Reducąõ do erro amostral na estimativa do volume de povoamentos de Eucalyptus ssp. por meio de escaneamento laser aerotransportado

Abstract

Forest inventory is an important activity to provide a wide range of information about the parameters from extensive forested areas based on sampling designs. This study evaluates the use of ALS (Airborne Laser Scanning) metrics as part of a double sampling design for stand volume estimation in Eucalyptus plantation, comparing the results to traditional sampling designs such as simple random sampling and stratified random sampling. In one scheme, the double sampling first phase was the height metric of the 90th percentile (P90) derived from LiDAR (Light Detection And Ranging) data. In the second double sampling scenario we adopted the 90th percentile (P90) and the density metric percentage of all returns above the mean of first returns (ARMFR), also derived from LiDAR data. Through simulations we sought to determine the sampling intensity required to obtain a sampling error at most 5%. The study was conducted in 401,6 hectares of Eucalyptus plantation located between the municipalities of Saõ Miguel Arcanjo and Pilar do Sul (Saõ Paulo), where 37 plots were measured in the field. The smallest sampling error was obtained from the double sampling with multiple regression (±1,8%) followed by the double sampling with simple regression, the stratified random sampling and the simple random sampling; confirming the use of ALS data to improve the volume estimation and enabling the sampling intensity reduction. The error of double sampling with multiple regression was ±3,4% considering only 10 field plots.