Influencing Factors And Prediction Simulation Of Soil Salinity In The Coastal Area Of The Yellow River Delta

Land is an important part of the terrestrial ecosystem,but with the global shortage of resources and the surge of population,leading to the increasing shortage of land resources,Soil salinization is an important aspect of land resource degradation.As the most potential delta in China,the Yellow River Delta is rich in natural resources,but at the same time,the regional ecological environment is fragile,and the soil salinization phenomenon is common,which seriously restricts the local agricultural development.Therefore,it is of great practical significance to find out the status and causes of soil salinization and grasp the trend of soil salinization.In the Yellow River delta,through data collection and field investigation,determine the 12 soil salinity influence indexes,including 8 quantitative indicators(evaporation,precipitation,groundwater depth,groundwater salinity,soil particle content,relative level,distance of vegetation coverage,from river and sea)and four qualitative indicators(soil configuration,landform type,vegetation type,human activities),systematic analysis of the influence of each index on soil salt.The driving factors of soil salt were screened by gray association analysis and geographic detector method,and the soil salt statistical analysis(MLR,PLSR)and machine learning(BPNN,SVM)prediction models were constructed respectively,and the best model was applied to the corresponding data of Kenli area,and the comparative analysis of the prediction results was conducted.Finally,three scenarios of precipitation,groundwater buried depth and vegetation cover change were set,and the soil salinity was predicted in different scenarios.The main conclusions are summarized as follows:(1)In the quantitative index of soil salinity,evaporation and groundwater salinity;precipitation,groundwater depth,clay content,relative elevation,vegetation cover and distance from sea and soil salinity;the correlation between groundwater factors and soil salinity is the highest,followed by vegetation cover.Among the four qualitative indicators,the mean salt content of soil body configuration is ranked from high to low as sand body →sand bottom →sand inclusion → clay inclusion → soil body → clay bottom;the mean salinity of different geomorphic types is ranked from high to low as beach land →depression → slightly inclined flat land → gentle hill → river beach highland;the order of soil salt content of different vegetation types from high to low is bare land → Suaeda salsa→ Tamarix → reed → thatch →cotton → rice → wheat → corn;the average salt content of different land use types is ranked from high to low as salt wasteland → beach →forest and grassland.(2)In the selection of soil salinity,the ranking results of the grey correlation method are groundwater salinity> groundwater depth> groundwater depth> vegetation coverage> particle content> distance to sea> surface elevation> precipitation> evaporation;the ranking results of the geographical detector method are groundwater salinity> vegetation coverage> groundwater depth> surface elevation> evaporation;through comparison of different methods,groundwater salinity,groundwater depth,vegetation coverage and sea distance are the driving factors of soil salinity.In the construction of soil salinity prediction model,BP neural network model modeling set R~2 0.8847,RMSE 1.1350,validation set R~2 0.7999 and RMSE 1.1204,the best prediction effect.Analysis of the prediction results of soil salinity: the prediction results of mild saline soil are 0.27% and the measured area is 0.38%;the prediction results of moderate saline soil are25.72% and the measured area is 27.91%;the prediction results of severe saline soil are 25.73%and the measured area is 23.90%;the predicted results of saline soil are 48.29% and the measured area is 47.81%.(3)In the simulation of soil salinity scenario,the results showed that proper increase of precipitation,reduction of groundwater level and increase of vegetation cover can reduce the degree of soil salinization.The precipitation change is divided into dry period and wet period.Under the three scenarios of control group,precipitation of 20 mm and precipitation of 40 mm,the change rates of mild,moderate and severe saline soil and saline soil in dry period are 1.51%,0.73%,3.90% and-6.14% respectively;The corresponding change rates in wet period are2.29%,0.13%,1.19% and-3.61% respectively.The change of groundwater level was set in three scenarios: the control group,the water level decreased by 0.5m,and the water level increased by 0.5m.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 groundwater level decreased;When the groundwater level rises,the change rates of mild,moderate and severe saline soil and saline soil are-0.02%,-14.77%,22.51% and-8.02% respectively.The change of vegetation cover is set in three scenarios: control group,low increase of vegetation cover and high increase of vegetation cover.The change rates of mild,moderate and severe saline soil and saline soil are 5.84%,-8.35%,6.91% and-4.40% respectively when the vegetation cover is low increase,and the corresponding change rates are 10.55%,-7.35%,5.19% and-8.39% respectively when the vegetation cover is high increase.This paper explores the influencing factors of the Yellow River delta coastal soil salt,clear the main driver factor of soil salt,construct the prediction model of soil salinity,quantitative simulation of soil salt analysis,realize the dynamic prediction simulation of regional soil salinization,the prevention and monitoring of soil salinization of the Yellow River delta is of great significance.