A 2024 Nature Communications study tested whether globally available daily weather datasets can recover crop responses to extreme heat as reliably as fine-scale national data.
U.S. yield responses were similar across three weather datasets
County-level U.S. corn and soybean yields were linked to PRISM, ERA5-Land and GMFD. PRISM had the best out-of-sample predictive skill, but all three datasets recovered a nonlinear temperature response: moderate temperatures can help yield, while extreme heat causes sharp losses.
Models retaining daily temperature extremes projected much larger warming impacts than models using seasonal average temperature. The functional form was more important than the choice among the daily datasets.
The African analysis did not transfer a U.S.-trained yield model
For Sub-Saharan Africa, reliable subnational yield data were unavailable. The researchers therefore used cropland Enhanced Vegetation Index (EVI) as a yield proxy for 2000–2010 and compared ERA5-Land/GMFD with monthly CRU weather data. The two daily global datasets performed better than CRU.
EVI is not direct crop-specific harvest yield, and weather-station coverage was sparse and uneven. The result supports daily global weather datasets as useful tools in data-poor regions rather than proving universal yield transfer from U.S. models.
For related context, see Predicting Individual Chinese Cabbage Weight by Drone: R² > 0.72 Even 53 Days Before Harvest.
For related context, see How Much Crop Choice Could Low Latitudes Lose? Climatic Niches of 30 Food Crops Under 1.5–4°C Warming.
For related context, see Large Trees Are Increasing in the United States and Amazonia—but That Does Not Prove Warming Made Trees Bigger.
Reference
- Hogan D, Schlenker W. Non-linear relationships between daily temperature extremes and US agricultural yields uncovered by global gridded meteorological datasets. Nature Communications. 2024;15:4638. https://doi.org/10.1038/s41467-024-48388-w
Comments