Design Implementation And Application Of Data Spatial Discretization Algorithm In Geographical Detector Model

As a spatial analysis model that effectively reveals the driving factors behind spatial differentiation,the geographic detector model has been widely used in many research fields.The geographic detector model is good at analyzing discrete spatial data.Continuous spatial data needs to be converted into discrete spatial data before it can be imported into the geographic detector model for analysis.In the current application research using geographic detector models,the discretization of continuous spatial data is mostly based on prior knowledge or traditional data discretization algorithms.Although these algorithms can discretize spatial data,they are not specific to the discretization algorithm of data in the geographic detector model.The spatial data processed by different discretization algorithms are imported into the geographic detector model for analysis,and the results obtained have certain differences.Considering the spatial characteristics of spatial data during the discretization process,researching data discretization algorithms suitable for geographic detector models is a fundamental and critical issue in the application of geographical detector models.A suitable discretization algorithm can effectively improve the accuracy of the geographical detector model and reduce the uncertainty in the model application process.Draw support from spatial statistics,geographic information systems,data structures and algorithms and other related theories,this paper designs and implements two new data spatial discretization algorithm in geographical detector.The main research contents and results of this paper are as follows:(1)Two new data spatial discretization algorithms in geographical detector are designed and implemented.On the basis of spatial data spatial feature measurement,data structure and algorithm,combined with the principle of geographic detector model,the ZI and OB methods are designed.The overall technical process of the algorithm is: initializing candidate breakpoints,determining the optimal set of breakpoints,determining the optimal interval,discretizing continuous attributes.Through the algorithm description,the Arc GIS platform is used to build a custom script tool to implement the algorithm.Thus completing the process from the theoretical research of the algorithm to the concrete realization.(2)The effectiveness of the algorithm is verified.By citing three predecessors using geographic detector models for applied research and published research results,as case data,experimental design was conducted to compare the experimental results between different discretization algorithms.The obtained geodetector model results are better than the traditional discretization algorithm to a certain extent,and a relatively reasonable number of discretization intervals can be obtained.(3)Use the discretization algorithm in this paper for application research.Applying the two algorithms proposed in this paper to the study of the spatial correlation of the gully density and its environmental factors in Guangxi.The 12 continuous environmental factors involved in the study need to be discretized,and 6 of these factors have obtained the most Excellent model results.After processing various environmental factors,geographical detector analysis was carried out to obtain that slope,geological lithology,and soil type were the dominant factors affecting the spatial differentiation of ravine density.The affected areas were mainly concentrated in southern Liuzhou,southern Laibin,western Baise,and Guigang Most areas in the east,Nanning,Chongzuo,Fangchenggang,Qinzhou,Yulin and Wuzhou.Each environmental factor does not affect the gully density alone,and there is an enhanced relationship between the factors,mainly including two-factor enhancement and nonlinear enhancement.