Theor. Appl. Climatol. 67, 115±121 (2000)
1 2
Indian Institute of Tropical Meteorology, Pune, India National Center for Medium Range Weather Forecasting, New Delhi, India
Stratospheric zonal wind and temperature in relation to summer monsoon rainfall over India S. D. Bansod1 , K. D. Prasad1 , and S. V. Singh2 With 5 Figures Received July 30, 1999 Revised March 17, 2000 Summary The interannual variability of the Indian summer monsoon (June±September) rainfall is examined in relation to the stratospheric zonal wind and temperature ¯uctuations at three stations, widely spaced apart. The data analyzed are for Balboa, Ascension and Singapore, equatorial stations using recent period (1964±1994) data, at each of the 10, 30 and 50 hPa levels. The 10 hPa zonal wind for Balboa and Ascension during January and the 30 hPa zonal wind for Balboa during April are found to be positively correlated with the subsequent Indian summer monsoon rainfall, whereas the temperature at 10 hPa for Ascension during May is negatively correlated with Indian summer monsoon rainfall. The relationship with stratospheric temperatures appears to be the best, and is found to be stable over the period of analysis. Stratospheric temperature is also signi®cantly correlated with the summer monsoon rainfall over a large and coherent region, in the north-west of India. Thus, the 10 hPa temperature for Ascension in May appears to be useful for forecasting summer monsoon rainfall for not only the whole of India, but also for a smaller region lying to the north-west of India.
1. Introduction The Indian summer monsoon (June to September) rainfall contributes about 75% of the total annual rainfall over India and is characterized by large interannual variability causing ¯ooding and droughts over large parts of the country. The agricultural economy of India depends mainly on
this seasonal rainfall and hence there is an increasing demand for accurate forecasts of the Indian summer monsoon rainfall (henceforth called as ISMR) on long range or seasonal scale. Generally, empirical techniques have been developed for such forecasting by relating predictands with predictors based on historical relationships. Attempts have also been made to relate the ISMR with the antecedent stratospheric zonal wind which is found to exhibit the Quasi-BiennialOscillation (QBO). In this oscillation (i.e., QBO) the stratospheric zonal wind varies alternatively from easterly to westerly with a period varying from 24±30 months (Edbon and Veryard, 1961; Reed et al., 1961). The zonal wind assumes maximum amplitude of about 20±30 meter/sec. near 20±30 km. The oscillation shows downward phase propagation with a speed of about 1 km/ month (Reed and Rogers, 1962; Wallace and Holton, 1968; Holton, 1968; Quiroz, 1981) and attenuates around the height of the tropopause. The oscillation seems to be due to the interaction between upward propagating waves (Holton, 1968) and the mean atmospheric ¯ow. Mukherjee (1985) found that the average (June to August) zonal wind at 30 hPa for Balboa has a signi®cant relationship with the ISMR. Raja Rao and Lakhole (1978) noted that the phase of the QBO over the equatorial Indian region gives
