Abstract: Gaining a timely and accurate understanding of regional soil salinization information holds great significance for salinized soil management and sustainable land use. This study investigated the Qingtongxia irrigation district in Ningxia. First, employing Landsat7 ETM+ and Landsat8 OLI images as data sources, along with the salinity index (SI), normalized difference vegetation index (NDVI), surface albedo, and modified soil-adjusted vegetation index (MSAVI), this study constructed remote sensing-based salinity models for extracting the soil salinization information in the irrigation district. Then, using the transfer matrix, change map, and centroid migration methods, this study analyzed the spatial characteristics of soil salinization in the irrigation district from 2004 to 2024. The results indicate that the SD1 (SI-NDVI) model exhibited a goodness of fit (R²) of 0.840 69 against the measured soil electrical conductivity and a classification accuracy of 75.00%. From 2004 to 2024, the areas of non-salinized, slightly salinized, and highly salinized soils decreased by 214.1 km², 127.1 km², and 186.3 km², respectively. In contrast, the areas of moderately salinized and saline soils increased by 72.7 km² and 454.8 km², respectively. Saline soil expanded the most remarkably from 2019 to 2024. Various soil types exhibited different centroid migration directions, with the centroids of non-salinized, slightly salinized, moderately salinized, highly salinized, and saline soils migrating toward the northeast, southeast, southwest, northwest, and northwest by 11.517 km, 5.182 km, 3.657 km, 6.299 km, and 5.519 km, respectively. The SD1 model is suitable for inversion of soil salinization in irrigation districts. Geoscience information maps and the centroid migration analysis can effectively reveal the spatiotemporal evolutionary patterns of soil salinization. The results of this study will provide a reference for the management, prevention, and control of soil salinization in irrigation districts.