Document
Transcription
Document
JOURNAL OF CLIMATOLOGY, VOL. 8. 283 -296 (1988) 55 1.553.21:551.577.32(548) THE ONSET OF THE SOUTHWEST MONSOON OVER KERALA: 1901-1980 R . A N A N T H A K R I S H N A N A N D M. K. S O M A N Indian lnslirute of Tropical Meleorology. Pune 41 1005, India Receioed 6 A U ~ U S1987 I Rc,r>ised28 OcIoher 1987 ABSTRACT Utilising daily mean rainfall from dense rain gauge networks, the dates of onset of the southwest monsoon over south and north Kerala have been derived on the basis of objective criteria for the years 1901 to 1980. These dates have been compared with the onset dates as per records of the India Meteorological Department. Statistics of the onset dates are presented. The mean onset date for south Kerala is found to be 30 May and for north Kerala 1 June with a standard deviation of about 9 days in both cases. The sharp increase in rainfall that heralds the onset of the monsoon is highlighted by superposed epoch analysis. The prevailing notion that rainfall from pre-monsoon thunderstorms progressively increases and merges with thc monsoon rainfall is shown to be not valid. KEYWORIIS Southwest monsoon Onset dates Kerala. 1. INTRODUCTION The date of onset of the southwest monsoon over Kerala is a crucial date in the meteorological calendar of India as it marks the beginning of the rainy season for the country. Although the onset of the monsoon is associated with changes in the circulation features in the lower and upper troposphere, a sustained increase in rainfall at the observatory stations over Kerala and the island stations over the southeast Arabian Sea is an essential feature of the monsoon onset. It is difficult to quantify this precisely and so the experience of the forecaster plays a key role in declaring the date of monsoon onset in individual years. Some two decades ago a critical study was undertaken to examine the monsoon onset dates as per records of the India Meteorological Department (IMD) from 1901 onwards in relation to the daily rainfall from 1 May to 30 June at seven observatory stations over Kerala and the adjoining sea area (Ananthakrishnan et al., 1967). Based on this study some broad criteria were drawn up for operational use. It should be noted that in general, the forecaster works with a measure of caution. The post-mortem study of the onset dates in relation to daily rainfall distribution showed that in many years the onset dates should have been somewhat earlier than the declared dates; the discrepancies were appreciable in some years. Apart from the small number of observatory stations, Kerala has a dense network of rain gauge stations which have been in existence from the end of the last century. It was considered to be of interest to examine the behaviour of mean daily rainfall of Kerala utilizing the data from this network in relation to the monsoon onset. The aim was to arrive at a uniform set of onset dates based on objective rainfall criteria since the existing dates of onset as per IMD records are based on subjective estimates by several forecasters over the years. In recent years there have been attempts to forecast the dates of monsoon onset using regression techniques based on past data (Kung and Sharif 1980,1982). For such studies it is desirable to have a uniform set of onset dates free from subjectivity. 2. MEAN RAINFALL SERIES AND OBJECTIVE CRITERIA FOR MONSOON ONSET The rain gauge network over Kerala utilized in the present study is shown in Figure 1. Latitudinally the state of Kerala extends from 8" 15" to 12" 50". For the present study we have considered south Kerala and 01961748/88/030283- 14$07.00 the Royal Meteorological Society 0 1988 by 284 R AKAKTHAKRISHNAN A K D M. K SOMAN 13 li ii Id 4 0 Figure 1. Rain gauge network over Kcrala north Kerala separately, the dividing line being around 10 N. The division is based on the fact that the monsoon rainfall progresses from south to north along the west coast of India and also on the known difference in the rainfall characteristics between the southern and northern parts of the state. South Kerala has 44 rain gauge stations and north Kerala 31, more or less uniformly distributed over the respective regions. Utiliing daily rainfall recorded at the rain gauge networks, daily rainfall series were constructed for south and north Kerala for the years 1901 to 1980 by averaging the rainfall at the individual stations. Examination of these rainfall series revealed alternating patterns of spells of light and heavy rain amounts of varying durations. Light