Soil salinization threatens global agriculture and ecosystems, posing a critical challenge for sustainable development. Understanding how seasonal variations and environmental factors influence salinity dynamics is essential. However, current research relies heavily on single-time-point remote sensing, which offers limited temporal insights and fails to uncover the mechanisms driving seasonal changes. This study proposed the dynamic time warping-based model transfer-structural equation model (DBS) framework, which integrates dynamic time warping (DTW), base model transfer, and structural equation modeling (SEM), to explore the regulatory mechanisms of environmental factors on soil salinity dynamics. The framework includes building a stacking-electrical conductivity (EC) base model, aligning multi-month data with DTW, and analyzing environmental factors through SEM. Key predictors identified were normalized difference vegetation index (NDVI), normalized difference water index (NDWI), air temperature (AT), and precipitation. NDVI reduced salt accumulation by lowering evaporation and stabilizing soil moisture, while NDWI reflected precipitation-driven dilution and leaching. Temperature influenced salinity indirectly by regulating NDVI and NDWI. SEM confirmed NDVI and NDWI had direct effects on EC, while AT and precipitation acted indirectly. Model validation showed high accuracy and adaptability, with R-squared (R2), Nash–Sutcliffe efficiency coefficient (NSE), and Kling–Gupta Efficiency (KGE) values of 0.93, 0.94, and 0.89 for training and 0.86, 0.85, and 0.79 for validation, respectively. After DTW optimization, R2 improved by 0.12–0.22, NSE by 0.07–0.18, and KGE by 0.02–0.12, demonstrating significant performance gains. The framework demonstrated strong migration capability across different soil types and vegetation covers, achieving R2 of 0.73–0.96, The root mean squared error (RMSE) of 1–20, and residual prediction deviation (RPD) of 1.22–1.95. This study highlights the dominant role of climate-ecological interactions in salinity regulation and offers a robust, transferable method for multi-temporal salinity prediction. The findings provide critical insights for precision soil salinity management, sustainable agriculture, and climate resilience strategies, particularly in regions vulnerable to salinization.