The rains have been good during this year's south-west monsoon. The season ended with the country, as a whole, receiving one per cent more rain than the long-period average. The rains have been geographically well distributed too, with the north-west, the central region, and the southern part of the country getting more than average rainfall. Only the north-east recorded a deficit. Consequently, the country is looking forward to a bumper harvest in the kharif season. In this optimistic scenario, it is easy to forget that the India Meteorological Department's forecasts had gone awry. In mid-April, the IMD put the countrywide seasonal rainfall at 98 per cent of the long-period average, with an error bar of five percentage points. In June, the agency revised it as 95 per cent with an error bar of four percentage points. While the country received excess rain in June, the rains in July were considerably below par. By the end of July, the nationwide deficit for the season stood at about four per cent. On August 1, the IMD declared that countrywide rainfall during August-September was likely to be 90 per cent of the long-period average, with an error bar of eight percentage points. This forecast raised the possibility of the monsoon fizzling out. But, as things turned out, the rains were well above average during both August and September.

In its end-of-season report, the IMD agreed that its operational long-range forecasts had not been “very accurate.” It explained that the sudden re-emergence of a La Nin˜a — the cooling in the equatorial Pacific Ocean that is usually beneficial for the monsoon — led to increased rainfall in the second half of the season. What is clear is that the department needs to re-evaluate the parameters that go into the statistical model used for its seasonal predictions. These parameters have remained unchanged for the past five years. It is possible that the correlation of some of them with the monsoon's outcome has dropped sharply and must therefore be replaced with new ones. There is also a good case for focussing on probabilistic forecasts. The IMD's April forecast indicated a 93 per cent probability that the monsoon would be ‘normal', with the nationwide rainfall between 90 per cent and 110 per cent of the long-period average; in June, that probability had dropped only to 80 per cent. Rainfall data for over a century show that seven years out of ten fall into this category. Thus, the probabilistic forecasts showed that this year the odds favoured a normal monsoon. And that, mercifully, is just how things have turned out.