RAINFALL VARIABILITY OVER BANGLADESH AND NEPAL: COMPARISON AND CONNECTIONS WITH FEATURES OVER INDIA

Transcription

1 INTERNATIONAL JOURNAL OF CLIMATOLOGY, VOL. 16, (1996) RAINFALL VARIABILITY OVER BANGLADESH AND NEPAL: COMPARISON AND CONNECTIONS WITH FEATURES OVER INDIA R. H. KRIPALANI, SUSHAMA INAMDAR AND N. A. SONTAKKE Indian Institute of Tropical Meteorology. Pashan, Pune-41 I 008, India Received 15 March 1995 Accepted 6 September 1995 ABSTRACT In this study monthly rainfall data for 14 stations over Bangladesh for the period are used to investigate and understand the interannual variability of the summer monsoon rainfall. Monthly, seasonal, and annual spatial rainfall patterns, and the spatial patterns of variability, are presented. Dominant structures of seasonal rainfall are determined through the empirical orthogonal functions. A homogeneous series for All Bangladesh Monsoon Rainfall is prepared and its temporal characteristics are studied. It is observed that the standardized rainfall for this series shows random fluctuations up to 1963, thereafter the standardized values are much above the normal values. Further the rainfall variations over Bangladesh are not related to large-scale variables such as the Northern Hemisphere surface temperature, Darwin pressure tendency, and the subtropical ridge over the Indian region. However, the rainfall variations over Bangladesh are related well with rainfall variations over northeast India. Similar analysis is done for the Nepal region by examining the monthly rainfall data over Kathmandu for a 105 year period ( , ). Results reveal that Nepal rainfall is well related with rainfall variations over northern and central parts of India. KEY WORDS: monsoon rainfall variability; Bangladesh; Nepal; empirical orthogonal functions 1. INTRODUCTION The START (Global Change System for Analysis, Research and Training) activities supported by the International Biosphere Programme (IGBP), World Climate Research Programme (WCRP) and Human Dimensions of Global change (HDP) provides the necessary structure of the networks, centres, and sites to address the regional origin and impacts of global change (IGBP, 1994). The START comprises a number of regional research networks (RRNs). The first planning meeting of the South Asia RRN took place in Colombo, Sri Lanka in February It was pointed out that although in India considerable work has been done on variability of climate and particularly rainfall by analysis of the extensive data-sets of the India Meteorological Department (IMD), the nature of the climate variability over the other parts of the Indian monsoon regime, such as Bangladesh, Sri Lanka, Nepal, and Pakistan, is not as well documented. It was therefore suggested that climate variability of the South Asia region be one of the foci of the South Asia RRN. Secondly, the Asian monsoon is one of the important components of the coupled ocean-land-atmosphere system. The Indian monsoon has often been used as a proxy for the Asian monsoon as a whole. This may create problems in understanding the broad-scale features of the Asian monsoon. The relationship between the Indian monsoon and monsoons of the other regions in the Asian region will enhance our understanding of the large-scale features of the Asian monsoon. A start in this direction has been made by investigating the connection between Indian and Chinese rainfall (Kripalani and Singh, 1993) and India-Thailand rainfall (Kripalani et al., 1995). Although the general climatological information over Bangladesh and Nepal is available (Robinson, 1976; Rawson, 1963; Yoshino, 1984), but a detailed analysis is lacking. Recently Ahmed and Karmakar (1993) have studied the onset dates over Bangladesh. It is worthwhile to study the variations in the monsoon rainfall over Bangladesh in a comprehensive manner on temporal and spatial scales. CCC /96/ by the Royal Meteorological Society

