The weather mart

Met department is not the only one predicting weather in India. Forecasts are now sold by businesses to businesses

At the peak of summer weathermen in Pune sat down to predict when the monsoon will hit India and how much rainfall it will bring. They fed data into forecast models, and out came the result: rainfall will be 96 per cent of the normal, they announced on April 17. But in June the country received only 57 per cent of the normal rainfall for the month.

The government-run India Meteorological Department (imd) in its internal communication had said the June rainfall would be considerably less than normal, yet neither imd nor farmers across the country expected it to be so low. "The last time such a low rainfall was recorded in June was in 1926," said D S Pai, scientist at imd Pune who is in charge of forecasts.

On June 24, imd modified its forecast, saying the monsoon will be 93 per cent of the normal. Its scientists said their earlier forecast followed statistical probability that relies on historical data, not the prevailing weather conditions, to predict the trend. They did suspect that El Nino, a periodic weather anomaly, and heavy snowing in Eurasia would play spoilsport. But, according to Pai, two things happened in May that took them by surprise: Aila cyclone hit West Bengal and western disturbance hit north India. Aila took away a large part of the moisture in the air and western disturbance brought drizzle cooling down the northern parts of the country. This weakened the monsoon, said the scientist.

Is weather forecast on the basis of historical data relevant any longer? Research shows trends of decreasing rainfall in June and July (see 'Monsoon earlyburly', Down to Earth, May 16-31, 2009). "We try to take the results from 10 dynamical forecast models, and often each model gives a different version of the weather," said Pai. "Unsure of the results we settle for the age old statistical model." Dynamical models take into account weather conditions like cloud movement and humidity.

Several imd scientists, who do not want to be named, say dynamical models are not reliable because they are not customized to Indian conditions. These models also need regular updates, which require research. imd scientists complained that though such research is happening in organizations like the National Centre for Medium Range Forecasting, iits, Indian Institute of Science and the Indian Institute of Tropical Meteorology, there is not enough interaction between the researchers and forecasters.

"There is almost a caste-like division among meteorological researchers in other organizations and forecasters in imd. So much so they do not publish in same journals," said an imd scientist. "Interaction is limited to seminars."

imd is not very happy about its seasonal predictions meant for the government to formulate its agricultural strategies. But it is confident about its medium- and short-range forecasts given five days and one day ahead respectively.

Enter the private forecaster
Knowing which way the wind blows-or does not-is not left to imd any more. Over the past six years a clutch of companies in India has entered the business of predicting and recording the weather every day, every hour, every 10 minutes. It depends on how frequently the client wants the data. These companies cater to the specialized needs of increasingly competitive sectors like power.

Source:: http://www.downtoearth.org.in