Regional drought monitoring based on the Temperature Vegetation Drought Index (TVDI) holds significant potential in efforts to ensure food safety. However, its empirical determination of dry and wet edges introduces subjectivity and uncertainty, limiting its accuracy and applicability. An improved TVDI (iTVDI) was developed by optimizing boundary parameters using reinforcement learning, based on maximizing the correlation between the TVDI and the ERA5-Land soil moisture dataset. The findings are as follows: (1) The enclosed area and the absolute value of dry edge slope of iTVDI was 34.83–39.97% and 0.79–33.75% larger than TVDI, indicating that the iTVDI can be used to achieve better representation of drought conditions. (2) The iTVDI showed stronger correlations with ERA5 soil moisture (r: −0.416 to −0.174), with average |r| values 17.25% higher than TVDI; its correlations with Standardized Precipitation Index (SPI), the Standardized Precipitation Evapotranspiration Index (SPEI), and the Vegetation Condition Index (VCI) were also 12.69–75.43% higher. (3) From 2005 to 2024, the spring drought in the Huaihe Basin intensified, with the annual iTVDI increasing by 0.008–0.011, primarily driven by rising temperature, potential evapotranspiration, and vapor pressure deficit. Overall, the iTVDI is proved to be more accurate and reliable for monitoring drought dynamics and driving factors.
Spatial heterogeneity and sparse management observations complicate regional irrigation planning. We constrained AquaCrop-OSPy with remotely sensed canopy cover (CC) and actual evapotranspiration (ETa), together with gridded profile soil moisture (SM), using Gaussian-process Bayesian optimization. Regional simulations covered 500 m cells with at least 70% winter wheat cover in the Huaihe Basin, China. Four stage-specific irrigation thresholds were optimized to maximize incremental irrigation water productivity while maintaining at least 95% of a model-defined reference yield. Three individually calibrated historical growing seasons represented dry, normal, and wet conditions. Separate evaluation against field observations from four growing seasons yielded normalized root mean square errors (NRMSE) of 18.29% for aboveground biomass and 4.54% for grain yield; county-level yield NRMSE ranged from 12.58% to 15.88%. Using annual winter wheat fractions as area weights, mean seasonal irrigation depths over irrigated wheat area were 154.98, 119.48, and 64.09 mm, with corresponding net irrigation volumes of 5.73 × 109, 3.81 × 109, and 1.16 × 109 m3. Excluding cells that violated the SM or ETa error criterion reduced these volumes to 5.01 × 109, 3.30 × 109, and 1.06 × 109 m3 over the remaining domain. Irrigation was concentrated during the sowing-emergence and flowering-grain filling stages under dry conditions, and during the sowing-emergence and jointing-heading stages under normal and wet conditions. Mean simulated yield reached 96.67% to 97.57% of the reference yield. These model-derived estimates describe optimized net irrigation under the specified objective and yield constraint, rather than minimum irrigation requirements or independently validated schedules. Scenario differences jointly reflect climate, initial conditions, calibrated parameters, and annual spatial coverage.