2 690 R. H. KRIPALANI, S. INAMDAR AND N. A. SONTAKKE Keeping the above in view, in this paper the spatial and temporal characteristics of monthly, seasonal, and annual rainfall over Bangladesh have been investigated in detail. For Nepal, data were available for Kathmandu only, hence the analysis is done for Kathmandu rainfall only. It should be emphasized that Kathmandu cannot be assumed to represent the whole of mountainous Nepal. The geographical location of Bangladesh and Nepal with respect to India is shown in Figure 1. The details of the data used in this study are described in section 2. Section 3 presents the environmental conditions and monthly rainfall patterns over Bangladesh. The spatial characteristics of interannual variability of annual and monsoon season rainfall over Bangladesh is presented in section 4. A reliable homogeneous Bangladesh monsoon rainfall series has been prepared using the available data. Statistical properties and fluctuation characteristics of the Bangladesh rainfall series are examined in section 5. Connections between India- Bangladesh rainfall are also examined in this section. In section 6 the environmental conditions and the results of the analysis of Nepal rainfall are presented. Here we also examine the connections between India-Nepal rainfall. The main conclusions of this study are given in section 7. The following data sets have been used. 2. DATA (i) Monthly rainfall data for 14 stations (Figure 2) over Bangladesh for months January through to December for the period were obtained from the India Meteorological Department, Pune (data for 1947 were missing). (ii) The monthly data for the period over Bangladesh have been acquired from Oak Ridge National Laboratory, USA from their Global Historical Climatology data sets. (iii) Data for six stations from 1979 up to 1990 for Bangladesh have been taken from the publications of the Monthly Climate Data of the World, sponsored by the World Meteorological Organization in cooperation with the US National Oceanic and Atmospheric Administration. For the period only 9 per cent of the data are not available. However, for the period 1978 to 1990 nearly 90 per cent of the data were not available to the authors, hence the results presented here are based on the data period for the period (except 1947) only. (iv) For Nepal the monthly rainfall data were available for Kathmandu only. Monthly rainfall data for and for Kathmandu are taken from the India Meteorological Memoirs (Eliot, 1902) and climatological records of Nepal (DIHM, 1977). (v) Rainfall data ( ) for the 52 blocks over India are computed from 306 stations with data available in processed form from Dr B. Parthasarathy, who obtained the original data from IMD. Details of this data set are available in Kulkami et al. (1992). (vi) Northern Hemisphere surface temperature (NHSTfiaverage for January and February for the period (Jones et al., 1982). (vii) The 500 hpa subtropical ridge position (latitude) along 75"E during April (ridge) for the period (Banejee et al., 1978). (viii) Darwin pressure tendency (DPT)-MSL pressure difference between January and April for Darwin in Northern Australia representing the state of the Southern Oscillation for the period (Shukla and Paolino, 1983). (ix) The time series of All India Monsoon Rainfall (AIMR) for the period has been taken from Parthasarathy et al. (1994). 3. ENVIRONMENTAL CONDITIONS AND MONTHLY RAINFALL PATTERNS OVER BANGLADESH 3.1. Environmental conditions Climatologically, Bangladesh is located in the tropical monsoon region. In shape, Bangladesh is a rather irregular quadrilateral area lying across the northern tropics. It is surrounded on all sides by India (see Figure 1) except the

3 RAINFALL OVER BANGLADESH AND NEPAL Figure 1. Map showing the positions of Kathmandu (Nepal) and Bangladesh with respect to India 88' 89' 90' 91' 92O 93O 2 6" ME G H A C AYA 2 5 O 2 4" 2 3" 2 2" BAY OF BENGAL 2 1' 88' 89' 90' 91' 92' 93O Figure 2. Names and locations of the 14 stations over Bangladesh

4 692 R. H. KRIPALANI, S. INAMDAR AND N. A. SONTAKKE coast and the south-eastem part where it touches Myanmar (formerly Burma). Its climate is characterized by a cool dry season, a hot summer season and the rainy monsoon season. The dry season roughly prevails from mid-october through to February. Less than 10 per cent of the annual rainfall occurs during this season. On account of the strong influence of the monsoons, the climates of India and Bangladesh are similar. Bangladesh has a heavy rainfall averaging cm annually and cm for the monsoon season (June to September). (These figures are based on the data period , except 1947.) Although Bangladesh as a whole receives plentiful rainfall, the amount may vary appreciably from year to year. For example, in the data period analysed, the minimum monsoon season and annual rainfall were cm and cm, respectively, for the year On the other hand the maximum seasonal and annual rainfall were 2 15 a6 cm and cm, respectively, for the year In the late summer Bangladesh suffers from destructive tropical cyclones. Bangladesh has been affected by natural disasters such as cyclones and floods on a huge scale. Except the south-east region around the town of Chittagong, Bangladesh comprises the deltaic plains of the lower Ganges and Brahmaputra rivers. Topographically, it is a flat area, much of it a little above sea-level, with innumerable rivers and streams, especially in the southern half of the country. In the south-eastem region there is a pattern of longitudinal hills running in parallel with the coast, although these are nowhere very high (Robinson, 1976) Monthly rainfall patterns Figure 3 shows the spatial distribution of rainfall (in mm) over Bangladesh for all the 12 months. Monthly rainfall may be described by considering four climatological periods. The rainfall distribution patterns for each month are similar and in general the isohytes display a gradient from east to west. (i) March-May. During March some areas, in particular the north-east, receive moderate rainfall ( mm), although in most of Bangladesh, the rainfall is still below 50 mm. By April the eastern half of the country Figure 3. Spatial distribution of the monthly rainfall (mm) over Bangladesh

