The proposed move by the India Meteorological Department (IMD) to issue only monthly rainfall predictions, instead of the long-range monsoon forecast for the entire four-month (June to September) monsoon season, will disappoint many monsoon watchers, even though it makes sense for several reasons. Given the track record of seasonal monsoon projections, despite the growing sophistication of the data collection and data processing technology, not offering a misleading forecast may be a good idea.
Besides, putting out an estimate of the total quantum of rainfall (by averaging all-India rainfall to a single number), without indicating its spread in terms of time and space, is of little use for agriculture, reservoir water management or any other purpose. The credibility of the IMD’s predictions has taken a beating because of its poor record of accuracy in recent years. The only brief period when the IMD got its projections more or less right was between 1988 and 1999 when its forecasts were computed with the 16 parameters-based power regression model developed by a team of scientists led by Dr VK Gowarikar, originally of the Indian Space Research Organisation (Isro) and later Secretary of the Department of Science and Technology. This model, too, began faltering subsequently and could not forewarn the near 19 per cent rainfall deficiency in 2002. It was ultimately given up after it totally failed to foresee the severe drought of 2004. The new models tried out since then have generally failed to offer reliable forecasts. This year’s fiasco in monsoon forecasting was the worst of all.
This decline in accuracy of prediction stands in contrast to the improvement in information network, technology and data processing systems. Huge investments have been made in Doppler weather radars, meteorological satellites and high speed data communication and computing wherewithal. The IMD now has in place a satellite-based dense network of automatic rain gauge stations and weather data collecting systems for surface and upper atmosphere as well as oceans. What is wanting is a suitable model — statistical or dynamical — for gainfully utilising this data for foreseeing the performance of the monsoon. While the weather forecasting centres in developed countries switched over to dynamical weather prediction models (based on mathematical equations simulating the physics of the global atmosphere) in the 1990s, Indian meteorologists have stuck to the statistical models (based on empirical statistical formulations) since the 1920s. These models are incapable of generating forecasts of higher spatial and temporal resolutions which the users of these forecasts, including agriculturalists, need. While it is true that the simulation of year-to-year variation of a weather system as complex as the monsoon is a challenging task, but this challenge has to be met as it is a prerequisite for the success of any forecasting model, more so for dynamical models. While IMD will no doubt work at improving its capabilities, there may be a case for reaching out globally for technical assistance and support.