116
S. D. Bansod et al.
prior indication of the amount of rainfall over central parts of India. Thapliyal (1979) found that the 10 hPa easterly (westerly) zonal wind at Balboa during January is followed by bad (good) monsoon rainfall. Bhalme (1972) reported a dominant QBO spectral peak in Indian monsoon rainfall and in the number of monsoon storms/ depressions. Bhalme et al. (1987) also found that monsoon rainfall is likely to be less (more) than normal during easterly (westerly) zonal wind anomalies at 10 hPa for Balboa during January. The above studies are based on analyses of the stratospheric zonal wind at selected stations using limited periods of data. The present study analyses the stratospheric zonal wind, as well as the stratospheric temperature, at three equatorial stations for each of the 50, 30, 10 hPa levels in order to investigate in detail their relationship with the ISMR. 2. Data The summer monsoon (June±September) rainfall from 306 stations uniformly distributed over the Indian subcontinent (Parthasarthy et al., 1992) forms the basic rainfall data used in this study. The ISMR series has been prepared by weighting each rainfall station by the area of the district in which the rainfall station lies. A summer monsoon rainfall series for each 2 lat./lon. grid boxes overs contiguous India and has been prepared by simply averaging the summer monsoon rainfall for those stations falling within each grid box. The monthly stratospheric zonal wind and temperature series analyzed in the present study are from the three equatorial stations, Balboa, Ascension and Singapore at the pressure levels 10, 30, and 50 hPa. These stations lie in the Paci®c ocean around 9 N, 80 W, in the Atlantic ocean, 8 S, 14 W, and in the Indian ocean, 1 N, 103 E, respectively. These data were obtained from the Climate Analysis Center, U.S.A. The stratospheric parameters analyzed were for the period January 1964 to December 1994, except for the zonal wind and temperature at 10 hPa for Singapore and Balboa (the data for which were available from January 1980 to December 1994 and January 1964 to December 1990, respectively). 3. Stratospheric zonal wind and ISMR The monthly (January to December) stratospheric zonal wind for Balboa, Singapore and Ascension
Fig. 1. The correlation coef®cient of the India summer monsoon rainfall with the monthly (January to December) stratospheric zonal wind at (a) 50 hPa (b) 30 hPa and (c) 10 hPa. Horizontal dash line shows correlation coef®cient signi®cant at 5% level
at three stratospheric levels (i.e., 10, 30 and 50 hPa) was analyzed for a possible predictive relationship with the ISMR by simple correlation analysis. In estimating the signi®cance of the correlation coef®cient (C.C.), Quenouille's (1952) method was adopted to account for the reduction in degrees of freedom due to persistence. The annual C.C. between stratospheric zonal wind and the ISMR is presented in Fig. 1. At 50 hPa (Fig. 1a) the C.C. of the ISMR with the stratospheric zonal wind is found to have small negative value from January to April at all three stations. The C.C. is positive thereafter and then increases gradually, with the maximum C.C. (signi®cant at 5% level) reached during July±
Stratospheric zonal wind and temperature in relation to summer monsoon rainfall over India
August. This shows that the 50 hPa equatorial zonal wind has concurrent association with the ISMR at all three stations. The positive relationship found between them suggests that the ISMR is likely to be below (above) normal during the easterly (westerly) phase of the 50 hPa equatorial zonal wind. The relationship of the ISMR with 30 hPa stratospheric zonal wind (Fig. 1b) is found to be positive from January to December at all three stations. However, a signi®cant positive C.C. with ISMR occurs with the Balboa 30 hPa zonal wind during April and May (with the maximum in April). This shows that the 30 hPa zonal wind for Balboa during April has some predictive implications for the ISMR. The 10 hPa Singapore zonal wind is available for a limited period of 15 years and hence the result for the 10 hPa zonal wind for Singapore is not presented. The 10 hPa zonal wind for the other two stations (Fig. 1c) shows a signi®cant (positive) C.C. with the ISMR during January. The C.C. is found to decrease thereafter, reaching nearly zero during July and August. The C.C. is also found to be negative, and a high negative C.C. occurs during October to December. Hence, it seems that the ISMR is predictable 5 months in advance using Ascension and Balboa 10 hPa zonal wind. It was noted by Bhalme et al. (1987) that the 10 hPa zonal wind anomaly for Balboa during January is highly correlated with the subsequent ISMR. However, this study shows that in addition to the zonal wind for Balboa (January), Ascension 10 hPa zonal wind (January) is also useful for prediction of the ISMR. The above results indicate that the ISMR has the maximum association with the zonal wind at the 10, 30 and 50 hPa during January, April and July respectively. This appears to be due to the downward phase propagation of the QBO in the stratospheric zonal wind. Such downward phase propagation of the QBO in the stratospheric zonal wind was also noted by Reed et al. (1961). It can be also noted from Fig. 1(a,b,c) that the QBO in the stratospheric zonal wind is in phase for the equatorial stations. 4. Stratospheric temperature and ISMR As with the analysis of stratospheric zonal wind, the stratospheric temperatures at Balboa, Singapore and Ascension for the 10, 30 and 50 hPa
117
Fig. 2. Same as Fig. 1 but for the stratospheric temperature
levels were analyzed for their relationship with the ISMR. The annual variation of the relationship of these parameters with the ISMR is presented in Fig. 2. The 50 hPa temperature (Fig. 2a) for all three stations shows positive association with the ISMR, except for Ascension, during August to November and for Singapore during September. It can be noted that the concurrent relationship between the ISMR and the 50 hPa temperatures is generally not found to be signi®cant (as was the case for the 50 hPa zonal wind). However, a signi®cant negative association between the ISMR and the 50 hPa temperature occurs for Ascension, during the post-monsoon month (i.e. in October). The 30 hPa temperature (Fig. 2b) generally has positive association with the ISMR from January to May and is negative thereafter. A signi®cant positive C.C. between the ISMR and temperature
