LingVo.club
Level
AI weather forecasts help India's farmers — Level B2 — A couple of people that are in the water

AI weather forecasts help India's farmersCEFR B2

30 Oct 2025

Adapted from Ranjit Devraj, SciDev CC BY 2.0

Photo by EqualStock, Unsplash

Level B2 – Upper-intermediate
5 min
284 words

AI-enhanced forecasting is changing how seasonal weather risks are predicted and shared. In India this summer 38 million farmers received forecasts produced by NeuralGCM, a hybrid model that merges traditional physics-based methods with machine learning. The forecasts arrived about four weeks before the normal monsoon onset and correctly signalled a three-week pause after the monsoon made landfall in early June and moved north. Researchers say the model matched or outperformed conventional physics models and other AI approaches while requiring less computing power.

The University of Chicago team, including Pedram Hassanzadeh, argues that independent benchmarking is essential. Chicago recently received Gates Foundation support to benchmark existing models over East and West Africa, with attention to rainy seasons and heatwaves. The Human-Centred Weather Forecasts initiative launched this year and now partners five countries; the team plans to add ten more countries in 2026 and 15 more in 2027, so the project could benefit some 30 countries and millions of farmers.

Project partners note clear economic potential. The forecasts were used to advise farmers on planting and timing, and Michael Kremer estimated more than US$100 for farmers for every dollar invested by the government. Agricultural scientists urged stronger links between rainfall signals and measures such as soil moisture, vapour pressure deficit, heat-stress forecasts and crop-stage sensitivity, warning that a wrong early-onset forecast can cause seedling loss, re-sowing costs and lost growing time.

Researchers are also training meteorologists in low- and middle-income countries to use AI models. Hassanzadeh called the current progress a scientific achievement and the start of wider AI-driven change in forecasting, while noting that technical and practical challenges remain as the project scales up.

Difficult words

  • hybridA model that combines two different methods
  • physics-basedBased on physical laws and scientific equations
  • benchmarkTo compare a model's performance against standards
    benchmarking
  • initiativeA new organized program or project
  • vapour pressure deficitA measure of air dryness affecting plant water loss
  • seedlingA young plant recently sprouted from seed

Tip: hover, focus or tap highlighted words in the article to see quick definitions while you read or listen.

Discussion questions

  • How could these AI-enhanced forecasts change farmers' decisions about planting and timing? Give reasons and examples from the article.
  • What are the main benefits and risks of scaling this forecasting project to many countries?
  • Why is independent benchmarking important for weather models like NeuralGCM, and how could it affect trust in forecasts?

Related articles