After losing face last year when the monsoon prediction was completely off the mark, this year the Indian Meteorological Department (IMD) has completely revamped its monsoon forecast system. Using the new model, they have predicted that the South West monsoon for 2003 will be below normal — 96 per cent of the Long Period Average (LPA).The new models developed by 24 scientists working for nearly a year has enabled IMD to advance the date of the forecast from May 25 to April 16.For this year, the new models indicate 39 per cent probability of below normal rainfall and 21 per cent probability of drought. The probability of above normal rainfall is 23 per cent, near normal rainfall 14 and excess rainfall is 3 per cent.These findings are based on the newly adopted 8-parameter power regression model that replaces the 16 parametric model which was being used since 1988. The parameters are based on wind, temperature, pressure, El Nino and Snow cover and combined with historical data to arrive at the forecast at the beginning of each year.According to S R Kelkar, director general IMD ‘‘a critical re-valuation of the existing 16-parameter model was carried out by IMD and it was found that correlation of 10 models had rapidly declined in recent year.’’ Four new parameters have been found and six have been retained from the earlier set.Instead of the three categories which they used to give earlier, the new model will give the probability of the monsoon rainfall for the country as a whole in five different categories — drought (less than 90 per cent of the long period average), below normal (90-97 per cent), near normal (98 to 102 per cent), above normal (103-110 per cent) and excess (more than 110 per cent).Even more revolutionary is the fact that a new 10-parameter model requiring data up to June has been developed so that an update can be given on July 15. Earlier, once the forecast was issued on May 25, there was no possibility of a modification. Now a quantitative forecast will be given in the middle of July.More importantly, it will be able to incorporate the advance of the El Nino, which is seen as one of the likely factors behind the 2002 drought. One of the most worrying facts was that the earlier model only allowed the recordings of El Nino temperatures when it was far away from the Indian coast. The other factor is the wind direction and speed in June that can also be accounted for in the new model.Considering the importance of July rainfall to agriculture, a new model that will incorporate data till June has been developed for July rainfall that will use 38 years historical data and current data till June to arrive at separate prediction for July rain.According to the Met department, the two new models have given much better results in hindsight and in hindsight has proven correct in 8 out of 9 drought years.The Met office had came under severe criticism as despite its prediction of normal monsoon for 13 consecutive years, the country suffered its worst drought last year with 14 states reeling under a dry spell. The new models are being introduced with caution. ‘‘This is a test year and if they are not found to be working, we will modify next year,’’ said Kelkar.
