Modelagem do crescimento diamétrico de árvores individuais do estrato arbóreo de uma floresta ombrófila densa.

Abstract

The study objective was to model the diametric growth of individual trees of the tree stratum of a Tropical Rain Forest from two approaches, to model the diametric increase and model the autocorrelation of growth, and to design the DAP (cm) through estimation of the diametric increment obtained by the modeling. This study was conducted with data from the Good Management Project (EMBRAPA / CIFOR / ITTO) - which relates to sustainable management of production forests on a commercial scale in the Brazilian Amazon. , It was undertaken by Embrapa Amazônia Oriental, in the area of Jari Florestal SA, Victoria municipality of Jari, State of Amapa. Three permanent plots were used of one hectare each (100x100 m), on which were measured all trees with DBH ≥ 20 cm in a period of 27 years. The competition between the trees was evaluated by independent indices (G, BAL (basal area of larger trees), Stage and Glover Hool) and dependent (Hegyi and modified Hegyi) distance considering several portions sizes (20 x 20m, 50 x 50 m and 100 x 100m) and different competition rays (3m, 5m, 10m, 15m and 20m). They were evaluated by the Pearson correlation between each index and the IPA DAP (cm-1). In the evaluation of the autocorrelation of diametric growth, two consecutive periods of 10 years of measurement (1984-1994 and 1994-2004) were considered. The grouping of species was constructed from a graph that relates the diametric increment of the trees and the 95th percentile of the diameter. The models were evaluated by R², Syx% and Waste Analysis. Projections were evaluated by Bias (%) and Precision (%).The study concluded that all variations of competition indices tested resulted in very weak correlations with diametric growth (10 to 20%). The BAL index was the one with the highest correlation (r = -0.212 and p = 0.000) and in plots of 50m x 50m. The study confirmed the autocorrelation of growth between two equal and successive periods, presenting a linear correlation of r = 0.73. Both approaches of modeling, did not present satisfactory results, from the point of view of the adjustment of the data. However, when the DAP projection (cm) was performed, the results were adequate, as they produced projections very close to the actual values, indicating a hit of more than 99%. Among the two modeling approaches, autocorrelation was more accurate about 42% more than the diametric projection resulting from the increment modeling.