Abstract: Soil salinization refers to the phenomenon where salt accumulates continuously in the surface layer of soil due to natural and human activities, exceeding a certain threshold. This phenomenon is a major factor contributing to soil structure deterioration, reduced crop yields, and poses a threat to the sustainability of agricultural ecosystems. Effective monitoring and management of soil salinization have become urgent needs in saline-alkali areas. Hyperspectral technology has important application value in soil salinity estimation due to its rich spectral information and high resolution. Soil salt content (SSC) is a key indicator for evaluating the degree of soil salinization. To precisely identify salinized areas and optimize farmland management, this study focuses on the soil of the Weigan-Kuqa River Oasis in Xinjiang, China. A total of 193 soil samples were collected from the field, then brought to the laboratory for salt content measurement. Additionally, the spectral reflectance data of these samples were measured outdoors. The spectrum of each soil sample was measured ten times and averaged to obtain the spectral reflectance. The data were then preprocessed to remove noisy spectral bands at the ends of the curves and two water absorption bands, followed by smoothing using the Savitzky-Golay method. The smoothed spectral data were further transformed using mathematical transformations, continuous wavelet transformation (CWT), and discrete wavelet transformation (DWT). Mathematical transformations were also combined with CWT and DWT to further process the smoothed original spectral data. Based on Pearson correlation analysis, spectral bands passing the 0.01 significance test were extracted as SSC characteristic bands. A random forest model was then constructed to quantitatively estimate SSC. This study provides a scientific foundation for agricultural production and environmental protection. The results indicate that: 1) The spectral curve shapes of soils with different salinization levels are generally consistent, with reflectance showing distinct patterns of variation with wavelength and SSC. As the wavelength increases from short (350 nm) to long (2 450 nm), the reflectance increases within the 350～800 nm range, remains relatively unchanged in the 800～2 130 nm range, while around 2 130 nm, an absorption peak corresponding to Na₂SO₄ appears, after which the reflectance starts to decrease. Throughout the entire wavelength range, reflectance increases with SSC, showing a positive correlation, but decreasing at extremely high salinization levels. 2) First-order differential transformation, CWT, DWT, as well as first-order differential combined with CWT and first-order differential combined with DWT, can all significantly enhance the sensitivity of the spectra to SSC, with DWT being the most effective. The highest correlation coefficient between the processed spectra and SSC reached −0.621 (P<0.01), indicating a negative correlation. This was followed by first-order differential combined with CWT, first-order differential combined with DWT, CWT, and first-order differential transformation. 3) The use of first-order differential transformation, CWT, DWT, first-order differential combined with CWT, and first-order differential combined with DWT can all improve the model's accuracy and stability. The combination of first-order differential with CWT significantly enhanced the SSC estimation model’s accuracy. The model based on the full-band spectral of (1/R)'_CWT_2⁸ proved optimal, with coefficients of determination (R²) of 0.821 and 0.715 for the training and validation sets, respectively. It achieved root mean square errors (RMSE) of 16.049 and 17.467 g/kg, and relative predictive determinant (RPD) values of 1.65 and 1.48, indicating effective SSC estimation for the study area. The study demonstrates that combining mathematical transformations with wavelet transformations to process the smoothed original spectral data can significantly enhance the sensitivity of spectra to SSC. Compared to using either mathematical or wavelet transformations alone, this combined approach is more effective, and providing a valuable reference for the quantitative monitoring of SSC.