118
S. D. Bansod et al.
for Balboa is found during May. Also, the 30 hPa temperature at Ascension during September shows a signi®cant negative C.C. with the ISMR. As noted in section 3, the 30 hPa zonal wind for Balboa is also found to have positive association (signi®cant at the 5% level) with the ISMR during April and May. Because the Singapore temperature at 10 hPa is only available for a period of 15 years, the results for Singapore are not presented. The 10 hPa temperature (Fig. 2c) for Balboa and Ascension shows negative association with the ISMR from January to November and the association is found to be positive during December. The 10 hPa temperature for Balboa shows a signi®cant C.C. with the ISMR during March, whereas in the case of Ascension the signi®cant C.C. occurs from March to May, with the maximum C.C. (significant at 1% level) during May. It can be seen that in the case of stratospheric temperatures at Ascension, signi®cant negative correlations with the ISMR are found during October, September and May at the 50, 30 and 10 hPa levels, re-
spectively. As found with the stratospheric zonal wind, the stratospheric temperature also appears to have downward phase propagation of the QBO in the stratospheric temperature. However, the downward phase propagation of the QBO in stratospheric temperature is not so well re¯ected, as appears in the case of stratospheric zonal wind. This may be due to the fact that the stratospheric temperature is also in¯uenced by trospheric processes. As found for the equatorial stratospheric zonal wind, the equatorial stratospheric temperature is also generally in phase (Fig. 2a,b,c) for the equatorial stations. 5. Stability of the relationships In previous sections, four stratospheric parameters were identi®ed as showing signi®cant association with the ISMR. These parameters are the 10 hPa zonal wind for Balboa and Ascension during January, the 10 hPa temperature for Ascension during May and the 30 hPa zonal wind for Balboa during April. The stability of the relationship
Fig. 3. The correlation coef®cient of the Indian summer monsoon rainfall with (a) Balboa wind at 10 hPa (January) (b) Ascension wind at 10 hPa (January) (c) Ascension temperature at 10 hPa (May) (d) Balboa wind at 30 hPa (April) for 15 year running window. Horizontal dashed line shows correlation coef®cient signi®cant at 5% level
Stratospheric zonal wind and temperature in relation to summer monsoon rainfall over India
between these four parameters and the ISMR is now analyzed by applying a 15 year running window. In this process, 15 years of data (starting from the ®rst year) from both series are considered in a sliding window for calculating the C.C. s. The C.C. s obtained are plotted against the middle year of the sliding window (Fig. 3). The 10 hPa zonal wind for Balboa during January (Fig. 3a) shows a decline in the relationship with ISMR and looses its signi®cance during the most recent decade. The 10 hPa zonal wind for Ascension during January (Fig. 3b) shows a comparatively stable relationship with the ISMR, with a slight decrease in its relationship during the recent decade. However, the 10 hPa temperature for Ascension during May (Fig. 3c) shows a stable relationship with the ISMR over the period of analysis. Also, the relationship of this parameter with the ISMR is found to be highest (signi®cant at 1% level) among all the stratospheric parameters examined. The 30 hPa zonal wind for Balboa during April (Fig. 3d) shows almost a stable relationship with the ISMR with slight a decrease in its strength during the last decade of the analysis. Since Balboa data are no longer reported, the parameter does not have practical utility for long range prediction of ISMR. The above analysis thus shows that the association of the 10 hPa temperature (May) at Ascension with the ISMR is quite
119
stable over time and it can be useful for the longrange prediction the ISMR. The interannual variability (standardized) of the ISMR, as well as the 10 hPa temperature for Ascension (May), is presented in Fig. 4. It can also be seen from Figure 4 that the variability of the ISMR is in closer agreement (in opposite phase) with the variability of the 10 hPa temperature for Ascension (May). The parameter is further examined for the predictive importance of drought or ¯ood over the Indian subcontinent by means of Contingency Tables. For this purpose the ISMR is classi®ed as drought (¯ood) if the standardized ISMR is found to be less (more) than 0.5 standard deviation, and normal otherwise. Similarly, for Ascension at 10 hPa, temperature (May) is classi®ed as cold (warm) if the standardized Ascension 10 hPa temperature is found to be less (more) than 0.5 standard deviation and normal otherwise. The Contingency Table of the frequency of occurrence of these three categories of the ISMR in relation to the corresponding frequency of occurrence of Ascension 10 hPa (May) temperature is shown in Table 1. The 2 computed from the contingency Table is found to be signi®cant at the 5% level. Hence, it appears that the 10 hPa temperature for Ascension during May can be useful for predicting the occurrence of ¯oods and droughts over the Indian subcontinent.