rain spells are a feature of the pre-monsoon months which give place to heavy rain spells hcralding the onset of the southwest monsoon. The transition is sharp and abrupt. January and February are the driest months in Kerala with very little rainfall. Thunderstorm activity sets in from March and progressively increases with the advance of the season. This activity is more pronounced over south Kerala. The long term (1901-1 980) normal daily rainfall of south Kerala for the months of March and April is 1.9 and 4.8 mm respectively; the corresponding amounts arc lcss than 1 mm for north Kerala. The area-averaged rainfall associated with pre-monsoon thunderstorm activity is found to be less than 10 mm per day. Averaged over individual spells whose durations are variable, the amounts are found to be between 2 and 7 mm for south Kerala and less for north Kerala. We designate these as light rain spells. Thc long term normal daily rainfall for May and June as found from our study is 8.9 and 19.5 mm respectively for south Kerala and 7.3 and 24.0 mm for north Kerala. These are contributed by light rain spells OKSF7' O F T H t SOIJTIIWEST M O N S O O K 28 5 before the monsoon onset and heavy rain spells during and after the monsoon onset. The observational data showed that during such heavy spells the daily area-averaged rainfall exceeds 10 mm per day on most 0 1 the days while the rainfall averaged over the spell duration varies between 15 and 50 mm per day. In view of this, a limit of 10 mm per day was arbitrarily chosen as the demarcation between light and heavy rain spells. Since we are dealing with the average rainfall from a large network of stations distributed over the regions, it is obvious that the heavy spells arc associated with widespread moderate to heavy rainfall while light spells imply scattered rainfall of small amounts with perhaps isolated heavy falls. This distinction in the nature of the two types of rain spells is important. It implies that the physical mechanisms associated with them are different. Heavy rain spells imply deep convection in association with synoptic scale systems indicating ITCZ activity; light spells imply sporadic convective activity associated with prc-monsoon thunderstorms or weak monsoon conditions after monsoon onset. The date of monsoon onset is taken as the first day of the transition from light to heavy rain spell category with the proviso that the average daily rainfall during the first 5 days after the transition should not be less than 10 mm. In several years the heavy rain spell which heralds the onset of the monsoon continues for more than 5 days with daily rainfdl in excess of 10 mm on all days. There are, however, some years in which the daily rainfall on 1 or 2 days during the onset spell may fall below 10 mm, the average still remaining in cxccss of 10 mm per day. This is associated with the pulsatory nature of monsoon activity. As an illustration, the daily rainfall sequences from 1 May to 30 June for south Kerala for the years 1917, 1918, 1979 and 1980 are shown in Figure 2 in which the monsoon onset dates fixed by the above-mentioned criteria are also indicated. The average daily rainfall for the first 5 days including the onset date in these years is 18.1, 19.1, 27.7 and 17.3 mm respectively. I t may be noted that in the first 3 years the daily rainfall was less than 1 0 nim on 1 day during the first 5 days following the onset. The rainfall sequences associated with the monsoon onset over Kerala and the further northward advance of the monsoon vary from year to year. In some years the onset spell of heavy rains is followed by a prolonged spell of light rainfall with subsequent revival of rain activity. In such years it has been customary in the IMD to regard the first onset as a temporary phase and consider the revival date as the onset date. We have adopted the same convention in the present study. Figures 3 and 4 illustrate the nature of the rainfall transition associated with the onset and advance of the monsoon over Kerala for two years, 1936 and 1951. In these diagrams, the distribution of rainfall over the station network is shown for 4 consecutive days starting from the date prior to the onset. The rainfall is plotted correct to 1 mm over the station locations shown in Figure I . Stations with zero rainfall are left blank. In 1936 the monsoon set in over south Kerala on 20 May. On the previous day the rainfall over south Kerala was sporadic with the daily mean of 8.7 mm. By the next day widespread heavy rainfall with an average amount of 41.6 mm was recorded. The monsoon advanced into north Kcrala on 21 May and by the next day widespread heavy rainfall occurred all over the State with average amounts of 61.4 mm and 80.3 mm respectively over south and north Kcrala. In 1951 the onset over south Kerala occurred on 30 May and over north Kerala on 1 June. The spectacular increase in rainfall over north Kerala from 31 May to 1 June may be noted. 