ALBERT

Description: This study considered airborne laser scanning (ALS) based aboveground biomass (AGB) prediction in mountain forests. The study area consisted of a long transect from southern Norway to northern parts of the country with wide ranges of elevation along a long latitudinal gradient (58°N–69°N). This transect was covered by ALS data and field data from 238 plots. AGB was modeled using different types of predictor variables, namely ALS metrics, variables related to growing conditions (elevation, latitude, and climatic variables), and tree species information. Modelling of AGB in the long transect covering diverse mountainous forest conditions was challenging: the RMSE values were rather large (37%–70%). The effects of growing conditions on model predictions were minor. However, species information was essential to improve accuracy. The analysis revealed that when doing inventories of spruce-dominated areas, all plots should be pooled together when the models are developed, whereas if pine or deciduous species dominate the area in question, separate dominant species-wise models should be constructed.

Description: Inferences for forest-related spatial problems can be enhanced using remote sensing-based maps constructed with nearest neighbours techniques. The non-parametric k -nearest neighbours ( k -NN) technique calculates predictions as linear combinations of observations for sample units that are nearest in a space of auxiliary variables to population units for which predictions are desired. Implementations of k -NN require four choices: a distance or similarity metric, the specific auxiliary variables to be used with the metric, the number of nearest neighbours, and a scheme for weighting the nearest neighbours. The study objective was to compare optimized k -NN configurations with respect to confidence intervals for airborne laser scanning-assisted estimates of mean volume or biomass per unit area for study areas in Norway, Italy, and the USA. Novel features of the study include a new neighbour weighting scheme, a statistically rigorous method for selecting feature variables, simultaneous optimization with respect to all four k -NN implementation choices and comparisons based on confidence intervals for population means. The primary conclusions were that optimization greatly increased the precision of estimates and that the results of optimization were similar for the k -NN configurations considered. Together, these two conclusions suggest that optimization itself is more important than the particular k -NN configuration that is optimized.

Description: Estimation of wood volume and biomass is an important assignment of any National Forest Inventory. However, the estimation process is often expensive, laborious and sometimes imprecise because of small sample sizes relative to population variability. Remote sensing techniques are an option to assist in surveying large areas by providing data that can be related to the forest attribute of interest through mathematical models of relationships. Light Detection and Ranging (LiDAR) is a technology that can provide data that are closely related to forest wood volume and biomass. With these data, linear regression is often used to estimate forest attributes. If the relationship provides evidence of nonlinearity, a transformation in the variables can be considered. However, modern computation allows fitting nonlinear regression models without transformations of the variables. Nonlinear least squares (NLS) techniques also give more freedom to assure satisfaction of natural conditions such as non-negativity and/or lower and upper asymptotes. Like any estimation technique, NLS is subject to overfitting when using a large number of predictor variables. Because NLS is more computationally intensive than linear regression, stepwise selection techniques may require considerable programming effort. We compared three methods to select predictor variables for nonlinear models of relationships between forest attributes and LiDAR metrics, two of them based on genetic algorithms (GAs) and one based on random forest (RM). GAs were implemented to optimize a cost function that yields root mean square error or the Akaike Information Criterion (AIC), while RM was based on variable importance in decision trees. A model with the predictor variable most correlated with the response variable was also considered. We compared the results of overall estimation for two datasets using the model-assisted, generalized regression estimator and concluded that the combination of GAs and AIC was the most efficient and stable procedure for selection of variables. We attribute this result to the penalty that AIC applies to models with large numbers of variables, which leads to a more efficient model with a minimum loss of information.

Description: A single a priori chosen linear regression model with two alternative error structures is proposed for model-assisted (MA) and model-dependent (MD) estimation of state and change in aboveground tree biomass (AGB, Mg·ha−1) in three forest strata in the Våler forest in southeastern Norway. Field data of tree height and stem diameter were collected in 145 permanent 200 m2circular plots. Concurrent LiDAR data were collected for the entire forest. The regression model includes two LiDAR-based explanatory variables: the mean of canopy height raised to a power of 1.5 and the standard deviation of canopy heights. A nearest-neighbour thinning of the 2010 LiDAR data to the density of the 1999 data was implemented to counter density effects in the explanatory variables. Estimates of change based on a single regression model were more accurate than estimating change from year-specific models (and no data thinning). A canopy height dependent correlated error structure was preferred over a partitioning of the error to temporary and “permanent” plot effects. For point estimates of AGB in 1999 and 2010, MA and MD estimates of errors were numerically comparable, but MD errors of change were much smaller than corresponding MA errors.

Description: A profiling airborne LiDAR is used to estimate the forest resources of Hedmark County, Norway, a 27390 square kilometer area in southeastern Norway on the Swedish border. One hundred five profiling flight lines totaling 9166 km were flown over the entire county; east-west. The lines, spaced 3 km apart north-south, duplicate the systematic pattern of the Norwegian Forest Inventory (NFI) ground plot arrangement, enabling the profiler to transit 1290 circular, 250 square meter fixed-area NFI ground plots while collecting the systematic LiDAR sample. Seven hundred sixty-three plots of the 1290 plots were overflown within 17.8 m of plot center. Laser measurements of canopy height and crown density are extracted along fixed-length, 17.8 m segments closest to the center of the ground plot and related to basal area, timber volume and above- and belowground dry biomass. Linear, nonstratified equations that estimate ground-measured total aboveground dry biomass report an R(sup 2) = 0.63, with an regression RMSE = 35.2 t/ha. Nonstratified model results for the other biomass components, volume, and basal area are similar, with R(sup 2) values for all models ranging from 0.58 (belowground biomass, RMSE = 8.6 t/ha) to 0.63. Consistently, the most useful single profiling LiDAR variable is quadratic mean canopy height, h (sup bar)(sub qa). Two-variable models typically include h (sup bar)(sub qa) or mean canopy height, h(sup bar)(sub a), with a canopy density or a canopy height standard deviation measure. Stratification by productivity class did not improve the nonstratified models, nor did stratification by pine/spruce/hardwood. County-wide profiling LiDAR estimates are reported, by land cover type, and compared to NFI estimates.