The skewed pattern of rainfall experienced across the country during this year's monsoon appears to be an anomaly.
UNPREDICTABLITY of the monsoon, unusual spatial and temporal distribution of rainfall over the four-month period, shifting rainfall patterns, sustained low rainfall activity, drought-like conditions in some areas and excessive rainfall in others, and the probable impact of global warming have, in the past, given rise frequently to questions about whether or not we are witnessing permanent or quasi-permanent changes in monsoon behaviour. These questions have resurfaced in the wake of the unforeseen droughts of 2002 and 2004 and the skewed distribution of 2005 and 2006.
Some recent studies on long-term trends do indicate that there are discernible changes in the rainfall pattern over the country, if not in the gross scale for the country as a whole and for the entire monsoon period, but at smaller scales of space and time. One such significant study is by P. Guhathakurta and M. Rajeevan of the National Climate Centre (NCC) of the India Meteorological Department (IMD) in Pune. Whether these changes have anything to do with climate change and its impact on regional and global meteorological forces, and whether they are here to stay, are questions requiring more detailed and continued investigation.
The basic principle of the NCC study is that past performance of the monsoon may hold clues for future scenarios. The scientists looked at the rainfall time series of 36 meteorological subdivisions for the period 1901-2003 constructed from a fixed but large subset of 1,476 rain gauge stations (in 524 districts) from the bigger national network. The criterion adopted for the selection of a given station was that the number of years for which data are missing or unreliable from the station should be 10 per cent or less. The missing data in these were replaced by data from neighbouring rain-gauge stations.
According to the scientists, the network considered in this study is more uniformly spaced and temporally homogeneous than earlier studies, which were all based on the rainfall series constructed by B. Parthasarathy and associates of the Indian Institute of Tropical Meteorology (IITM), Pune, in 1994 using data from 306 stations that excluded hilly regions. Though the data set under study has been periodically refined, the network size, and hence the spatio-temporal coverage and the sensitivity of the analyses based on it, remains limited, argue Guhathakurta and Rajeevan.
The rainfall for a meteorological subdivision was calculated in the present study as the area-weighted average of the rainfall of districts within a sub-division. Long-term trends in rainfall over different subdivisions and monthly contributions of each of the monsoon months to seasonal and annual rainfall were examined. The study has found that while rainfall for the country as a whole does not show any significant changes in long-term trends, there are discernible changes with regard to annual and seasonal rainfall and in the monthly rainfall activity in some subdivisions and areas.
As per the data used in the study, the mean monsoon rainfall (June 1-September 30) over 1901-2003 was 87.72 centimetres, which accounted for nearly three-fourths of the mean annual rainfall of 118.28 cm. July rainfall contributed the maximum (of 24.2 per cent) to the annual rainfall while August contributed 21.2 per cent, September 14.2 per cent and June 13.8 per cent. Pre-monsoon and post-monsoon rain contributed nearly equal amounts to the balance. (In terms of monsoon rainfall, the respective contributions of July, August, June and September works out to 32.6 per cent, 28.6 per cent, 18.6 per cent and 19.1 per cent.)
The findings of the NCC study suggest that, the erratic monthly rainfall distribution of the 2006 monsoon was only an aberration and not reflective of any definitive long-term trend. Through analysis using different statistical tools on the time series data, the study has found that for the country as a whole, the monsoon rainfall and monthly rainfall for the monsoon months do not show any significant trend. However, trend analysis of the monthly rainfall series and the season as a whole for all the 36 subdivisions shows "significant and remarkable variations on the regional scale".
The authors have found that in July, six subdivisions show decreasing rainfall trends and eight subdivisions increasing rainfall trends. The regions with decreasing trends include most parts of central and peninsular India while a significant increasing trend can be discerned in the northeastern parts of the country. In August, four subdivisions show decreasing rainfall trends and 10 increasing rainfall trends. Subdivisions showing "significant" (95 per cent statistical significance) decreasing trends include Chhattisgarh, Jharkhand and Bihar, and those showing significant increasing trends include Konkan and Goa, Marathwada, central Maharashtra, Vidarbha, western Madya Pradesh, Telengana and western Uttar Pradesh. The months of July and August together contribute over 60 per cent of the monsoon rainfall. Rainfall in June, which accounts for about 19 per cent of the seasonal rainfall, has been found to have increased in the western and southwestern regions and decreased trend for Central and Eastern parts of the country.
