Donald M. Hooper | USGS Science

USGS2021

Linear deconvolution applied to ASTER imagery of terrestrial dune analog sites

Advanced Spaceborne Thermal Emission and Reflec-tion (ASTER) radiometer onboard NASA’s Terra satellite has nearly complete global coverage in the 8 – 14 µm thermal infrared (TIR) atmospheric window and is the highest resolution sensor providing TIR emissivity data at 90-m spatial resolution and five multispectral bands. ASTER imagery enables mapping of spatial variations in the distribution of mineral abundances across terrestrial dune fields. We compared simulated ASTER emissivity spectra derived from higher resolution laboratory measurements, with those simulated for various higher and lower resolution multispectral and hyperspectral sensors. Linear deconvolution was then applied to simulated ASTER as well as seven other convoluted laboratory spectral measurements of eight samples that we collected from seven different dune fields throughout the Western United States and Alaska: (1) Algodones, CA; (2) Big Dune, NV; (3) Bruneau, ID; (4) Great Kobuk Sand Dunes (GKSD), AK; (5) Great Sand Dunes National Park and Preserve (GSDNPP), CO; (6) Sunset Crater, AZ; and (7) White Sands National Monument, NM. We evaluate the utility of each of these seven dune fields as potential Martian aeolian analog sites by comparing their compositional similarities and differ-ences between morphologically similar dune land-forms found on Mars.

Conference Paper↗

USGS2018

Determining mineralogical variations of aeolian deposits using thermal infrared emissivity and linear deconvolution methods

We apply linear deconvolution methods to derive mineral and glass proportions for eight field sample training sites at seven dune fields: (1) Algodones, California; (2) Big Dune, Nevada; (3) Bruneau, Idaho; (4) Great Kobuk Sand Dunes, Alaska; (5) Great Sand Dunes National Park and Preserve, Colorado; (6) Sunset Crater, Arizona; and (7) White Sands National Monument, New Mexico. These dune fields were chosen because they represent a wide range of mineral grain mixtures and allow us to gauge a better understanding of both compositional and sorting effects within terrestrial and extraterrestrial dune systems. We also use actual ASTER TIR emissivity imagery to map the spatial distribution of these minerals throughout the seven dune fields and evaluate the effects of degraded spectral resolution on the accuracy of mineral abundances retrieved. Our results show that hyperspectral data convolutions of our laboratory emissivity spectra outperformed multispectral data convolutions of the same data with respect to the mineral, glass and lithic abundances derived. Both the number and wavelength position of spectral bands greatly impacts the accuracy of linear deconvolution retrieval of feldspar proportions (e.g. K-feldspar vs. plagioclase) especially, as well as the detection of certain mafic and carbonate minerals. In particular, ASTER mapping results show that several of the dune sites display patterns such that less dense minerals typically have higher abundances near the center of the active and most evolved dunes in the field, while more dense minerals and glasses appear to be more abundant along the margins of the active dune fields.

Alaska, Arizona, California, Colorado, Idaho, Neva↗