TERRAIN MAPPING USING INTEGRATED VISIBLE TO NEAR INFRARED, SHORT WAVE INFRARED, AND LONG WAVE INFRARED SPECTROSCOPY

Identification of geologic materials using visible to near infrared (VNIR), short wave infrared (SWIR), and long wave infrared (LWIR) spectroscopy is well established; however, the technique is not commonly used to full advantage. Most previous studies have focused predominantly on individual spectral ranges, and little has been done to utilize the complementary information available when the full range is considered. This common approach may be particularly limiting where multiple surface materials and complex lithologies cannot be adequately characterized using a single spectral range.

We have developed a new method for enhanced terrain identification and mapping that integrates VNIR-SWIR and LWIR multispectral and hyperspectral imagery. The spectral analysis is conducted in two main stages. An initial classification using mixture-tuned matched filtering (partial unmixing) is performed independently for each spectral range. These results are then joined for each co-registered pixel and undergo a second classification to produce a final terrain map derived from the full spectral range.

Initial research combined hyperspectral Airborne Visible/Infrared Imaging Spectrometer (AVIRIS) data (VNIR-SWIR, 0.4 – 2.5 µm, 224 spectral bands) and multispectral MODIS/ASTER airborne simulator (MASTER) data (LWIR, 7.8 – 12.8 µm, 10 spectral bands) of the Mountain Pass, California area. A wide variety of rock types with diverse spectral features are exposed here, and Mountain Pass is particularly notable for its unique deposits of rare earth element (REE)–bearing minerals. AVIRIS and MASTER spectra were adjusted to surface reflectance and emissivity, respectively, before analysis. The spectral and spatial dimensions were reduced using a minimum noise fraction transformation and “pure-pixel” extraction approach to select image endmembers for spatial mapping and abundance determination. The full-range analysis approach was applied and results show improvements with respect to analysis of individual spectral ranges.

Additional research using AVIRIS and the Aerospace Corporation’s hyperspectral LWIR Mako instrument (7.6 – 13.2 µm, 128 spectral bands) illustrates mapping improvements resulting from use of hyperspectral LWIR data in the full-range spectral terrain analysis.