Abstract
Accurate monthly monitoring of surface water area (SWA) is essential for characterizing hydrological variability and supporting water-resource management in arid regions. However, persistent observation gaps caused by clouds, cloud shadows, and sensor-related missing pixels hinder the generation of temporally continuous monthly SWA records from optical imagery. To address this limitation, we developed a 30 m gap-free monthly SWA dataset for Central Asia for 1990–2022 by integrating the full Landsat archive with an image-similarity-based gap-filling (ISGF) framework. The resulting dataset was evaluated using MODIS and Sentinel observations and achieved an overall accuracy of 93.0% with a Kappa coefficient of 0.89. Additional consistency assessments against authoritative datasets yielded coefficients of determination (R²) of 0.96 for lakes and 0.98 for reservoirs. Relative to the JRC Global Surface Water product, the ISGF dataset increased the effective monitored monthly inundation area by 100–150 × 103 km², indicating a substantial reduction in omission errors associated with missing observations. At the regional scale, the maximum and minimum inundation extents were 175.6 × 103 km² and 101.4 × 103 km², respectively, with the largest inundation generally occurring from March to May and the smallest from September to November. We further identified three characteristic intra-annual SWA regimes across Central Asia: a spring flood pattern in downstream plains, a summer flood pattern in the Syr Darya and Amu Darya basins, and a summer deficit pattern in upstream rivers and mountainous reservoirs. By providing a long-term, high-resolution, and gap-free monthly SWA record, this study advances regional surface-water monitoring and offers a geospatial data foundation for sustainable water management, irrigated-agriculture resilience, and climate adaptation, with relevance to SDG 6, SDG 2, and SDG 13.