Airborne laser scanning data and corresponding field data were acquired from boreal forests in Norway and Sweden, coniferous and broadleaved forests in Germany and tropical pulpwood plantations in Brazil. Treetop positions were extracted using six different algorithms developed in Finland, Germany, Norway and Sweden, and the accuracy of tree detection and height estimation was assessed. Furthermore, the weaknesses and strengths of the methods under different types of forest were analyzed. The results showed that forest structure strongly affected the performance of all algorithms. Particularly, the success of tree detection was found to be dependent on tree density and clustering. The differences in performance between methods were more pronounced for tree detection than for height estimation. The algorithms showed a slightly better performance in the conditions for which they were developed, while some could be adapted by different parameterization according to training with local data. The results of this study may help guiding the choice of method under different forest types and may be of great value for future refinement of the single-tree detection algorithms.
The aim of this study was to test different stem volume predictors that are capable of utilizing laser scanning-based canopy height metrics as independent variables. The three laser scanning-based methods compared were (1) a direct prediction model for the stem volume at plot level, (2) a volume prediction system based on the modelled percentiles of the basal area diameter distribution, and (3) a parameter prediction method used to determinate Weibull-based basal area diameter distributions for the plot-level stem volume prediction. The predicted volumes were also compared with field-measured volumes obtained with the Finnish conventional inventory by compartments. The best results were obtained with the first method, i.e. the model that predicts plot-level stem volumes directly, which is logical. Furthermore, the simulated reduction of point density of laser data had no effect on the accuracy of stem volume predictions. The percentile-based modelling of diameter distributions was applied, in particular, to the determination of non-homogenous stand structure; using this method, it is even possible to fit multimodal distributions. In terms of the accuracy of the predicted plot-level stem volumes, the volume prediction method based on modelled percentiles of basal area diameter distributions was the second best, whereas the volume prediction method based on the parameter prediction of the Weibull-based basal area diameter distributions resulted in slightly worse results. However, the accuracies of the three laser-based volume prediction methods tested were superior to the published results of spectral value-based remote sensing studies implemented using data collected from Finland. Furthermore, the accuracy of plot-level stem volume estimates calculated from field assessments was considerably weaker than the accuracy of the three volume prediction methods that utilized measures obtained with laser scanning.
1 July 2008 in Canadian Journal of Forest Research
The use of diameter distributions originates from a need for tree-level description of forest stands, which is required, for example, in growth simulators and bucking. Diameter distribution models are usually applied, since measuring empirical diameter distributions in practical forest inventories is too laborious. This study investigated the ability of remote sensing information to predict species-specific diameter distributions. The study was carried out in Finland in a typical managed boreal forest area. The tree species considered were Scots pine ( Pinus sylvestris L.), Norway spruce ( Picea abies (L.) Karst.), and deciduous trees as a group. Growing stock was estimated using the k-MSN method using airborne laser scanning data and aerial photographs. Two approaches were compared: first, the nearest neighbour approach based on field measured trees was used as such to predict diameter distribution, and second, a theoretical diameter distribution approach in which the parameters of the Weibull distribution are predicted using the k-MSN estimates was applied. Basically, all test criteria indicated that the diameter distribution based on nearest neighbour imputed trees outperforms the Weibull distribution, but care must be taken to ensure that the modelling data are comprehensive enough.