Abstract:

Understanding the mechanism of soil salinization is an important foundation for the improvement and utilization of saline soil. This paper selects Kenli District, the coastal area of Yellow River Delta, as the study area. First, the influences of 8 factors on soil salinity, including evaporation, precipitation, groundwater burial depth, groundwater mineralization, soil clay content, relative elevation, vegetation coverage and distance from the sea, were analyzed through geographical detectors, and then the main driving forces were screened; subsequently, MLR, PLSR, BPNN and SVM models were constructed, and the most accurate model was selected to construct the soil salt prediction model; finally, under the situation of groundwater changes, three scenarios (a control group, a 0.5 m decrease in groundwater level and a 0.5 m increase in groundwater level) were set up to simulate soil salinity. The factor detection results by the geographical detector showed that the influence of different factors on soil salinity was groundwater salinity>vegetation coverage>groundwater burial depth>distance from the sea>clay content>surface elevation>precipitation>evaporation; the interaction detection results showed that the interaction between groundwater factors, vegetation coverage and distance from the sea had a strong influence. It determined that groundwater mineralization, groundwater depth, vegetation coverage and distance from the sea were the main driving factors for soil salinity; the most accurate prediction model was the BP neural network model, with a modeling set R² of 0.8847 and RMSE of 1.1350, a validation set R² of 0.7999 and RMSE of 1.1204; the simulation results of groundwater scenarios showed that appropriately lowering the groundwater level could improve soil salinization. When the water level dropped, the change rates of mild, moderate, and severe saline soil and saline soil were 0.22%, -5.46%, 15.28% and -10.04%, respectively; when the water level rose, the change rates of mild, moderate and severe saline soil and saline soil were -0.02%, -14.77%, 22.51% and -8.02%, respectively. This paper screened out the main driving factors of soil salinity, constructed the best prediction model, and simulated and analyzed the soil salinity according to the set scenarios, providing a basis for the control and prevention of soil salinization in Yellow River Delta.

Key words: the Yellow River Delta, soil salinization, driving force, predictive simulation