Research On Retrieval Model Of Surface Geochemistry Composition Based On Remote Sensing Data

Exploration geochemistry uses geochemistry data and geochemistry to study surface and subsurface conditions in an area,but the collection of geochemistry data is very labor intensive,especially in some bad conditions of the natural environment to collect samples,and then need to collect samples for analysis,the process is also very cumbersome.In this paper,a remote sensing geochemistry inversion model is constructed,which combines the advantages of geochemistry method and remote sensing technology,the advantages of remote sensing data acquisition in time and space,and the chemical element distribution law.Because of the large amount of remote sensing data,redundant data and strong correlation,it is necessary to extract the features of remote sensing data.Principal Component Analysis(PCA)can reduce the data dimension without losing the main information,while kernel PCA can deal with the nonlinear data better by mapping the kernel function to the higher dimension space.At the same time,because of the discontinuity of the abnormal distribution of the geochemistry and the non-linear characteristics of the remote sensing data bands,we choose the support vector regression model optimized by bee colony in the choice of the machine learning method,support vector machine uses inner product kernel function to map to high-dimensional space,which has better generalization ability and robustness,but the choice of parameters will affect its performance greatly,inspired by the foraging behavior of honeybees in 2005,Karaboga and others proposed that by simulating the activity of Honeybee foraging,the high-dimensional problem and multi objective optimization problem optimization problem can be solved quickly and the local optimal solution can be obtained,it is combined with the support vector regression model to optimize the model parameters and optimize its performance.In this paper,Kernel Principal Component Analysis(KPCA)and swarm optimization(Support vector machine)are combined to establish the geochemistry model.In this paper,the mawu-zhaishang region of Gansu Province was selected as the research area,and 1:50000 river sediment geochemistry data and multi-spectral Landsat 8OLI data were selected as the experimental data.Using swarm optimized support vector regression to establish the relationship between geochemistry data and remote sensing data,compared with bayesian ridge regression model,general linear regression model,elastic network regression model and support vector machine regression model,it is found that the parameter optimized support vector machine regression model has the highest score.In order to verify the validity of the model,the outliers delineated from the model output are compared with the outliers delineated from the original data.The analysis results show that the geochemical data delineate the areas with relatively enriched elements,but indicate the low-abnormal ore(chemical)points,and the anomalies delineated by the inversion data are better for this purpose,for better indication.At the same time,the distribution and intensity of the corresponding abnormal areas found that the abnormal areas delineated by the inversion data basically contain the abnormal areas delineated by the original data,and the anomalies located at the ore spots are obviously enhanced,it shows that the support vector machine model of artificial bee colony optimization can establish the relation between geochemistry data and remote sensing data,can supplement the original data effectively,and can also provide the direction for the next mineral prospecting works.