New Delhi: State-run weather forecaster India Meteorological Department (IMD) is soon going to take a call on shifting from a statistical model that it has used for almost 100 years to make its monsoon forecast to the coupled dynamical model, deemed to be more accurate and run on a supercomputer. While the dynamical model has been used on an experimental basis in the past few years, Indian Institute of Tropical Meteorology (IITM) scientists in Pune are now confident that it can be used as the main operational model to make India’s monsoon forecast. IITM works closely with IMD on forecast technology.

“In another couple of months, IMD will take a call on whether it will shift to a dynamical model after a review of its performance," said M. Rajeevan, secretary in the ministry of earth sciences.

“Earlier, the dynamical model had poor accuracy, but over the years it has become much better and scientists at IITM have become confident that it can be used," Rajeevan added.

IMD issues its forecast for the June-September monsoon season in April, which is based on the Ensemble Statistical Forecasting System that uses five predictors. These predictors along with some equations are used by scientists to carry out a statistical analysis to finally come out with a forecast.

The problem is that this method uses data from the past 100 years to come up with a forecast when many of the relationships between global climate factors have changed.

The dynamical forecast system, on the other hand, is based on current physical observations of the atmosphere, cloud properties and oceans, which can then be used to get a forecast. The observations are used to run the model on a supercomputer at IITM Pune to arrive at a rainfall forecast.

IITM has been working along with various climate research centres in India and abroad on the development of a coupled dynamical model for forecasting the Indian summer monsoon under the aegis of the Earth System Science Organization’s Monsoon Mission project.

“We have already used the dynamical model for temperature forecasts. We have been working on improving the dynamical model step by step over years. It has performed well in the deficient monsoons in the past two years," said a senior IMD scientist, who did not want to be named. “This model can also give a better forecast for spatial distribution of rainfall," the scientist added.

The experimental forecast based on the coupled dynamical model said that the monsoon rainfall during 2016 is likely to be 111% of the long-period average (LPA) with a margin of error of 5%. The operational forecast using the statistical method forecast that the monsoon this year would be 106% of the LPA with a model error of 5%.

Close