Quantifying utility-scale photovoltaic impacts on eastern Tibetan alpine grasslands through RSEI and LSTM approaches

Utility-scale solar photovoltaic (PV) are expanding across the Qinghai–Tibet Plateau, yet their ecological consequences for alpine grasslands remain poorly quantified. We integrated the Remote-Sensing Ecological Index (RSEI) with a Long Short-Term Memory (LSTM) network to examine spatiotemporal responses of alpine grasslands to a 1.2 GW PV plant constructed in 2016. The results indicated that relative to the pre-construction period (1990–2015), the mean RSEI within PV system increased significantly by 33.4% (p < 0.01) in the post-construction period (2016–2024), and the proportion of pixels exhibiting positive RSEI anomalies rose from 44.1% to 77.6%. Ecological improvements propagated beyond the panel arrays, producing statistically significant effects (p < 0.05) within 120 m and detectable influences up to 720 m. Elevation and slope modulated the magnitude of these spill-over effects. Our findings demonstrate that PV system can enhance grassland quality in this alpine region and provide a transferable framework for evaluating renewable-energy impacts in fragile ecosystems.