| Debris flow disasters are very common in Zhouqu, causing great injures to the lives and property of the local people, and also cause significant damages to the local economy, transportation, ecological environment. Because of the sudden happening and heavy destruction of debris flows, the work of monitoring and early warning of disasters on debris flows become extremely important. Existing governances for debris flows are to build dams to intercept the materials or to construct the debris flows through drainage channels. The work of monitoring and early warning of debris flows is to monitor rainfall which is the main motivating factor for debris flows. One of the most important conditions that debris flows happen is the loose deposit. The researches on loose deposit are less. The hyperspectral remote sensing technology has been applied mainly on geology since being developed and has made great success, such as mineral species identification and so on. weathering processes of rocks in the nature is a continuous procedure. The rocks have different structures and mineral he mineral components in different phases. The weathering phase of rock could be used to determine whether it is loose deposit or not. So the hyperspectral remote sensing technology can be applied on mountainous loose deposit identification because of its advantages on mineral species identification. The region of loose deposit could be outlined and even the volume could be estimated. The result can provide support to rain recorder settings and dams and drainage channels’ building. Hyperion is a hyperspectral instrument on the American Earth-Observing1(EO-1) spacecraft. Hyperion data has a total of242bands spreading from357nm to2576nm. The interval between them is10nm. Hyperion data of Quwa in Zhouqu was used in this paper. Two methods based on the characteristic bands and complete waveform were both used to identify mineral species in the study area. In the method based on the characteristic bands we used spectral absorption index (SAI) created by Wang Jin to unmix the pixels and to map the minerals. In the method based on the complete waveform Mixture Tuned Matched Filtering method was used for unmixing of mixed pixels and mapping of minerals extracted. Then the region of loose deposit in the study area could be outlined. Here are results achieved in this paper:(1) Some preprocessing had been done before the Hyperion image being used. The bands that are not calibrated and the atmospheric water vapor absorption bands were removed. Bad pixels and vertical stripes were also removed. Spectral "smile" was corrected. Atmospheric correction and geometric correction were done. Then the date was ready for further analysis.(2) Eight samples were collected in places where landslide or avalanche had happened in the study area. XRD diffraction experiments was carried out to analysis mineral composition. The results showed four minerals chlorite, quartz, muscovite, albite were commonly existed. We got spectral parameters of three minerals chlorite, muscovite, albite by using the continuum removal method, found their " non-absorption baseline " and unmixed pixels by spectral absorption index method (SAI) created by Wang Jin. Then distribution of these three minerals was mapped.(3) we implemented minimum noise fraction transform and calculated the pure pixel indices of all pixels, then the ten endmembers were chose out. Four of them were mineral spectrums. Their lithology and mineral species were got through spectral analyst. Mixture Tuned Matched Filtering method was used for unmixing of mixed pixels and mapping of four minerals extracted.(4) vegetation coverage in study area was estimated based on normalized differential vegetation index. The correlation between mineral identification and vegetation coverage was analyzed. The results showed that the identification and extraction of minerals in low vegetation coverage area were effective, but not good in high vegetation coverage area. |
PDF Full Text Request