AI-Based Monsoon Forecast Pilot Guides Crores of Farmers Across 13 States

Key facts

• An AI-based pilot used an open-source blended model combining Google’s NeuralGCM, the European Centre for Medium-Range Weather Forecasts’ ECMWF Artificial Intelligence Forecasting System (AIFS), and 125 years of India Meteorological Department (IMD) rainfall data to generate probabilistic forecasts of local monsoon onset.
• The Development Innovation Lab-India collaborated on the pilot that covered agriculturally relevant areas across parts of 13 states for Kharif 2025.
• Probabilistic forecasts of local monsoon onset were sent via SMS through the M-Kisan portal to 3,88,45,214 farmers in five regional languages—Hindi, Odia, Marathi, Bangla, and Punjabi.
• Telephonic farmer feedback surveys conducted through Kisan Call Centres in Madhya Pradesh and Bihar revealed that 31–52% of farmers adjusted planting decisions, primarily altering land preparation and sowing timing, including crop and input choice.