Single tree aboveground biomass models for native birch in Iceland

Abstract

In Iceland, mountain birch dominates indigenous woodlands and scrub communities. For use in inventories of the natural birch population, we derived single parameter aboveground biomass functions from a stratified random sample encompassing the entire native birch population. We evaluated the accuracy of these models on independent data from the same population and used regressions of log-transformed predicted versus observed values and compared slope and intercept parameters against the 1:1 line. We propose that the level of accuracy of allometric models might be quantified by the size of Theil’s random error component (Ue) and the normality of residual variances might be a decisive test of acceptable functions. The commonly used allometric power function without intercept proved highly accurate for diameters at ground level but was biased for diameters measured at 0.5 m up the stem. We compared both non-linear regressions and log-transformed linear regression techniques. The latter produced more accurate models especially for applications to small diameter trees. Power functions with intercept and diameters measured 0.5 m above ground produced accurate estimates, except for trees with diameters less than 50 mm. We suggest allometric models for general use in Iceland for inventories of native birch woodlands and scrub.