Simulation models of ecosystem processes may be necessary to separate the long-term effects of climate change on forest productivity from the effects of year-to-year variations in climate. The objective of this study was to compare simulated annual stem growth with measured annual stem growth from 1930 to 1982 for a uniform stand of ponderosa pine (Pinus ponderosa Dougl.) in Montana, USA. The model, FOREST-BGC, was used to simulate growth assuming leaf area index (LAI) was either constant or increasing. The measured stem annual growth increased exponentially over time; the differences between the simulated and measured stem carbon accumulations were not large. Growth trends were removed from both the measured and simulated annual increments of stem carbon to enhance the year-to-year variations in growth resulting from climate. The detrended increments from the increasing LAI simulation fit the detrended increments of the stand data over time with an R(2) of 0.47; the R(2) increased to 0.65 when the previous year's simulated detrended increment was included with the current year's simulated increment to account for autocorrelation. Stepwise multiple linear regression of the detrended increments of the stand data versus monthly meteorological variables had an R(2) of 0.37, and the R(2) increased to 0.47 when the previous year's meteorological data were included to account for autocorrelation. Thus, FOREST-BGC was more sensitive to the effects of year-to-year climate variation on annual stem growth than were multiple linear regression models.
The study appraised nine models that predict total tree height from diameter by species within individual stands. Models were fitted with nonlinear least squares by species within individual stands using inventory data from western Washington. Stand-level models were examined with respect to species, geographic regions, dominance characteristics, and sample sizes. Models were evaluated for mean square error, bias by diameter class, overfitting, and consistency in relative ranking. No substantial differences in model performance were noted with respect to geographic regions, but small differences were evident by species, dominance characteristics, and sample sizes. Model bias occurred with some but not all models. Overfitting was detected and considered a problem in fitting three-parameter models with the often small height sample in some stands. Some models were consistently good across species and sample sizes, whereas others were consistently poor. Yet the performance of other models varied by species and sample sizes. For predicting heights by species within individual stands, a single model was recommended Height = 1.37 + b0eb1dbh-1.0 A method was examined for constraining height predictions for trees beyond the range of sample data. West. J. Appl. For. 13(4):109-119.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.