COMPARISON BETWEEN RANDOM FOREST AND LINEAR REGRESSION FOR TROPICAL FOREST ABOVEGROUND BIOMASS ESTIMATION

Vol 19, 2019 - 96128
Pôster
Download
Favoritar este trabalho
Como citar esse trabalho?
Resumo

The objective was to compare two methods for estimating aboveground biomass (AGB) in tropical rainforest using airborne LiDAR data. The study was conducted at Fazenda Cauxi in northern Brazil. Data from LiDAR and field inventory collected in 2014 were used. A total of 85 plots were used for the modeling. In the R environment, Random Forest (RF) and Linear Regression (lm) were compared in terms of RMSE, Bias and adj.R² through a LOOCV process with 500 replicates. The best performance was verified for the LM algorithm.

Instituições
  • 1 Universidade Federal do Paraná
  • 2 Universidade Federal do Paraná - UFPR
  • 3 Universidade Federal de São João Del Rei
  • 4 USDA Forest Service
  • 5 Universidade Federal do Rio Grande do Sul
  • 6 Universidade Federal Rural de Pernambuco
  • 7 University of Maryland\NASA Goddard Space Flight Center
Eixo Temático
  • LIDAR: sensores e aplicações
Palavras-chave
LiDAR
Machine Learning
Remote Sensing