Drought-to-flood transitions (DTF) show regional contrasts across China's major maize-producing regions, yet whether the flood phase mitigates or compounds drought impacts remains unresolved. 1-km observations (2000-2019) are used to identify soil-moisture-defined DTF and track canopy function with 8-day gross primary productivity (GPP) percentiles. Responses are quantified with two metrics: Compound Resistance (CR; whole-sequence GPP suppression) and Resistance Difference (RD; flood-phase effect beyond drought legacy; RD>0 mitigation, RD<0 compounding). Three vulnerability archetypes clarify flood-phase effects on drought legacy. Across climatic gradients, outcomes depend on the hydrothermal window: low-energy (cool/cloudy) and/or waterlogging-prone rewetting tends to compound drought legacy, whereas warm, energy-favorable rewetting promotes recovery. A compounding stress trap characterizes the Northern Arid-Semiarid Region, where exposure is highest (1.2 events per unit area) and impactful events (drought-stage GPP suppression) are numerous (5150). Random Forest-SHAP shows flood-phase solar radiation and temperature limit recovery and CR, while RD varies with drought-stage constraints and climate background. The Northeast China Plain shows a conditional response: events are frequent (5033) and impactful events increase over 2000-2019 (p < 0.01), while RD spans both signs, indicating that flooding can offset or deepen drought impacts depending on flood-phase solar radiation and temperature. The Yunnan-Guizhou Plateau exhibits buffered recovery: events remain long (mean duration 63 days), events are fewer (3019) yet impactful events increase over 2000-2019 (p < 0.001), and RD is predominantly positive, consistent with longer flood duration and favorable flood-phase energy conditions supporting rebound. These archetypes clarify when flooding mitigates or compounds drought legacy, informing region-specific adaptation.