AI weather forecasts and their economic value
Weather forecasting, precipitation forecasting and typhoons appear on 45 recent papers, mostly in AI journals, and only 9 carry an economic, energy, agricultural, insurance or market keyword. Machine learning models now beat the physics-based systems on many measures at a fraction of the cost. Who captures the value, and does it reach the farmers, utilities and insurers who need it?
What prompted this
Since October 2025, 45 papers carry a weather-forecasting, weather-prediction, precipitation-forecasting, numerical-weather or typhoon keyword (35 in AI journals, 7 in economics): "weather forecasting" on 21, "precipitation forecasting" on 8, "typhoons" on 5, "numerical weather prediction" on 3. Only 9 of the 45 also carry an economics, finance, energy, agriculture, insurance, commodity, value, market, cost or decision keyword.
Questions
- For which decisions (electricity trading, crop planting, flood warnings, airline operations) does the accuracy gain of machine learning forecasts translate into money, and how much?
- If a few firms own the best forecasts, do they sell them, trade on them, or both, and does public forecasting lose funding?
- Do the new models forecast the extremes that matter for insurance and disaster response as well as they forecast the averages?
Data that could help
Public forecast archives from both physics-based and machine learning systems, decision records from utilities and farms, and market prices sensitive to weather.
Next step
Compare the value of physics-based and machine learning forecasts for day-ahead electricity trading in one market over a year, using realized prices.