5 RAINFALL OVER BANGLADESH AND NEPAL 693 receives over 100 mm of rain and the north-eastern part receives over 300 mm. In May the whole country receives well over 170 mm with a maximum over the north-east region (more than 500 mm). On an average this season contributes 19 per cent of the annual rainfall. (ii) June-August. During this period the south-west monsoon is at its peak. During June the whole country receives over 300 mm of rain with a maximum over the north-east and south-east part of the country. The rainfall distribution patterns for July and August are similar to June. During this period rainfall is especially heavy in the Chittagong region because it is exposed to the full force of the south-west monsoon and Cox's Bazar receives more than 700, 900, and 700 mm of rain during June, July, and August respectively. These three months together contribute about 57 per cent of the annual rainfall. (iii) September-October. These are the months of the withdrawal of the south-west monsoon. Although the rainfall pattern remains similar as the pattern during the peak of the monsoon, the rainfall over the eastern parts of the country has become half that during the peak of the south-west monsoon. These two months contribute about 20 per cent of the annual rainfall. (iv) November-February. This is the season of the north-east monsoon and Bangladesh is practically dry during this period. In November the whole of the country receives well below 50 mm of rain, except the Chittagong region. During December and January the rainfall is around 10 mm over the entire country. During February the rainfall is between 20 mm and 30 mm. These four months contribute about 4 per cent of the annual rainfall. At individual stations, the contribution of the south-west monsoon (June-September) to annual rainfall varies from 62 per cent at Brahmanbaria to 78 per cent at Cox's Bazar. This spatial distribution pattern showing the contribution of seasonal to annual rainfall is depicted in Figure 6 (bottom right-hand panel). For Bangladesh as a whole the contribution of the south-west monsoon rainfall to annual rainfall is about 70 per cent. For India the south-west monsoon contributes about 78 per cent of the annual rainfall (Parthasarathy et al., 1994). Hence in the subsequent sections, mainly results of the analysis of the southwest monsoon are presented. 4. DOMINANT SPATIAL STRUCTURES OF INTERANNUAL VARIABILITY OVER BANGLADESH In this section the dominant modes of interannual variability are studied through seasonal (June-September) empirical orthogonal fimctions (EOFs). Before the EOFs are presented some features of the seasonal/annual rainfall over Bangladesh are shown. Here the relationship with some large-scale parameters, such as the NHST, DPT, and subtropical ridge are investigated. The spatial relationship with Indian monsoon rainfall is also investigated Seasonal and annual rainfall patterns (hased on data) The annual and monsoon season spatial distribution of rainfall patterns are shown in Figure 4. The isohytes over the country display an east to west gradient for the annual and the monsoon season patterns. Both the spatial patterns are similar, with maximum rainfall over the north-east and south-east parts of the country and minimum rainfall over the west-central parts (around 23.5"N, 89.O"E). The standard deviation varies from 30 cm to 90 cm for the annual pattern and from 20 cm to 50 cm for the monsoon season pattern. The coefficient of variation (CV) varies from 16( 18) per cent to 22(26) per cent for the annual (seasonal) rainfall. In general the CV is less over the central parts of the country. This pattern of CV is different from that observed over the Indian region, where we observe maximum CV (50 per cent) over regions of minimum rainfall, i.e. north-west India and minimum CV (20 per cent) over the west coast and north-east India (Kripalani et al., 1995) EOFs of monsoon season rainfall Empirical orthogonal functions are the most effective tool for describing parsimoniously the spatio-temporal variability of meteorological fields which explain maximum variability with the constraints of orthogonality. It is also a statistical approach towards the regisnalization of rainfall regimes.