3. ONSET DATES OVER SOUTH AND WORTH KERALA (1901 1980) Following the criteria discussed in the previous section, the dates of monsoon onset over south and north Kerala were fixed for the individual years 1901 1980. These dates are listed in Table I in which the monsoon onset dates for Kerala as per IMD records are also given. The IMD records give only a single date for Kerala as a whole. The onset dates vary between 7 May and 22 June. Comparison of the corresponding onset dates for south and north Kerala show that, in many years, the onset over north Kerala was either simultaneous or I to 3 days later than over south Kerala. In 1913, 1952, 1953, 1965 and 1977 the northward progress of the heavy rain spells was slow, resulting in delayed onset of the monsoon over north Kerala by 10 days o r more compared with south Kerala. Table I shows that for the six years 1932, 1943. 1949, 1955, 1957 and 1959 the IMD onset dates were later by 10 days or more than the dates we have found from the rainfall time series: in 1969 the IMD onset date 286 R A N A N T H A K R I S H N A N A N D M . K. S O M A N mm 2; +-MAY- c 30 + -, -*loJu,:' 4ol-++-!+4--- T--- r - I l - I I9*O L O 20 10 I I 30 30 I 20- 20 I 10- o-dll..lll' I**Al 30 - 20- 10- 0- 20 ' I ' I9l7 I 6 I1 k----Figure 2. Mean daily rainfall over south Kerala from 1 May to 30 Junc for selected years showing onset dates of the southwest monsoon was 8 days earlier. Scrutiny of the rainfall data shows that the IMD onset dates have to be revised for all these years. The daily rainfall sequence from 1 May to 30 June for some of these years is depicted in Figure 5. The mean daily rainfall during the first 5 days including the onset dates for the ycars 1932, 1943, 1959 and 1969 are 19.5, 16.2, 14.2 and 13.4 mm respectively, satisfying the onset criteria. Table I1 gives the number of onset years in 3-day intervals for south and north Kerala through May and June. For the 80-year period the onset dates are equally distributed between May and June for south Kerala; for north Kerala the corresponding numbers are in the ratio 35. The mean, median and standard deviations of the onset dates are given in Tablc 111. The mean onset date for south Kerala is 30 May and for north Kerala 1 June with a standard deviation of about 9 days in both cases. The median onset dates are found to be a couple of days after the mean dates. The mean onset date for Kerala as a whole as per IMD data in Table I is 2 June which is later by 3 days and 1 day respectively from the mean onset dates for south and north Kerala of the present study. 4. SUPER-POGED EPOCH ANALYSIS OF THE MEAN DAILY RAINFALL SERIES IN RELATION T O THE MONSOON ONSET It is often stated that the pre-monsoon thunderstorm rains over Kerala progressively increase and merge with the monsoon rains on account of which the fixation of the monsoon onset date is rendered difficult KERALA DAILY RAINFALL I m m l 2 2 MAY 1936 A R A B I A N S E A Figure 3. Rainfall sequencc associated with the onset and advance of the southwest monsoon over Kerala (1936) KERALA DAILY R A I N F A L L ( m m ) 2 9 M A Y 1951 I0 A R A ~ I A N S E A IS 7- * KERALA 157 DAILY RAINFALL l m m ) I JUNEIS51 12. 123 39 I1 I0 10 A R A B I A N S E A 09 0, 75 Figure 4. Rainfall sequence associated with the onset and advance of the southwest monsoon over Kerala (1951) 36 37 43 32 39 33 36 39 33 37 35 31 40 32 36 26 35 25 33 45 27 27 07 28 33 1901 1902 I903 1904 1905 1906 1907 1908 1909 1910 1911 i912 1913 1914 1915 1916 1917 1918 1919 1920 Year 1921 1922 1923 1924 1925 1926 1927 1928 1929 1930 1931 1932 1933 1934 1935 1936 1937 1938 1939 1940 IMD 38 37 43 38 41 45 39 42 33 33 37 39 33 35 46 33 31 11 34 34 NK 37 31 41 32 27 38 27 34 32 38 30 15 22 39 45 21 34 26 37 45 SK 38 31 36 31 27 36 23 34 32 36 31 14 21 37 45 20 33 32 36 43 33 31 42 33 27 37 27 34 29 39 35 33 22 39 43 19 35 26 36 45 IMD 1941 1942 1943 1944 1945 1946 1947 1948 1949 1950 1951 1952 1953 ,1954 1955 1956 1957 1958 1959 1960 Year 22 32 12 33 32 29 33 41 13 27 30 20 37 28 17 18 18 43 12 14 SK 15 15 44 22 41 13 34 36 34 33 40 13 27 32 33 48 32 16 20 18 NK Year 1961 1962 1963 1964 1965 1966 1967 1968 1969 1970 1971 1972 1973 1974 1975 1976 1977 1978 1979 1980 IMD 23 41 29 34 36 29 34 42 23 27 31 20 38 31 29 21 32 45 31 14 SK, Onset date over south Kerala; NK, Onset date over north Kerala; IMD, Onset date for Kerala as per I M D records. Dates are given sequentially as follows: 1 May=Ol . . . 