Accurate assessment of landscape ecological resilience-the capacity of landscapes to resist, recover from, and adapt to disturbances-is fundamental to sustainable basin governance under rapid environmental change; however, an integrated multidimensional framework remains elusive. Here, we develop a novel Landscape Ecological Resilience Index (LEREI) that synthesizes multi-source geospatial data (1985-2023) to explicitly integrate resistance, recovery, and adaptability across the Yellow River Basin, China. Implemented through a reproducible, cloud-native workflow on the Google Earth Engine and Colab platforms, the framework integrates explainable AI using the LightGBM and SHapley Additive exPlanations (SHAP), ecosystem service modeling via the InVEST Python API, and structural-functional connectivity analysis using Morphological Spatial Pattern Analysis (MSPA) and Conefor. Results reveal pronounced spatiotemporal heterogeneity in landscape resilience and a net 28.3% increase in mean LEREI, accompanied by a 12.5% expansion of core habitat area. The explainable AI models achieved high predictive accuracy (R-2 = 0.855-0.887; RMSE < 0.035) and identified landscape connectivity (delta Probability of Connectivity, dPC) and restoration intensity as the dominant positive drivers of resilience, whereas erosion susceptibility and geomorphometric ruggedness were primary constraints. Vegetation productivity and soil organic carbon were also substantial positive contributors. The derived LEREI showed strong alignment with synergistic improvements in ecosystem service delivery (e.g., +18.2% carbon storage), supporting the functional validity of the index. This study delivers a scalable, interpretable, and transferable resilience assessment framework that links resilience theory with spatially explicit diagnostics, providing actionable insights for precision landscape management and sustainable basin-scale governance under ongoing environmental change.