Kellndorfer, J.M., M.C. Dobson, and F.T. Ulaby
Presented at the All Scientist Meeting (1996-07-16 to 1996-07-17 )
Accurate regional to global scale land cover classification is needed as input to many models concerning global change research. Current land cover characterizations are inconsistent in method and result. Remote sensing has long been championed to redress this problem. To overcome problems with optical remote sensing (e.g. atmospheric conditions, seasonal variations in illumination, sensor calibration), image data acquired by existing orbital radar systems (European ERS-1, C-band, vv-polarization; Japanese JERS-1, L-band, hh-polarization) are used to define a unique classifier. Unlike optical remote sensing techniques, radar remote sensing can provide calibrated data where the image signal is solely determined by the physical (structural) and electrical properties of the targets on the Earth’s surface and near subsurface.To date we have classified radar image composites from test sites in different eco-zones throughout a network of LTER sites. We have found that similar structural types of vegetation from different test sites show similar (absolutely calibrated) backscatter values and are classified correctly by applying the same set of decision rules. Temporal variations of backscatter values, resulting in a constraint of the applicability of a single classifier. are found to be due to freeze/thaw conditions, defoliation or when intercepted rain covered the vegetation during data acquisition.
Back to meeting | Show |