Radar Remote Sensing of Forests Using CARABAS and ERS
Forests cover more than 30% of the Earth's land surface, and represent an important resource. In order to manage this resource effectively, accurate and up-to-date information is required on the extent and state of the forests.
Radar remote sensing offers the possibility of providing some of this information. This thesis studies the potential of two SAR (synthetic aperture radar) systems for measuring forest stem volume. Case studies, of Scandinavian coniferous forests, have been used to develop models, which can be used to help interpret the results, to improve parameter retrieval accuracy, and to make predictions about the capabilities of the techniques.
Firstly, repeat-pass interferometry with the ERS (European Remote Sensing) satellite has been investigated. The coherence measurements presented show the potential for estimating stem volume. Unfortunately, the accuracy of stem volume retrieval is unreliable, depending on weather conditions. In the best cases, standwise estimates of stem volume show accuracies similar to those obtained with ground-based methods. The use of coherence for detecting disturbance of the forest (clear-cutting) is also investigated, showing the possibility of using repeat-pass satellite interferometry for monitoring changes in forest cover.
CARABAS is a very low frequency (VHF-band), high-resolution, airborne SAR system. Using long wavelengths (3-15 m), CARABAS penetrates forest canopies, and measures backscattering from the tree trunks. The theories and observations presented indicate that the scattering strength is directly related to stem volume, and that stem volume retrieval with CARABAS has potentially high accuracy. In particular, CARABAS is suited for dense forests, where other remote sensing techniques fail. The backscatter is very sensitive to terrain slopes; which must be corrected for when estimating stem volume. A model has been developed for this purpose, indicating the feasibility of compensating for the effects of slopes.