•  
  •  
 

Abstract

Foresters and landowners have a growing interest in carbon sequestration and cellulosic biofuels in southern pine forests, and hence need to be able to accurately predict them. To this end, we derived a set of aboveground biomass models using data from 62 small-diameter loblolly pines (Pinus taeda) sampled on the Crossett Experimental Forest in southeastern Arkansas. Of the 25 equations initially evaluated, we chose 17 that best fit our dataset and compared them using a suite of conventional test statistics, including pseudo-R2 , root mean squared error (RMSE), and bias. Because most of the 17 models varied little in pseudoR 2 (ranging between 0.96 and 0.99), bias (all were within ± 0.01), and RMSE, an additional comparison was done using Akaike’s Information Criterion corrected for small sample size (AICc). This test statistic produced considerably more discrimination between the biomass models. Of the 17 models evaluated, six produced ΔAICc scores that met or exceeded the threshold for substantial support. To recommend a single preferred model, we then extrapolated beyond our actual data and qualitatively compared model predictions with those from the National Biomass Estimator. Our “best” model did not have the minimum AICc score, but rather predicted logically consistent aboveground biomass values at both the upper and lower ends of our extrapolation.

Download

Share

COinS