基于植被岩溶比重指数的岩溶流域岩溶信息提取——以江西省长河流域为例

Information Extraction of Karst in Karst Watershed Based on Vegetation- Karst Proportion Index——A Case Study of Changhe River Basin in Jiangxi Province

利用多光谱遥感数据构建合适的指数提取岩溶信息,对于岩溶地表水文参数(如CN值)的确定、流域水文模型的应用、岩溶流域水土流失和石漠化的监测等具有重要作用。以江西省长河流域为研究区,利用Landsat 8遥感数据获取NDVI、FV(植被覆盖度)、LST(地表温度)以及卷云波段,将FV、LST和卷云波段进行主成分分析,提取第一主成分PFLC并构建PFLC-NDVI特征空间,依据该特征空间提出植被岩溶比重指数(VKPI),采用分级阈值法对VKPI值进行等级分值生成VKPI分级图,结合野外实地采集的岩溶数据进行精度验证。研究表明:野外实地采样的岩溶数据与VKPI分级图重合度比较高,总精度为77.44%,Kappa系数为0.713 4。综合了植被指数、地表温度、植被覆盖度等生物物理参数的VKPI,能够较好地提取岩溶信息,并将其与其它地物信息区分开来。

The multi-spectral remote sensing data are applied to construct suitable index for information extraction of karst. It is of great significance to Karst surface hydrological parameters （ such as CN value） the identification, application of watershed hydrological models, monitor soil erosion and monitoring of rocky desertification in karst watershed. In this paper, as a study area of Changhe river basin in Jiangxi province, NDVI, Fv （vegetation coverage） , LST （land surface temperature） and cirrus band were obtained by using Landsat 8 remote sensing data, principal component analysis was done with Fv, LST and cirrus band. Fv, LST and cirrus band were used for principal component analysis, and the extraction of first principal component PFLc was to construct the PFLc-NDVI space, vegetation-karst proportion index （VKPI） was proposed on the basis of PrLc-NDVI space, VKPI were ranged on grade values by grade threshold method. Verified with field karst data, the values results show that VKPI classification diagram data has a high coincidence degree with field sampling data, total accuracy of 77.44%, Kappa coefficient of 0. 713 4. Finally, VKPI classification diagram was generated. Research shows that VKPI which integrates the biophysical parameters such as vegetation index,land surface temperature, vegetation coverage can better extract karst information and distinguished different from other information.