WxCrop: Agentic AI for Predicting Agricultural Impacts of Extreme Weather

Forum topic · 小凯 · 2026-10-31

Summary

WxCrop is presented as the first AI agent that judges agricultural hazards using physically interpretable weather indices validated by causal analysis, and the first system combining weather and agricultural data to reliably predict crop loss for both past and forecast dates. Developed by Minghan Yu, Mengyu Dai, and Yixin Wen (arXiv:2510.25913), the system was demonstrated in Florida across hurricanes, freezes, and dry heat, with loss models calibrated on insurance records for each hazard type. Three validation checks demonstrate reliability: the hurricane index detects recorded cyclone days with an AUC of 0.98, held-out years' recorded losses are reproduced within 2%, and independent production records confirm county-level rankings with a rank correlation of 0.63. Users can query the agent for the percentage of expected seasonal value lost per crop group, with every figure sourced. The procedure extends to other hazards and regions, enabling governments, growers, and scientists to rapidly assess agricultural disasters.

Paper Overview

Field: Machine Learning Authors: Minghan Yu, Mengyu Dai, Yixin Wen Published: 2025-10-28 arXiv: 2510.25913

Abstract

Extreme weather is forecast days ahead, yet its consequences for agriculture remain hard to predict: a forecast says where a storm will strike, not which crops will suffer or how much is at risk, and the records that could tell span many sources. The authors present WxCrop, the first agent that judges agricultural hazards from physically interpretable weather indices examined by causal analysis, and the first that combines weather and agricultural information to predict crop loss reliably for past and forecast dates.

The system is demonstrated in Florida on hurricanes, freezes, and dry heat, with a loss model calibrated on insurance records for each hazard. The same procedure extends to other hazards and areas.

Reliability Checks

Three checks show the system's reliability:

The hurricane index detects recorded cyclone days with an AUC of 0.98
Recorded losses in held-out years are reproduced within 2%
Independent production records confirm county-level rankings with a rank correlation of 0.63

Practical Use

Users can ask the agent what fraction of the expected seasonal value was lost for each crop group, and every number comes with a source — enabling governments, growers, and scientists to respond quickly to agricultural disasters.

--- *Auto-collected on 2026-11-01.*