Improving the forecasts of commercial timber volume in transition forest in the northern Brazilian Amazon

Abstract

The commercial volume in tropical forests is fundamental to the implementation of sustainable forest management plans. However, in the absence of local or generic volumetric equations, most forest managers and forest services are still using traditional form factors (FF) to estimate the volume of commercial tree species in the Amazon. Our study shows the modelling to estimate the commercial volume of wood in a transition forest in the Brazilian Amazon. Specifically, volumetric models were tested using data from 181 fallen trees and measured in three plots (primary units) in Amapá, Brazil. We tested generic volumetric models and compared them with a specific model (i.e., in all plots), and with FF. Models that include shaft diameter and height have returned the most accurate volume estimates. The Schumacher-Hall generic equation allowed accurate predictions (root mean square error = 0.47 m³ and bias = 0.04). The minimum sampling design for developing volumetric models consists of measuring ~70% of trees considering a random sampling of the entire population. The locally developed FF generated substantial bias, but presented better performance for small trees. These results can be used by forest managers as a technical tool in predicting the commercial volume of natural forests in Amapá.

Keywords: form factor, rainforest, regression analysis, relative error, volume models