Modelagem biométrica para árvores individuais a partir do lidar em área de manejo de precisão em florestas tropicais na Amazônia Ocidental

Abstract

The studies in this thesis aim to obtain equations for estimating bole volume, impact of gaps and bole biomass of dominant and codominant trees cut under conditions of precision forest management. These equations used morphometric variables of the canopy obtained by the airborne LIDAR. The study area is in the Antimary State Forest (FEA), located in the municipality of Bujari in the state of Acre, Brazil. The studies were grouped into three chapters. The first has as its object the construction of equations for estimating the bole volume of individual trees considering two situations of forest inventory: a) with the collection of diameter at breast height (DBH), and crown morphometric variables obtained from LIDAR data and b) using only the crown morphometry variables. For the selection of models the factors considered were: the correlation matrix of predictor variables and the combination of variables that generates the best results by statistical criteria Syx, Syx (%) and Pressp, and that were homoscedastic and had normal and independent distributions of errors. Influence analysis was perfomed for the best equations. Results for the statistical fit of the equations for the two situations allowed selection of models with and without DBH, with R2 aj.(%) values of a) 92.92 and b) 79.44, Syx(%) values of a) 16.73 and b) 27.47, and, Pressp criterion values of a) 201.15 m6 and b) 537.47 m6, respectively. The second chapter describes the studies for estimating the areas of gaps of individual dominant and codominant trees, since this information is neglected in the forest planning process and its prior analysis is an important tool for selection of trees that can maximize the forest volume maintained and reduce the impacts on forest cover. On two separate occasions profiles were made in an annual forestproduction unit in the Antimary State Forest (FEA). The first was carried out a few days before the start of logging and the second was done after completion of harvest activities. With field measurements and processing of the cloud of LIDAR points, dendrometric and morphometric variables were obtained for the canopy in order to develop equations for estimating gap areas. After evaluation of the explanatory variables with the highest correlations with gap area, the method used considering all possible models and including 2-4 parameters. Ten equations were selected, of which two were chosen for use; these had R2 aj > 75%, Syx < 23%, the sum of the residuals tending to zero and a graph of the distribution of the residuals indicating no bias. The third and final chapter presents the development of allometric models for estimating green and dry biomass stored in the boles of dominant and codominant trees. The method used selected from among all possible models and performed identity testing of models in order to consider the different groups of basic wood density (low, medium and high). The morphometric variables of the crown showed high explanatory power for predicting bole biomass and allometric equations can be associated or not with varying DBH. When considering bole biomass according to the basic wood density, the best estimate is obtained using allometric equations with variables on both morphology of crown and DBH. To form a single group involving the three classes of basic density, one must adhere exclusively the explanatory variables represent the crown, or, in the case of dry biomass, variables basic density (DB) and total height (Ht). By of morphometric variables of the tree crown obtained with the airborne LIDAR it was possible to develop equations capable of accurately estimating the area of gap and the volume and bole biomass of dominant and codominant trees in tropical forests, which demonstrates the potential of using forest profiling for improving precision management.