CLIMATOLOGY OF MONSOON RAINS OF MYANMAR (BURMA)

Size: px

Start display at page:

Download "CLIMATOLOGY OF MONSOON RAINS OF MYANMAR (BURMA)"

Jasmine Theodora Perry
5 years ago
Views:

1 INTERNATIONAL JOURNAL OF CLIMATOLOGY Int. J. Climatol. 20: (2000) CLIMATOLOGY OF MONSOON RAINS OF MYANMAR (BURMA) N. SEN ROY* and SURINDER KAUR India Meteorological Department, Lodi Road, New Delhi , India Recei ed 16 March 1999 Re ised 20 July 1999 Accepted 25 July 1999 ABSTRACT Based on 33 years rainfall data of Myanmar for the summer monsoon months (June September), the detailed rainfall climatology of the country has been studied. Seasonal rainfall series are found to approximate to a Gaussian distribution. By using the rainfall distribution and coefficient of variation, it has been possible to divide the country into five homogeneous rainfall regions. Different statistical characteristics of the seasonal, monthly and zonal rainfall, as well as the whole country s rainfall, have been determined. Analysis of interannual and intraseasonal variability highlights the fact that the correlation between the rainfall of different months and zones is rather weak. Trend and periodicity of the rainfall series have been examined by different statistical techniques, indicating little evidence of a trend. The power spectrum of the rainfall series appears to show only marginal significance at the 95% level for an 11 year cycle. The rainfall series of Myanmar shows little correspondence with neighbouring Bangladesh and Northeast India, even though all of them are influenced by similar weather systems. Copyright 2000 Royal Meteorological Society. KEY WORDS: Myanmar; climatology; summer rainfall 1. INTRODUCTION Myanmar, formerly known as Burma, occupies a vast landmass (about km 2 ) in the tropical belt of Southeast Asia. Its economy is heavily dependent on agriculture. It has an old civilization. With its annual rainfall of more than twice the global average of about 100 cm and its rich farmland, it has the potential for developing into a regional bread basket. It is predominantly a monsoon country and falls in the tracks of east and west moving weather systems. Some of the weather systems originating in the China Sea or the Pacific Ocean traverse the Myanmarese landmass and emerge into the Bay of Bengal, where they often intensify into depressions or cyclonic storms, which are devastating at times. It has a long recorded rainfall history, and has a well-distributed network of rain gauges. It is, therefore, somewhat annoying to find that there is not much internationally accessible published material on the climatology of this important country. Some of the earlier works, like Maung (1945) and Huke (1965), are important. Myanmar also figures in a small way in many of the climatological studies of the monsoon belt, primarily because of its association with the main topic, e.g. Lyde (1938), Rao and Ramamurthy (1968), Ramaswamy (1969), Rao (1981), Ramage (1971), Nieuwolt (1981), Fein and Stephens (1987), Kripalani and Kulkarni (1997, 1998), and Pant and Rupa Kumar (1997). Climatological details of the country, however, are found only in works devoted exclusively to Myanmar. The present study attempts to fill part of this gap by studying salient features of the climatology of the summer monsoon rainfall of Myanmar using recent data. This choice is dictated largely by the fact that about 75% of the country s annual average rainfall is during this season. The period chosen is from June to September, as practically the entire country is under the influence of the monsoon during that period even though the monsoon reaches the southern part of the country by around the third week of May; there is some rainfall in October too. * Correspondence to: India Meteorological Department, Lodi Road, New Delhi , India. Copyright 2000 Royal Meteorological Society