6 Figure 4. Spatial distribution of the annual and seasonal rainfall, standard deviation, and coefficient of variation over Bangladesh The EOFs of the monsoon season rainfall over Bangladesh are presented in Figure 5. The EOFs over the Indian region and the method of computation are available in Kulkarni et al. (1 992). The first four EOFs explain 60 per cent of the variance. Out of this, EOF 1 itself explains 30 per cent variance and EOF 2,3 and 4 each explain about 10 per cent variance. Empirical orthogonal function 1 shows loadings of the same sign over the entire country, suggesting simultaneous interannual variability of monsoon rainfall over the whole country. The central region around Dhaka shows maximum loadings. Over the Indian region the loadings for EOF 1 over north-east India are opposite to those over central India (Kulkami et al., 1992). Empirical orthogonal function 2 shows a dipole type of structure, with one centre around Mymensingh area and the other around Cox s Bazar area. This type of feature could be associated with the onset or the withdrawal phase of the monsoon. Positive loadings are shown for EOF 3 over the eastern part of the country (around Comilla) and negative loadings over the western parts (around Jessore). Smaller scale features are shown by EOF 4, with negative/positive/negative loadings from north to south. The EOF analysis reveals that the most dominant EOF (EOF 1) suggests simultaneous interannual variability over the whole of Bangladesh Relationship with large-scale parameters Spatial predictive relationships of three selected parameters representative of different features of the atmospheric-ocean system with the monsoon season rainfall of the 14 stations over Bangladesh are analysed. The NHST represents the global climatic variability and includes to a large extent the large-scale trends in the global climate system as well as surface heating, particularly along the middle latitudes. The DPT represents the Southern Oscillation, the most important cause of the short-term fluctuations of the tropical climate system. The ridge represents the most important regional circulation parameter (Singh et al., 1986). From dynamic considerations these parameters appear to be independent of each other.

7 RAINFALL OVER BANGLADESH AND NEPAL * Figure 5. The first four empirical orthogonal functions of seasonal rainfall (VE = variance explained) The time series of these three parameters is correlated with monsoon season rainfall of each of the 14 stations over Bangladesh. These spatial correlation patterns, based on data for the period (except 1947) are shown in Figure 6. The data for the ridge were available from 1939 only, hence the common period for these three parameters is taken. None of these parameters show any significant relationship with monsoon season rainfall at any station over Bangladesh. However, these parameters show significant relationship with rainfall variations over India (Kripalani and Singh, 1993) and are important parameters for predicting monsoon season rainfall over India (Gowariker et al., 1991). It is surprising that Bangladesh, being adjacent to India, shows no significant relationship with these parameters. Even remote regions, such as north China, show a significant relationship with All India Monsoon Rainfall (AIMR) (Kripalani and Singh, 1993), and over north-west Thailand (Kripalani et al., 1995). Figure 6 also shows the spatial distribution of the correlation coefficients over Bangladesh with AIMR. Again no significant relationship is noticed. Just as the AIMR series has been prepared by treating India as a whole (Parthasarathy et al., 1994), monsoon season rainfall series for Bangladesh as a whole is prepared and is designated as All Bangladesh Monsoon Rainfall (ABMR). The ABMR series is prepared using the simple arithmetic mean of the 14 stations. The fact that EOF 1 shows loadings of the same sign over the entire country justifies using the simple arithmetic mean of the 14 stations as a representative of the Bangladesh rainfall series.

