Ecohydrological impacts of photovoltaic construction across aridity gradients in China: Divergent responses of vegetation and evapotranspiration

Study Region China, with 14,849 natural-surface photovoltaic (PV) stations distributed across a complete aridity gradient. Study Focus Through using the Kernel Normalized Difference Vegetation Index ( kNDVI ) and actual evapotranspiration (ET) retrieved via a modified Surface Energy Balance System (SEBS), a spatiotemporal comparative framework was constructed to quantitatively assessed the climate-dependent ecohydrological impacts of large-scale PV construction. Machine learning (CatBoost) was applied to decode the non-linear driving mechanisms. New Hydrological Insights for the Region PV deployment leads to divergent ecohydrological trade-offs. Retrieval from the coupled SEBS and PV efficiency model showed that PV-induced ET suppression strengthens significantly with aridity ( slope =0.81, p < 0.01 ). In arid zones, strong ET suppression (mean relative change in ET, ∆ETr ≈ − 0.5 ) conserves moisture, driving positive vegetation responses relative change in vegetation growth ( ∆kNDVI ) that increase with PV area. Conversely, humid zones exhibit weak ET reductions (mean ∆ETr ≈ − 0.08 ) and negative scale effects on vegetation. Furthermore, CatBoost modeling explained 46% ( ∆kNDVI ) and 76% ( ∆ETr net ) of the variance. Elevation serves as the primary driver, while secondary drivers diverge: vegetation relies heavily on precipitation, whereas ET is highly sensitive to solar radiation, exhibiting non-linear thresholds that reveal regional ecohydrological response mechanisms.