Editorial: The hope of rain

For the India Meteorological Department (IMD), it has been an annual ritual for the past 20 years to issue its long-range monsoon forecast around this time of the year, though the credibility of these predictions has been on the wane. The projection that the rainfall during this year's monsoon season (June to September) will be equal to 99 per cent of the long-period average, should therefore be viewed with some caution, even if prima facie it is a good news for the country's farmers, economy and the hydrological balance. Last year, the IMD in its long-range forecast had reckoned the rainfall to be below normal by 6 per cent. In the event, it turned out to be above normal by the same margin. Since 1988, the IMD's prognosis hit the mark in only the first six years, when the 16-parameter power regression statistical monsoon prediction model (often called the Gowarikar model) proved accurate. Subsequent projections tended to be off the mark, leading to the discarding of this model in 2002, when it failed to forecast the drought that ensued.

None of the statistical models evolved and tried out since then has been able to inspire much confidence in the accuracy of the monsoon prediction. The most significant modification in the monsoon prediction procedure was effected last year. Instead of a single mode, two new statistical ensemble systems were deployed "" one for compiling the preliminary forecast issued in April, and the other for subsequent, more fine-tuned forecast in June that contained the region-wise rainfall estimates as well. However, this failed in its very first year to make a correct prediction, particularly for the central and southern regions, but also for the country as a whole. While the model had projected a 4 to 6 per cent shortfall in monsoon precipitation in these two regions, the actual was 8 per cent above normal in the central zone and a whopping 26 per cent excess in the southern peninsula. Chastened by this experience, the IMD has now used last year's model for computing the prediction for only the north-west and the north-east, the regions for which it gave better results last year. For the central region and southern peninsula, it has deployed a dynamic model of the kind in use in most developed countries but still in the trial stage in India. The results of the two different systems have been amalgamated to arrive at the rainfall reckoning for the country as a whole. The effectiveness of this new approach will be known in a few months.

Meanwhile, one thing is clear. Despite fresh investments in data collection and data-processing capabilities, the IMD is still experimenting and does not have a reliable monsoon prediction system. Its disabilities in this respect become all the more glaring when contrasted with meteorological agencies in other countries which are able to come out with far more precise forecasts for the Indian monsoon than the IMD has been managing to do, though it is the IMD that has access to much more country-specific field data. The question that arises, therefore, is whether the IMD should continue with its solo experimentation in monsoon prediction, or collaborate with other agencies and learn from the experience. There is no great merit in being nationalistic about weather forecasting; the important thing for farmers, agro-based industries, economic administrators and the country in general is that the forecasts should be reliable.