A Phenology-Aligned Assessment of Flowering Heat and Soil Moisture Deficit and the Associated Remote Sensing Responses of Summer Maize Across the Huang–Huai–Hai Plain

Flowering-stage heat and water limitation are important environmental constraints for summer maize, but their co-occurrence and remotely sensed expression vary with the phenological timing and the indicators used. We divided the Huang–Huai–Hai Plain into six phenological regions with fixed representative flowering windows and integrated ERA5-Land air temperature and 0–20 cm soil moisture for 1980–2025, MODIS variables for 2002–2025, a 2025 field survey, and long-term agricultural station records. The hourly heat exposure was calculated within each region-specific 10-day window after converting ERA5-Land timestamps from UTC to China Standard Time. Two-way fixed-effects models with 1° spatial-block clustered standard errors were used for the ERA5–MODIS panel. The mean flowering exposures were 22.09 h above 32 °C and 2.71 h above 35 °C; both the H32 and H35 increased significantly, whereas the soil moisture deficit indices showed no significant trend. Concurrent flowering heat and soil moisture deficit accounted for 13.56% of weighted grid-years, compared with 6.45% expected under independence from the pooled marginal frequencies. Antecedent soil moisture deficit occurred in 60.50% of concurrent grid-years, compared with 43.14% of flowering-deficit-only grid-years. In the state fixed-effects models, concurrent stress was associated with higher land surface temperature and lower NDVI, LAI, and evapotranspiration values. However, joint models containing heat, soil moisture deficit, and their interaction provided little evidence of a general multiplicative interaction. In the full 2025 survey, the flowering soil moisture deficit, concurrent index, and daytime land surface temperature were negatively associated with kernel number. None of these associations remained FDR-significant in the Region 2 (R2)-only sensitivity analysis, indicating that the full-survey associations could not be clearly separated from regional differences. Long-term station models did not detect FDR-significant associations between the environmental-stress indices and yield components, but seven flowering or post-flowering MODIS indicators were associated with theoretical-yield anomalies after FDR correction. These findings support phenology-aligned regional monitoring, while the reported relationships remain observational.