Accurate diagnosis of soil moisture and canopy water content in citrus orchards based on synergistic multi-source remote sensing

Accurate prediction of soil moisture is essential for precise agricultural irrigation decision-making and regional water resource optimal scheduling. However, regional soil moisture inversion commonly suffers from three major limitations, including overreliance on single remote sensing data sources, fixed canopy parameters adopted in conventional physical models, and severe cloud contamination in the cloudy mountainous areas of Southwest China. In this study, a synergistic multi-source remote sensing framework integrating Landsat-8/9, Sentinel-2, and Sentinel-1A datasets was established to improve the traditional water cloud model by introducing phenology-driven dynamic canopy parameters. Specifically, canopy coverage and canopy water content of late-maturing citrus across different growth stages were accurately retrieved from Landsat-8/9 and Sentinel-2 imagery to optimize the simulation of canopy backscattering signals. The all-weather C-band radar data effectively mitigate the persistent cloud-cover limitations prevalent in the hilly regions of Southwest China. The results indicated that the vegetation water content in large-scale citrus orchards retrieved by optical remote sensing exhibits significant spatial variability, mainly distributed in the range of 1.65–5.71 kg·m −2 and 1.55–5.78 kg·m −2 , respectively. The soil moisture inversion results using multi-temporal Sentinel-1A data in VV polarization outperformed those in VH polarization, with R² ranges of 0.785–0.854 and 0.515–0.759, RMSE ranges of 6.125–8.351% and 6.942–8.412%, respectively. The two polarizations data have the highest inversion accuracy in citrus fruit maturation stage, with R² ranges of 0.811–0.842 and 0.682–0.711, RMSE ranges of 6.125–6.132% and 6.942–7.025%, respectively. This study proposes a novel methodology for high-resolution and precision inversion of surface soil moisture at the regional farmland scale.