Estimation of forest stem volume in sloping terrain using CARABAS-II VHF SAR data

In this paper a heuristic method is presented for reducing topography effects on forest stem volume estimation at stand level using a combination of low-frequency CARABAS-II very high frequency (VHF) synthetic aperture radar (SAR) images acquired from several flight directions. The method has been tested on stand-wise backscatter averages and on segment (derived from digital elevation models) backscatter averages aggregated to stand level. The investigation was performed at two test sites in Sweden characterized by coniferous forests on sloping terrain (ground slope <22°). At the two test sites, Tönnersjöheden and Brattåker, objectively inventoried stand-wise stem volumes were in the range of 80–690 m3·ha–1 (average 324 m3·ha–1) and 80–320 m3·ha–1 (average 182 m3·ha–1), respectively. At Tönnersjöheden the results show that the accuracy, in terms of root mean square error (RMSE), obtained from linear regression analysis was in the range of 70–105 m3·ha–1 using single-image data from four flight directions compared with the segment approach resulting in an RMSE of 67 m3·ha–1 using the images in combination. The corresponding results at Brattåker were 45–58 m3·ha–1 and 42 m3·ha–1 based on images acquired from four flight directions. In both cases, the RMSE obtained is about 22% of the average stem volume. The applied methodology significantly improved stand-wise stem volume estimation in sloping terrain compared with that using single-image data.