| This paper take Xue Jiping and Chun Du mining area as study area, by comparing and analyzing the ecological and biochemical characteristics of dominant vegetation inside and outside of the mining area, taking advantage of ground measured hyperspectral data and Hyperion imaging data, combining with geological maps and geochemical data, basing on statistics methods to select vegetation unusual spectral character, analysis it's rule and realize the abnormal information extraction of Chun Du from Hyperion image, and provide the theoretical support for use hyperspectral remote sensing technology to find copper deposits in high plantation coverage area. The results show:(1)The leaves of dominant plant in mining area is more yellow than background area, besides there are many black points in these leaves which show significant abnormal ecology character. Because of the barrier effect of vegetation, Quercus of the two areas didn't show obvious enrichment phenomenon, it is difficult to find out the mineralize abnormal by Quercus. While fir in Xue Jiping is of low barrier factor and high contrast coefficients, which is a kind of effective indicative plant.(2)The reflectivity of vegetation in mining zone is the highest, the reflectivity of vegetation in porphyritic mass take the second place, the value of background area is the lowest, and the infrared band is sensitive to geological change. Quercus and fir is sensitive to ore body, so they can be used to detecting ore by remote sensing, while pinus is not suit for that. Different mining area and vegetation have different sensitive vegetation index. The variable coefficient of m ARI for Quercus and fir is greater than 0.5, and the variable coefficient of m ARI for Pinus yunnanensis is greater than 0.3, so the vegetation m ARI can effectively reflect how is the vegetation's spactral being affected by metallogenic element.(3) In Chun Du mining area, the Hyperion's 750nm-1300 nm infrared reflectance of Quercus in ore body was significantly lower than those in the porphyry body and the background area. In infrared spectrum, Quercus canopy's reflectance spectra changes is consistent with the trend of the blade in the ore body, which is significantly lower than those in the periphery. The canopy spectral characteristics of Hyperion image are similar to leaves' hyperspectral characteristics in the mining area, leaves are more sensitive to geochemical anomalies. Quercus leaves' spectral variation coefficient of MCARI4, NWI, PRI2, PWI in the ore body and its correlation coefficient with soil Cu element content of the absolute values are large, these hyperspectral index can effectively reflect mineralized anomalies information. The NWI and NDWI of Quercus' leaf and Hyperion imaging canopy spectral are sensitive to Cu element content in soil, Quercus canopy spectral's MTCI has high correlation with Cu element content of soil but Leaf's MTCI is not sensitive to Cu element in soil. NWI, NDWI and MTCI index of plant canopy spectra is extremely sensitive to mining abnormal, the calculation of NWI, NDWI and MTCI index use all or partly bands within 750nm-1300 nm infrared spectral range, this range of the spectrum is the key to study of vegetation canopy spectral abnormal information range.(4) As for Hyperion images, using hyperspectral vegetation index NWI, NDWI and MTCI, direct classifying the abnormal plant or getting on false color image synthesis process as R, G, B bands, we find that NWI vegetation anomalies grading can be a good indication of vegetation anomalies. At last, we consider NWI, NDWI and MTCI of Hyperion image's sampling points as the independent variable and ground mineralization abnormalities as the dependent variable, build a multiple linear regression model. The model can effectively extract abnormal vegetation which represents unusually symbol and restrain the influence of the periphery area. Therefore, the method can be used to predict prospecting targets in areas which vegetation specie is single and of high vegetation coverage. |
PDF Full Text Request