Comparing height-age and height-diameter modelling approaches for estimating site productivity of natural uneven-aged forests

Abstract

Accurate estimates of forest site productivity are an essential part of forest management. Only a few approaches exist for estimating site productivity of natural uneven-aged forests. This study compared two approaches: dominant height (DH)-dominant diameter (DD) modelling and DH-stand age (A) modelling in terms of their prediction accuracy for site productivity. We developed the models based on both the algebraic difference approach (ADA) and ADA with mixed-effects modelling. Data originated from four sets of continuous measurements on natural Mongolian Oak (Quercus mongolica Fisch.) and Korean Larch (Larix olgensis Henry.) in northeastern China. The leave-one-out cross-validation was applied to evaluate the models. The results showed that the prediction accuracies of the DH-DD models were significantly higher than that of the DH-A models. The inclusion of block-level random effects significantly increased the accuracies of both the DH-A models and DH-DD models. Compared with the DH-A models, the DH-DD models did not require age measurements which is time-consuming and difficult, but the DD and DH measurements are available from the routine inventories and compatible with the existing forest inventory data. Therefore, the nonlinear mixed-effects DH-DD models are recommended to estimate site productivity of the natural forests.