Tree Height-Diameter Allometry and Implications for Biomass Estimates in Northeastern Amazonian Forests

Tree height-diameter allometry varies among forest types and uncertainties in high estimates may be propagated to above ground biomass in tropical forests. Here we measured diameter and height of 1,964 trees in two representative forest types in the Amazon (non-flooded terra-firme and seasonally-flooded várzea forests) in order to provide local allometric models and compare their performances with an early published pantropical height model. We also determined which estimates of height (or the absence of height) resulted in best estimates of biomass at different spatial scales. Asymptotic height-diameter models (rectangular-hyperbola and Weibull) showed the best fit for both terra-firme and várzea forests at local and regional scales. Local models performed better than the pantropical for each forest physiognomy separately. The height estimated with asymptotic local models increased consistently the precision of biomass estimates for both forest types (terra-firme: < 1% and várzea: 4.6% [differences between biomass estimates with field heights and allometric local models]). However, biomass estimates using heights from pantropical model or the absence of height was underestimated for terra-firme (16%) and overestimated for várzea forests (23%). Using pooled data of both forests the height from the pantropical model provided better biomass estimates (only 5% of difference) than local height models. Uncertainties in biomass estimates were mainly due to error of large trees (> 90 cm of diameter) which vary widely in height. We conclude that modeling of tree height is necessary for distinct forest types and more effort must be needed to sample large trees in order to improve biomass estimates in northeastern Amazonia