Overview
- India’s Ministry of Agriculture, working with University of Chicago researchers, operated a nationwide SMS service that sent localized month-ahead monsoon guidance to 38 million farmers this season.
- The system blended Google Research’s NeuralGCM with ECMWF’s AIFS and historical records to generate efficient, high-resolution forecasts.
- Project reports say the models anticipated monsoon onset up to a month in advance and identified a midseason dry spell during an unusual season.
- Messages converted technical forecasts into decisions for farmers, including when to plant, whether to buy more seed, switch crops, or wait.
- Backers note the low computing needs that enable forecasts on smartphones and see potential for replication in other developing countries, while income gains cited stem from prior research rather than independent evaluation of this rollout.