RESUMOO semivariograma possibilita uma avaliação visual da dependência espacial, mas sem resultar diretamente em um valor numérico único que expresse a mensuração de tal dependência. Contudo, a partir dos parâmetros estimados do modelo teórico ajustado ao semivariograma experimental, é possível construir uma medida dessa dependência espacial. Atualmente, há dois índices na literatura, com uso cada vez mais frequente; porém, há inadequações nesses índices existentes. O objetivo deste trabalho foi propor um novo índice para medir a dependência espacial de dados geoestatísticos, que supere as incipiências dos atuais. Esse novo índice utiliza a relação existente entre o semivariograma e o correlograma, contemplando dessa forma todos os aspectos da dependência espacial. Realizaram-se uma comparação, por simulação, entre o índice proposto e os índices já existentes e também verificação da aplicabilidade do índice proposto utilizando pesquisas reais publicadas, em que ocorreram ajustes dos modelos teóricos esférico, exponencial e gaussiano. Verificou-se que o índice proposto foi melhor que os índices existentes. Além disso, observou-se que os índices existentes podem levar a equívocos nas interpretações do grau de dependência espacial, evidenciando que devem ser evitados. Em decorrência, recomenda-se a utilização do novo índice proposto para medir o grau da dependência espacial.Termos de indexação: semivariograma, alcance prático, fator de modelo, medidas resumo.

ABSTRACT:In geostatistical studies, spatial dependence can generally be described by means of the semivariogram or, in complementary form, with a single index followed by its categorization to classify the degree of such dependence. The objective of this study was to construct a categorization for the spatial dependence index (SDI) proposed by Seidel and Oliveira (2014) in order to classify spatial variability in terms of weak, moderate, and strong dependence. Theoretical values were constructed from different degrees of spatial dependence, which served as a basis for calculation of the SDI. In view of the form of distribution and SDI descriptive measures, we developed a categorization for posterior classification of spatial dependence, specific to each semivariogram model. The SDI categorization was based on its median and 3rd quartile, allowing us to classify spatial dependence as weak, moderate, or strong. We established that for the spherical semivariogram: SDI Spherical (%) ≤ 7 % (weak spatial dependence), 7 % < SDI Spherical (%) ≤ 15 % (moderate spatial dependence), and SDI Spherical (%) > 15 % (strong spatial dependence); for the exponential semivariogram: SDI Exponential (%) ≤ 6 % (weak spatial dependence), 6 % < SDI Exponential (%) ≤ 13 % (moderate spatial dependence), SDI Exponential (%) > 13 % (strong spatial dependence); and for the Gaussian semivariogram: SDI Gaussian (%) ≤ 9 % (weak spatial dependence), 9 % < SDI Gaussian (%) ≤ 20 % (moderate spatial dependence), and SDI Gaussian (%) > 20 % (strong spatial dependence). The proposed categorization allows the user to transform the numerical values calculated for SDI into categories of variability of spatial dependence, with adequate power for explanation and comparison.

Tibraca limbativentris (rice stem bug) is an insect highly injurious to the rice crop in Brazil. The aim of this research was to define the spatial distribution of the T. limbativentris and improve the sampling process by means of geostatistical application techniques and construction of prediction maps in a flooded rice field located in the "Planalto da Campanha" Region, Rio Grande do Sul (RS), Brazil. The experiments were conducted in rice crop in the municipality of Itaqui - RS, in the crop years of 2009/10, 2010/11 and 2011/12, counting fortnightly the number of nymphs and adults in a georeferenced grid with points spaced at 50m in the first year and in 10m in the another years. It was performed a geostatistical analysis by means adjusting semivariogram and interpolation of numeric data by kriging to verify the spatial dependence and the subsequent mapping population. The results obtained indicated that the rice stem bug, T. limbativentris, has a strong spatial dependence. The prediction maps allow estimating population density of the pest and visualization of the spatial distribution in flooded rice fields, enabling the improvement of the traditional method of sampling for rice stem bug

This research was conducted to propose a classification of the coefficient of variation (CV%) in many categories of variables of production and carcass of beef cattle experiments. The data was collected from theses and dissertations. We used the methods of classification considering mean and standard deviation, and considering median and pseudo-sigma. The two methods showed similar results so both can be used to classify CV%. We propose only three categories to rank CV%: low, medium and high.

No abstract