31 May=31; I J u n e = 3 2 . . . 30 June=61. 37 41 33 37 35 36 35 35 42 28 30 08 35 34 44 NK SK Year Table I. Onset dates of the southwest monsoon over Kerala (1901-1980). 10 36 36 24 31 39 38 25 25 25 53 34 23 32 30 27 27 42 31 18 SK 40 32 26 25 53 37 23 31 31 38 29 43 31 32 44 18 17 31 37 26 31 40 39 17 26 27 49 35 26 31 31 30 28 20 10 35 35 37 31 40 IMD NK z 3 C I 290 R . A N A N T H A K R I S H N A N A N D M. K. S O M A N 1. -MAY mm 40,+ ::[ 6 11 I 16 21 26 1 1 I 311 I S 5 10 JUNE I5 I 1 1 25 20 I ' I . Y I I 30 I , mm -40 - 30 I I - 20 - 10 0 0 I- 30 - 20 - 10 0 - 30 -20 b- 10 0 -30 - 20 10 0 Figure 5. Mean daily rainfall over south Kerala from 1 May to 30 June showing IMD and refixed onset dates. (IMD, 1943).Our study shows that the types of rainfall associated with the two events are distinctly different, as stated in Section 2. Merging occurs when the rainfall data for several years are averagcd for calendar dates because of the dispersal in the onset dates over a period of years. Rainfall curves constructed by such averaging give the impression of the merging of pre-monsoon and monsoon rainfall. To highlight the abrupt nature of the rainfall transition during the onset, the daily mean rainfall amounts for the 80-year period were composited in two ways: (a) by averaging the rainfall amounts for calendar dates (Figures 6a, b). (b) by averaging the daily rainfall amounts after superposing the onset dates for the individual years (Figures 7a, b). Figures 6a, b present the variation of the calendar date averaged daily rainfall for south and north Kerala from the third week of March to the end of August. The mean onset dates are indicated on these diagrams. It 29 1 ONSET OF THE SOIJTHWEST MONSOON Table 11. Number of onset years in 3-day intervals S.K. Interval May 5-7 8-10 11-13 14-16 17 19 2&22 23-25 26 28 29 31 Total 40 N.K. IMD 0 2 2 4 0 0 1 5 2 6 7 4 3 2 9 14 29 34 1 1 Interval June 1-3 4- 6 7-9 10 12 13-15 16 18 19 21 22-25 S.K. N.K. 1MD 15 13 15 16 8 6 4 16 8 9 7 5 1 1 5 4 2 0 0 0 1 1 40 51 46 SK, South Kerala; NK, North Kerala. Table 111. Mean, median and standard deviation of onset dates Parameter Mean date Standard deviation Median date Earliest onset date Latest onset date S. Kerala 30 May 8.8 days 1 June 7 May 1918 22June 1972 N. Kerala 1 June 9.2 days 3 June 8 May 1918 22 June 1972 IMD 2 June 7.8 days 2 June 1 1 May 1918 18 June 1972 may be noted that from about the second week of May till the onset date there is a progressive increase in the mean daily rainfall. This arises from the superposition of the monsoon rainfall in the years of early monsoon onset on the pre-monsoon thunderstorm rain of the remaining years. After the mean onset date, there is a further increase in the mean daily rainfall which is more pronounced over north Kerala. The superposed epoch diagrams, Figures 7a, b, bring out a sharp and spectacular increase in rainfall accompanying the monsoon onset, compared to the pre-onset phase. For south Kerala the level of daily rainfall prior to the onset is around 5-6mm which is due to pre-monsoon thunderstorm activity. It is important to note that the pre-monsoon thunderstorm rain does not progressively increase but levels off by the middle of April. On the onset date the rainfall abruptly jumps up to the level of 17 mm and continues to increase reaching a peak value of about 28 m m pcr day during the next 5 or 6 days. This is followed by a gradual oscillatory decline. The mean pre-monsoon rainfall level for north Kerala is about 4 to 5 mm per day which shoots up to a little over 20 mm per day during the onset with further increase during the next 5 to 6 days. It should be emphasized that the mean rainfall variation depicted by Figures 6a, b is not a correct representation of the actual rainfall variation in individual years. This is more correctly represented by Figures 7a, b. Thus, the present study categorically shows that the prevailing notion that the pre-monsoon thunderstorm rainfall over Kerala progressively increases and merges into the monsoon rain resulting in a gradual transition is not in conformity with facts. The onset of the monsoon is heralded by a sudden and spectacular increase in the rainfall from the pre-monsoon level. 