8 696 R. H. KRIPALANI, S. INAMDAR AND N. A. SONTAKKE RIDGE A BUR A PERCENT ( SEASONAL eo Be Figure 6. Spatial distribution of the correlation coefficient with large scale parameters NHST, DPT, RIDGE, AIMR and ABh4R. Percentage of seasonal rainfall to annual rainfall is also shown Figure 6 (bottom centre panel) also shows the spatial distribution of the correlation coefficients of ABMR with each of the 14 stations. All the correlation coefficients are statistically significant (significant correlation coefficient for a sample of 70 is 0.3 at the 1 per cent level), this again justifies the use of simple arithmetic mean of the 14 stations as being representative of the All Bangladesh Monsoon Rainfall. Even the correlation coefficients of the above three large-scale parameters with ABMR are insignificant. The Correlation coefficient with NHST, DPT, and subtropical ridge being -0.12, 0.20, and -0.03, respectively, based on the data for the period The correlation coefficient between ABMR and AIMR is based on data for the period To determine whether the pre-monsoon rainfall over Bangladesh can give a clue to the subsequent monsoon rainfall over India, monthly rainfall over Bangladesh has been correlated with AIMR. Results reveal that May Bangladesh Rainfall is significantly correlated with AIMR Relationship of ABMR with rainfall over various locations in India To investigate whether the rainfall variations in ABMR are related with rainfall variations over various regions of India, the monsoon season rainfall series of 52 blocks (for the location of these 52 blocks, refer to Kulkarni et al., 1992) over India is correlated with ABMR. This spatial pattern, shown in Figure 7, reveals interesting features. It is observed that the rainfall over Bangladesh shows significant positive relationship with rainfall over north-east India. The correlations with rainfall over other parts of India are insignificant and even negative. The maximum negative relationship is centred around 20"N, 84"E along the east coast of India, which is very near to Bangladesh. There is a steep gradient from north-east India (correlation coefficient=0.5) to northern parts of the east coast of India (correlation coefficient = -0.3).

9 30 1. RAINFALL OVER BANGLADESH AND NEPAL COR WITH ABMR to 1 I I 1 I 1 J Figure 7. Spatial distribution of the correlation coefficient over the Indian region with ABMR and Nepal 5. INTERANNUAL VARIABILITY AND STATISTICAL CHARACTERISTICS OF ABMR It is seen (Figure 3) that the monsoon months June through to September are the main rainy months, contributing about 70 per cent of the annual rainfall. The CV for these four months varies from 18 to 26 per cent (Figure 4). The year has been divided into four seasons and the average seasonal rainfall amounts are 39.3 mm, mm, mm, and mm for the winter (DJF), summer (MAM), south west monsoon (JJAS), and post-monsoon (ON) contributing 1.7, 19.3, 70.0, and 9 per cent, respectively, to the annual rainfall amount. As stated earlier a time series of the All Bangladesh Monsoon Rainfall (ABMR) has been prepared for the period by using the simple arithmetic mean of the rainfall of the 14 stations. This series is homogeneous according to Swed-Eisenhart s run test (WMO, 1966a). The frequency distribution of this series is Gaussian according to the chi-square, Fisher s g statistic and Kolmogorov-Smimov (K-S) tests. This series is also free fkom Markovian type persistence because the lag one autocorrelation is insignificant. This series has a mean of cm, standard deviation of 20.2 cm and coefficient of variation 12.2 per cent. Figure 8 shows the ABMR series expressed in standardized units. It is difficult to define a drought (dry) or flood (wet) year precisely. A year has been classified as an excess (wet) monsoon rainfall year when R; > R + S and deficient (dry) when R; < R - S, where R; is the monsoon rainfall of the ith year, R the mean and S the standard deviation of the series. There were 10 dry (drought) and 12 wet (flood) years in the series and although their occurrence appears to be distributed randomly up to 1960, there are more dry years during the period However, it is observed that there were relatively more frequent wet years during the period Whether this increase in rainfall is connected with the recent global warming (Jones, 1994), increase in frequency of severe cyclonic storms over the Bay of Bengal (Mooley, 198 l), or the Bay of Bengal storms having a more northerly track needs further investigation. 5. I. Trends and periodicities in ABMR To examine the significant long-term changes, if any, in the series, the Mann-Kendall rank statistic (WMO, 1966b) has been applied. The test statistic is significant at the 5 per cent level but not at the 1 per cent level. The short-term fluctuations have been studied by applying Cramer s tk test for the 11-year running means (WMO, 1966b), and the results are shown in Figure 9. Running means for 11 years up to 1960 show a negative tendency, with the 193&1932 period significant at the 5 per cent level. From 1961 onwards it shows a positive tendency and from 1964 they are significant at the 5 per cent level.

