Predicting site index with a physiologically based growth model across Oregon, USA

Abstract

With expanded interests in sustaining productivity under changing climate, management, and disturbance regimes, we sought a means of mapping the potential productivity of forests across the state of Oregon in the Pacific Northwest, USA. We chose the mapping tool 3-PG, a simplified physiologically based process model that can be driven with monthly averaged climatic data (DAYMET) and estimates of soil fertility based on soil nitrogen content. Maximum periodic mean increment (MAI, m3·ha–1·year–1), a measure of the forest's productive potential, was generated by the 3-PG spatial model and mapped at 1-km2 resolution for the most widely distributed tree species, Douglas-fir (Pseudotsuga menziesii (Mirb.) Franco). Maximum MAI is linearly correlated with yield table site indices and therefore comparable with field-derived estimates of site indices obtained from measurement of tree heights and ages at 5263 federal forest survey points. The model predicted 100-year site index (SI) reasonably well (R2 = 0.55; RMSE = 9.1), considering the difference in spatial resolution between the modeled (1 km2) and field-measured SI (<0.1 ha) and that field plots were offset for confidentiality by 1–3 km. We created a map of the differences between modeled and field-measured SI and found that the 3000 points within ±6 m error were relatively evenly distributed across Oregon. Improving the accuracy in modeling and mapping forest productivity using 3-PG will likely require refinements in soil surveys, the quality of climatic data, the location of field plots, and the model functions and species parameters.