Part of the government's R400 crore monsoon mission, the Met department has been concurrently using the dynamical as well as the existing statistical model for providing long-term monsoon rain forecast since 2012.
The department plans to use only the dynamical model for long as well as short range forecast by 2017, thus replacing the statistical model which had been proven to lack consistency and reliability. This is an experimentation stage and last year, the dynamical model's forecast was accurate. We are building our capacities for operating the dynamical model at present, Shailesh Nayak, secretary, ministry of earth sciences, told FE.
Nayak said that last year, while the statistical model had predicted monsoon rains to be 98% of the long period average (LPA), calculated on the basis of a 50-year average rainfall of 89 cm, the dynamical model had predicted monsoon rains to be 106% of LPA, which was right on target.
This year, the dynamical model has predicted rains to be 96% of LPA in its second stage forecast released earlier this month, while the statistical model has predicted the monsoon to be 93% of LPA. IMD, in its second stage forecast, has predicted monsoons to be 95% of LPA.
The statistical models had not been proving too accurate in the last few years, Nayak noted. The statistical model makes seasonal climate forecast using data generated using information received from agencies in the US, Europe and Australia. Dynamical model of prediction involves mathematical simulation of the atmosphere on a computer, which needs skilled manpower.
The government had decided to upgrade the monsoon forecast system following IMD's inability to predict the severity of the deficiency in monsoon rains in 2009.
In 2009, the IMD had given a forecast that monsoon would be 'normal' at 96% of LPA in its first prediction in April. In its second prediction, the department had said rainfall would be 93% of LPA, which was far off the mark.