Browsing by Subject "laaja alue"

The large area forest mapping is traditionally based on satellite images and laborious field plots. The newer lidar plot concept is the Airborne Laser Scanning (ALS) and a small amount of field plots based plot-level attribute estimation, which is treated similar to the field plots in satellite based mapping. The target of the study was to evaluate the satellite-based, the ALS-based and their combination based forest mapping in the large area by comparing the accuracy of the forest attribute estimations per different amounts of plots. The study area of 3 128 000 hectares in southern Finland included 3630 field plots from the Metsäkeskus of Finland. The attribute estimates were computed from the Landsat 8 image and from the low-density lidar dataset to the field plot locations. The plot-level attribute estimates were calculated for the volume, height, basal area and biomass using k-NN-method (k=5). The accuracy of the mapping methods was evaluated by relative RMSE (Root Mean Square Error), which measured the differences between the estimated and the measured attributes of the validation plots. From the results it can be seen that the relative RMSEs of the attribute estimates from the Landsat-image based mapping on plot-level were at their best 47.6 % for the volume, similarly 26.2 % for the height, 34.6 % for the basal area and 47.7 % for the biomass. For the ALS-based mapping, they were 23.2 %, 9.8 %, 21.0% and 23.7 % respectively. By using 400 field plots to estimate attributes for 2900 lidar plots, and these lidar plots were used further for Landsat-based attribute estimations, these results were quite close to the accuracy of field plots-based Landsat attribute estimations. In addition a small sample of the lidar cover of the area was needed, in this case having less than 1 % area coverage. The lidar plots in Landsat-based mapping could be used for the biomass and carbon mappings in large area even if there are no roads available in the area and only a small amount of plots can be measured. Also, it is suitable for the areas where forest mapping has not started. It can be used also to follow up the changes in large forest areas, due the frequent updates of the satellite image measurements.