Designing Wood Supply Scenarios from Forest Inventories with Stratified Predictions

Abstract

Forest growth and wood supply projections are increasingly used to estimate the future availability of woody biomass and the correlated effects on forests and climate. This research parameterizes an inventory-based business-as-usual wood supply scenario, with a focus on southwest Germany and the period 2002-2012 with a stratified prediction. First, the Classification and Regression Trees algorithm groups the inventory plots into strata with corresponding harvest probabilities. Second, Random Forest algorithms generate individual harvest probabilities for the plots of each stratum. Third, the plots with the highest individual probabilities are selected as harvested until the harvest probability of the stratum is fulfilled. Fourth, the harvested volume of these plots is predicted with a linear regression model trained on harvested plots only. To illustrate the pros and cons of this method, it is compared to a direct harvested volume prediction with linear regression, and a combination of logistic regression and linear regression. Direct harvested volume regression predicts comparable volume figures, but generates these volumes in a way that differs from business-as-usual. The logistic model achieves higher overall classification accuracies, but results in underestimations or overestimations of harvest shares for several subsets of the data. The stratified prediction method balances this shortcoming, and can be of general use for forest growth and timber supply projections from large-scale forest inventories.

Generating Tree-Level Harvest Predictions from Forest Inventories with Random Forests

“…Baden-Württemberg covers an area of around 35,750 km 2 , of which approximately 38.4% is considered to be forest [28,29]. A more detailed description of the case can be found in Kilham et al [4] and MLR (Ministry of Rural Affairs and Consumer Protection Baden-Württemberg) [30]. Models were trained on the period 2002-2012, which corresponds to the most recent inter-NFI period.…”

“…National forest inventories (NFIs) are conducted worldwide-some using angle-count sample methods [1,2]. Based on these inventories, forest growth and wood supply modeling is used to highlight consequences of potential management decisions and to show future prospects as well as limitations of the status quo [3][4][5].…”

“…Instead, NFI data becomes representative for higher aggregations of inventory plots. For this reason, the fact that no harvests occurred on the plot cannot be taken as evidence that no harvests occurred in the stand [4]. In the following, an NFI plot is considered as harvested if a minimum of one tree is removed from the plot in the period of interest.…”

“…One consequence is that the extracted timber is harvested from a lower number of plots treated with higher intensities than average figures would suggest. Neglecting this phenomenon would result in inventory predictions clearly different from BAU [4].…”

“…In a deterministic approach, a clear yes-no decision is applied to each plot for every predicted period. In this context, Kilham et al [4] demonstrated that the replication of NFI harvest patterns on plot level can be improved with a stratified procedure based on a classification and regression tree (CART) introduced by Breiman et al [12] and random forest (RF) algorithms [13]. Antón-Fernández and Astrup [6] used harvest probabilities to select plots randomly in numerous repetitions.…”