Spatial Prediction And Comprehensive Evaluation Of Soil Nutrients Based On Different Geospatial Technologies

Soil nutrient indicators are the key indicators for soil quality evaluation.Studying the spatial distribution information of soil nutrients is the basis for understanding the regional soil quality status,adjusting management measures and various material inputs,and obtaining maximum benefits.It is of great significance for the sustainable use and management of soil.However,due to the combined effect of structure and randomness,soil nutrients have a high degree of spatial heterogeneity and dependence,which makes it necessary but difficult to grasp more accurate spatial distribution information of soil nutrients in the region.In this study,Wutai County,Shanxi province was used as the research area.Based on the 2010 soil testing formula project data and environmental variable data,soil nutrient spatial prediction models based on different geospatial technologies were constructed firstly,including Multiple Linear Stepwise Regression model(MLSR),Regression Kriging model(RK),Geographically Weighted Regression model(GWR),Geographically Weighted Regression Kriging model(GWRK).The accuracy of different models were evaluated according to the average error,average absolute error,and root mean square error.Secondly,Pearson correlation analysis and geographically weighted regression models were used to explore the influence mechanism of environmental variables on soil nutrients on the global and local scales respectively.Finally,the improved TOPSIS model was used to calculate the comprehensive soil nutrient index of Wutai County and the comprehensive horizontal spatial distribution characteristics of soil nutrients was mapped by the Cokring interpolation methods.1.The diagnostic results of soil nutrient content indicated that the average content of soil organic matter,total nitrogen and available phosphorus in Wutai County is at a medium level,and only the content of available potassium is relatively abundant.All nutrient indicators are of moderate degree of variation,of which the coefficient of variation of available phosphorus is the largest.2.The best prediction model for soil organic matter,available phosphorus,and available potassium is Regression Kriging or Geographically Weighted Regression Kriging,followed by Geographically Weighted Regression model,the prediction accuracy of Multiple Linear Stepwise Regression model is the lowest.However,total nitrogen can’t use mixed interpolation methods due to the weak autocorrelation,it’s optimal model is Geographically Weighted Regression model.The mapping effects of different models are different.The detailed variation information displayed by the local models are more refined than the global models,but the overall trend is basically the same: the content of soil organic matter in the northeast is significantly higher than that in the southwest,which are consistent with the spatial pattern of topographic factors.The distribution pattern of total nitrogen is basically similar to that of organic matter.The content of available phosphorus is lower in the northwest of the study area,but generally higher in the east.The content of available potassium is higher in the east of the study area,and it present the characteristics of the staggered distribution of high and low values in the west and middle of the study area.3.On the global scale,soil nutrients are basically positively correlated with topographic factors,positively correlated with annual average precipitation,negatively correlated with annual average temperature,and positively correlated with vegetation factors,indicating that higher altitudes in Wutai County are more conducive to the accumulation of soil nutrients.The Geographically Weighted Regression model is used to output the spatial distribution map of the regression coefficients of different soil nutrients on a local scale,showing that there exists a complex feedback mechanism between environmental variables and soil nutrients.However,due to the special geographical environment of Wutai County,this regularity is not very obvious.4.Use the improved TOPSIS model to calculate the comprehensive soil nutrient index,and use the most relevant available phosphorus index as a synergistic variable for Cokriging interpolation to obtain the approximate distribution trend of the comprehensive soil nutrient index.It’s spatial pattern is similar to the topography distribution of Wutai County,which may be caused by the topographical characteristics of the southwestern part of Wutai County from the southwest at an altitude of 624 m to the top of Beitai at an altitude of 3061.1 m.