A Novel Approach For The Classification Of Geological Body Based On ASTER Image And DEM Within Sparsely Outcropping Terrain,Northwest Yunnan,China

Remote sensing technology,as an efficient and practical information acquisition method,has been widely used in geological surveys,such as linear structure extraction,lithological mapping and mineral exploration.However,there are few researches on sparse outcrops area and the reason mainly lay in three factors: 1)In the sparse outcrop area,there are fewer rock outcrops,and the background information is fierce,which makes it difficult to accurately identify lithologic outcrops.2)The lithological weathering of rock is fierce in the sparse outcrop area,which causes huge interference to the identification of the original rock.3)The method of lithology interpretation is not perfect,so that the false anomalies become quite obvious with a low degree of recognition accuracy.Therefore,it is imperative to explore and use remote sensing technology for lithological mapping in sparse outcrop areas.In order to solve the problems mentioned above,this study takes the Quaternary deposits,sandstone,carbonate rocks and basalts in the Lijiang-Dali area of northwestern Yunnan as research objects,using ASTER image,GOOGLE satellite image and DEM data as the basic data to interpret different lithological mapping units in the study area and establish a methodological system named “matched filter—fractal theory—spatial interpolation” for outcrop identification and classification.This methodological system mainly includes six aspects: 1)Pre-process the selected ASTER data,including crosstalk correction,atmospheric correction,band combination and image cropping.2)Select a few outcrops with known lithology in the study area as Regions of Interest(ROIs)and establish “standard” spectral characteristic curves of different rock types on the preprocessed remote sensing image.3)Use Matching Filtering Method(MF)to enhance the image information and take the Fractal Theory Method to decide the threshold values of the spectral curves of different lithologies so that most of the background information can be eliminated.4)Combine different bands and make ratios to eliminate false anomalies caused by vegetation,water bodies,artificial buildings and mountain shadows.5)Analyze the terrain of ground objects combined with the weathering characteristics and terrain characteristics of different lithologies in the study area to reduce the impact of terrain factors on the interpretation results.6)Conduct a field survey in combination with the 1: 50,000 geological map in the study area to evaluate the accuracy of the interpretation results and use the classification accuracy evaluation criteria of kappa coefficient to evaluate the accuracy of the interpretation results and finally draw the distribution of lithology units in the study area.The results show that the Kappa accuracy coefficients of quaternary deposits,sandstones,carbonates,and basalts are respectively 0.9305,0.8211,0.9003 and 0.4986,which verifies that the identification of the sparse outcrop in complex terrain is feasible and also provides an effective reference scheme for remote sensing geological mapping work in other sparse outcrop areas.At the same time,the results also show that this method does not have a significant effect on the identification of basalt,especially in strongly weathered areas,so it has certain limitations and still needs further research.