2 914 N. SEN ROY AND S. KAUR The choice of this period also enables one to compare the Myanmarese summer monsoon with the Indian southwest monsoon. Such a comparison is useful since the southwest monsoon of India, spanning the period June September, has been studied extensively and some of the results, which are well documented, could be considered for possible extension to the Myanmar case. 2. DATA USED Monthly rainfall data were obtained from the Department of Meteorology and Hydrology of Myanmar for the 33 year period This study will only consider rainfall behaviour in the months from June to September. The data covered 50 rain gauges, with a good countrywide distribution. The rainfall series have been subjected to tests for homogeneity, trend and randomness. Its median value of 170 cm and the number of runs above and below is 17. According to the run test, the series appears to be strongly homogeneous. The rainfall series of 50 individual stations were also found to be homogeneous. Thus, the data are suitable for further rigorous time series analysis. A few short gaps in some of the station data series have been filled in by assuming that the missing data of the individual station will be approximately in the same ratio with their average as that of the neighbouring stations. The length of data series was limited by the availability of data. However, considering that the World Meteorological Organization (WMO) has adopted 30 years as the period for climatological normals, it is felt that the 33 year period is adequate to filter out climatic transients and capture broadly the essential climatic characteristics of the summer monsoon rainfall of Myanmar. 3. GEOGRAPHICAL FEATURES Myanmar is strategically located in Southeast Asia and extends from 10 N to 28 N and from 92 E to 102 E. The north south running mountain chains of the Arakan Yoma (mean elevation 1800 m, peaks 3000 m) and the Shanmassif (mean elevation m) divide the country into three different physical zones, iz. Western, Central and Eastern zones. Politically, the country is divided into 13 states. The three northwestern states of Kanchin, Sagaing and Chin Hills have a highly variable terrain, while the easternmost states of Shan, Kayah, Karen and Tena-see-rim are plateau-like. The other states have a relatively flat terrain. The main river, the Irrawaddy, runs from north to south and empties into the Bay of Bengal. 4. SEASONAL CIRCULATION FEATURES The monsoon reaches the southern parts of Myanmar by the third week of May and covers the entire country by the beginning of June. Its withdrawal is almost completed by the beginning of October. During the onset phase, the Inter Tropical Convergence Zone (ITCZ) moves northward, bringing rains to Myanmar. The monsoon trough on the surface, which starts from Pakistan and extends southeastwards, can often be traced over Myanmar as the continuation of the trough or a low pressure area over Myanmar. As the monsoon sets in, the ITCZ merges with the monsoon trough in a complex interaction. The coastal wind at 850 hpa is thus mostly from the southwest, as the India Tibetan anticyclone appears over Myanmar and Southeast Asia by May and then moves northwestward over the Tibetan Plateau at the peak of the monsoon in July. From September onwards, during the withdrawal phase, the anticyclone moves southeastward, losing its identity by October. Its location, orientation and intensity greatly influence the rainfall. Similarly, the monsoon trough also has a strong influence over the rainfall. Unlike India, there are very few monsoon depressions, which normally form over the Bay of Bengal and move along northwesterly or northerly tracks. The typhoons of the south China Sea or the Pacific Ocean that influence Indo-China and Thailand in their westward movement have relatively little impact on Myanmar.

3 MYANMAR RAINFALL CLIMATOLOGY 915 Table I. Monthly mean rainfall, S.D. and coefficient of variation (CV) of Myanmar s rainfall Month Mean S.D. CV (cm) (cm) (%) June July August September Seasonal (June September) RAINFALL DISTRIBUTION The area weighted average rainfall of Myanmar over the 33 year ( ) period ranges from 156 cm in 1956 to 224 cm in The average is 179 cm, which can be regarded as the country s normal rainfall for the summer monsoon season (June September). During the same period, India receives on average 88 cm of rainfall. The mean monthly rainfall of the country along with their standard deviation (S.D.) and coefficient of variation (CV) are given in Table I. The results show that August is the month with most rainfall, with July coming a close second. The lowest CV is in July. Myanmar s seasonal rainfall appears to be more stable than Indian monsoon seasonal rainfall, which has a CV of about 10%. Table II gives the mean monthly rainfall for different states, which exhibit similar patterns of high rainfall in July and August. Isopleths of the monthly and seasonal rainfall of Myanmar are shown in Figure 1(a) (e). All the maps exhibit a similar pattern. The higher seasonal rainfall region ( 300 cm) lies towards the western coast. A second maximum ( 250 cm) is observed in the northern states. The central states are the arid zones, where less than 60 cm rainfall is received during the monsoon. The central plains show relatively flat gradients. This spatial distribution is caused by orography, and the behaviour of the monsoon low and the Tibetan anticyclone. 6. RAINFALL ZONES AND VARIABILITY Maps showing the spatial distribution of CV with respect to monthly and seasonal rainfall are given in Figure 2(a) (e). All four months exhibit lowest variability ( 30%) over the eastern and northern frontier Table II. Mean monthly rainfall (cm) of different states State Number of June July August Septmeber stations Kachin Shan Chin Sagaing Mandalay Magme Arakan Pegu Irrawaddy Kayah Karen Mon Tena-see-rim

4 916 N. SEN ROY AND S. KAUR Figure 1. (a) (e) Mean monsoon rainfall (cm) of Myanmar

