Relative importance of predictor variables in the forest productivity modeling

Authors

  • Daniel Henrique Breda Binoti danielhbbinoti@gmail.com
    Eldorado Brasil
  • Helio Garcia Leite hgleite@gmail.com
    Universidade Federal de Viçosa
  • Valdir Carlos Lima de Andrade vclandrade@mail.uft.edu.br
    Universidade Federal de Tocantins
  • Marcio da Conceição marcio.conceicao@thetimbergroup.com
    The Timber Group
  • Nairam Felix de Barros Filho nairam.filho@thetimbergroup.com
    The Timber Group
  • Leonardo Machado Pires leonardo.pires@thetimbergroup.com
    The Timber Group
  • Luiz Otávio Rodrigues Pinto luiz.rodrigues@eldoradobrasil.com.br
    Eldorado Brasil
  • Thuliany Fernandes Araujo Paes thuliany.paes@eldoradobrasil.com.br
    Eldorado Brasil

DOI:

10.34062/afs.v9i4.13519

Abstract

Modeling forest growth and production is a major challenge for forest managers due to the large number of variables involved and the importance of the generated estimates for decision making in the forestry enterprise. Several statistical and artificial intelligence methods can be used to verify the importance of variables and their selection for the forest modeling process. This study demonstrates the use of the perturbation method in defining the relative importance of predictor variables (silvicultural, climatic and management) in predicting the productivity of eucalyptus stands at the end of the rotation. Data from 320 eucalyptus plantations located in the north of the State of Minas Gerais, aged over seven years, were used. Precipitation distributed at different ages and soil clay content were the most important variables for the prediction of volume at cutting age.

Published

2022-12-27