Data for Ecosystem quality assessment of national parks: a case study of the Shaanxi section of Giant Panda National Park

Existing studies are limited by coarse remote sensing resolution, which restricts the fine discrimination of ecological disparities between functional zones, and lack quantitative analysis of driving mechanisms. In this study, multi-temporal Landsat datasets from 1986 to 2025 were employed to reconstruct Normalized Difference Vegetation Index (NDVI) time series, retrieve land-type-specific Leaf Area Index (LAI), estimate fractional vegetation cover (FVC), and simulate gross primary productivity (GPP).

A 30 m Ecosystem Quality Index (EQI) was established following national technical specifications to characterize its spatiotemporal variations. Furthermore, a combined framework of random forest and SHapley Additive exPlanations (SHAP) was adopted to quantify the contribution of diverse factors to EQI dynamics. The results show that: (1) the retrieved long-term ecological parameters achieve satisfactory accuracy, with an average R2 of 0.87 for reconstructed NDVI, which satisfies the demands of refined ecological assessment in small mountainous regions; (2) over the past 40 years, the EQI of the study area increased significantly at an annual growth rate of 1.02 yr-1, indicating continuous positive ecological succession, with a remarkable rise in the area proportion of excellent and good-grade ecosystems; (3) the core protection zone maintains superior baseline ecological conditions, while the general control zone exhibits a faster restoration rate, demonstrating remarkable effectiveness of zoned management; (4) anthropogenic factors dominate EQI variations.