On simplifying allometric analyses of forest biomass

Tree biomass plays a key role in sustainable management and in estimating forest carbon stocks. The most common mathematical model in biomass studies takes the form of the power function M = aD(b) where a and b are the allometric coefficients to be determined by empirical data, and M the total aboveground tree dry biomass for a specific diameter at breast height, D. In this study the development and comparison of three methods for simplifying allometric equations of aboveground biomass estimation are reported. Based on the criterion of the relative difference (RD) between observed and predicted biomass data, the small trees sampling scheme (SSS) predicted quite accurate estimates for raw data reported in 10 studies. The SSS equation was based on the hypothesis that information provided in published allometric equations, in conjunction with two pairs of empirical M-D values, are enough to obtain reliable predictions for aboveground stand biomass. In addition, predictions of M based on theoretical values of b were also tested with the RD criterion, but reliability of predictions in 10 studies is questioned. Finally, fractal geometry was used to develop a 'reductionist' model for M estimation and implications from its implementation in biomass studies are discussed. We totally based our investigation on a metadata set derived from published aboveground biomass allometric studies conducted for different species spanning the globe. (C) 2003 Elsevier B.V. All rights reserved.