5 MYANMAR RAINFALL CLIMATOLOGY 917 Figure 1 (Continued)

6 918 N. SEN ROY AND S. KAUR Figure 2. (a) (e) Coefficient of variation of rainfall

7 MYANMAR RAINFALL CLIMATOLOGY 919 Figure 2 (Continued) regions with heavy rainfall, while the highest CV ( 60%) is observed in the central parts of the country with relatively less rainfall. The seasonal rainfall also exhibits a similar pattern, with the magnitude of CV falling to half its monthly value. A similar pattern is observed in India, where heavy rainfall regions also show a minimum variability ( 20%) and arid zones of Rajasthan show a maximum variability ( 60%).

8 920 N. SEN ROY AND S. KAUR Figure 3. Homogeneous rainfall zones of Myanmar It is useful to define regions in which rainfall behaviour is homogeneous in time. The most commonly used technique for such zonation is principal components analysis (PCA). In this study, we have utilized a rainfall distribution pattern and its CV to classify the entire country into five homogeneous rainfall zones, as in Figure 3. This makes it easy to use such zonation for working out optimum agricultural and water management strategies. A few statistical details of each region, as identified in this study, are given in Table III to indicate rainfall variability. Table III. Rainfall characteristics of different rainfall regions Region States covered Area Mean SD of CV m ms rainfall rainfall (km 2 ) (cm) (cm) (%) (%) (%) North Mayanmar West Myanmar Kanchin, Sagaing above 24.5 N Chin Hills, Arakan Central Mandalay, Magwe, Sagaing below Myanmar 24.5 N. East Myanmar Shan, Kayah South Peru, Irrawaddy, Mon, Karen, Mayanmar Tena-see-rim WholeMyanmar SD denotes standard deviation; CV, coefficient of variation. m denotes SD of monthly departure from normal, and ms, SD of monthly departure from normal after substracting seasonal departure.

9 MYANMAR RAINFALL CLIMATOLOGY 921 Apart from S.D. and CV, the table gives the S.D. of percentage departure from normal ( m ) for June, July, August and September combined and S.D. of monthly percentage departure from normal ( ms ) for the same 4 months combined, noting that ms has been calculated after subtracting the seasonal percentage departure from normal from the monthly percentage departure from normal for each year. This has been done in order to take into account the contribution of seasonal anomalies towards interannual variability and monthly means (Shukla, 1987). 7. NORMALITY TEST The statistical nature of the seasonal rainfall data for the country can be summarized as follows: Mean=179 cm S.D.=14 cm Skewness= 1 =0.49, g 1 =0.7 Kurtosis= 2 =3.84, g 2 =0.84 To test normality, Fisher s test was used. The test statistics are Z 1 =1.77, Z 2 =0.25. The insignificance of these values indicates that the seasonal rainfall series approximates to a Gaussian distribution. 8. INTERANNUAL VARIABILITY Yearly rainfall departures from normal for each zone and the country as a whole are given in Figure 4. It can be seen that rainfall was more than +10% in excess during 1949, 1952, 1961 and 1963, while it was deficient by more than 10% in 1954 and Figure 4, along with Table IV, indicate that there is a good correspondence between the rainfall of the south zone and that of the country as a whole, especially when the country receives rainfall within 10% of the long-term average. This has implications for seasonal predictions of rainfall. It will be of interest to examine the correlation between the seasonal rainfall (June September) of Myanmar with that of Bangladesh and Northeast India, as similar weather systems influence them. In fact, Bangladesh shows a positive correlation with Northeast India (Kripalani et al., 1996). However, analysis indicates no significant relationship between Myanmar s rainfall and that of Bangladesh or Northeast India. Perhaps an orographic barrier between Myanmar and these two zones effectively isolates them. Furthermore, it is known that there is little correlation between the annual monsoon rainfalls of the whole of India and of Myanmar (Kripalani and Kulkarni, 1997). 9. El NIN O AND MONSOON During the period , there were five El Niño years. In 1957, 1972 and 1976, the rainfall over Myanmar was deficient by 7%, 10% and 10%, respectively, while in 1953 and 1965, it was above average by +3% and 1%, respectively. Deficiency was identical ( 10%) during moderate (1972) and severe (1976) El Niño years. Further, during non-el Niño years, both deficient and excess rainfalls have been observed. Thus, El Niño does not have a one-to-one relationship with the seasonal rainfall of Myanmar. In India, also, out of 17 El Niño years between 1901 and 1996, only nine experienced deficient rainfall, while in other years, rainfall was either normal or excess, indicating a weak correspondence between El Niño and seasonal rainfall (Pant and Rupa Kumar, 1997).

