Estimation height level of Copaifera sp. (Leguminosae) by Artificial Neural Networks

Abstract

Abstract The knowledge of tree attributes of the genus Copaifera sp. (copaiba), such as the height of the trunks, helps to estimate the productive potential of oleoresin and to propose more suitable ways of handling, aiming at optimizing production. This research aimed to test hypsometric equations and deterministic methods of Artificial Neural Networks (ANN) to estimate the total heights levels of the trunks of 31 copaiba trees of the Western Brazilian Amazon, at unknown ages. However, the ANN correlation coefficients obtained were greater than 0,99, demonstrating that they are appropriate for the estimation of height level (h100%). Among the ANN architectures, ANN 3 with 2 neurons in the hidden layer stood out. The application of ANN to estimate the total height of the trunk of Copaifera sp. native trees is a viable tool that can contribute to optimize modeling of the different important aspects to determine the productive potential of oleoresin.

Keywords

• Artificial intelligence
• Forest inventory
• Forest measurement
• Forest technology
• Hypsometry
• Non-timber forest products