The International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences
Publications Copernicus
Download
Citation
Articles | Volume XLII-2/W13
Int. Arch. Photogramm. Remote Sens. Spatial Inf. Sci., XLII-2/W13, 1893–1897, 2019
https://doi.org/10.5194/isprs-archives-XLII-2-W13-1893-2019
Int. Arch. Photogramm. Remote Sens. Spatial Inf. Sci., XLII-2/W13, 1893–1897, 2019
https://doi.org/10.5194/isprs-archives-XLII-2-W13-1893-2019

  05 Jun 2019

05 Jun 2019

APPLICATION OF HYPERSPECTRAL THERMAL EMISSION SPECTROMETER (HYTES) DATA FOR HYSPIRI OPTIMAL BAND POSITIONING TO CHARACTERIZE SURFACE MINERALS

S. Ullah and A. Iqbal S. Ullah and A. Iqbal
  • Department of Space Science, Institute of Space Technology (IST), Islamabad, Pakistan

Keywords: Hyperspectral, Thermal Infrared, Surface minerals, HyTES, HyspIRI, Dimensionality reduction and Genetic Algorithms

Abstract. This study aimed to characterize surface minerals from high dimensional HyTES (Hyperspectral Thermal Emission Spectrometer) data comprised of 256 spectral bands between 7.5 and 12 μm (i.e., TIR domain of the electromagnetic spectrum). The HyTES is across-track imager and can image 512 pixels with spatial resolution varies between 5 to 50 m depending upon aircraft flying height. HyTES is developed to support the HyspIRI (Hyperspectral Infrared Imager) mission by acquiring TIR data at much higher spectral and spatial resolutions in-order to define the optimum band positions for the TIR instrument of HyspIRI. For earth compositional mapping, the HyTES images of Cuprite and Death Valley regions were acquired in summer 2014 and spectral emissivities of fifteen minerals classes were extracted from regions of known mineral compositions and were randomly divided into training and testing sets (each mineral class com-prised of 100 spectra). These extracted emissivity signatures were then used for categorizing minerals and for finding HyspIRI's optimal band positions for earth composition mapping using Genetic Algorithm (GA) coupled with Spectral angle mapper (SAM). The GA-SAM was trained for fifteen mineral classes and the algorithms were run iteratively 40 times. High calibration (> 95 %) and validation (> 90 %) accuracies were achieved with limited numbers (seven) of spectral bands selected by GA-SAM. Knowing the important band Positions will help scientist of HyspIRI group to place spectral bands at regions were accuracies of Earth compositional mapping can be enhanced.