• Rural Digital Europe
• 2012-2021close
• NO close
• Canadian Journal of Forest Research close

Model-based inference is an alternative to probability-based inference for small areas or remote areas for which probability sampling is difficult. Model-based mean square error estimators incorporate three components: prediction covariance, residual variance, and residual covariance. The latter two components are often considered negligible, particularly for large areas, but no thresholds that justify ignoring them have been reported. The objectives of the study were threefold: (i) to compare analytical and bootstrap estimators of model parameter covariances as the primary factors affecting prediction covariance; (ii) to estimate the contribution of residual variance to overall variance; and (iii) to estimate thresholds for residual spatial correlation that justify ignoring this component. Five datasets were used, three from Europe, one from Africa, and one from North America. The dependent variable was either forest volume or biomass and the independent variables were either Landsat satellite image bands or airborne laser scanning metrics. Three conclusions were noteworthy: (i) analytical estimators of the model parameter covariances tended to be biased; (ii) the effects of residual variance were mostly negligible; and (iii) the effects of spatial correlation on residual covariance vary by multiple factors but decrease with increasing study area size. For study areas greater than 75 km2 in size, residual covariance could generally be ignored.

Managing boreal forests for effective climate change mitigation requires comprehension of the full spectrum of climate regulation services that they provide, which includes both the storage of carbon and exchanges of heat and moisture with the atmosphere. It is increasingly recognized that surface albedo is the most important biogeophysical mechanism by which the boreal forest directly influences the global energy balance. Forest management decisions that influence age class and species distributions affect not only the carbon sink capacity, but also the albedo (and hence climate services) of the forested landscape. Disregarding albedo and how it is influenced by management decisions can have profound implications for the effectiveness of any climate change mitigation policy involving active forest management. Here, we explore, analyze, and compare the albedo predicted by simple empirical models with in situ and remotely sensed albedo observations in regions outside the region in which the models were originally developed (southeastern Norway), including boreal Canada and Europe. We find that the models are robust in their ability to predict the longer term interannual trends in the mean winter–summer albedo amplitude, the rapid albedo evolution in young stands, and the timing of seasonal transitions and weak with respect to capturing interannual albedo changes linked to seasonal climate variability and phenology.

This study proposes a method to perform spatially consistent imputations of forest data to serve simulation studies where spatial autocorrelation is expected to have an effect (e.g., sampling simulations and forest scenario analysis). Starting with a nearest neighbour imputation, an optimization process brings the spatially comprehensive data to a desired state, controlled by a target semivariogram and a target histogram. The target values for both parameters may be approximated using empirical data and are combined in the objective function used by the optimization algorithm. Here, we demonstrate a case study using wall-to-wall airborne laser scanner data, satellite data, and field observations for an 852 ha forest area in southern Norway. Different combinations of data types and target parameters were tested, and the target values were reached in most cases. In some cases, with a more restrictive objective function, the semivariogram did not completely converge to its target values, yet still had a convergence of at least 93%, expressed by the difference reduction between initial and target values. The results recommend the proposed method as a practical means to generate spatially explicit forest data when a particular distribution and well-defined spatial dependence are required.

By using more sophisticated sampling designs in forest field inventories, it is possible to select more representative field samples. When full cover auxiliary information is available at the planning stage of a forest inventory, an efficient strategy for sampling is formed by making sure that the sample is well spread in the space spanned by the auxiliary variables. We show that by using such a sampling design, we can improve not only design-based estimation, but also estimation based on nearest neighbour techniques. A new technique to select well-spread probability samples, in multidimensional spaces, from larger populations is introduced. As an application, we illustrate how this strategy can be applied to a forest field inventory. We use an artificial dataset corresponding to a full cover forest remote sensing inventory of a 30 000 ha area of Kuortane, western Finland. The target variable (growing stock volume) has been generated for the entire area by a copula technique. The artificial population has been validated by utilizing the Finnish National Forest Inventory.

