Abstract: Accurate monitoring of the spatiotemporal distribution of soil moisture content is crucial for geotechnical engineering monitoring and geological disaster prevention and control. Given the limitations of passive distributed temperature sensing (PDTS) technology in monitoring soil moisture content, the Spearman correlation coefficient method was introduced to quantitatively analyze the correlations among radiation, air temperature, warming slope, soil temperature, salinity, and moisture content. By incorporating the back propagation (BP) neural network, a passive sensing model for soil moisture is proposed. The model considers the comprehensive effects of water, heat, and salt and can replace the complex numerical iterative algorithm in traditional PDTS technology. This model not only expands the application scope of PDTS technology, but also significantly improves the accuracy of moisture content prediction. Long-term observations on the Loess Plateau in China verified the effectiveness of the proposed model using in-situ data. The analysis results indicate a strong positive correlation between loess moisture content and salinity, temperature, which can complement each other in depth. The input variables maintain a shallow soil moisture content with a root mean square error below 0.006 8 m³·m⁻³. The model’s errors mainly arise from rainfall and soil freeze-thaw processes, which tend to be smaller in winter and larger in summer. This study provides important theoretical support and practical reference for applying PDTS technology to soil moisture content monitoring and reveals the water-salt migration mechanism in loess.

Key words: passive distributed temperature sensing, artificial neural network, moisture content, hydrothermal-saline transport, fiber Bragg grating