Thesis--University of Wisconsin--Madison. / Typescript. Vita. eContent provider-neutral record in process. Description based on print version record. Includes bibliographical references.

The potential of remote sensing in upland soil erosion studies has been examined on Glaisdale Moor, North Yorkshire Moors. The study considers four different remote sensing sources, viz. sequential air photographs, ground radiometry, Landsat Thematic Mapper (TM) and SPOT simulation. Sequential air photographs have been interpreted in order to elucidate the land use/land cover changes and the drainage development and associated erosion problems in the region. A series of statistical analyses were employed in an effort to establish the relationships between the different spectral variables and the soil/ground variables. Attempts have also been made to evaluate the spectral separability performance of the Ground radiometer, the Landsat TM and the SPOT simulation wave bands. The Landsat TM and the SPOT simulation imagery have been further analysed in order to gather information about the best band and band combinations that would be required to optimize the discrimination of moorland surface types including eroded areas. Digital image processing of the Landsat TM and the SPOT simulation subscene for Glaisdale Moor was performed using the DIAD image processing system. The land use/land cover classification information derived from the air photographs, the Landsat TM and the SPOT simulation, has been used as an input into a soil loss prediction model (USLE) to predict the soil erosion rate of the study area. Of the various remote sensing systems used, air photographs and TM data proved the most useful in this area.

This research examines the use of remote sensing techniques to quantify rill erosion in two agricultural fields in the Lower Fraser Valley. Soil erosion during the winter is particularly problematic in some of the sloping soils developed from loess over glacio-marine parent materials. New techniques are needed to quantify rill erosion on a timely basis, and this research focuses on measuring the extent and rate of rill erosion from field and aerial photograph measurements. A model which used rill measurements as input, was used to determine the rill plan areas, rill volumes, and thus rill erosion rates in the test area. Using field rillometer measurements of rills as input into the model resulted in a soil loss estimate of 49m³ /ha/yr or 38.4 t/ha/yr for the test site. This soil loss estimate is deemed to be more reliable than erosion plot and Universal Soil Loss Equation estimates of soil loss for the test area. The rill volume and plan area of three main rills, using three different rill measurement methods for input into the model, were compared. Using field measuring tape measurements of rills as input into the model, resulted in a soil loss estimate which was 16 % greater than the estimate from rillometer measurements. Using photo rill width measurements and an estimation of rill depths and bottom widths from field data as model input, resulted in a soil loss estimate which was 22 % less than the estimate from rillometer measurements. Spectral reflection measurements made in rill, interrill and depositional areas were found to be significantly different, confirming that rill erosion could be assessed in a quantitative manner using digital image analysis techniques. The spectral separation was largely due to differences in organic matter, surface roughness and imaging geometry. The latter is of particular importance in creating darker shadowed rill sides opposite bright sun-facing rill sides within a single rill. A maximum likelihood classifier, used as part of the computer based image analysis, determined the rill plan area for a sample area to be 9 % less than the rill plan area obtained from the model using rillometer input. This indicates the potential of digital analysis to quickly determine the plan area of larger rills. Digital elevation and moisture content data confirmed that the topographic shape of the field is important in determining the spatial pattern of rill formation. The combination of such data with image analysis and geographic information systems (GIS) have great potential in the timely quantification of erosion in the future. / Land and Food Systems, Faculty of / Graduate