Investigation of hillslope processes and land cover change using remote sensing and laboratory spectroscopy (Arizona)

Remote sensing and laboratory spectroscopy can be used to characterize surficial properties and processes of geological and ecological interest on scales ranging from centimeters to kilometers. Linear deconvolution of mid-infrared emission spectra presents a useful technique for rapid mineralogical characterization of soils. A suite of 54 laboratory soil spectra was compiled and deconvolved using a mineral and particle-size endmember library. Average accuracy of total endmember detection is 49% while major endmember (>5% abundance) detection is 72%. Clay endmembers are detected with 75% accuracy on average, however misidentification is common due to similarity of endmember spectra. The results obtained using this technique are readily interpreted using the mineralogical families portion of the Soil Taxonomy System.; The methodology is applied to Thermal Infrared Multispectral. Scanner (TIMS) data of the McDowell Mountains, Scottsdale, Arizona, to characterize semiarid hillslope soil development and sediment transport processes. Deconvolution of laboratory soil sample and image pixel emission spectra was performed to generate surficial soil mineralogical maps of three compositionally distinct (felsic to mafic bedrock lithology) study areas. Comparison of degree of soil development and gravel + bedrock abundance between the study areas indicates that upper and lower slopes tend to be dominated by gravelly soils and exposed bedrock. Sediment accumulation and soil development is preferentially concentrated on lower slopes and midslopes. These results provide a synoptic view of hillslope geomorphic processes that validates and extends the results of smaller scale studies.; Remotely sensed data also records the spatial and temporal distribution of land cover (both natural and manmade) necessary for study of urban ecological systems. An expert system (or hypothesis-testing) model is used to classify land cover in the Phoenix, Arizona metropolitan region. The expert system integrates Landsat Thematic Mapper (TM) data with other geospatial information (such as land use, vegetation indices, and spatial texture) to obtain individual class user accuracies ranging from 73–99%. User accuracy for the Commercial/Industrial class is poor (49%) due to high degrees of subpixel mixing. The expert system is adaptable to any type of geospatial data.