Flowering‐Stage Heat Stress Assessment in Summer Maize Using Multi‐Source Data and Machine Learning

Flowering‐stage heat stress in summer maize ( Zea mays L.) can substantially reduce grain number per ear (GNP), yet traditional assessments based on daily temperature or single indicators often fail to capture short‐duration extreme events and their physiological impacts. This study develops a mechanism‐oriented framework that integrates multi‐source datasets—including hourly meteorological observations, remote‐sensing indicators, and field surveys—with machine‐learning approaches to quantify flowering‐stage heat stress in Henan Province, China. An hourly‐scale heat stress index (HI) was constructed to represent both intensity and cumulative exposure. Key meteorological predictors included hourly temperature, daily maximum temperature, heat accumulation, and soil moisture, while remote‐sensing variables encompassed leaf area index (LAI), normalized difference vegetation index (NDVI), canopy temperature, and water stress indicators. GNP was used as the dependent variable to reflect reproductive‐stage yield loss. Feature selection combined Pearson correlation, random forest (RF) importance ranking, and embedded methods, and five regression models—multiple linear regression (MLR), ridge regression, least absolute shrinkage and selection operator (LASSO) regression, RF and Extreme Gradient Boosting (XGBoost)—were evaluated. Results indicate that tree‐based models outperformed linear models in test‐set evaluation, with RF ( R 2 = 0.619) and XGBoost ( R 2 = 0.588), demonstrating an improved ability to capture nonlinear interactions among thermal, water, and canopy‐related factors. Application to the 2022 heatwave revealed that 12.4% of the study area experienced >20% reduction in GNP, consistent with field observations. Notably, the hourly HI enabled detection of short‐duration extreme events that daily indices overlooked. This framework provides a foundation for precise early‐warning, regional disaster mitigation, and climate‐resilient management of summer maize.