Soil salinization mapping across different sandy land-cover types in the Shiyang River Basin: A remote sensing and multiple linear regression approach

作者：Maurice Ngabire, Tao Wang, Xian Xue, Jie Liao, Ghada Sahbeni, Cuihua Huang, Hanchen Duan, Xiang Song · 发表于：Remote Sensing Applications Society and Environment · 年份：2022 · DOI：10.1016/j.rsase.2022.100847 · 被引用次数：41 · 研究领域：Soil Geostatistics and Mapping、Remote Sensing in Agriculture、Soil and Land Suitability Analysis

Soil salinization has a critical impact on land in arid and semi-arid regions. Consequently, mapping and monitoring saline soils have been the subject of growing attention for many scientists due to the irregular spectral reflectivity related to geographic locations and landscape characteristics, which are challenging for policy-makers to improve and sustain the ecological balance. This study applied geoinformation techniques combined with multiple linear regression modeling to map soil salinity over different land-cover types in the Shiyang River Basin. Based on field knowledge and area accessibility, eighty samples were collected, then Kennard-Stone (K–S) algorithm was used for sample partition, 70% for training, and 30% for validation. Variance Inflation Factor (VIF) was applied to identify multicollinearity and adjust the model covariates to ensure proper specification and functioning. Mobile sand, water channels, built-up, and mountainous areas were masked. The results revealed the model's high performance with a coefficient of determination R2 of 0.898, a probability level of 95%, a Root Mean Square Error (RMSE) of 1.653, a Ratio of the Performance to the Interquartile range (RPIQ) of 4.182, and standard error of Electric conductivity (EC) laboratory measurements to the standard error of the predicted EC (SE Lab/SE Pred) of 1.048. Overall, an area of 13834.59 km2, accounting for 33.54% of the entire study region, is under salinization threat. The basin's lower reach is ...