COMPARATIVE-ANALYSIS OF LANDSAT-5 TM AND SPOT HRV-1 DATA FOR USE IN MULTIPLE SENSOR APPROACHES

In view of multiple sensor approaches in remote sensing, we have analyzed the geometric and, in particular, radiometric accuracy of multispectral SPOT HRV (high resolution visible) data in comparison to Landsat-5 Thematic Mapper images. We used concurrent system corrected TM and SPOT images which were recorded almost simultaneously over the southern Ardeche region (France). The TM and SPOT scenes were registered to map projection with subpixel accuracy, but the mountainous character of the study region required the use of digital elevation data for the compensation of relief induced distortions. Our radiometric comparison of the TM and SPOT sensors involved an analysis of the calibration accuracy and a comparison of spectral greenness and brightness features which can be obtained from the two systems. The calibration assessments were primarily based upon the apparent reflectances of bare soil reference targets, and the in-flight calibration of both sensor systems was found in accordance with the respective specifications. For the comparative analysis of spectral features we used an extended set of targets, representing a wide range of vegetated and nonvegetated cover types. In order to directly relate the image-based relations to those obtained from field radiometry data, we had to correct the TM and SPOT data for atmospheric absorption, scattering and pixel adjacency effects. Both systems provide slightly different brightness and greenness information (Tasseled Cap equivalent brightness and greenness, normalized difference vegetation index). But these indices appear to be linearly related, and the respective transfer functions permit mutual data adjustments. Therefore, a wide range of applications in agricultural monitoring and vegetation observation can be approached by using multiple sensor data sets which involve TM and SPOT imagery.