The Study Of Farmland Productivity Potential Evaluation Based On Weight Of Evidence Method

In the process of long-term and rapid urbanization,a large number of cultivated land has been occupied or unreasonably utilized,which seriously endangers the food security of the country.To carry out reasonable and effective evaluation of farmland productive potential can scientifically predict the productive potential of cultivated land resources and provide theoretical basis for alleviating the contradiction of limited land supply.The workflow of farmland productivity potential evaluation mainly includes selecting evaluation target,establishing evaluation index system and determining index weight.In order to solve the problems and shortcomings in the traditional evaluation work,this paper constructs an evaluation model of farmland productivity potential based on the weight of evidence method,and conducts in-depth research and Analysis on each link of the evaluation process.The main research contents and achievements are as follows:(1)After analyzing the main links and contents of the workflow of farmland productivity potential evaluation,this paper introduces the evidence weight method into the evaluation of farmland productivity potential,and constructs a complete farmland productivity potential evaluation model based on the evidence weight method,and gives full play to the layer data-driven mechanism of the evidence weight method to make up for the lack of subjectivity of the weight in the evaluation of farmland productivity potential.(2)The objective of the study is to delineate the potential level of farmland productivity.The potential productivity of farmland is influenced by natural conditions and location conditions.The potential productivity of farmland is affected by natural conditions and location conditions,and the quality grade patches of cultivated land and basic farmland patches are delineated on the basis of these factors.In this paper,we use the methods of centroid extraction,spatial thinning and random sampling to extract target sample points from these speckles.The sample point set includes training point set and test point set.The training point set is used for the training input data of the evidential weight method evaluation model,and the test point set is used for the accuracy test of the evaluation results.(3)When establishing the index system,this paper uses the methods of literature research and quantitative analysis to screen evaluation factors from both qualitative and quantitative aspects.Through the analysis of frequency histogram and the use of fuzzy membership conversion algorithm,the dimensionality and numerical normalization of evidence driving factors are completed.Finally,seven driving factors are selected:slope,soil texture,soil erosion intensity,distance from road,distance from residential area,distance from water source and land continuity.(4)Taking the evaluation of farmland productivity potential in Weihai City as an example,the feasibility of the model and the accuracy of the experimental results are verified.According to the test data of four sets of test points,the discriminant accuracy of the experimental results can reach 91.3%.After that,the spatial overlay analysis of the evaluation results and the cultivated land map patches in Weihai City is carried out to generate the spatial distribution map of cultivated land potential evaluation grade in Weihai City.This paper makes a scientific and reasonable analysis of the evaluation results,and based on the analysis of the data results,puts forward relevant suggestions for the delimitation of basic farmland.Finally,on the basis of the AutoGIS geographic information platform independently developed by Beijing Forestry University,this paper develops a farmland productivity potential evaluation system based on the evidence weight method.