Three-Dimensional Identification of Compound Heat–Drought Events in Maize in Northeast China Based on the Process-based Cumulative Heat–Drought Index (PCHDI)

Abstract. Under global warming, compound heat–drought events pose a serious threat to maize production in Northeast China. Existing indices are insufficient to capture the synergistic interactions among multiple stresses and their cumulative effects during the growing season. In this study, a novel Process-based Cumulative Heat–Drought Index (PCHDI) was developed based on soil–plant–atmosphere continuum (SPAC) theory by integrating the Vegetation Condition Index (VCI), Soil Temperature Index (STI), and Standardized Soil Moisture Index (SSMI).The response thresholds of maize to compound stress were optimized using the XGBoost–SHAP framework, and the intensity and synergistic effects of stresses were quantified through three-dimensional vector analysis. The spatiotemporal evolution of compound events from 1990 to 2023 was further analyzed using the 3D DBSCAN algorithm. The results show that high-intensity events are concentrated in the central and western agricultural-pastoral transition zones and the western Songnen Plain, exhibiting a "central-western concentration, peripheral diffusion" pattern. The event centroid follows a "southwest-northeast-southwest" oscillation, with the migration step length reaching a peak of nearly 900 km from 2015 to 2022. The intensification of stress propagates from the later growth stages to the earlier stages: the earliest mutation occurred in the flowering stage (2008), followed by the ear stage (2010) and the seedling stage (2015), with no significant mutation observed during the emergence stage. The annual pixel mutation rate for the flowering stage peaked between 2010 and 2018.Validation results demonstrate that the PCHDI has a stronger correlation with maize yield reduction than traditional indices, with improvements of 62 % compared to empirical threshold-based indices and 33 % compared to existing datasets. The PCHDI also shows high consistency with historical disaster records. This study provides a new methodological framework for assessing compound climate disasters and supporting agricultural adaptation.