Prediction and model-assisted estimation of diameter distributions using Norwegian national forest inventory and airborne laser scanning data

Abstract

Diameter at breast height (DBH) distributions offer valuable information for operational and strategic forest management decisions. We predicted DBH distributions using Norwegian national forest inventory and airborne laser scanning data in an 8.7 Mha study area and compared the predictive performance of parameter prediction methods using linear-mixed effects (PPM) and generalized linear-mixed models (GLM), and a k nearest neighbor (NN) approach. With PPM and GLM, it was assumed that the data follow a truncated Weibull distribution. While GLM resulted in slightly smaller errors than PPM, both were clearly outperformed by NN. We applied NN to study the variance of model-assisted (MA) estimates of the DBH distribution in the whole study area. The MA estimator yielded greater than or almost equal efficiencies as the direct estimator in the 2 cm DBH classes (6, 8,..., 50 cm) where relative efficiencies (REs) varied in the range of 0.97$-$1.63. RE was largest in the DBH classes $\leq$ 10 cm and decreased towards the right tail of the distribution. A forest mask and tree species map introduced further uncertainty beyond the DBH distribution model, which reduced REs to 0.97$-$1.50.