Identification of phenology-aligned sensitive windows to extreme-rainfall events for maize in north China Plain and actionable irrigation–drainage triggers

Abstract

Agricultural water-risk assessment requires the identification of exposure windows that accurately capture extreme rainfall during crop-sensitive stages. Using county-level data from the North China Plain from 2001 to 2015, phenology-aligned exposure windows were reconstructed and the Exposure Mismatch Index (EMI) was defined as the directional difference in cumulative threshold-exceeding precipitation between phenology-aligned and static windows. At the reporting benchmark threshold of 50 mm, static W2 (late vegetative to reproductive transition window) exhibited missed-detection and false-alarm rates of 9.7% and 10.0%, respectively, relative to phenology-aligned windows, resulting in a total misclassification rate of 19.7%. Total misclassification ranged between 17.3% and 19.9% across reporting thresholds of 30–70 mm. Stagewise model comparisons identified W2 showing the clearest contrast among long-term irrigation-coverage strata. Among counties with long-term irrigation coverage of < 0.90, the adjusted W2 response contained adverse association bands at EMI values of −120 mm to −50 mm and + 40 mm to + 110 mm, whereas the association was attenuated among counties with irrigation coverage at ≥ 0.90. Analysis of root-zone soil moisture data from the Global Land Data Assimilation System provided limited hydrological support for the positive band in the irrigation-coverage stratum of < 0.90 but did not establish a water deficit in the negative band or direct physiological mediation. These outputs distinguished extreme-rainfall occurrence from the statistical visibility of its phenological alignment with yield. Weak associations under extensive irrigation coverage should therefore not be interpreted as evidence of no rainfall exposure or intrinsic crop tolerance. EMI can serve as a regional screening indicator of exposure-window mismatch. A positive mismatch should prompt a review of drainage readiness, whereas a negative mismatch should prompt an assessment of soil-water availability. Both signals require confirmation using forecasts, soil moisture, drainage conditions, and actual irrigation information.