Remote Sensing Lithologic Classification of Multispectral Data Based on the Vegetation Inhibition Method in the Vegetation Coverage Area
-
摘要: 植被的发育限制了遥感在地质学方面的应用, 在植被覆盖区进行岩石填图, 首先要考虑去除植被干扰影响.以内蒙古东乌旗地区为例, 选择先进星载热发射和反射辐射仪(advanced spaceborne thermal emission and reflection radiometer, ASTER)数据, 分别计算研究区内含土壤因子植被指数和不含土壤因子的植被指数, 并对两类不同的植被指数进行主成分分析, 挑选出植被信息被抑制和岩石-土壤信息突出的主成分进行岩性分类, 和利用最大似然法的分类结果进行对比分析, 评价两种方法的岩性分类性能, 植被抑制法的总体分类正确率为82.946 8%, 最大似然法的总体分类正确率为76.364 3%.结果说明在植被覆盖区, 利用植被指数来抑制植被信息是可行的, 和常规分类方法中的最大似然法相比, 大大提高解译的准确性.Abstract: It is the top priority for rock mapping in the vegetation coverage area to eliminate the vegetation interference effect since the growth of vegetation limits the application of remote sensing in geology. Taking Dong Ujimqin Banner of Inner Mongolia as the study area, this paper compares vegetation inhibition method and maximum likelihood method in lithologic classification. Firstly, ASTER (advanced spaceborne thermal emission and reflection radiometer) data are chosen for vegetation index calculation with the soil factor and the vegetation index without the soil factor for principal component analysis respectively in the study area. Then, the principal component which shows the vegetation information is suppressed for lithologic classification. Furthermore, a comparative analysis is conducted and the lithology classification performance of the two methods is evaluated. It is found that the overall classification precision of the vegetation inhibition method reaches 82.946 8%, while that of the maximum likelihood classification reaches 76.364 3%. It shows that it is feasible to use the vegetation index to suppress the vegetation information in the vegetation coverage area. Compared with the conventional classification method of maximum likelihood method, the vegetation inhibition method greatly improves the accuracy of interpretation.
-
图 1 研究区区域地质图
1.全新统；2.下更新统复台堆积；3.砖红色粉砂质泥岩；4.大磨拐河组；5.满克头鄂博组上端；6.满克头鄂博组一段；7.安格尔音乌拉组二段；8.安格尔音乌拉组一段；9.塔尔巴格特组；10.泥鳅河组二段；11.泥鳅河组一段；12.细粒似斑状黑云母二长花岗岩；13.中粒白云母二长花岗岩；14.中细粒白云母二长花岗岩；15.中细粒正长花岗岩；16.中粒正长花岗岩；17.细粒二长花岗岩；18.中细粒花岗闪长岩；19.中细粒二长花岗岩；20.中细粒正长花岗岩；21.细粒正长花岗岩；22.二长斑岩
Fig. 1. Regional geological sketch of experimental area
图 2 ASTER数据最大似然法岩性分类结果
1.泥鳅河组一段；2.细粒正长花岗岩；3.细粒二长花岗岩；4.满克头鄂博组一段；5.下更新统复台堆积；6.中细粒二长花岗岩；7.中细粒二长花岗岩；8.中细粒白云母二长花岗岩；9.细粒似斑状黑云母二长花岗岩
Fig. 2. The lithology classification sketch of maximum likelihood method about the ASTER data
图 3 ASTER数据植被抑制法岩性分类结果
1.泥鳅河组一段；2.细粒正长花岗岩；3.细粒二长花岗岩；4.中细粒正长花岗岩；5.满克头鄂博组一段；6.下更新统复台堆积；7.中细粒二长花岗岩；8.砖红色粉砂质泥岩；9.中细粒白云母二长花岗岩
Fig. 3. The lithology classification sketch of vegetation inhibition method about the ASTER data
表 1 添加土壤因子植被指数的主成分特征贡献值
Table 1. Principal component feature contribution of the vegetation indexes with soil factors
