The Met forecast that this year’s monsoon is going to be “normal” will provide reassurance to many, since the country has just come through a drought year. But the reliability of the Met office’s long-range monsoon predictions has been so poor that fingers should continue to be crossed. Bear in mind that met officials failed to predict the droughts of 2002, 2004 and 2009. Last year, the Met’s first-stage long-range rainfall projection was 96 per cent of normal rainfall. An updated forecast of 93 per cent precipitation, made after the onset of the monsoon in June, also turned out to be too optimistic. The unprecedented step of revisiting the forecast in August, when the monsoon season was already more than half way through, resulted in an anticipated rainfall figure of 87 per cent. In the event, actual rainfall was just 77 per cent of normal. So, when the same Met people forecast 98 per cent of normal precipitation this year, bear in mind this track record.
The truth is that the forecasting ability of the Met department has shown little or no improvement over decades. As much has been acknowledged by the weather scientists themselves in some of the research papers published in recent years. The only bright spot was a brief period, between 1988 and 1993, when the Met generated fairly accurate long-range forecasts by using a 16-parameteric power regression model. That model too proved unreliable subsequently, and had to be discarded after it failed to give any indication of the 2002 drought, which saw a 19 per cent deficiency in rainfall, against the Met prophecy of it being 1 per cent above normal. Since then the Met has been tinkering with various monsoon prediction models. The model used for generating this year’s rainfall prediction of 98 per cent is the same five-parameteric statistical model that was used last year. Introduced in 2007, this model has already erred twice. What confidence can such a record inspire?
The Met department has invested in strengthening its data-collection infrastructure, instrumentation and forecasting techniques after the 2002 drought. It now possesses a fairly dense network of satellite-based automatic rain gauge stations for online monitoring, an augmented network of upper air observations, an intensive infrastructure of S-band Doppler radars for complete coverage of coastal areas and more C-band storm detection radars, besides, of course, better computing facilities for faster data processing. Failure after all this points to plain incompetence.
