Title: Testing the Application of Process-Based Forest Growth Model Prebas to Uneven-Aged Forests in Finland
Abstract: The challenges of applying process-based models to uneven-aged forests are the difficulties in simulating the interactions between trees and resource allocation between size classes. In this study, we focused on a process-based forest growth model PREBAS which is a mean tree model with Reineke self-thinning mortality and was originally developed for even-aged forests. The primary aim was to test the application of PREBAS model to uneven-aged forests by introducing different diameter at breast height (DBH) size classes to better represent the forest structure. Additionally, we introduced a new mortality model MOR new to PREBAS which is developed for uneven-aged stands and compared with the current PREBAS version in which a modification Reineke rule is used. The tests were conducted in 26 old Norway spruce dominated stands in southern and central Finland with three consecutive measurements (on average a 25-year study period). To evaluate the model performance, we calculated average model bias (AMB), model efficiency (EF) and normalised mean square error (MSE) by comparing the estimations of stand averaged diameter at breast height ( D ), stand averaged tree height ( H ), stand averaged crown base height ( H c ), stand basal area ( B ) and density ( N ) with measurements. Moreover, biomass estimations of each tree component (foliage, branch and stem) were compared to estimations from empirical models. Results showed that EF values of D , H , H c , N and B were over 90% (except for N in Norway spruce and volume, V, in Silver birch) and they were higher than corresponding EF values without DBH size classes. As for biomasses, model estimations considering the DBH size classes showed better agreement with the estimations from empirical models in all the species by showing higher EF. In addition, MOR new was better for predicting mortality of larger DBH classes while the Reineke model was superior for smaller DBH classes. To conclude, introducing size classes generally improved the model predictions and MOR new gave qualitatively more realistic results especially among the largest tree sizes. However, the results also revealed that further calibration of the PREBAS model with size classes should be done to better extend the model applicability to uneven-aged forests.