Potential of airborne radar to support the assessment of land cover in a tropical rain forest environment

The potential of airborne radar systems as tools for collecting information in support of the assessment of tropical primary forests and derived cover types was examined. SAR systems operating with high spatial resolutions and different wavelengths (i.e., X-, C-, L- and P-band) acquired data in Guyana and Colombia. Three fundamentally different information sources from the radar return signal were considered in the study: its strength or backscatter, polarization and phase, and spatial variability or texture. Radiometric, polarimetric, and textural attributes were computed from predefined image regions selected to represent five types of primary forest, selectively logged forest, secondary forest, and a mixture of nonforest cover types. Texture was found to be the most important source of information in high resolution X- and C-band images. Textural attributes computed per region made modest to good bases for automated classifications of the land cover types studied. Primary forests and logged-over forests were found to display particularly distinctive textural patterns. Backscatter values computed per region from L- and P-band radar images also made modest to good classification bases. Backscatter measurements in either a single L- or P-band channel enabled accurate classification of nonforest cover types. Reliable identification of secondary forests and logged-over forests generally asked for measurements in a minimum of two C-, L-, and/or P-band radar channels. Similarly, reliable assessment of primary forest types required observations in a minimum of three C-, L-, and/or P-band radar channels.