Assessing Long-term ecological environment quality dynamics and anthropogenic disturbance in the lower Jinsha River basin (LJRB): Integrating MRSEI with Hurst-based persistence analysis

The Lower Jinsha River Basin (LJRB) has experienced decades of concurrent cascade hydropower construction, mineral exploitation and large-scale Grain for Green (GFG) restoration. Comprehensive long-term multi-dimensional assessments linking these overlapping human activities to basin ecological sustainability remain limited in existing research. This study develops an integrated evaluation framework built on a 35-year (1991–2025) Modified Remote Sensing Ecological Index (MRSEI) dataset, Hurst exponent persistence analysis, a composite Ecological Stability Index (ESI) developed for this study, and Convergent Cross Mapping (CCM) causal attribution to quantify long-term ecological trajectories and identify the ecological regime shift. Results show sustained basin-wide ecological improvement with the mean MRSEI rising from 0.201 to 0.386, with Sen's slope = 0.0055 yr −1 and Kendall's τ = 0.476 ( p < 0.01). Three phases of uninterrupted MRSEI growth are distinguished throughout the study timeline: slow baseline improvement from 1991 to 2000, moderate steady gains between 2001 and 2012, and notably accelerated ecological recovery after 2012 coinciding with reservoir impoundments where EVI and WET increased markedly after reservoir impoundment, by 0.456 and 0.567, respectively. While 93.08% of the basin presents positive MRSEI trends, only 42.51% corresponds to stable long-term recovery; 45.21% displays anti-persistent trajectories (Hurst <0.5), a latent vulnerability that single-indicator trend analysis may not identify. Mining disturbance follows a clear distance-decay gradient (R 2 = 0.63), with functional ecosystem recovery observable 8–10 km from open-pit mine sites. CCM analysis confirms reservoir impoundment as the primary trigger of post-2012 ecological improvement (ρ = 0.617), with cross-map skill 39.3% stronger than GFG revegetation (ρ = 0.443) and 22.7% greater than the reverse feedback pathway (ρ = 0.503). Geodetector modelling shows the interaction between land cover and evapotranspiration delivers the highest explanatory power (q = 0.534), exceeding all single independent variables and reflecting coupled land-water-climate feedbacks shaping regional ecological conditions. The integrated MRSEI–ESI workflow supports delineation of spatial management zones, including stable conservation areas, mining degradation corridors, and empirical 2 km disturbance buffers around reservoirs and mine sites. This analytical framework may be adapted to assess ecological sustainability in other mountain river basins undergoing rapid anthropogenic transformation.