Sheltering the rain men

No forecast model developed for the prediction of Indian monsoon at any time, at any place, and by anybody else in the world has come anywhere near the accuracy of the IMD model.

Am I exaggerating? You decide. Here are three fairly representative reality checks for comparison. Some eminent American scientists working on Indian monsoon sent their forecast to the Prime Minister8217;s Office for three years in succession8212;1986, 1987 and 1988.

According to them, the 1986 monsoon was supposed to be 8216;8216;normal8217;8217;; it wasn8217;t. The 1987 rains were to be on the lower side of 8216;8216;normal8217;8217;; they weren8217;t8212;actually, 1987 was one of the worst monsoon years of the century.

Finally, they predicted 1988 to be an average monsoon year. For the first time in its century-old history, IMD disagreed publicly. Its recently operationalised model had indicated that 1988 was not to be an average but an excellent monsoon year. And it indeed was8212;1988 was one of the best monsoon years of the century.

I recall this not to denigrate the American scientists, who are by any standard outstanding. My intention is to underscore the Indian capability in rising to the extreme complexities of the monsoon dynamics.

But will this capability make the forecast come true this year, the 15th year? Let me first recall some relevant points. Unlike African rains, the Indian monsoon rains never fail 100. That8217;s a big big plus, endowed by nature to this country. In the past 123 years 1878-2001, for instance, the rainfall has always been over 72.

Indeed, barring some six years or so 1899, 1918, 1951, 1972, 1979, and 1987 out of the past 123 years, the country has always had more than 80 rains the percentage being of the long period average value of the rains. The average rainfall of the last 31 years 1971-2001 is, in fact, a hefty 98.5.

Many environmentalists say 8216;8216;global warming8217;8217; has given us a good rainfall for the last 14 years consecutively. If that8217;s so, the same 8216;8216;global warming8217;8217; should have given us good rains this year too.

Indeed, if 8216;8216;global warming8217;8217; causes good rainfall, how is it that in the 14 years from 1958-1971, the average rainfall was 102 without global warming ? Again, what about the 14 years from 1931-1945, when the rainfall was again normal despite no 8216;8216;global warming8217;8217;?

Is there any natural climate change or cyclicity in the weather systems? One simple parameter can tell you yes or no: the number of deficient monsoon years in each decade of the last 100 years. Starting from 1901-1910 to 1991-2000, the numbers of deficient monsoon years are 3, 2, 1, 0, 0, 1, 3, 3, 3, 0 respectively. There is obviously no cyclicity nor any indication of a climate change insofar as India is concerned.

The question is, whether anything unusual is occurring this year that wasn8217;t the case before. Clearly this year, the rainfall deficiency till the end of July is high at 30. In 1992, the trend of deficiency till July end was similar but the severity was less minus 12. Yet, the 1992 season ended with a 8216;8216;normal8217;8217; rainfall of 93, compared to the predicted 92. It is thus not uncommon for the second half of the season making up for the deficiency of the first half.

Can this happen this year in the case of a much larger deficiency? It will depend on how the low pressure systems and depressions get formed over the north Bay of Bengal, their movement along the central parts of India and the placement, orientation and movement of the monsoon trough in August and September. None of these factors is predictable. The first 10-12 days of August will be extremely crucial.

There is one positive aspect which is often ignored. In the last few years, the monsoon activity didn8217;t fold up at the end of September; it continued well into October. The rainfall in October doesn8217;t get formally included in the June-September rainfall figures but the extra rains in the extended southwest monsoon, as it were, are always a very valuable bonus for the country as a whole.

There are some newspaper reports saying that the IMD model is untested for extreme drought conditions. Not so. The validity of the model was checked with data of 30 years prior to 1988 which included six years each of the 8216;8216;Century8217;s worst monsoon years8217;8217; category and of the 8216;8216;Century8217;s best monsoon years8217;8217; category see box. In fact, the validity of the IMD model was tested not just in extreme bad weather conditions but also in some of the best weather conditions.

A forecast model is obviously for predicting the future. In its formulation, one has to use past data to determine the various constants of the power regression equation. In determining the constants we used, wherever possible, the data with a wide spectrum of variations to increase the reliability of the model and make it amenable for wider use.

The model has a potential for wider application. E Ray Garnett of Canadian Wheat Board, Winnipeg, Manitoba says, 8216;8216;The Long-Range Weather and Crop Forecasting Group, formed in Canada in 1993, is working toward forecasting the Canadian spring wheat and possibly the USA corn crop a season or quarter in advance, modelling the success that has been achieved in India in forecasting the monsoon in recent years Gowariker et al., 1989; Thapliyal and Kulshreshta, 1992. The approach used in India relies heavily on teleconnections8212;it is felt that a similar seasonal forecasting technique should be developed for the Canadian prairies and possibly the USA corn belt.8217;8217; E. Ray Garnett, The Use of EI Nino Information in Forecasting Grain Yields in the Canadian Prairie Provinces.

The 44 years in a row of the IMD model8217;s 100 success notwithstanding, it is quite possible that there remain some hitherto unnoticed atmospheric forcings which need to be appropriately incorporated in the model. This can be addressed only after seeing the actual data at the end of the season. When that is done, the model will hopefully be even further refined. Until then, let8217;s hope this refinement becomes unnecessary and the country ends up with a normal monsoon for the fifteenth year in succession.

Gowariker was the first author of the paper on the 16-point monsoon model in the IMD8217;s journal Mausam. He is at present a member of the Govt8217;s Consultative Group of Eminent Senior Scientists