A soil moisture prediction method for mixed soils under multiple measurement conditions

Soil moisture prediction is crucial for agriculture, water resource management, and climate research. However, the accuracy of optical remote sensing methods is easily affected by soil type diversity and variations in measurement conditions. To address these issues, this study proposes an improved method, Baseline Arc Extension and Optimization (BAEO), based on existing approaches. Through mathematical modeling and algorithm optimization, BAEO enables accurate soil moisture prediction across different soil types and complex measurement conditions. In addition, by combining data interpolation with multi-objective optimization, the method selects and compresses the necessary spectral bands. Validation on multiple datasets, and comparison with Partial Least Squares (PLS) and the Normalized Soil Dryness Spectral Index (NSDSI) methods, show that BAEO achieves effective predictions using only six wavelengths, with determination coefficients (R²) ranging from 0.82 to 0.95, significantly outperforming traditional methods. This provides new insights and methods for remote sensing applications. • Overcomes dependence on extreme samples through mathematical modeling. • Enables soil moisture prediction in areas with mixed soil types. • Requires only six spectral bands and is applicable across different datasets. • Achieves a maximum R² of 0.82–0.95 across multiple datasets. • Maintains reliable prediction performance under pre- and post-rainfield conditions.