Researchers at IIT Delhi create a model that forecasts monsoon conditions

The India PrintJune 16, 2023

Researchers from the DST Centre of Excellence in Climate Modelling at IIT Delhi, in collaboration with IIIT Delhi, MIT in the United States, and JAMSTEC in Japan, have created a machine learning model for predicting monsoon rainfall.

According to the All India Summer Monsoon Rainfall (AISMR) model created by a research team at IIT Delhi lead by Prof. Saroj K. Mishra, the country would see a typical monsoon in 2023 with an AISMR of 790mm.

It has been shown that the created and tested AI/ML model outperforms the nation’s current physical models for monsoon forecasting. It has shown a remarkable prediction success rate of 61.9% over the test period of 2002–2022 in this study. This is based on the model’s ability to anticipate the AISMR within +/-5% of the actual values observed each year.

A small group of people running these models quickly on a laptop computer may provide a more accurate monsoon rainfall prediction than the more resource-intensive process required for standard physical models.

“This study holds immense significance for the whole nation, as an accurate monsoon forecast well in advance is pivotal for making key decisions in various socioeconomic sectors, including agriculture, energy, water resources, disaster management, and health,” said Prof. Saroj K. Mishra.

Prof. Mishra said that state-by-state monsoon rainfall forecasting would be made possible in order to enhance the data-driven approaches’ use for regional applications.

A model that was trained using historical AISMR data, Nio3.4 index data, and categorical Indian Ocean Dipole (IOD) data for the years 1901–2001 is used to make the forecast.

The prediction using the AI/ML model may be made months in advance and updated properly based on how they are evolving, depending on the availability of the Nio3.4 index and IOD forecast. Data-driven models are thus more input-responsive, computationally less expensive, and better able to capture the nonlinear interactions among the monsoon components.