10 698 R. H. KRIPALANI, S. INAMDAR AND N. A. SONTAKKE 3 i 2 2 z z1 a n w o N - n 5-1 D z a I- -2 u? YEAR Figure 8. Year-to-year standardized deviation of All-Bangladesh monsoon rainfall (June to September) from 1901 to 1977 To detect significant periodicities, if any, in the ABMR series, the Schickedanz and Bowen (1975) power spectrum analysis has been applied. A high spectral estimate is noticed for the 5.3-year period (Figure 10) but it is not significant. This periodicity could be associated with the periodicity of the Southern Oscillation, which has a periodicity range of 2-7 years. It was seen earlier that out of the three large-scale parameters DPT shows a higher correlation coefficient than the NHST and the subtropical ridge, although not significant. However, the significant 40-year periodicity seen needs further investigation. 5 BANGLADESH 11- yr window I

11 RAINFALL OVER BANGLADESH AND NEPAL v) W F I- v) W J a a W v) 5 10% 0 I HARMONIC NUMBER Figure 10. Spectral estimates for the ABMR series 6.1. Environmental conditions over Nepal 6. RAINFALL VARIATIONS OVER NEPAL (KATHMANDU) Nepal stretches for a distance of about 850 km along the southern slopes of the Himalayas and 13G240 km from north to south. To the north lies Tibet and to the south, India. Geographically, Nepal may be thought of in terms of a series of roughly east-west belts or zones. In the south, separating Nepal from the plains of India, is a long narrow belt of fairly level land covered with jungle and swamp, known as Terai; north of this and running parallel with it, is the forested Siwalik Range, m high but with some peaks reaching 1800 m; north again is a zone of discontinuous east-west valleys; further north still is the Mahabharat Range, about m, and above, the Himalayas proper, which include Mounts Everest, Dhaulagiri, and Makula (Robinson, 1976). Some general climatological information, as taken from the Climatological Records of Nepal (DIHM, 1977) is as follows. (i) Although it lies near the northern limit of the tropics, there is a wide range of climate, from summer tropical heat and humidity of the Terai to colder dry continental and alpine winter climate through the middle and northern mountainous section. The amount of precipitation and range of temperature vary considerably because of the exceptionally rugged terrain. (ii) Mean annual precipitation ranges from more than 6000 mm along the southern slopes of Annapurna range in Central Nepal to less than 0 mm in the north-central portion near the Tibetan plateau. Varying amounts between 1500 and 00 mm predominate over most of the country, with a distinct maxima along the southern slopes of Mahabharat and the Himalayan ranges in the eastern two-thirds of the country. Minima stretch east-west through the mid+ection of the kingdom. On an average 80 per cent of the precipitation is during the monsoon season. (iii) The rains during the monsoon have a tendency to be more frequent at night. Spring months are characterized by showery and windy weather, higher humidity and pre-monsoon thunderstorms. Snowfall is confined to the northern and western mountainous region above 3500 m. (iv) May temperatures are the warmest. In summer the temperatures are more than 40 C in Tmi and 28 C in the mid-section of the country, whereas during winter Terai has 23 C as maximum temperature and 7 C as minimum temperature. In the central valley 12 C is the maximum temperature and below freezing the minimum temperature.

