Airborne low-frequency synthetic aperture radar and optical satellite data as complementary data sources for forest stem volume estimation

Forest stem volume estimation using a combination of airborne CARABAS-II low-frequency synthetic aperture radar (SAR) and multispectral optical Landsat ETM+ satellite data was evaluated and compared with aerial photo-interpretation. The study was performed in coniferous-dominated hemi-boreal forests at a test site located in southern Sweden. The stem volume for the stands included in the analysis was in the range of 15-585 m(3) ha(-1) (average 266 m(3) ha(-1)) and the average size of the stands was 3.5 ha. Regression analysis was used to develop stem volume models. The accuracy in terms of root mean square error (RMSE) was found to be 18.5% (of the average stem volume) for CARABAS-II, 24.8% for Landsat ETM+, and 14.3% for the combination. CARABAS-II gave the best results for high stem volumes, and Landsat ETM+ was more accurate for lower stem volumes. Hence, the two data sources supplement each other providing in combination better results over the whole range of stem volumes. Using conventional aerial photo-interpretation in panchromatic photographs for stem volume estimation the RMSE was found to be 23.7% (21.1% after correction for systematic errors). In conclusion, the method of combining airborne low-frequency SAR and optical satellite data for stem volume estimation gave an accuracy that was significantly better than aerial photo-interpretation and comparable with airborne laser scanning.