Ten-year periodic diameter model for uneven-aged ponderosa pine stands in the southwest reduces long-term error propagation

Abstract

Forecasting diameter growth accurately is essential for long-term forest management. The large-tree diameter growth submodel for ponderosa pine in the Southwest Ponderosa Pine model type of the Central Rockies variant of the Forest Vegetation Simulator (FVS) was evaluated using a long-term data set. Small trees were predicted least accurately, with a large proportion being overpredicted in long projections. Sensitivity analysis methods were used to assess how the model could be improved by ranking input variables in terms of their importance to model predictions. Based on these results, a new 10-year periodic diameter model was developed using regression techniques. Compared with the FVS model, the new model reduced both the short-term (10-year) prediction error by 0.18 in. and the overprediction bias and error propagation in the long-term (40-year) projection. The improvement of the new model is attributed to the addition of top height and quadratic mean diameter. These variables describe the vertical and horizontal distribution of the canopy and account for the effect of intertree competition for light and nutrients. Describing the stand structure comprehensively helped to more accurately assess the resource environment that affects tree productivity. Comprehensive measures of intertree competition were found to be more important in long-term projections.