10 922 N. SEN ROY AND S. KAUR Figure 4 (Continued)

11 MYANMAR RAINFALL CLIMATOLOGY 923 Figure 4. Percentage departures from seasonal rainfall for different zones (solid bars denote El Niño years)

12 924 N. SEN ROY AND S. KAUR Table IV. Frequency of rainfall departure (RD) from normal for the whole country (WC) in relation to its five zones WC RD Zone RD to to to North Myanmar to West Myanmar to Central Myanmar to East Myanmar to South Myanmar to It is useful to find out the mean successive difference (MSD) of the rainfall to assess quantitatively the interannual variability. Defining MSD as MSD= 1 n n 1 x i x i 1, i=2 where x i is the seasonal rainfall series, it is seen that for Myanmar, MSD=14 cm, which is about 8% of the mean rainfall. Thus, interannual variability is not pronounced. 10. TREND AND PERIODICITY The time series graph of monsoon rainfall (Figure 5) shows the stochastic nature of the rainfall. This series has been subjected to non-parametric tests for homogeneity, trend and randomness. It was further analysed using moving averages and a power spectrum technique to ascertain trend and periodicity. To test the null hypothesis No Trend using the Mann Kendall non-parametric test, let N=33 and N P= i=1 n i, where n i is the score corresponding to observation x i, which is obtained from the number of observations greater in magnitude following x i. The test statistic is defined as = 4P N(N 1) =0.057, with a standard error of 4N+10 =S.E.( )= 9N(N 1) =0.122 Therefore, the standard normal score (Z) is

13 MYANMAR RAINFALL CLIMATOLOGY 925 Figure 5. 5 and 10 year moving average of seasonal rainfall S.E.( ) =0.46 Z0.05 =1.96. Hence, the monsoon rainfall series of Myanmar supports the null hypothesis that there is no trend. The absence of any trend in seasonal data is also apparent from 5 and 10 year moving averages plotted in Figure 5. The non-existence of a trend is also indicated by Cramer s test for stability in a time series. 11. SPECTRAL ANALYSIS The seasonal rainfall data series over 33 years was analysed for randomness, periodicity and persistence using a power spectrum approach. The plot of autocorrelation r L against lag L is shown in Figure 6 (L=11). The value of r 1 =0.17. Since the series can be approximated by a Gaussian distribution, the significant value of r 1 at the 95% probability level for 32 observations should be more than With the computed value being less than this value the series is free from persistence. Raw and final spectral estimates S k were computed for k=0,1,2,...,11. S 0 corresponds to the fluctuations of infinite wavelength and gives indications of trend, while the last estimate S 11 corresponds to the shortest wavelength fluctuation. The power spectrum showing the distribution of spectral estimates for the maximum lag of 11 years is shown in Figure 7. Since we have analysed the yearly series, each lag year represents a 2 year wavelength. Hence, the x-axis represents the harmonic cycles for 22 years. The peak power spectrum is marginally significant at the 95% level, which shows an 11 year cycle. 12. INTRASEASONAL CORRELATIONS Intraseasonal and interzonal correlation studies are useful for possible application in long-range forecasts. Monthly rainfall series for the country for 33 years covering the four monsoon months and the seasons have been used to calculate the intraseasonal correlation. The correlation matrix is given in Table V.

14 926 N. SEN ROY AND S. KAUR Figure 6. Correlogram of rainfall series Figure 7. Power spectrum of rainfall series It can be seen from the table that rainfall in all 4 months is significantly correlated with seasonal rainfall. The months of July, August and September have a correlation of , but the month of June has a smaller value of Between months there is only one significant correlation value of 0.42, which is between the rainfall series of August and September. Thus, there is no significant interrelationship among the rainfall values of different months. Table V. Intraseasonal correlations June July August September Seasonal June l July August September Seasonal

