Mineral Mapping In Dongsheng District Based On Space Hyperspectral Remote Sensing

This research is about utilizing the first space hyperspectral remote sensing data, Hyperion to do some research on mineral mapping in DongSheng district of the Inner Mongolia Autonomous Region. First do some research on geology about DongSheng district, according to mineralogy rules we decided do mineral mapping about ten types of minerals, they are kaolinite, montmorillonite, sericite, dolomite, magnesite, calcite, siderite, goethite, hematite, jarosite. With help of standard spectral from USGS Spectral Library, I analysis special absorbs and reflections characteristics of these ten minerals, so I have a good start in mineral mapping study.EO-1 Hyperion is the first civil satellite spectrograph, its spectral resolution is less than 10nm with 242 bands. Spectral covers from 355 to 2577nm continuously. Because of the height of orbit, the data collected by Hyperion are influenced seriously by atmosphere. Also there is serious Swath Error and noise information, strong absorbing bands of water vapor and so on contended in Hyperion data. It is difficult to rebuild the spectral and extract the useful information. Using Fourier Transform, then defined the Filtering Operators which can be used for remove the Swath Error quickly and exactly. I didn't know the appropriate spectral Calibration methods of space hyperspectral data before this study. Through comparing many spectral Calibration Models, I use atmosphere radiation-based spectral Calibration Model, the core of FLAASH to Calibrate Hyperion data and rebuild the spectral, especially I didn't have the synchronous real spectral data. I get a good result about spectral rebuild. Because the space resolution of Hyperion data is the same as LandSat ETM+ data, also the orbit is not too much differences, I decided to registration Hyperion data based on ETM+ data of the same district.A serial of methods has been summarized in research: in the first instance analysis standard spectral, the different minerals have different spectral, the ratio of two bands could magnify this effect; of course the bands have special characteristics. So I could ascertain the approximate position of minerals, after that I collected the real spectral from Hyperion reflectance data to build a Library of given minerals. And then with the help of MNF transform, I extracted data without too much noisy. After using PPI method, I could get characteristic endmembers, what should I do is compare the spectral of these endmenbers with real spectral extracted form Hyperion reflectance data, choose and modify them.Using Spectral Angle Mapper, Linear Spectral Unmixing and Spectral Feature Fitting methods to mapping. With the help of geology map and geography information I do validate and conclusion on the result of mineral mapping.