|Abstract or Summary
- Chameleon is a physics-based landscape modeling software system designed for modeling and simulations applications. Hyperspectral laboratory, field, and imaging spectrometer measurements are collected as empirical foundation data. Linear spectral unmixing is performed to decompose each image pixel into spectral endmembers. Mathematical manipulation of these fractional abundances and introduction of new spectral information is accomplished with spectral editing tools. Image spectra are modified on a sub-pixel, per-pixel, or neighborhood basis, or the entire hypercube can be customized at once. Chameleon then regenerates synthetic, but spectrally accurate, terrain models using linear spectral remixing algorithms. By incorporating elevation, sun angle, and weather data, the landscape becomes a "Chameleon"- able to change hyperspectral properties based on multitemporal spectral measurements, requirements for developmental tests and operational training, or as required by specific simulation scenarios. Advanced knowledge of natural environments to be modeled is prerequisite to generating useful synthetic terrains. Our spectral research on and Quaternary geomorphic surfaces suggests that deserts (often assumed to be less difficult to study remotely than humid, temperate, and cold environments) are more complex than is generally accepted. A variety of rock coatings can significantly alter reflectance in the solar reflected spectrum. Weathering rinds and carbonate deposits inhibit lithologic reflectance altogether. However, manganese-rich rock varnish obscures rock reflectance in the visible and near infrared wavelengths, but transmits lithologic information in the 2,000 to 2,500 nanometer (nm) wavelengths. Surface soils on desert pavements consist of a layer of eolian dust that overlies an accreting vesicular (Av) horizon. These soils have same structure and chemistry, and, therefore, the same hyperspectral signature, regardless of landform age, geomorphic process, or parent material. From a remote sensing perspective, this has a normalizing effect on reflectance across the landscape. Spectral Mixture Analysis is a proven hyperspectral technique for mapping composition and abundance of surface materials characteristic of volcanic landforms that exhibit diagnostic absorption features. We found that desert pavement spectra are featureless in that they exhibit few distinct spectral features related to rock varnish, clast lithology, or soil. Image spectra of these surfaces are the result of intimate mixtures of heterogeneous materials, requiring nonlinear spectral unmixing solutions.