| With the rapid development of serious resources such as mountain slope collapse,and ecological coverage with good economic conditions such as mountain slope collapse,there are already land,arable land and excavation that endanger the environment,which is a serious resource mining problem,prison escape,monitoring and occupation,and excavation.In other phenomena circles,the problem of real estate resource supervision is prominent.Therefore,this paper uses multi-source remote sensing to study the temporal and spatial evolution of mining land,which provides a basis for promoting natural resource supervision and ecological environmental governance.Taking Anyang open-pit mine as the research area,this paper uses GF-1 WFV,GF-2 PMS and field survey and mining rights data from 2017 to 2020 to establish open-pit mine interpretation signs,and analyze the spectrum and texture of various ground objects in the study area.And terrain features,and optimizes the extraction feature set;A multi-feature open-pit mining area identification method and technical process based on the Cg parameter optimization libsvm model were constructed,and the mining land change status of the open-pit mining area from 2017 to 2020 was extracted accordingly.Providing a basis for natural resource regulation and scientific management.The main research work and conclusions are as following:(1)Using the GF-1 image features,the mining land interpretation markers were established;the spectral and texture differences as well as the spatial location,slope and slope direction of the mining land,residential areas,cultivated land,forest land and water features were analyzed,according to which the set of mining land extraction features established was optimized,the thresholds of topographic factors such as slope,slope direction and height for different mining land types were constructed,and then the spatial distribution pattern of mining land in the open pit mine in the study area was extracted based on the decision tree method.(2)To verify the superiority of the Cg parameter seeking libsvm model constructed in this paper,the features such as band,texture,vegetation index and topography in the optimized feature set are superimposed one by one to construct a multi-feature combination,and the mining land is extracted by using maximum likelihood,BP neural network,random forest algorithm and the method in this paper,and it is concluded through comparative analysis that the extraction accuracy of the method in this paper is optimal,and the overall accuracy,F1 function and The overall accuracy,F1 function and kappa coefficient are 91.76%,94.55% and 0.8972,respectively,thus demonstrating the superiority of the method in this paper.(3)Using the method of this paper,we extracted the changes of mining land in Anyang City from 2017 to 2020,and analyzed the spatial and temporal evolution of mining land by using spatial dynamic change analysis methods such as standard deviation ellipse,mean center and dynamic attitude,and concluded that: mining land is only distributed in the west of Anyang County and northeast of Linzhou City,and the mining land and cross-border mining are decreasing year by year;except for 2019,the distribution of mining land is "northwest-southeast diffusion and northeast-southwest aggregation",while the other years are "northeast-southwest diffusion and northwest-southeast aggregation".The dynamic degree from 2017 to 2019 was-11.97%and-26.62%,indicating that the decline rate of mining land continued to increase during this period;the dynamic degree from 2019 to 2020 was 9.10%,indicating that mining land showed a slight rebound.37 pictures,12 tables,73 references... |
PDF Full Text Request