| Thermal infrared (TIR) spectroscopy and remote sensing have been proven to be extremely valuable tools for mineralogical discrimination. The need for accurate processing, detection, and data reduction techniques becomes apparent when considering the large data volume soon to be returned from future Earth and Mars orbital missions. One such technique, known as spectral deconvolution using a linear retrieval algorithm, provides the ability for mineral (endmember) detection along with estimates of its percentage and accuracy. Translated to image format, the results produce abundance maps of the desired endmembers. Techniques such as these are not new and have been used for years in remote sensing. This study, however, is the first quantitative attempt to identify the limits of the model with specific emphasis on thermal infrared emission. To understand the results of different initial conditions, the algorithm was coded and rigorously applied to high-resolution laboratory data, testing the effects of particle size, noise, multiple endmembers, and instrument precision. In addition, it has been adapted to operate on airborne Thermal Infrared Multispectral Scanner (TIMS) image data. Testing was performed on fluvially-reworked impact alluvium at Meteor Crater, Arizona; the eolian sediments at Kelso Dunes, California; and the silicic lava domes at Medicine Lake, California. Detectability limits, residual errors, and mixing patterns at each of these areas were examined. Averaging the results from all the test cases indicate that linear deconvolution of emission spectra has a precision of 5% depending on the initial conditions. Mixtures in the laboratory were predicted to within 4% with RMS errors as low as $1.0/times10/sp[-6]$, whereas image results had average deviations of 5-10% from known abundances with RMS errors on the order of $1.0/times10/sp[-3].$ Further, lava vitrification and textural variations were able to be modeled. Spectral deconvolution via a linear retrieval approach where applied to thermal infrared data provides a technique for mineral and textural mapping. In a remote sensing context, its use should be constrained to arid regions where surface exposure and thermal flux are the greatest. The ability to predict endmember abundances to within several percent becomes extremely useful in monitoring sediment transport, desertification, and potential volcanic hazards. |