João Luis Ferreira Batista

Modeling the spatial structure of topical forests

Abstract The spatial structure of tropical forest stands under different management conditions was modeled as a series of different spatial point processes. Spatial patterns were first assessed by K -function analyses to help choose a point process appropriate for observed patterns. The homogenous Neyman–Scott process accurately described live tree distribution in clear cut areas, where tree patterns tended to be aggregated. Parameters were estimated by minimizing Diggles modified least squares criterion, and goodness-of-fit was assessed by comparison to confidence envelopes constructed by Monte Carlo simulation. Parameter estimates can be interpreted to help understand the ecological processes influencing re-colonization of disturbed areas. The inhomogeneous Poisson process was investigated for simulating the spatial pattern of ingrowth trees in lower canopy strata. The intensity function of this process was inversely proportional to variables representing canopy density. As assessed by Monte Carlo generation of confidence envelopes, the inhomogeneous Poisson process successfully portrayed the influence of canopy structure on understory plant distribution in most stands. Tree mortality was modeled as a thinning process in which the probability of individual tree mortality was conditional on subject tree attributes and competitive environment. The thinning function took the form of a generalized linear model with a binomial error distribution and logit link function. In most stands, tree neighborhood variables were powerful predictors of mortality, but they were not important predictors in all plots. This suggests that the surrounding forest structure of a subject tree has considerable influence on its morality, but competition is not the sole cause of tree morality in tropical forests.

Curva de acumulação de espécies e suficiência amostral em florestas tropicais

ABSTRACT – (Species accumulation curve and sampling sufficiency in tropical forests). The use of the species-arearelationship, or the accumulation species curve, to determine sampling sufficiency in phytosociological studies is a currenttechnique, despite of being a controversial issue. The definition of an optimum sample size is based on the idea that thelarger the sample size, the greater the number of species in the sample, but the rate of increase becomes progressivelysmaller so the curve tends to a flat line. The point where the curve becomes horizontal is the minimal area to represent theplant community: the sampling sufficiency. This concept assumes that the plant community is a spacially discrete entitywith fixed species composition. In tropical forests, the identification of communities boundaries is particularly difficult and,due to their high species richness, the species accumulation curves do not become flat, even with large sample sizes. Thespecies accumulation curve is unsuitable to determine optimum sample size in tropical forests because these curves do notbehave as expected, futhermore the concept of sampling sufficiency itself rests upon a problematic definition of plantcommunity.Keywords - phytosociology, sampling sufficiency, species-accumulation curve, species-area relation

Morse operators for digital planar surfaces and their application to image segmentation

This paper introduces the concept of digital planar surfaces and corresponding Morse operators. These operators offer a novel and powerful method for construction and de-construction of such surfaces in a way that global topological control of the resulting object is always maintained. In that respect, this paper offers a complete pixel characterization tool. Image handling is a natural application for such approach. We present a novel fast algorithm for image segmentation using Morse operators for digital planar surfaces. It classifies as a region growing technique with added topological control and is extremely useful for applications that need proper object description. Results from real data are stimulating, and show that the segmentation algorithm compares very well with other methods. The topological approach also forms a base for future expansion to applications such as volume segmentation.

PADRÃO ESPACIAL DE QUATRO FORMAÇÕES FLORESTAIS DO ESTADO DE SÃO PAULO, ATRAVÉS DA FUNÇÃO K DE RIPLEY

Investigar o padrao espacial das arvores, segundo suas classes de tamanho e segundo suas especies mais abundantes pode fornecer evidencias sobre a estrutura da comunidade vegetal, sendo que, o padrao espacial e uma questao-chave para estudos de ecologia florestal. O nivel de organizacao espacial das arvores no ambiente depende de diversos processos ecologicos e caracteristicas proprias de cada ambiente, de modo que a melhor compreensao deste quadro fornece subsidios importantes para o conhecimento sobre formacoes florestais. Neste trabalho objetivou-se estudar o padrao espacial das arvores, segundo suas classes de diâmetro e das quatro especies mais abundantes em diferentes florestas, visando fornecer evidencias sobre a ecologia de cada comunidade vegetal. A descricao do padrao espacial em cada formacao florestal foi realizada segundo a Funcao K de Ripley. As formacoes florestais estudadas foram: Floresta Ombrofila, Cerradao, Floresta Estacional e Restinga. Instalou-se uma parcela de 10,24 ha em cada formacao florestal, e todas as arvores com circunferencia na altura do peito (CAP) a partir de 15 cm foram medidas, georreferenciadas e identificadas. Os resultados obtidos ressaltam o carater agregado em florestas tropicais, como foi observado em todas as florestas estudadas. As arvores do Cerradao e da Restinga apresentaram padroes de agregacao muito proximos. Para a Floresta Ombrofila, o padrao agregado foi significativo em toda a escala de distâncias estudada. Na Floresta Estacional, foi observada tendencia a aleatoriedade, embora, uma agregacao significativa tenha sido notada para curtas distâncias. O padrao espacial por classes de diâmetro foi, em linhas gerais, agregado para arvores menores que 10 cm de diâmetro, e entre 10 e 20 cm, e aleatorio para as demais, evidenciando uma tendencia de que arvores jovens sao mais agregadas do que arvores adultas. O padrao espacial das especies dominantes e sempre muito semelhante ao padrao geral de cada formacao florestal. As diferencas entre o padrao espacial das especies dominantes coincidentes entre as formacoes florestais indicam que seu padrao e influenciado por cada ambiente. Ressaltando-se assim, a importância da sua autoecologia e dos processos ecologicos intrinsecos a cada comunidade que podem explicar os padroes observados.