Fig. 4. Interannual variability (standardized) of the Indian summer monsoon rainfall as well as Ascension temperature at 10 hPa (May)
120
S. D. Bansod et al.
Table 1. Contingency table (based on Fig. 4) of frequency of occurance of drought, ¯ood and normal rainfall over India during monsoon (June to September) season vs cold, normal and warm temperature in May for 10 hPa at Ascension Cold Normal Warm Total
Flood
Normal
Drought
Total
8.0 3.0 1.0 12.0
4.0 4.0 0.0 8.0
0.0 2.0 9.0 11.0
12.0 9.0 10.0 31.0
2 21.7, signi®cant at 5 per cent level.
The negative relationship found between ISMR and Ascension temperatures indicates that when the stratospheric temperature anomaly for Ascension at 10 hPa during May is positive (negative), the ensuing ISMR is likely to be below (above) normal. The physical cause of the relationship of this parameter with the ISMR is dif®cult to determine at this stage. However, it seems that cooling (warming) of the stratosphere at 10 hPa over Ascension during the pre-monsoon month (i.e., in May) enhances (weakens) the development of the monsoon circulation over the Indian subcontinent and consequently the ISMR is likely to be more (less) than normal. 6. The spatial in¯uence of stratospheric temperature on the ISMR The ISMR shows large spatial variability and as such it is not equally predictable over the entire region by one single empirical method. In order to examine the in¯uence of the 10 hPa Ascension temperature (May) on the spatial variation of the summer monsoon rainfall, the parameter has been correlated with the monsoon rainfall of all the 2 lat./lon. grid boxes for the contiguous India. The parameter shows signi®cant relationships (Fig. 5) with monsoon rainfall for a largearea of the Indian subcontinent, in North-west India. The ®rst principal component of the subdivisional scale summer monsoon rainfall also shows high loading for this region (Prasad et al., 1988) and the region is thus found to be highly spatially coherent. Hence, the 10 hPa temperature for Ascension during May appears to be a precursor of the summer monsoon rainfall not only for the whole of India but also for a smaller region in North west India. A suitable empirical model can be developed for the long-range
Fig. 5. The correlation coef®cient of Ascension temperature at 10 hPa (May) with the summer monsoon rainfall of each of the 75 blocks of India. Hatching shows correlation coef®cient signi®cant at 5% level
prediction of the monsoon rainfall for these two cases using the parameter in combination with other regional and global parameters. This will be the forms of future work. 7. Conclusions In this study the interannual variability of the ISMR has been examined in relation to variability in the stratospheric zonal wind and temperature. The zonal wind and temperature were analyzed for three equatorial stations at each of the 50, 30 and 10 hPa levels using recent period data. The principle ®ndings of the study are summarized as follows. Three stratospheric zonal wind parameters were identi®ed, showing signi®cant positive lagged relationships with the ISMR. These parameters are the 30 hPa zonal wind for Balboa during April and the 10 hPa zonal winds for Balboa and Ascension during January. The positive relationship found in all these cases implies that in the case of westerly (easterly) zonal wind, the ISMR is likely to be above (below) normal. The 10 hPa temperature for Ascension during May shows a signi®cant negative relationship with the ISMR and is relationship is found to be the
