Comparison of a physiological model and a statistical model for prediction of growth and yield in boreal forests

The structural and functional properties of a physiological model (FinnFor) and a statistical model (Motti), developed independently, were analysed in order to assess whether the former would provide the same prediction capacity as the latter, which is based on a huge body of long-term inventory data. The predictions were compared in terms of (i) stand-level variables, (ii) analysis of volume growth graphs, and (iii) stand structure variables (diameter and height distributions). Both unmanaged and managed (thinned) stands of Scots pine (Pinus sylvestris), Norway spruce (Picea abies) and silver birch (Betula pendula) growing on medium-fertility sites in central Finland were used for the comparison. In general, the outputs of the models agreed well in terms of relative growth rates regardless of tree species, with the implication that both predict competition within a stand and the effect of position on tree growth in a similar way. The statistical model was stable in its predictions, but not as sensitive to initial stand conditions and management as that based on physiological processes, but the two models agreed well in their dynamics and predictions. The process-based model may therefore be applied to practical management situations, in order to achieve more precise predictions under changing environmental conditions, as in the case of climate warming. On the other hand, some elements of process-model thinking could be incorporated into statistical models in order to make these responsive to changing conditions.