基于Sentinel-1和变化检测的灌区土壤水分反演适应性评估OA

Soil moisture(SM)is a fundamental variable in land surface processes and agricultural water management.Accurate monitoring of its spatiotemporal dynamics is crucial for water-saving irrigation and precision crop scheduling in arid and semi-arid regions.To address the sensitivity of traditional retrieval methods to vegetation interference and their limited adaptability,this study systematically evaluated the performance of three typical change detection algorithms—Short-Term(STCD),Advanced(ACD),and Long-Term Change Detection(LTCD)—across diverse underlying surfaces in the Qingtongxia Irrigation District,utilizing Sentinel-1 SAR imagery.Multi-source datasets including SMRFR(Soil Moisture via Random Forest Regression)and SMAP(Soil Moisture Active Passive)soil moisture products were integrated as prior constraints and validation,with both time-varying and fixed boundary conditions considered.Multi-year time series analysis was conducted to assess the accuracy,stability,and surface-type adaptability of each method.Results show that the ACD method achieved the best overall performance,with an average correlation coefficient(r)of 0.45 and unbiased root mean square error around 0.04 m3/m3.The STCD-V approach demonstrated strong responsiveness in cropland areas,while the LTCD-SM method exhibited high stability and reliable error control.Time-varying constraints consistently outperformed fixed boundary ones.Overall,by introducing vegetation index correction and multi-temporal information fusion,ACD effectively enhanced the applicability and accuracy of soil moisture retrieval across the irrigation districts.This study provides methodological insights for Sentinel-1-based soil moisture retrieval at the field scale and offers scientific support for precision irrigation and water resource management in arid regions.