Spatio-temporal evolution and simulation of soil salinization in typical oasis water-saving irrigation area based on long series data

Salinization in irrigation areas is a global environmental challenge. The complexity of the natural environment increases the uncertainty of salinization distribution. This study focuses on the Manasi River Irrigation Area, analyzing the relationships between soil salinity and various factors, including irrigation area (IAR), water-saving irrigation area (WSIA), surface water withdrawal (SWDA), groundwater withdrawal (UWDA), groundwater depth (GL), groundwater mineralization (MG), elevation (EL), soil bulk density (SBD), evaporation (AE), and precipitation (AR) over the temporal and spatial scales from 2013 to 2021 by correlation analysis. A geographically weighted regression (GWR) model was employed to predict the distribution of soil salinization at the irrigation scale. The results show that soil salinization in irrigation areas showed obvious spatial variation, and with time, the degree of soil salinization continued to improve, and the proportion of salinization area decreased from 98.9 % in 2013 to 63.3 % in 2021. The proportion of severe salinization and saline soil decreased to 0. On the spatial scale, there is a highly significant correlation between soil salinity and irrigation area (IAR), water-saving irrigation area (WSIA), surface water diversion (SWDA), underground water diversion (UWDA), groundwater level (GL), mineralization of groundwater (MG), elevation (EL) and soil bulk density (SBD). The correlation coefficient between soil salinity and WSIA MG is the highest, at −0.92 and 0.98, respectively. There is a highly significant correlation between soil salinity on the time scale and IAR, WSIA, SWDA, GL, MG, and SBD on the annual scale. The largest correlation coefficients are between soil salinity and WSIA (-0.99) and MG (0.99). The simulation result of soil salinity by GWR model has high precision, the slope of the simulation result is greater than 0.94, R 2 is greater than 0.98, and the relative root mean square error RRMSE is less than 13.08 %, which can well simulate the spatial distribution of soil salt in the irrigation area. The findings of this study are significant for understanding and controlling the distribution of soil salinization at the irrigation area scale. • Interpret the process of soil salinization from macro-environmental factors. • Clarify the main influencing factors of soil salinity on the spatio-temporal scale. • The GWR model can effectively simulate the spatial distribution of soil salinity.