This is not the first year in which the long-range monsoon rainfall prediction of the India Meteorological Department (IMD) has gone awry. Most years in the recent past have witnessed the same thing. The failure rate seems to have escalated sharply since the 16-parameter statistical power regression model (also called the Gowarikar model), introduced after the 1987 drought, lost its relevance after yielding accurate forecasts for a few years. This model was replaced with other models after its failure to predict the severe drought of 2002. But none of the new models tried out since then, including the two-pronged statistical model introduced this year, has been up to the mark, either. That having been said, the fiasco of the current model is rather unique. Its first forecast generated in April, based on five predictor factors, said the total rainfall would be 5 per cent below normal, whereas it has turned out to be in excess of the normal by the same margin. To make matters worse, the updated forecast in June, well after the onset of the monsoon and using six predictor factors this time, raised the anticipated deficiency level to 7 per cent "" thus widening the gap between forecast and eventual reality. | |
| The bitter truth is that the available records, dating back to 1932, indicate beyond doubt that monsoon-forecasting capabilities have not improved in these 75 years despite continuous alterations in prediction models. In the past, the IMD could put the blame on inadequacies in the observational network and in data-processing systems. It is difficult for it to do so now. After the 2002 drought, the government pumped in substantial funds to plug the gaps in infrastructure and technology. Gadgets like Doppler Weather Radars, meteorological satellites, and high-speed data communication and computing systems are now available to the IMD. Besides, a new integrated meteorological data reception and analysis system (IMDRAS) has been introduced. This facilitates the use of satellites meant for broadcasting data communication for meteorological purposes too, and to receive processed information from remote areas through portable computer-based workstations. This apart, fresh investments have gone into setting up satellite-based automatic weather surface stations for adequate coverage of weather data and into strengthening the network of upper air observations over land and sea surfaces. All this has been of little avail. |
| |
| The real problem, therefore, seems to be the IMD's persistence with statistical prediction models that the developed countries discarded long ago. It has been argued that these statistical models are incapable of generating forecasts of the high spatial and temporal resolutions that users now need and expect to get. A shift to dynamic prediction models, similar to those used in most developed countries, is inevitable. The question to ask under these circumstances, therefore, is whether the country should continue to make fruitless investments in enhancing the IMD's forecasting skills, or simply outsource this task to some other agency with a better track record. The needs of farmers, fishermen and others are so important that weather forecasting cannot be left in the hands of an organisation that seems incapable of getting its forecasts right most of the time. |
| |
|
| |
