Equivalence of allometric models and pantropical flexible allometry

Abstract

Estimating total aerial biomass, and carbon, in tropical forests is a critical task for forest management and carbon markets, for which various allometric models have been developed for this purpose. Pantropical, multi-species and multi-site models, try to develop estimates for the general case of lack of knowledge of specific local models. The application of allometric models is based on the selection of the "best" model using some metric of the estimation error. To put into perspective the equivalence of the main models used in the literature, under a simple algebraic approach, this paper presents the equivalences between the analyzed models, which are parameterized by simple linear regression in the log-log space, using a pantropical database (tropical forests) with 4004 measurements at 58 sites in the world, which includes measurements of diameter at breast height (D) and total height (H), as well as measurements or estimates of wood density (ρ). The results support the equivalence of the allometric models. Additionally, a parameterization of the allometric models was carried out utilizing the pantropical database using simple linear regression of the logarithmic transformations of the models. The results show that the model that uses D, H and ρ at the local level (density classes or sites) results in the estimation with the lowest error. The pantropical allometric model estimated from the database has been proposed as a global model, although with a discussion about its applicability in relation to local models. An approach to the global model is introduced by parameterizing a model as a function of the density of the wood and two approaches in relation to the exponent of the analyzed potential relationship. The results show that the pantropical models developed show slightly better estimates than the pantropical model in use, giving flexibility to the estimation error when associating it with the wood density of the tree species.