Comparação técnica e econômica entre harvesters de pneus e com máquina base de esteiras

Uma opcao muito empregada na colheita de arvores e a adaptacao de escavadoras de esteiras, como maquina base, equipadas com cabecote processador, sendo uma opcao as maquinas colhedoras tipicamente florestais. Sua utilizacao baseia-se principalmente no custo menor de aquisicao e a existencia de modelos fabricados no pais, facilitando a assistencia tecnica e a compra de pecas e manutencao. Argumentos contrarios referem-se ao custo elevado de manutencao das esteiras e uma maior restricao a operacao em terrenos acidentados, em comparacao com maquinas de pneus. O objetivo deste trabalho foi comparar, quanto a aspectos tecnicos e economicos, a utilizacao, na operacao de corte de arvores em terreno plano, de modelos de harvesters de pneus com escavadoras de esteiras adaptadas para uso florestal. O maior custo hora dos modelos de harvesters de pneus, utilizados nas condicoes deste estudo, favoreceu a indicacao do emprego de harvesters com maquina base de esteiras na operacao de corte raso de arvores em povoamentos localizados em terrenos planos, com os dois modelos de maquinas apresentando produtividades muito proximas em condicoes de povoamentos florestais teoricamente similares.

Visual Data Exploration to Feature Space Definition

Many image-related applications rely on the fact that the dataset under investigation is correctly represented by features. However, defining a set of features that properly represents a dataset is still a challenging and, in most cases, an exhausting task. Most of the available techniques, especially when a large number of features is considered, are based on purely quantitative statistical measures or approaches based on artificial intelligence, and normally are “black-boxes” to the user. The approach proposed here seeks to open this “black-box” by means of visual representations, enabling users to get insight about the meaning and representativeness of the features computed from different feature extraction algorithms and sets of parameters. The results show that, as the combination of sets of features and changes in parameters improves the quality of the visual representation, the accuracy of the classification for the computed features also improves. The results strongly suggest that our approach can be successfully employed as a guidance to defining and understanding a set of features that properly represents an image dataset.

Circulation and topological control in image segmentation

In this paper we present an image segmentation technique based on the concepts of circulation and topological control. Circulation is a mathematical tool widely used for engineering problems, but still little explored in the field of image processing. On the other hand, by controlling the topology it is possible to dictate the number of regions in the segmentation process. If we take very small regions as noise, the mechanism can be seen as an efficient means for noise reduction. This has motivated us to combine both mathematical tool in our algorithm. As a result, we obtained an automatic segmentation algorithm that generates segmented regions bounded by simple closed curves; a desireable characteristic in many applications.

Ausência de estabilização da curva de acumulação de espécies em florestas tropicais

A definicao do tamanho da amostra e um dos principais problemas em levantamentos fitossociologicos. A curva de acumulacao de especies e comumente utilizada para definir a suficiencia amostral, mas este metodo apresenta algumas limitacoes, como a ausencia de um ponto de estabilizacao que possa ser determinado objetivamente e a arbitrariedade da ordem de entrada das unidades amostrais na construcao da curva. Uma solucao para esse problema e o uso de procedimentos de aleatorizacao, como a permutacao, para obter curvas medias de acumulacao de especies e seus intervalos de confianca empiricos. Entretanto, esse procedimento enfatiza o carater assintotico da curva, que nao apresenta ponto de inflexao, tornando-se impossivel a determinacao de um ponto objetivo para avaliar o tamanho otimo da amostral.

Exatidão de dendrômetros ópticos para determinação do volume de árvores em pé

The aim of this paper is to verify the accuracy of the optical dendrometers Criterion 400 and RC3H by studying the quality of measures of wood volume determination of standing trees through these dendrometers. The study was developed at the Experimental Station of Forest Sciences, in Itatinga, Sao Paulo state, which belongs to the University of Sao Paulo (ESALQ / USP). It was sampled a total of 175 trees in three plots of Eucalyptus grandis. The stem diameter of standing trees was measured by the two optical dendrometers at distances of 0.1, 0.4, 0.7, 1.0, 1.3, 2.0 meters from this point meter by meter along the stem up to 8 meters height. After measuring the standing trees they were felled and the diameter was obtained in different positions with a caliper and a tape in the same positions that the non-destructive measures were taken. With the diameters the volume was calculated by section and by individual trees by the Smalian formula for the comparison of methods. Analyzing the measurements of stem diameter and the individual tree volume realizes that the two dendrometers provided measures generally underestimated. However, the Criterion provided the best estimates. For the diameter and individual tree volume the Criterion showed underestimated errors averaging approximately 1 cm (10%), while the RC3H resulted in errors greater than 5 cm (30%) on average.Thus, when measuring observed variables in a non-destructive way with reliability and accuracy, the Criterion showed better results.