主成分 PC1 PC2 PC3 PC4 PVI -0.559 290 0.828 982 -0.000 498 0.000 004 TSAVI 0.000 020 -0.000 100 -0.197 017 0.980 400 SAVI -0.000 138 0.000 495 0.980 400 0.197 017 DVI -0.828 982 -0.559 280 0.000 166 -0.000 001 总计 -1.388 367 0.270 098 0.783 051 1.177 420 
下载: 导出CSV
表 2 不添加土壤因子植被指数的主成分特征贡献值
Table 2. Principal component feature contribution of the vegetation indexes with not soil factors
主成分 PC1 PC2 PC3 PC4 PC5 PC6 PC7 SRI 0.179 793 0.317 875 -0.443 380 -0.401 495 -0.300 974 0.636 127 0.116 642 ARVI 0.271 234 -0.293 970 0.312 986 -0.474 883 0.628 784 0.313 792 -0.150 674 NDWI 0.050 802 0.782 098 0.619 458 -0.044 821 0.001 918 -0.000 785 -0.000 072 RVI 0.892 539 -0.022 930 -0.011 750 0.446 689 -0.055 993 0.006 585 -0.000 770 PSRI -0.084 350 0.388 712 -0.466 790 0.266 513 0.714 316 -0.027 097 0.204 650 SIPI 0.186 046 0.220 834 -0.321 030 -0.362 953 -0.002 182 -0.495 718 -0.660 366 NDVI 0.230 590 -0.021 600 -0.025 190 -0.457 147 -0.025 571 -0.500 363 0.696 941 总计 1.726 652 1.371 019 -0.335 700 -1.028 097 0.960 298 -0.067 459 0.206 351
下载: 导出CSV
表 3 研究区岩性分类正确率评价
Table 3. Lithology classification accuracy evaluation in the study area
Class 植被抑制法(%) 最大似然法(%) D1-2n1 77.87 87.19 xεγC 82.45 85.17 xηγJ 86.70 88.55 D3a1 78.78 72.27 N2b 83.73 Qp1 91.83 81.87 zxηγC 70.38 66.69 zxεγJ 67.83 zxmuηγJ 88.59 81.56 x(π)βηγJ 67.27 总体分类正确率(%) 82.946 8 76.364 3 Kappa系数 0.733 9 0.713 3
下载: 导出CSV
-
[1] Chen, S.B., Liu, Y.L., Yang, Q., et al., 2012. Lithologic Classification from Hyperspectral Data in Dense Vegetation Cover Area. Journal of Jilin University(Earth Science Edition), 42(6): 1959-1965(in Chinese with English abstract). http://en.cnki.com.cn/Article_en/CJFDTOTAL-CCDZ201206044.htm [2] Clark, R.N., Swayze, G.A., 1995. Mapping Minerals, Amorphous Materials, Environmental Materials, Vegetation, Water, Ice and Snow, and Other Materials: The USGS Tricorder Algorithm. Summaries of the Fifth Annual JPL Airborne Earth Science Workshop, 95(1): 39-40. [3] James, K.C., David, W.B., Lawrence, C.R., et al., 1989. Airborne Imaging Spectrometer Data of the Ruby Mountains, Montana: Mineral Discrimination Using Relative Absorption Band-Depth Images. Remote Sensing of Environment, 29(2): 121-134. doi: 10.1016/0034-4257(89)90021-7 [4] Roger, N.C., Trude, V.V.K., Matthew Klejwa, G.A., et al., 1990. High Spectral Resolution Reflectance Spectroscopy of Minerals. Journal of Geophysical Research, 95(B8): 12653-12680. doi: 10.1029/JB095Ib08p12653 [5] Wang, Y.D., Shen, X.L., Xie, J., 2006. A Review of Remote Sensing Image Classification Method. Remote Sensing Information, (5): 67-71 (in Chinese with English abstract). [6] 陈圣波, 刘彦丽, 杨倩, 等, 2012. 植被覆盖区卫星高光谱遥感岩性分类. 吉林大学学报(地球科学版), 42(6): 1959-1965. https://www.cnki.com.cn/Article/CJFDTOTAL-CCDZ201206044.htm [7] 王一达, 沈熙玲, 谢炯, 2006. 遥感图像分类方法综述. 遥感信息, (5): 67-71. doi: 10.3969/j.issn.1000-3177.2006.05.021 -
点击查看大图
计量
- 文章访问数: 2846
- HTML全文浏览量: 148
- PDF下载量: 379
- 被引次数: 0
