Study On Forest Resources Reconditioning In County-level Land Spatial Planning Supported By Machine Learning

Forest resources are important ecological background and key development resources of land space.With the development of society,science and technology,human’s cognition and utilization of forest resources have gone through the process of "from blind to orderly","from singly-purpose to multi-benefit","from qualitative to quantitative","from various to whole" and "from static to dynamic",based on which resource science and management planning came up.From the establishment of the ministry of natural resources to the proposal of the land spacial planning of "multiple plans integration",it’s focus topic and key point to comprehensively coordinate various resources and reform the methods of supervision and planning multiple resources under "multiple plans endowment".The spatial forest resources reconditioning,aiming at describing resource potential,energizing local development,coordinating resources and consolidate resources base,is the trend of spatial forest resources cognition and utilization under the multiple plans integration,which provides support for the integrated management of natural resources.In the era of big data,the initiative,integration and innovation of actively promoting the application of modern information technology in forestry construction are the focus of current forest resources research.In this context,this study provides a theoretical basis for exploring the scientific path of forest resource reconditioning by combing relevant theories.On the basis of forest ecosystem cognition,the operating mechanism behind forest space is understood from the perspective of "process-function-structure".The theory,system and mode of scientific management of forest resources were summarized based on forest classification management theory.Taking forest scientific evaluation method as the core of reconditioning,this paper explores the research trend of accurate and appropriate evaluation of forest resources under the target of resource reconditioning.This paper summarizes the key points and related methods of forest resource reconditioning in county-level land spatial planning,and selects two more mature machine learning methods,Back Propagation Neural Network and Random Forest,as the support of forest resource reconditioning by studying the principle of machine learning method and its practice in forest resource reconditioning.The construction method system is as follows: the forest resources endowment characteristics are recognized by "factors database" of key tree species,the "intelligent correlation" of site factors is realized by using Back Propagation Neural Network,and the potential of the key tree species suitable for forest land is quantitatively explored.Aiming at the problem of "subjectivity" of functional evaluation,based on the functional evaluation of forest ecological services,this paper uses Random Forest to realize "automatic screening" of service characteristics,predicts the dominant areas of forest development and utilization automatically and spatializes the reconditioning results with the help of GIS,so as to realize multi-technology integration to support resource reconditioning.Finally,combining with the scientific management approach of partition,classification and grading of resources,the final results of forest resources reconditioning under the county scale are "one library for supply conditions","one account for supply quantity" and "one map for supply pattern",so as to realize the quantitative implementation of the base number and the base map.The research on forest resources reconditioning provides the planning direction and spatial basis for optimizing the supply structure and spatial distribution of resources in county areas and coordinating the resources.At the same time,the application of machine learning also provides new ideas and methods for the scientific cognition of natural resources background of land space.