Stratospheric zonal wind and temperature in relation to summer monsoon rainfall over India
highest amongst all the stratospheric zonal wind and temperature parameters examined in the study. A temporal stability analysis revels that the relationship of the zonal wind with the ISMR generally deteriorates during the most recent decade. However, the 10 hPa stratospheric temperature for Ascension during May is found to have a stable relationship with the ISMR and hence the parameter seems to be useful for longrange prediction of ISMR. The result of a Contingency Table further indicates that the 10 hPa temperature for Ascension during May is useful for predicting the occurrence of drought/¯ood over the Indian subcontinent. This parameter is also found to be signi®cantly related to the summer monsoon rainfall of a large spatially coherent region lying in North-west India. Hence, the parameter also seems useful for predicting summer monsoon rainfall over a smaller region of the Indiansubcontinent. The study also reveals that the QBO in the stratospheric zonal wind and temperature pro®les is generally in phase for the three tropical equatorial stations. Acknowledgements The authors are grateful to Director and Dr. S.S. Singh, Head, Forecasting Research Division, Indian Institute of Tropical Meteorology, Pune, for providing the facilities and encouragement. The station rainfall data belonging to the Indian Meteorological Department were obtained from Dr. B. Parthasarthy in processed form. The authors are thankful to him. References Bhalme HN (1972) Trends and quasi-biennial oscillation in the series of cyclonic disturbance over the Indian region. Indian J Meteor Geophys 23: 355±358
121
Bhalme HN, Rahalkar SS, Sikdar AB (1987) Tropical quasibiennial oscillation of the 10mb wind and Indian Monsoon rainfall ± Implications for forecasting. J Climate 7: 345± 353 Ebdon RA, Veryand RG (1961) Fluctuation in equatorial stratospheric wind. Nature 189: 791±793 Holton JR (1968) A note on the propagation of the biennial oscillation. J Atmos Sci 24: 519±521 Mukherjee BK, Indira K, Reddy RS, Ramana Murthy BhV (1985) Quasi-biennial oscillation in stratospheric zonal wind and Indian monsoon. Mon Wea Rev 113: 1421± 1424 Parthasarthy B, Rupakumar K, Kothawale DR (1992) Indian Summer rainfall indices, 1871±1990. Met Mag 121: 174±186 Prasad KD, Singh SV (1988) Large-scale features of the Indian summer monsoon rainfall and their association with some oceanic and atmospheric variables. Adv Atmos Sci 5: 499±513 Raja Rao KS, Lakhole NJ (1978) Quasi-biennial oscillation and summer southwest monsoon. Indian J Met Hydrol Geophys 29: 403±410 Reed RJ, Campbell WC, Rasmussion EM, Rogers DG (1961) Evidence of downward propagating annual wind reversal in the equatorial stratosphere. J Geophys Res 66: 813±818 Reed RJ, Rogers DG (1962) The circulation of the tropical stratosphere in the years 1954±1960. J Atmos Sci 19: 127±135 Quiroz RS (1981) Period modulation of the stratospheric quasi-biennial oscillation. Mon Wea Rev 109: 665±670 Quenouille MH (1952) Associated measurements. London: Butterworth, 241 pp Thapliyal V (1979) Stratosphere circulation in relation to summer monsoon over India. Proc Symb on Hydrology Aspects of droughts. IIT, New Delhi 3±7 Dec 1979, 347± 350 Wallace JM, Holton JR (1968) A diagnostic numerical model of the quasi-biennial oscillation. J Atmos Sci 25: 280±292 Authors' addresses: S. D. Bansod, K. D. Prasad, Indian Institute of Tropical Meteorology, Pune, India (e-mail:
[email protected]); S. V. Singh, National Center for Medium Range Weather Forecasting, New Delhi, India.