12 700 R. H. KRIPALANI, S. INAMDAR AND N. A. SONTAKKE Figure 11. Histogram depicting monthly rainfall distribution for Kathmandu (v) Kathmandu. Situated in a large fertile valley at an altitude of 1300 m and with a seasonable but equable climate. Average summer temperature is between 27" and 19 C and winter between 21 C and 2 C. Annual rain exceeds 1300 mm. From October to February there are skies with calm windless nights, frequent dense early morning fog, and occasional frost Statistical analysis of Kathmandu rainfall Some analysis of rainfall climatology over the Nepal Himalayas has been done by Dhar and Mandal (1 982) and Dhar and Rakhecha (1986). Because long-period ( , ) monthly rainfall data was available for the station Kathmandu only, detailed analysis is restricted to Kathmandu. The series for the periods and are analysed separately. Both the series are homogeneous according to the Swed-Eisenhart's run test (WMO, 1966a). The frequency distribution of both the series is found to be Gaussian according to the chi-square, Fisher's g statistic, and Kolmogorov-Smimov (I-S) tests. Neither series shows any Markovian type of persistence. Based on the data for the above 105 years, the mean annual rainfall for Kathmandu is cm, with a maximum of cm during 1948 and a minimum of 84.2 cm during The monsoon season figures being a mean of 11 1 a 5 cm, a maximum of cm (1853) and a minimum of 63.6 cm (1921). Kathmandu receives about 80 per cent of the annual rainfall during the months of the south-west monsoon. Figure 1 1 shows a histogram depicting the monthly rainfall for each month at Kathmandu Temporal characteristics, trends and periodicities-kathmandu rainfall Figure 12 shows the year-to-year standardized monsoon season rainfall series. Out of the 105 years, there are 18 years when the standardized values were less than - 1.O, the most notable being 1864 (-2.4), 1877 (- 1.9), 1897 (-2.1), and 1921 (-2.4), whereas there were 17 years when the standardized values were greater than 1.O, the most notable being 1853 (2-0) and 1954 (2.0). To examine the significant long-term changes, if any, the Mann-Kendall statistic (WMO, 1966b) has been applied. The test statistic is significant at the 10 per cent level for the period To further understand the temporal characteristics of monsoon season rainfall, 1 1-year moving averages for the monsoon season anomalies for the periods and are presented in Figure 13. These short-term climatic fluctuations have been studied by applying Cramer's tk test for these 1 1-year running means (WMO, 1966b). It is seen that the running

13 RAINFALL OVER BANGLADESH AND NEPAL 70 1 Figure 12. Year-to-year standardized deviation of monsoon season Kathmandu rainfall means show an alternating systematic positive and negative tendency. A positive tendency is seen during the periods , , , , and , and a negative tendency during , , and , with some significant values. To detect significant periodicities if any, both the series have been subjected to the Schickedanz and Bowen (1975) power spectrum analysis. For the period , three cycles of 5.6,5.0, and 4.5 years are significant at the 10 per cent level (see Figure 14). All these periodicities correspond to the period of the Southern Oscillation. Figure 13. Values of the Cramer s tk statistic for the 1 I-year running means for the monsoon season Kathmandu rainfall series

14 702 R. H. KRIPALANI, S. WAMDAR AND N. A. SONTAKKE v) W 30 t a 20 I - I- v) : -I a w P '" KATHMANDU % 0 Lu I HARMONIC NUMBER Figure 14. Spectral estimates for Kathmandu monsoon season rainfall separately for the periods and Connections: India-Nepal (Kathmandu) rainfall The monsoon season rainfall of Kathmandu has been correlated with monsoon season rainfall of the 52 blocks over India for the period This spatial pattern of the correlation is shown in Figure 7 (right panel). This spatial pattern clearly reveals that Nepal rainfall is significantly related with rainfall over north and central India. It shows an inverse relationship with rainfall over north-east India. For the data period , the Kathmandu monsoon season rainfall shows a correlation coefficient of -0.14, -0.23,0.30, and 0.39 with NHST, DPT, subtropical ridge and AIMR, respectively. However, Kathmandu and Bangladesh show a correlation coefficient of 0.03 based on data for the period Thus Kathmandu rainfall shows some relationship with AIMR but not with ABMR. 7. CONCLUSIONS From the analysis of the monthly rainfall data over Bangladesh for the period and monthly rainfall data of Kathmandu for the periods and , the following main conclusions can be drawn. (i) Large-scale parameters such as the NHST, DPT, and the subtropical ridge show no relationship with monsoon season rainfall over Bangladesh. (ii) Although the monsoon season rainfall over India and Bangladesh as a whole shows no relationship, the rainfall variations over Bangladesh show a significant relationship with rainfall variations over north-east India. (iii) The occurrence of below and above the normal values of standardized ABMR is distributed randomly up to However, it is observed that the standardized values after 1963 are much above the normal value and statistically significant. (iv) The 11-year running means for the monsoon season at Kathmandu rainfall shows systematic alternate positive and negative tendencies. (v) Kathmandu seasonal rainfall shows significant relationship with All India Monsoon Rainfall. These relationships are particularly strong over the northern and central parts of India.