5. CHARACTERISTICS O F THE ONSET SPELL The onset spell has been defined as a heavy spell of rain with a duration of 5 days or more and mean rainfall in excess of 10 mm. We have examined the duration and average daily rain amount associated with the onset mm mm 135 3 51 D A I L Y M E A N RAINFALL OF SOUTH K E R A L A : 1901-1980 CALENOAR DATES : M E A N O N S E T D A T E 3 0 M A Y - -30 25- - 25 30 m m mm 35 DAILY M E A N R A I N F A L L OF N O R T H K E R A L A : 1901- 1980 C A L E N D A R D A T E S : M E A N O N S E T DATE I J U N E 3c 25 20 M E A N ONSET I5 I 5 0 Figure 6. (a) Mean daily rainfall over south Kerala averaged for calendar dates for the years 1901 -1980 showing mean onset date of the southwest monsoon. (b) Mean daily rainfdl over north Kerala averaged for calendar dates for the years 1901-1980showing mean onset date of the southwest monsoon 35 mm mm DAILY MEAN RAINFALL OF SOUTH KERALA: 1901-1980 SUPERPOSED EPOCH : ONSET DAY 0 --I DAILY MEAN RAINFALL OF NORTH KERALA ( 1901-l980) b- DAYS D E F O R E ONSET +DAYS AFTER Ol(fET Figure 7. (a) Mean daily rainfall over south Kerala after superposing onset dates of the southwest monsoon for the period 1901 1980.(b) Mean daily rainfall over north Kerala after superposing onset dates of the southwest monsoon for the period 190-1980 ~ 294 R. ANANTHAKRISHNAN A N D M. K. S O M A N spell for the individual years of the 80-year period for both south and north Kerala. The relevant data for south Kerala are shown graphically in Figure 8. Here the vertical pillars indicate the spell duration in days while the thick dots (joined by the broken curve) denote the mean daily rainfall for the respective spells. Note that both the duration as well as the rainfall associated with the onset spells show wide dispersion. Statistical data relating to the analysis of the onset spells are given in Table IV (a),(b). The spell duration in individual years varied from 5 days (1976) to 50 days (1922) with a mean duration of 15.4 days. The onset spells were generally of longer duration during the first-half of the period compared with the second half. The mean daily rainfall associated with the onset spell is 26 mm. This quantity showed a wide range from 1 0 3 mm in 1976 to 495 mm in 1933. It may be of interest to note that 1976 was the year of lowest rainfall for Kerala during the 80-year period covered by the present study. The data in Table IV provide justification for our criteria to fix the onset date of thc southwest monsoon on the basis of the mean daily rainfall over the rain gauge network over south Kerala. The data for north Kerala show similar behaviour. DAYS I Imm r n YEAll--r Figure 8. Duration and averagc rainfall of onset spells over south Kerala (1901 -1980) Table IV. Statistical data pertaining to onset spells (a) Spell duration Duration (days) 5-9 10-14 15-19 20-24 25-29 3 0 34 235 Total Frequency 24 23 14 7 6 3 3 80 Mean spell duration (days) Standard deviation 1901 -1940 1941 1980 1901-1980 17.9 10.9 12.9 7.0 15.4 9.5 (b) Mean daily rainfall Frequency 2 Mean daily spell rainfall (mrn) Standard deviation (mm) 18 19 18 13 5 2 3 1901--I 940 1941- 1980 1901-1 980 27.6 24.4 26.0 8.0 7.2 7.8 80 295 ONSET OF THE SOUTHWEST MONSOON 6. DISCUSSION In an earlier study by Ramdas et al. (1954) the mean daily rainfall data over south Kerala (referred to as Travancore Cochin in their paper) for May-June have been made use of to fix the date of establishment of the southwest monsoon for the period 1891-1950. Visual examination of the rainfall curve depicting the rainfall sequence was the basis on which the dates were arrived at. The date of ‘commencement of persistent heavy rainfall’ was taken by them as the date of establishment of the monsoon. Neither the threshold value of the rainfall nor the minimum duration of persistence has been specified. We have found it possible to fix such limits in the present study. The onset dates for south and north Kerala in Table I furnish a set of dates based on uniform criteria with the least measure of subjectivity. The delay in the mean onset date by 2 days between north and south Kerala seen in Table 111 is in accordance with the average rate of northward advance of the southwest monsoon along the west coast of India. Examination of the dates of monsoon onset for the first and second halves of the 80-year period shows some interesting features, which are shown in Table V. During the period 1901-1940 the monsoon onset was more