapliyal and Kulshrestha, 1992) and seem to give reasonably accurate predictions.



ACKNOWLEDGEMENTS

Thanks are due to Dr Parthasarathy for updated data for Indian rainfall and to Dr Angell and Dr Wrightfor updated data for SST. This work was partially supported by FNDCT, Brazil under contractFINEP-537/CT.

REFERENCES

Angell, J.K. 1981. ‘Comparison of variations in atmospheric quantities with sea surface temperature variations in the equatorialeastern Pacific’, Mon. Wea. Re6., 109, 230–243.

Bhalme, H.N., Rahalkar, S.S. and Sikdar, A.B. 1987. ‘Tropical quasi-biennial oscillation of the 10 mbar wind and Indian monsoonrainfall—implications for forecasting’, J. Climatol., 7, 345–358.

Chen, W.Y. 1982. ‘Assessment of Southern Oscillation sea level pressure indices’, Mon. Wea. Re6., 110, 800–807.Deser, C. and Wallace, J.M. 1987. ‘El Nino events and their relation to the Southern Oscillation: 1925–1986’, J. Geophys. Res., 92,

14189–14196.Dey, B. and Bhanu Kumar, O.S.R.U. 1983. ‘Himalayan winter snow area and summer monsoon rainfall over India’, J. Geophy.

Res., 88, 5471–5474.Dickson, R.R. 1984. ‘Eurasian snow cover versus Indian monsoon rainfall—an extension of Hahn–Shukla results’, J. Climatol.

Appl. Meteorol., 23, 171–173.Hahn, D.F. and Shukla, J. 1976. ‘An apparent relationship between Eurasian snow cover and Indian monsoon rainfall’, J. Atmos.

Sci., 33, 2461–2462.Khandekar, M.L. 1991. ‘Eurasian snow cover, Indian monsoon and El Nino/Southern oscillation—A synthesis’, Atmos.–Ocean, 29,

636–647.Khandekar, M.L. 1996. ‘El Nino/Southern Oscillation, Indian monsoon and world grain yields—a synthesis’, in El-Sabh, M.I.,

Venkatesh, S., Denis, H. and Murty, T.S. (eds.), Ad6ances in Natural and Technological Hazards Research, Kluwer, Dordrecht, pp.79–96.

Khandekar, M.L. and Neralla, V.R. 1984. ‘On the relationship between the sea surface temperatures in the equatorial Pacific andthe Indian monsoon rainfall’, Geophys. Res. Lett, 11, 1137–1140.

Kiladis, G.N. and Diaz, H.F. 1989. ‘Global climatic anomalies associated with extremes in the Southern Oscillation’, J. Climate, 2,1069–1090.

Krishna Kumar, K., Rupakumar, K. and Pant, G.B. 1992. ‘Pre-monsoon ridge location over India and its relation to monsoonrainfall’, J. Climate, 5, 979–986.

Mooley, D.A. and Paolino, D.A. 1989. ‘The response of the Indian monsoon associated with the change in sea surface temperatureove the eastern south equatorial Pacific’, Mausam, 40, 369–380.

Normand, C. 1953. ‘Monsoon seasonal forecasting’, Q. J. R. Meteorol. Soc., 79, 463–473.Pant, G.B. and Parthasarathy, B. 1981. ‘Some aspects of an association between the Southern Oscillation and Indian summer

monsoon’, Arch. Meterol. Geophys. Bioklimatol., B29, 245–252.Parker, D.E. 1983. ‘Documentation of the Southern Oscillation Index’, Meteorol. Mag., 112, 184–188.Parthasarathy, B., Sontakke, N.A., Monot, A.A. and Kothawale, D.r. 1987. ‘Droughts/floods in the summer monsoon season over

different meteorological subdivisions of India for the period 1871–1984’, J. Climatol., 7, 57–70.Parthasarathy, B., Rupa Kumar, K. and Kothawale, D.R. 1992. ‘Indian summer monsoon rainfall indices 1871–1990’, Meteorol.

Mag., 121, 174–186.Quinn, W.H., Zoff, D.F., Short, K.S. and Kuo Yang, R.T.W. 1978. ‘Historical trends and statistics of the Southern Oscillation, El

Nino and Indonesian droughts’, Fish. Bull., 76, 663–678.Quinn, W.H., Neal, V.T. and Antunes de Mayolo, S.E. 1987. ‘El Nino occurrences over the past four and a half centuries’, J.

Geophys. Res., 92, 14449–14461.Rasmusson, E.M. and Carpenter, T.H. 1983. ‘The relationship between eastern equatorial Pacific sea surface temperatures and

rainfall over India and Sri Lanka; Mon. Wea. Re6., 111, 517–528.Ropelewski, C.F. and Halpert, M.S. 1987. ‘Global and regional scale precipitation patterns associated with El Nino/Southern

Oscillation’, Mon. Wea. Re6., 115, 1606–1626.Ropelewski, C.F. and Jones, P.D. 1987. ‘An extension of the Tahiti–Darwin Southern Oscillaion Index’, Mon. Wea. Re6., 115,

2161–2165.Thapliyal, V. and Kulshrestha, S.M. 1992. ‘Recent models for long-range forecasting of southwest monsoon rainfall over India’,

Mausam, 43, 239–248.Wright, P.B. 1977. The Southern Oscillation Patterns and Mechanisms of the Teleconnections and the Persistence, Report HIG-77-13,

Hawaii Institute of Geophysics, 107 pp.Wright, P.B. 1984. ‘Relationship between indices of the Southern Oscillation’, Mon. Wea. Re6., 112, 1913–1919.


Extremes of the ENSO phenomenon and Indian summer monsoon rainfall

Documents

Transcript of Extremes of the ENSO phenomenon and Indian summer monsoon rainfall