India Launches AI-Based Pilot to Improve Local Monsoon Forecasts for Kharif Sowing
The Government of India has launched an artificial intelligence (AI)-based pilot programme to provide localised early monsoon forecasts to support decision-making for Kharif sowing. Implemented across selected districts in 13 states, the initiative aims to help farmers better plan land preparation and sowing dates. Feedback collected after the pilot indicates that a significant proportion of farmers adjusted planting-related decisions based on the forecasts, highlighting the potential policy value of integrating AI into agricultural advisory systems.
The pilot was implemented with support from the Development Innovation Lab–India and is based on an open-source hybrid forecasting model. According to details published by the Press Information Bureau, the system combined neural General Circulation Models with datasets from the European Centre for Medium-Range Weather Forecasts, its Artificial Intelligence Forecasting System, and the India Meteorological Department, drawing on 125 years of historical rainfall data.
Localised forecasts focused on monsoon onset
The probabilistic forecasts generated through the pilot focused specifically on predicting the local onset of the monsoon, a critical factor in determining sowing timelines. The initiative covered agricultural regions in 13 states for the Kharif 2025 season, targeting actionable guidance rather than general seasonal outlooks.
Forecast information was disseminated to 38845214 farmers through the mKisan portal using SMS alerts in five regional languages: Hindi, Odia, Marathi, Bengali and Punjabi. This approach reflects broader efforts to embed digital tools into extension services, as outlined in ongoing discussions on AI-enabled smart agriculture in India.
Farmer feedback and observed behavioural changes
Following the dissemination of forecasts, feedback surveys were conducted via Farmer Call Centres in Madhya Pradesh and Bihar. The surveys indicated that between 31% and 52% of respondents adjusted their sowing-related decisions after receiving the information.
Reported changes included modifications to land preparation schedules, sowing dates, crop selection and input investments. These findings suggest that timely, localised climate intelligence can influence on-farm decision-making, reinforcing policy arguments for scaling AI-driven advisory services within India’s agricultural ecosystem.
Integration into national forecasting systems
Based on the outcomes of the pilot, the government has decided to operationalise this capability within a national system. The approach will leverage collaboration between the Indian Institute of Tropical Meteorology, the India Meteorological Department and the Indian Space Research Organisation to develop an equivalent internal technical solution.
Dynamic models developed by the Indian Institute of Tropical Meteorology demonstrated stronger performance in capturing local conditions and are being integrated into the AI framework for the 2026 season. This step aligns with broader policy thinking on how frontier technologies could reshape agricultural planning, as explored in analyses of AI-driven agricultural transformation and India’s wider frontier technology roadmap for agriculture.
The information was provided by the Minister of State for Agriculture and Farmers Welfare, Bhagirath Choudhary, in a written reply to the Lok Sabha, underscoring the government’s intent to embed AI-based forecasting tools into formal agricultural support systems.
