The spectral properties of desert soils, including visible to short-wave infrared reflectance (VSWIR, 0.4-2.5 μm) and thermal infrared emission (TIR, 8.0-14 μm), must be accurately known in order to make remote sensing determinations of soil, rock, and vegetation compositions. However, collecting soil samples for laboratory spectral measurements commonly disturbs the surface characteristics and causes substantial spectral changes. Disturbance effects include the coating of rock fragments by fine dust particles, and the destruction of soil crusts that are weakly cemented by carbonate or sulfate minerals. To avoid spectral analysis errors related to soil disturbance we collected spectral data in situ at ten sites representing a variety of aridisol types near Las Cruces, New Mexico. The field spectral data were then used in conjunction with imagery from the Advanced Spaceborne Thermal Emission and Reflectance Radiometer (ASTER) and the Advanced Land Imager (ALI) to study soil spectral variations throughout the Las Cruces area. Soils having well-developed desert pavements showed a close spectral resemblance to local lithologic source areas, and displayed VSWIR and/or TIR spectral absorption bands related to quartz, feldspars, micas, clays and carbonates. Some soils with exposed chemical horizons, such as gypsic and calcic horizons, showed complex spectral behaviors. For example, an eroded gypsum-rich soil exhibited unusually strong VSWIR backscattering at low sun angles, presumably because of the soil's distinctive granular texture of small, faceted, crystallites. Certain spectral and mineral components in soils are related to large scale surface processes. For example, caliche (carbonate) horizons are widely distributed in the study area and the accumulation of carbonate in any given soil tends to increase with time. The exposure of shallow caliche horizons through local soil dissection is common, and may enable remote sensing distinctions to be made between geomorphic surfaces having different relative ages. Preliminary analysis of the ALI and ASTER remote sensing data indicates that key surface compositional aspects of soils in the Las Cruces study area can be discerned. These include distinctions between alluvial fans and desert pavements having different source areas, and between quartz-, gypsum- and carbonate-rich soils. Ongoing work is exploring the efficacy of spectral unmixing analysis for recognizing soil mineral constituents and their relations to geomorphic features.