Comparison of selected Nordic stand growth models for Norway spruce, Scots pine and birch

Abstract

Thirteen Nordic stand growth models have been validated by use of a test data set from long-term research plots in Norway. The evaluated data was from time-series of even-aged, pure stands of Norway spruce, Scots pine and birch (silver birch and downy birch). In selected models from Finland, Norway and Sweden measures of site productivity, mean tree size and various stand characteristics are represented. Different models display both strengths and weaknesses in their predicting ability. Several measures of precision and bias have been calculated and the models are ranked due to their performance. We observed site quality, stand density and average tree size as the three main components in the models. Basal area increment model for spruce from Sweden had the lowest standard deviation with 23%. The mean R2 between residuals and stand characteristics from this model was also low (1.3%), which indicates that independent variables are well included. For Scots pine and birch, Finnish volume increment models showed the best fit to the Norwegian test data, with a R2 between residuals and stand characteristics of 2.8 and 6.7%, respectively. Several of the models from Sweden and Finland predicted the growth as well as stand models frequently in use in Norway. The results indicated that similar forest conditions and traditional even-aged forest management practice in the Nordic countries could be seen as a suitable basis for developing a joint family of growth models. By careful recalibration of existing models, a reasonable accuracy could be achieved and the prediction bias could be reduced.