Radar remote sensing of a semi-arid environment : a case study in central Tunisia
This work examines the potential of spaceborne microwave remote sensing for the discrimination and analysis of morphological and surface cover-features in semi-arid Tunisia. The study area in central Tunisia comprises a region of overlap between two satellite radar passes: namely Seasat and SIR-A. This allows the influence of two different radar depression angles, look directions and resolutions upon image appearance to be determined. Botn these systems operated at a wavelength of 23cm and hence the characteristic responses of semi-arid surfaces at this wavelength are assessed. This is achieved through visual and digital image interpretation and discriminant analysis of image data. As SIR-A data is available only in optical format, the image had to be digitised before digital image processing could be undertaken. Several radiometric and geometric pre-processing procedures have to be accomplished. Despite the time-lag and calibration difficulties involved, the dominant ground controls on radar backscatter are identified through statistical analysis of information collected in the field. Surface relief, feature geometry and surface roughness are the most important parameters for both systems. The difference in depression angle causes relief information to dominate the Seasat image, while roughness information dominates the SIR-A image. The availability of Landsat Thematic Mapper data for part of the SIR-A swath west of the coastal study area provides a valuable opportunity to assess the complementary nature of data from the visible, infra-red and microwave parts of the electromagnetic spectrum. This is examined through digital and statistical analysis of image data. In the mountain environments, slope angle and aspect are found to be the dominant parameters influencing SIR-A backscatter through their effect on local radar incidence angle. The special quality of radar is its sensitivity to relief and roughness information. This is exploited in a number of applied studies that assess the contributions of radar to environmental management in semi-arid areas. Finally, recommendations are made for future research in the light of the multi-parameter radar systems due to be launched in the 1990s.