15 MYANMAR RAINFALL CLIMATOLOGY 927 Table VI. Interzonal correlations N W C E S WC N W C E S WC 1.00 N, north Myanmar; W, west Myanmar; C, central Myanmar; E, east Myanmar; S, south Myamnar; and WC, whole country. 13. INTERZONAL CORRELATIONS The rainfall series for the country and for its five zones were analysed to ascertain the degree of interzonal correlation. The correlation matrix is shown in Table VI. It is observed that all five zones have a significant correlation with the rainfall of the country as a whole. This has some utility in long-range predictions of zonal rainfall if a prediction model is available for the country s seasonal rainfall. The correlations, however, do not exhibit any regular pattern. The significant correlations obtained from the available dataset are r WS =0.66, r CE =0.54 and r WN =0.31. The symbols N, W, C, E and S represent north, west, central, east and south, respectively. North, west and south are comparatively heavier rainfall zones, while the central and east are low rainfall zones. Thus, one heavy rainfall zone (N, W and S) is better correlated with another heavy rainfall zone, while the low rainfall zones (C and E) have also a good correlation among themselves. A heavy rainfall zone does not show a significant relationship with a low rainfall zone. 14. CONCLUDING REMARKS Analysis of the summer monsoon rainfall of Myanmar for a period of 33 years yields some interesting results. The country s rainfall series is found to be Gaussian in nature and exhibits no trend or periodicity. El Niño, known to be the most important factor controlling inter-annual climate variations in the tropics, does not have a one-to-one relationship with Myanmar s rainfall, thus indicating that there are other important influences. Rainfall distribution exhibits maxima over the western coastal region and north Myanmar. Based on the distribution pattern of rainfall and its CV, the country can be divided into five homogeneous rainfall zones, each of which has a significant correlation with the rainfall of the country as a whole. Heavy rainfall zones have a better correlation among themselves. This is similar to the case with low rainfall zones. However, there is poor correlation between high and low rainfall zones. Even though India and Myanmar are geographical neighbours and are influenced by the same monsoon system, Myanmar s rainfall has no significant relationship with the rainfall of India or the contiguous areas of India. The results of this study are likely to be useful for validating climate models for Myanmar and neighbouring areas, as well as providing basic climatic information on Myanmar s rainfall. The present study is based on 33 years of data with a few short breaks in the individual station series. The length of data series was limited by data availability. However, the broad conclusions derived from this study and mentioned above are likely to be borne out when a more detailed analysis is carried out on a larger and more complete dataset. REFERENCES Fein, J.S. and Stephens, P.L Monsoons, Wiley, New York. Huke, R.E Rainfall in Burma, Geographical Publications, Dartmouth; as quoted in Ramage (1971). Kripalani, R.H. and Kulkarni, A Rainfall variability over Southeast Asia connections with Indian Monsoon and ENSO extremes: new perspectives, Int. J. Climatol., 17,

16 928 N. SEN ROY AND S. KAUR Kripalani, R.H. and Kulkarni, A The relationship between some large scale atmospheric parameters and rainfall over Southeast Asia: a comparison with features over India, Theor. Appl. Climatol., 59, Kripalani, R.H., Inamdar, S. and Sontakka, N.A Rainfall variability over Bangladesh and Nepal: comparison and connections with features over India, Int. J. Climatol., 16, Lyde, L.W The Continent of Asia, MacMillon, London. Maung, M.K Forecasting the coastal rainfall of Burma, Q. J. R. Meterol. Soc., 71, Nieuwolt, S The climate of the Indian subcontinent, in Takahashi, K. and Arakawa, H. (eds), World Sur ey of Climatology. Volume 9, Climates of Southern and Western Asia, Chapter 1. Pant, G.B. and Rupa Kumar, K Climates of South Asia, Wiley, New York, pp Ramage, C.S Monsoon Meteorology, Academy Press, New York. Ramaswamy, C The Problems of the Indian Southwest Monsoon, Indian Geophysics Union, Hyderabad, India. Rao, Y.P Climate of the Indian subcontinent, in Takahashi, K. and Arakawa, H. (eds), World Sur ey of Climatology. Volume 9, Climates of Southern and Western India, Chapter 2. Rao, Y.P. and Ramamurthy, K.S Forecasting Manual Part I: Climatology of India and Neighbourhood 2, Climate of India, India Meteorological Department, Poona, India. Shukla, J In Fein, J.S. and Stephens, P.L. (eds.), Monsoons, Wiley, New York.

Seasonal Climate Outlook for South Asia (June to September) Issued in May 2014

Ministry of Earth Sciences Earth System Science Organization India Meteorological Department WMO Regional Climate Centre (Demonstration Phase) Pune, India Seasonal Climate Outlook for South Asia (June