Geotecnologias aplicadas ao inventário de florestas plantadas

Abstract

The Continuous Forest Inventory (CFI) consists of a network of permanent plots established between two and three years in Eucalyptus plantations and monitored annually until the cutting age. However, the information obtained from the processing of these plots are not enough to generate accurate estimates of the volume per stand at the cutting age. Thus, there is a need to perform a pre-harvest inventory. In the pre-harvest inventory, new plots are located aggregating a new component in the cost of the forest production. An alternative without adding costs to obtain greater accuracy from the CFI would be the use of remote sensing imagery combined with geostatistical interpolation in the stratification of the forest plantations. In this sense, this dissertation was conducted with the following objectives: (i) to identify spectral bands and vegetation indices derived from Landsat 5 TM images that best correlated with the dendrometric variables basal area, dominant height and volume for all measured years in the CFI as well as in the age of the pre-harvest inventory; (ii) to select the spectral variable that best correlated with the dendrometric variable basal area in the cutting age and evaluate the spatial structure dependence of these variables; (iii) to compare the estimates of Systematic Sampling (SS) to the estimates of Random Stratified Sampling (RSS) for the variable of interest volume (m³/ha). The stratification was performed by ordinary kriging and cokriging in different sampling intensities at the cutting age. The results showed that TM4 and TM5 spectral bands were better correlated with basal area and volume, while TM2 spectral band was better correlated with dominant height. Vegetation indices that include in their composition the TM4 and TM5 spectral bands, as the ND54 and TM5/4, were better correlated with the dendrometric variables. The TM5 spectral band was selected as auxiliary variable in the ordinary cokriging procedure. The geostatistical interpolation ordinary kriging and ordinary cokriging showed to be efficient in the definition of strata in Eucalyptus spp. plantation. The ordinary kriging and ordinary cokriging showed to be efficient in defining the strata because the estimators of RSS were more accurate than the estimators of SS. The basal area and the reflectance in TM5 band can be used individually (ordinary kriging) or together (ordinary cokriging) as stratification tools by forestry companies.