Model-Based Stem Volume Retrieval and Windthrow Detection using CARABAS and P-band SAR
Radar remote sensing is an efficient method for monitoring forests over large areas. Radar waves transmitted from a system carried by a satellite or an aircraft interact with trees on the ground, and the reflected waves are used to form images which are independent of weather and light conditions. Information in the images is then used to retrieve forest parameters.
CARABAS is an airborne synthetic aperture radar (SAR) system. Due to the long wavelengths of the system (3-15 m), the radar signal is able to penetrate the forest canopies and for boreal forests the main backscattering component is the ground-trunk double-bounce. The scattering strength is directly related to the stem volume and hence stem volume retrieval with CARABAS has higher accuracy than SAR systems using shorter wavelengths. However, the ground-trunk double-bounce is very sensitive to ground slope and if not corrected for, ground slope will cause the stem volume to be underestimated.
By using multiple images acquired with different flight headings and combining the image information with ground topography data in a model-based inversion method, most of the ground topography influence on the retrieved stem volume is removed. With four CARABAS images and a coarse digital elevation model with 50 m horizontal grid, the stem volume can be retrieved with a root mean square error
(RMSE) of 48 m3/ha for stem volumes in the range 80-700 m3/ha. The proposed retrieval method gives better accuracy and is more robust than existing heuristic methods. By using this method, the retrieval accuracy in areas with ground topography is similar to that of similar forests standing on flat and horizontal ground.
Another forestry application where CARABAS and other airborne SAR systems with long-wavelengths and high resolution can provide timely and useful information is in the detection of windthrown trees. A method for detecting windthrows in CARABAS images and a discussion about when windthrows can be detected is presented in this thesis.
model-based stem volume retrieval
synthetic aperture radar (SAR)
EA-salen, Hörsalsvägen 11, Chalmers, Göteborg.
Opponent: Prof. Gunilla Borgefors, Centre for Image Analysis (CBA), Uppsala University and the Swedish University of Agricultural Sciences, Uppsala, Sweden.