Mapping intermittent river surface water in Arid and Semi-Arid China using Sentinel-2 and a novel seasonal intermittency index

Intermittent river surface water is a crucial feature in dryland river ecosystems, underpinning biodiversity, nutrient exchange, and hydrological connectivity, yet its spatiotemporal dynamics remain poorly quantified by existing ecological indicators. This study proposes a framework for mapping and quantifying the intermittency of river surface water across the Arid and Semi-Arid China (ASAC) using Sentinel-2 imagery from 2019 to 2024. A Spatial–Temporal Similarity Interpolation (STSI) method was developed to reconstruct a gap-filled monthly river surface water extent dataset at 10-m resolution by recovering observations obscured by cloud cover and topographic shadows, increasing detected river surface water extent by approximately 11% on average. Validation against PlanetScope imagery yielded an overall accuracy of 0.93 and an F1 score of 0.92. To characterize seasonal intermittency, we further developed a Seasonal Intermittency Index ( I s ), a spatially continuous ecological indicator that integrates water occurrence frequency with the seasonal intensity of water–non-water transitions. Using I s combined with water frequency, we found that intermittent river water bodies dominate the ASAC (89.3%), whereas perennial river surface water constitutes only 10.7%. Notably, 52.0% of the total river surface water area experienced interannual regime variability during 2019–2024, with transitions between seasonal and non-seasonal intermittent regimes (18.72%) occurring nearly three times more frequently than transitions between perennial and intermittent regimes (6.71%). These results indicate that hydrological change in the ASAC was primarily characterized by reorganization within intermittent regimes rather than expansion or contraction of perennial river surface water, with the semi-arid sub-region exhibiting the highest level of regime variability. As a spatially continuous, transferable indicator, I s enables consistent monitoring of non-perennial river dynamics and reveals substantial regime variability in dryland river systems, providing spatially explicit information to support ecological assessment, monitoring, and adaptive management.