In arid-region oasis agriculture, the interactions among water, salinity, and organic carbon crucially influence farmland quality and sustainability. However, conventional remote sensing inversion faces a trade-off between temporal and spatial resolution, limiting continuous monitoring of Water–Salt–Carbon (WSC) dynamics. To address this, a daily-scale inversion system for the Water–Salt–Carbon Coordination Index (WSC-CI) was developed using fused Unmanned Aerial Vehicle (UAV) multispectral and PlanetScope imagery in a high-standard cotton field in Wujiaqu City, Xinjiang. The Bootstrap Soft Shrinkage (BOSS) algorithm identified key spectral features of soil salinity (SS), soil water content (SWC), soil organic carbon (SOC), and WSC-CI at different growth stages. A Genetic Algorithm (GA)-optimized Random Forest (RF) model was then constructed for multi-stage inversion. The optimal spatiotemporal fusion algorithm was selected and combined with a spatiotemporal adaptive result fusion strategy to achieve daily-scale monitoring. Results showed that: (1) BOSS exhibited robust feature selection in high-dimensional spectral spaces, effectively identifying key indices and reducing redundancy, enhancing model generalization. (2) The GA-RF model achieved R 2 values above 0.5 in all stages, confirming strong adaptability and stability in capturing nonlinear relationships. (3) The optimal fusion with adaptive integration improved spatiotemporal continuity and revealed spatial heterogeneity compared with raw data. (4) WSC-CI was low during the Pre-sowing stage, increased during Squaring, fluctuated during Flowering–bolling, and peaked at Boll-opening. This multi-source fusion and model integration framework enables dynamic, high-precision monitoring of water-salt transformation processes in cotton fields, providing a scientific basis for precision management, water–salt regulation, and optimized irrigation scheduling in high-standard farmland in arid areas, thereby supporting sustainable agricultural water resource management and improving water-use efficiency. • Developed a daily-scale inversion framework for Water–Salt–Carbon coordination using multi-source fusion. • Proposed a BOSS–GA–RF-based system for feature selection and inversion modeling of water, salt, and carbon. • Applied optimal image fusion and spatiotemporal adaptive result integration to enhance monitoring accuracy. • Revealed daily-scale dynamic patterns of the Water–Salt–Carbon coupling process in high-standard cotton fields.
Zhang et al. (Sun,) studied this question.