Robust Streamflow and Drought-Flood Hazard Signals from Criteria-Based Verification of Hybrid Hydrological Projections Under Climate and Land-Use Change

Streamflow projections under climate and land-use change rarely state which results are robust to model uncertainty. This study developed a hybrid QSWAT–ANN workflow with four verification criteria—model performance (C1), signal direction (C2), spatial ranking (C3), and reference-series sensitivity (C4)—for the Huai Luang River Basin in Northeast Thailand. We used six CMIP6 trajectories from three climate models under SSP2–4.5 and SSP5–8.5, plus a single land-use pathway to 2070. The corrected model passed C1 with an overall NSE of 0.80 and 0.79. Under ensemble precipitation increases of 10.5–13.7% and a 55.1% decrease in paddy rice toward cash crops, annual streamflow increased by 32.0–49.9%, mainly in the monsoon season, as a combined response in which the drivers were not separated. Within-trajectory Standardized Streamflow Index (SSFI) signals agreed in direction in four of six trajectories in all cases (C2 passed), and ACCESS-CM2 was the only dissenting model. The Drought Hazard Index (DHI) and Flood Hazard Index (FHI) ranks of 31 sub-basins passed C3 in three of six cases (one of four future cases), while screening identified non-overlapping robust cores of seven (DHI) and ten (FHI) sub-basins. C4 showed that index magnitudes shifted by typically 1.2–1.5 SD. Robust signal direction, seasonal structure, and spatial cores support reservoir management and sub-basin targeting, whereas absolute index magnitudes should not serve as design values. This framework is transferable to other monsoon basins.