often in June than in May while the reverse feature occurred during 1941-1980. This feature is reflected in the mean onset dates for the two periods which differ by 4 days. A significant feature brought out by the present study is the clear demarcation between the rainfall associated with the pre-monsoon thunderstorm activity over Kerala and the rainfall that heralds the onset of the monsoon. The super-posed epoch rainfall diagrams (Figures 7a, b) show that the average daily rainfall from pre-monsoon thunderstorm is of the order of 5 mm. This limit is reached by about the middle of April and there is little increase thereafter. Following the monsoon onset the mean daily rainfall steeply increases to about 30 mm. The limit of 10 mm chosen for demarcating the light and heavy rain spells roughly corresponds to twice the average pre-monsoon rainfall level. In two recent papers, Subbaramayya et al. (Subbaramayya and Bhanukumar, 1978; Subbaramayya et ul. 1984) have discussed various features associated with the onset and advance of the southwest monsoon over India. For the 25-year period 1956- 1980 these authors have, in one of their diagrams, indicated a mean onset date of 20 May over Kerala. This is totally at variance with the mean onset dates for the same period as per IMD records (30 May) and as per the present study (29 May for south Kerala and 30 May for north Kerala). It is obvious that these authors distinguished not between the pre-monsoon thunderstorm rains and the monsoon rainfall over Kerala. 7. CONCLUSIONS The main points brought out by the present study are: (i) The mean daily rainfall series for south and north Kerala enable formulation of objective criteria for fixing monsoon onset dates in individual years. From the dates thus arrived at, the mean onset dates for south and north Kerala are found to be 30 May and 1 June respectively for the period 1901-1980 with a standard deviation of about 9 days. Table V. Comparison of monsoon onset over Kerala during 1901- 1940 and 1941 -1980. ~ 1901 1940 Item Mean date Std. Dev. (days) May onset (No. of years) June onset (No. of years) S.K N.K 1 June 7.5 3 June 7.9 15 11 25 29 1941-1980 IMD S.K N.K IMD 4 June 28 May 30 May 31 May 7.0 9.6 10.1 7.9 9 31 25 18 25 15 22 15 296 R. A N A N T H A K R I S H N A N AND M. K. SOMAN (ii) The mean onset date for 1901-1940 is later than for 1941-1980 by 4 days. (iii) Super-posed epoch analysis of the mean daily rainfall reveals a sharp and spectacular increase heralding the monsoon onset. (iv) The statement often made that the pre-monsoon thunderstorm rain over Kerala progressively increases and merges with the monsoon rains resulting in a gradual transition is not valid. (v) The spell that heralds the onset of the monsoon has a mean duration of about 15 days and the associated daily mean rainfall is 26 mm. These have large variations in individual years. ACKNOWLEDGEMENTS We wish to thank the Director, Indian Institute of Tropical Meteorology for his interest in this study and for providing the necessary facilities. We also wish to express our thanks to Shri R. M. Soni for preparation of fair copies of the diagrams, Shri R. P. Mali for photographic assistance and Miss S. R. Kamble for typing the manuscript. The data utilised in this study were kindly made available by the National Data Centre of IMD. We arc grateful to the India Meteorological Department for a research grant. REFERENCES Ananthakrishnan, R., Acharya, U. R. and Ramakrishnan. A. R. 1967. 'On the criteria for declaring the onset of the southwest monsoon over Kerala', I M D , Forecasting Manual, No. IV-18.1. India Meteorological Department. 1943. CIimarological Atlas jiir Airmen. Kung. E. C . and Sharif, T. A. 1980. 'Regression forecasting of the onset of the Indian summer monsoon with antecedent upper air conditions', J . A p p l . Me!., 19, 37C380. Kung, E. C . and Sharif, T. A. 1982. 'Long range forecasting orthe Indian summer monsoon onset and rainfall with upper air parameters and sea surface tempcratures', J . Mer. Soc. Japan, 60,672481. Ramdas, L. A., Jagannathan. P. and Gopal Rao. S. 1954. 'Prediction of the date of the establishment of southwest monsoon along the west coast of India', Ind. J . Mer. Geophys. 5, 305 314. Subbaramayya. 1. and Bhanukumar, 0 .S. R. U. 1978. 'The onset and the northern limit of the southwest monsoon over India', M e f . M a g . . 107, 3 7 4 4 . Subbaramayya, I., Babu, S. V. and Rao, S. S., 1984. 'Onsct of the summer monsoon over India and its variability', Met. Mag., 113, 127-135.