15 RAINFALL OVER BANGLADESH AND NEPAL 703 ACKNOWLEDGEMENTS The authors express their sincere thanks to Professor R. N. Keshavamurty, Director, I.I.T.M. and Dr S. S. Singh, Deputy Director, for encouragement and for the facilities provided. Thanks are also due to Dr B. Parthasarathy for reviewing the initial draft of the paper, to the anonymous referees for fruitful suggestions, Shri V. H. Sasane for draughting the diagrams and Shn V R. Deshpande for computer-graphics work. REFERENCES Ahmed, R. and Karmakar, S Arrival and withdrawal dates of the summer monsoon in Bangladesh, Int. 1 Climatol., 13, Banejee, A. K., Sen, F! N. and Raman, C. R. V On foreshadowing southwest monsoon rainfall over India with mid tropospheric circulation anomaly of April, Ind. 1 Meteoml. Hydml. Geophys., 29, DIHM Climatological Records of Nepal, Vols I, I1 and 111. Department of Irrigation, Hydrology and Meteorology, Ministry of Food, Agriculture and Irrigation, Government of Nepal. Dhar, 0. N. and Mandal, B. N A pocket of heavy rainfall in the Nepal Himalayas-a brief appraisal, Nepal Himalaya, Dhar, 0. N. and Rakhecha, P. R Precipitation climatology of Nepal of Nepal Himalayas, Nepal Himalaya, Eliot, J India and the neighbouring countries, Ind. Meteoml. Mem., XIV, 709 pp. IGBP IGBP in action. Work Plan , Global Change Report No. 28, International Geosphere-Biosphere Programme, Stockholm, 151 pp. Jones, I? D Recent warming in global temperature series, Geophys. Res. Lett., 21, Jones, P. D., Wigley, T. M. L. and Kelly, P. M Variations in surface air temperature. Part I: Northern Hemisphere, Mon. Wea. Rev., 110, 59. Kripalani, R. H. and Singh, S. V Large scale aspects of India-China summer monsoon rainfall, Adv. Atmos. Sci., 10, Kripalani, R. H., Singh, S. V, Panchawagh, N. and Brikshavana, M Variability of the summer monsoon rainfall over Thailandcomparison with features over India, Int. 1 Climatol., 15, Kulkami, A,, Kripalani, R. H. and Singh, S. V Classification of summer monsoon rainfall patterns over India, Int. 1 Climatol., 12, Mooley, D. A Increase in annual frequency of the severe cyclonic storms of the Bay after Possible causes, Mausam, 32, Parthasarathy, B., Munot, A. A. and Kothavale, D. R All-India monthly and seasonal rainfall series , Theo,: Appl. Climatol., 49, Rawson, R. R The Monsoon Lands ofasia, Hutchinson Educational Ltd., London, 6 pp. Robinson, H Monsoon Asia, 3rd edn, Macdonald and Evans Ltd., Estover, Plymouth, 510 pp. Schickendanz, F! T. and Bowen, E. G Computation of climatological power spectra using variable record lengths, Fourth Conference on Pmbabiliw and Statistics in Atmospheric Sciences, Tallahassee, FL, pp Shukla, J. and Paolino, D. A The Southern Oscillation and long range forecasting of the Summer monsoon rainfall over India, Mon. Wea. Rev., 111, Singh, S. V, Inamdar, S. R., Kripalani, R. H. and Prasad, K. D Relationship between 500 hpa ridge position over the Indian and the west Pacific region and the Indian summer monsoon rainfall, Adv. Ahnos. Sci., 3, WMO 1966a. Some Methods in Climatological Analysis, Technical Note No. 81, WMO No. 199-TP-103, World Meteorological Organization, Geneva, 53 pp. WMO 1966b. Climatic Change, Technical Note No. 79, WMO No. 195-TP-100, World Meteorological Organization, Geneva, 53 pp. Yoshino, M. M Climate and Agriculfuml Land Use in Monsoon Asia, University of Tokyo Press, Tokyo, 398 pp.