For the season as a whole, while Jharkhand, Chhattisgarh and Kerala have experienced a significant decreasing trend, Gangetic West Bengal, western Uttar Pradesh, Jammu and Kashmir, Konkan and Goa, central Maharashtra, Rayalaseema, coastal Andhra Pradesh, and north interior Karnataka have been found to have significant increasing trends. However, as noted earlier, the skewed pattern of 2006 does not seem to fit into the changing long-term trend revealed by the study and, therefore, must be regarded as a statistical anomaly.
Monthly contributions to the subdivisional annual rainfall too seem to show significant changes. Results suggest that contributions of July and August to the annual subdivisional rainfall have gained importance because of their increasing trend, while July rainfall exhibited a decreasing trend. June rainfall contribution to annual rainfall was found to increase in as many as 19 subdivisions while decreasing in the remaining 17 subdivisions. The contribution of July rainfall, on the other hand, seems to be decreasing in central and west peninsular India. Interestingly, August rainfall seems to be increasing in those very subdivisions where there is a decreasing July contribution. "Therefore," observe Guhathakurta and Rajeevan, "we see a major shift in rainfall pattern spatially during the recent years."
Though the southwest monsoon months form the major rain-producing season over the country, other months of the year produce significant rainfall in some specific areas. For instance, the northeast monsoon (October-December) makes a significant contribution to the rainfall in the southern States. Similarly, western disturbances and convective processes produce significant rainfall during winter and pre-monsoon months in some other areas. The NCC study also includes a trend analysis of subdivisional rainfall time series for winter (January-February), pre-monsoon (March-May) and post-monsoon (October-December) months.
There is a decreasing trend in rainfall in all subdivisions except for Himachal Pradesh, Jharkhand, Nagaland, Manipur, Mizoram and Tripura during the winter season. In fact, this is significant in as many as 18 subdivisions, particularly in the southern States. During the pre-monsoon season, there is a decreasing trend in rainfall in most parts of central India, implying thereby that the convective activity responsible for pre-monsoon showers in these parts is decreasing. However, post-monsoon rains seem to be increasing in most of the subdivisions.
The combination of trends in the different seasons obviously affects the total annual rainfall in the various subdivisions. For Chhattisgarh, Jharkhand and Kerala, there is a decreasing trend in rainfall at 99 per cent statistical significance. At 95 per cent level, as many as 12 subdivisions show a decreasing trend. Besides the above three, these include Himachal Pradesh, Uttaranchal, eastern Rajasthan, eastern Uttar Pradesh, eastern Madhya Pradesh, Vidarbha, Bihar, Orissa and Tamil Nadu.
A decrease in annual rainfall can be observed in the case of Konkan and Goa, central Maharashtra, north interior Karnataka, Rayalaseema, coastal Andhra Pradesh, Gangetic West Bengal, Assam and Meghalaya and Jammu and Kashmir.
The Indian monsoon rainfall is known to display multi-decadal variations with clustering sets of wet and dry monsoons. This epochal variation was examined again with the new and better data set. A calculation of 31-year running means of monsoon rainfall revealed this pattern of alternating sequence of multi-decadal periods of frequent drought and frequent flood years: 1901-30 (dry), 1931-60 (wet), 1961-90 (dry) and 1991-2020 (wet, possibly).
During the dry period of 1901-1930, there were eight rainfall-deficient monsoons and three rainfall-excess monsoons. During the next wet period of three decades, three monsoons were deficient and five in excess years. In the dry period of 1961-90, seven monsoons were deficient and four in excess. Also, significantly, in the entire period 1901-2003 considered (see table), there were more deficient monsoons (19) than excess ones (13). The first decade (1991-2000) has already witnessed a wet period. "Therefore, there is," notes the NCC study, "a chance of wet period for the subsequent two decades [up to 2020]."
This implies that, regional imbalances notwithstanding, the country as a whole will have more normal monsoon seasons than droughts, even though the decade 2000-2010 has already witnessed two drought years, 2002 and 2004. Of course, the long-period data are yet to reveal the impact of a global climate change, if any, on the monsoon. If that were to happen, all the trends that such studies reveal could change radically.