Once again Indian Meteorological Department (IMD) has changed its monsoon forecast model. The Met agency for the first time will use a dynamic model to forecast the July rainfall for the southern peninsula and the North-East. But it will use a six-parameter statistical model for the North-West and Central zones. This is the third change in the past six years. The 16-parameter statistical model, which IMD used for 14 years  since 1988, was dumped after its failure to predict the 2002 drought. The number of parameters was reduced to eight and ten for April and July forecast respectively and the model was overhauled. When it also failed in central India in 2006, the number of parameters was further reduced to five and six.  But the reduction could not anticipate the heavy 2007 rainfall in the peninsula. In what seems to be a knee–jerk reaction, IMD now dumps the statistical model for the southern states and embraces the dynamic model.

But instead of such frequent modelling changes, which chips away at people’s confidence with IMD forecasts, introspection is required to find out if the Met agency has necessary wherewithal and human resources for the onerous task of predicting the southwest monsoon, which is a complex land-sea breeze phenomenon. There is a misplaced conception that after IMD made long range forecast public, it successfully predicted “normal” monsoon for 13 years. But except 1989, 1991, 1992 and 1995, the forecasts were always beyond the model error of four per cent. Since the 1990s witnessed bountiful rains nobody complained. But with a number of extreme weather events in the last six years, forecasting flaws are becoming evident as none of the forecasts could foretell any of those events.

The shift towards dynamic models is unlikely to solve the problems. Though dynamic models are more scientific, they require extensive data, which are simply not available with the IMD. In the absence of an accurate model, IMD needs to consult with other scientific agencies to find out an acceptable alternative.  Even if it takes a few years to work on the alternative, IMD must focus on the substitute and perhaps even consider suspending the long range forecast temporarily. In fact, even before 1988 the forecasts were made but they were never meant for public consumption. The  Met agency will do well if it tries to find out how the public announcement of monsoon forecast helped the farmers and the economy.