“The absolute error of the IMD’s monsoon forecasts for the seasonal rainfall over the country has shown significant decrease after the implementation of the new statistical ensemble forecasting system in 2007 compared to previous years,” said Dr Harsh Vardhan, minister for earth sciences, government of India.
The average difference between what the IMD predicted and what actually happened has decreased to 5.9% (of average) since the new statistical model was introduced in 2007 compared 8.5% before it, Vardhan added.
“This clearly indicates improvement made in the operational forecast system in the recent period compared to the earlier period,” he said, adding that IMD was able to anticipate the rain deficit seen in 2014 and 2015.
The agency makes a prediction of rainfall based on data from the last 100 years on the relationship between various factors, such as overland temperature and sea temperature.
However, there has been some move towards replacing the statistics-based model with an input-based dynamic model.
A new dynamic prediction system, which uses realtime data fed into a computer, has been in testing for several years, though the agency is yet to switch to the new method for making its primary forecast in April.
“IMD is working to improve the skill of this dynamical prediction system in collaboration with Indian and foreign academic institutions,” Vadhan said today.