Recognition Method And Application Of Spatial Characteristics Of Soil Nutrients Based On Farm And Field Scale

The main problems of arable soil degradation,low water and fertilizer utilization,and poor crop quality in the summer maize-winter wheat rotation area in the northern Huanghuaihai Sea often improve soil fertility and improve soil quality through formula fertilization measures.However,the scattered field management model and the complex spatial distribution of soil nutrients at different scales.The spatial distribution information of soil nutrients in large areas obtained by traditional methods,the selection of sampling points is accidental and limited,and it cannot accurately reflect the relative The content of soil nutrients in smaller units poses difficulties for the implementation of precision agriculture.Correctly and comprehensively understand the spatial variability of soil nutrients and their spatial distribution patterns for the development of precision agriculture,digital soil mapping,reasonable soil nutrient management zoning and agricultural products The high quality and high yield are of great significance.In view of the current lack of comparative research data on small and medium scale soil nutrient spatial variability prediction methods in China,in order to obtain more accurate spatial distribution characteristics of soil nutrients and nutrient management zoning,this paper selects the summer corn harvest period in Baimu Village,Ningjin County,Xingtai City,northern Huanghuaihai Plain The unreasoned 50m ×50m field scale(take 82 soil samples)and 1000m × 1000m(take 108 soil samples)farm-scale cultivated land are two research plots,using ordinary Kriging interpolation method and RBF neural network interpolation method to explore The spatial variability of soil nutrients,soil quality and the prediction accuracy of the two spatial interpolation methods in the two-scale study area are provided.In order to obtain more accurate spatial distribution characteristics of soil nutrients,complete soil nutrient management zoning,and provide methods and data support for precise fertilization.The main research results are as follows:(1)The analysis of the spatial variability of the nutrient contents of the cultivated soil shows that the total nitrogen,organic matter,available phosphorus and available potassium in the study area are 0.33?2.9 g·kg-1,8.39?20.59 g·kg-1,20.15?119.88 mg·kg-1,27?175.4 mg·kg-1 respectively,the pH change is not alkaline,which shows that it is slightly alkaline.The farm-scale soil total nitrogen,organic matter and available phosphorus generally showed a spatial distribution trend with the content of the northwest and southeast symmetry line decreasing gradually to both sides,and the spatial distribution pattern of available potassium showed the overall spatial distribution characteristics of high content in the north and low content in the south.The field-scale soil total nitrogen,organic matter,available phosphorus and available potassium show the spatial distribution characteristics that the content in the northeast region is more and the overall decrease in the southwest.The ratio of gold to abutment for each nutrient nugget in the surface soil of the two-scale research plots was 0.448?0.746,indicating that the soil nutrients showed moderate spatial variability in the two-scale study plots.(2)The application of ordinary Kriging and RBF neural network interpolation methods to explore the spatial distribution prediction results of soil nutrients at two scales shows that,based on the ordinary Kriging interpolation method,the farm-scale study of soil total nitrogen,pH,available phosphorus and The optimal semi-variance fitting model of available potassium is an exponential model,while the spherical model is the optimal semi-variance fitting model of soil organic matter.The optimal semi-variance fitting models of soil nutrients in the field plot scale study area are all exponential models..The determination coefficient R2(0.755?0.877)of the spatial distribution characteristics of soil nutrients in the field plot-scale research sample area is larger than the determination coefficient R2(0.705?0.837)of the spatial distribution characteristics of soil nutrients in the farm-scale study sample area,based on RBF neural network interpolation Each error of the field plot study area of the method is smaller than the farm scale,and the parts of the soil nutrient content high and low values in the prediction results are more accurately expressed,which indicates that the prediction accuracy of the spatial distribution of soil nutrients is affected by the sampling scale.The fitting accuracy of the scale is better than the farm scale.(3)Based on the ordinary Kriging interpolation method and RBF neural network interpolation method,the accuracy of the prediction results of the spatial distribution of soil nutrients in the two-scale research sample area was evaluated.It was found that the soil in the sample area was studied at the same scale based on the RBF neural network interpolation method.All the errors in the prediction of the spatial distribution of nutrients have been reduced.The "smoothness" of the ordinary Kriging interpolation results is overcome in the prediction result map,and the parts of the high and low values of soil nutrient content are more clearly expressed and closer to the distribution of soil nutrients The actual prediction situation has higher prediction accuracy,which indicates that the RBF neural network interpolation method has better prediction ability of the spatial distribution of soil nutrients in the study area at the same scale.For farm-scale research plots with larger sampling scales,the RBF neural network interpolation method is used to predict the spatial distribution of soil quality better when evaluating the rationality of soil nutrient management zoning results.(4)Using principal component analysis and k-means clustering analysis,the farm-scale research sample is divided into three soil nutrient management zones,and the soil nutrient content in each zone tends to be homogeneous.The differences in organic matter and available potassium showed significant differences.The pH values were significantly different between Zone 3 and Zones 1 and 2.The available phosphorus content in the soil was significant between Zone 1 and Zone 2 and Zone 3.The overall distribution pattern of high and low value areas of soil fertility quality index is consistent,which indicates that the purpose of reasonable nutrient management zoning is achieved.Accurate fertilization can be carried out according to the spatial distribution of soil quality in the study area and the results of soil nutrient management division.Through the research on the farm-scale and field-scale farmland soil nutrients spatial variability and spatial distribution characteristics of summer maize-winter wheat rotation area in the northern Huanghuaihai,and the accuracy of the prediction results of the spatial distribution of soil nutrients under different interpolation methods,in fitting The research on the spatial distribution characteristics of soil comprehensive fertility quality in the study area and the study of soil nutrient management zoning under the interpolation method with higher precision revealed that the fitting accuracy of the RBF artificial neural network interpolation method is higher than that of the ordinary Kriging interpolation method.The spatial distribution map of soil comprehensive fertility quality based on RBF artificial neural network interpolation method combined with k-means clustering analysis method divides the farm scale into 3 soil nutrient management zones,provides data reference for the promotion and implementation of precision agriculture,and the development of sustainable agriculture And theoretical support.