Forests are important contributors to the global carbon cycle and mitigate climate change through carbon sequestration and the supply of wood that substitutes for fossil fuels and greenhouse gas (GHG)-intensive building materials. However, current climate policies only partially credit forest carbon sequestration and bioenergy policies are handled independently of forestry. Using Norway as a case study, we analyze two sets of simulated carbon tax/subsidy policies, one crediting forest carbon sequestration while maintaining predetermined harvest levels and utilization of wood, and another targeting GHG fluxes in the entire forest industrial sector allowing harvest levels and wood markets to change in response to the policy. Results indicate that GHG emission reduction potentials differ substantially between the two policies, being several times higher for the latter than the former policy at a given carbon price. This suggests that (i) previous research efforts in Europe have not captured the full mitigation potential as they have not included adaptations in the harvest level and the wood market and (ii) climate policies should target GHG fluxes in the entire sector to utilize its potential contribution for mitigating climate change.

Deadwood can represent a substantial portion of forest ecosystem carbon stocks and is often reported following good practice guidance associated with national greenhouse gas inventories. In high-latitude forest ecosystems, a substantial proportion of downed deadwood is overgrown by ground vegetation and buried in the humus layer. Such burial obfuscates the important process of deadwood carbon transfer to other pools (e.g., litter and soil) and emission to the atmosphere (i.e., rates of decay). Using data from the Swedish National Forest Inventory, we found that the proportion of downed logs that is buried increased from temperate to boreal forests. Several factors affect the probability of burial, including log attributes (e.g., decay class), ground vegetation (e.g., moss dominance, type of moss cover), and edaphic conditions (e.g., soil type, depth of organic layer). Combined assessments suggest that about 24% of the carbon in the aboveground downed deadwood pool was found to be buried in boreal forests. Deadwood burial has important implications for forest carbon dynamics and associated monitoring (e.g., United Nations Framework Convention on Climate Change reporting) as such a pool typically decomposes much slower compared with aboveground deadwood.

Modern cut-to-length harvesters are recording information about each harvested tree, and with accurate positioning, this information can be used as field reference data, replacing manually measured reference data. In the present study, models developed from accurately positioned harvester data were compared with a reference model. A set of ∼55 000 accurately positioned trees was used as the basis for a division into 792 reference plots of 400 m2 each. A set of manually measured field plots was used for validation. Regression models were developed based on the relationship between airborne laser scanning data and the reference plot volumes. Separate models were developed for two strata: medium and high site productivity. Several modelling methods were compared, including nonparametric models; at the plot level, predictions for the validation dataset yielded RMSEs of 32%–60% for the medium productivity stratum and 19%–22% for the high productivity stratum. A reference model was fitted to the manually measured validation data in each stratum, and RMSEs of 45% and 25% were obtained for the medium and high productivity strata, respectively. The results show that the models based on the harvester data yield prediction errors at the same level as the reference model.

For a study area in the Brazilian state of Santa Catarina, the utilities of local and global forest maps in combination with poststratified and model-assisted estimators for increasing the precision of estimates of forest area were compared. Auxiliary information was in the form of local maps, the recent Global Forest Change map, and combinations of these maps. The poststratified estimators produced estimates of greater precision than the model-assisted regression estimators for maps of categorical variables, but the model assisted estimators produced estimates of greater precision for maps of continuous variables. The Global Forest Change map was the least accurate of all the maps, but it produced estimates of forest area that were similar to those for the other maps and that were more precise than if the map had not been used. Thus, the Global Forest Change map may be an attractive option if local maps are not available or cannot be constructed. The primary contributions of the study are two-fold. First, this is one of the first case studies that rigorously assess the utility of global maps for national estimation. After accumulation of a few more such studies, broader generalizations should be forthcoming. Second, a statistical basis is provided for the previously unexplained greater precision for poststratified estimators than for model-assisted estimators.

Survey sampling with model-assisted estimation has been gaining popularity in forest inventory recently, as the availability of cheap, good-quality remotely sensed data that can be used as auxiliary information has improved. Most of the studies have been carried out using parametric (linear or nonlinear) models. However, nonparametric and semiparametric models such as k nearest neighbor, kernel, and generalized additive are widely used in forest inventory. The results are usually calculated using the difference estimator (i.e., assuming an external model), even though the models used are based on the sample (i.e., an internal model). In that case, variances will likely be underestimated. In this study, we analyze how well the difference estimator works for different types of models, both internal and external. The study is based on simulated populations produced using a C-vine copula model with empirical marginals. The external model is based on real data, and the internal models are estimated from samples from the simulated population. The results show that the analytical variance estimates for a difference estimator based on an overfitted kernel model can seriously underestimate the true variance.