ABSTRACT The goal of this study was to use the spatial bootstrap method to model the spatial dependence structure of soybean yield and soil chemical attributes in an agricultural area. The study involved developing confidence intervals in probability plots to determine the probability distributions assumed by the data; determine the empirical distributions of the semivariances and model parameters, allowing to obtain statistics and confidence intervals; and to construct maps for the variables. The quantile-quantile plots indicated that the data follows a normal distribution. The confidence intervals for the semivariances helped to model the spatial dependence structure, and the descriptive statistics of the bootstrap replicates of the model parameters allowed to test the consistency of the estimates. The soil chemical attributes (calcium, potassium, and organic matter) were at levels suitable for soybean cultivation. However, the pH was below the ideal range in most of the study area, and water stress during cultivation decreased the mean yield. Therefore, according to the results, a recommendation to the farmer is to correct the soil pH to increase the yield.

ABSTRACT This study aims to quantify the uncertainties associated to the parameters of a Gaussian spatial linear model (GSLM) and the assumption of normality residuals in the modeling of the spatial dependence of the soybean yield as a function of soil chemical attributes. The spatial bootstrap methods were used to determine the point and interval estimators associated with the model parameters. Hypothesis tests were carried out on the significance of the model parameters and the quantile-quantile probability plot was elaborated to verify the data normality. The uncertainties associated to the parameters of the spatial dependence structure were quantified and the potassium content, phosphorus content and soil pH covariates were significant to explain the soybean yield mean. These covariates were used in the elaboration of a new model, which provided the elaboration of a contour map of soybean yield. Analysis of the quantile-quantile plot indicated that soybean yield data follow a normal probability distribution.

ABSTRACT: In geostatistical modeling of soil chemical properties, one or more influential observations in a dataset may impair the construction of interpolation maps and their accuracy. An alternative to avoid the problem would be to use most robust models, based on distributions that have heavier tails. Therefore, this study proposes a spatial linear model based on the slash distribution (SSLM) in order to characterize the spatial variability of soybean yields as a function of soil chemical properties. The likelihood ratio statistic (LR) was applied to verify the significance of parameters associated with the model. We evaluated the sensitivity of the maximum likelihood estimators by means of local influence analysis for both the soybean response and the linear predictor. In the proposed model, we analyzed data gathered from a commercial grain production area (127.18 ha) located in the western part of the state of Paraná (Brazil). The results showed that the slash distribution allowed us to adjust the high kurtosis of the data set distribution and the LR test confirmed that the soil chemical properties of phosphorus, potassium, pH, and organic matter were significant for the SSLM. Diagnostic analysis indicated that the atypical value of the sample set was not influential in the parameter estimation process. Construction of the interpolation map based on the proposed model is not affected when considering the atypical and/or influential observations. Thus, SSLM becomes a robust alternative in the study of soybean yield variability as a function of soil chemical properties, making it possible to investigate the productive potential of the areas.

ABSTRACT: The t-Student distribution has been used to the spatial dependence modelling of soybean yield as an alternative to the normal distribution, being used for data with heavier tails or discrepant values. However, a usual Student t-distribution does not allow direct comparisons of geostatistical methods with a normal distribution. The aim of this study was to assess the soybean yield spatial variability through a reparameterized t-Student linear model, comparing the results with those of a Gaussian linear model. For parameter estimation, a complete maximum likelihood (CML) method was used through an expectation-maximization (EM) algorithm. The maps constructed with both reparameterized t-Student and normal distributions are dissimilar and present a kappa index (K) equivalent to 0.64. The reparameterized t-Student distribution is an alternative in studying data with discrepant values, showing the ability to decrease the influence of these points.

D.M. Grzegozewski, M.A. Uribe-Opazo, F. De Bastiani, and M. Galea. 2013. Local influence when fitting Gaussian spatial linear models: an agriculture application. Cien. Inv. Agr. 40(3): 523-535. Outliers can adversely affect how data fit into a model. Obviously, an analysis of dependent data is different from that of independent data. In the latter, i.e., in cases involving spatial data, local outliers can differ from the data in the neighborhood. In this article, we used the local influence technique to identify influential points in the response variables using two different schemes of perturbations. We applied this technique to soil chemical properties and soybean yield. We evaluated the effects of the influential points on the spatial model selection, the parameter estimation by maximum likelihood and the construction of thematic maps by kriging. In the construction of the thematic maps in studies with and without the influential points, there were changes in the levels of nutrients, allowing for the appropriate application of input, generating greater savings for the producer and contributing to the protection of the environment.

D.M. Grzegozewski, M.A. Uribe-Opazo, F. De Bastiani y M. Galea. 2013. Influencia local a modelos espaciales lineales Gaussianos: Aplicación a la agricultura. Cien. Inv. Agr. 40(3): 523-535. Los valores discrepantes pueden afectar negativamente el ajuste de un modelo. El análisis de datos dependientes es diferente al de datos independientes. En el primer caso, envuelven datos espaciales que pueden tener valores discrepantes localmente y que tienen algunas características diferentes de los datos vecinos. En este artículo, el objetivo fue detectar los puntos influyentes por medio de la técnica de influencia local en la variable de respuesta, mediante el uso de dos esquemas diferentes de perturbaciones denominados: perturbación aditiva y perturbación de Zhu. Se aplicó esta técnica a las propiedades químicas del suelo y a la productividad de la soja. Se evaluaron los efectos de los puntos influyentes en la elección del modelo, en la estimación de parámetros de máxima verosimilitud y la construcción de mapas temáticos mediante "kriging". En la construcción de mapas temáticos, se pudo observar alteraciones en los niveles de nutrientes al realizar el estudio con y sin los puntos de influencia, de tal forma que permite una aplicación apropiada de los insumos, lo que genera un mayor ahorro para el productor y en la contribución a la protección del medio ambiente.

O uso das ferramentas da geoestatística, aliadas à agricultura de precisão permitem o acompanhamento das áreas agrícolas produtoras de soja, estabelecendo as relações de dependência espacial entre os pontos amostrados. A modelagem da estrutura de variabilidade espacial possibilita a construção de mapas temáticos dos atributos estudados, utilizando como método de interpolação a krigagem. Porém, a presença de valores atípicos entre os elementos amostrais pode influenciar na construção e interpretação desses mapas. A distribuição de probabilidades t-Student tem sido utilizada na tentativa de diminuir a influência dos valores atípicos durante a estimativa dos parâmetros de dependência espacial, por ter caudas mais pesadas que a distribuição normal. A detecção dos valores influentes na área em estudo, por meio da análise de diagnósticos de influência local, confere maior confiabilidade na utilização dos mapas gerados, corroborando a aplicação de insumos. Deste modo, o objetivo deste trabalho foi aplicar as técnicas de influência local em dados espacialmente referenciados, com os modelos de perturbação aditiva e utilizando a matriz escala, considerando a distribuição t-Student n-variada. Foi utilizado um modelo espacial linear para o estudo de dados da produtividade da soja em função da altura média de plantas e do número médio de vagens por planta. As técnicas de influência local foram eficientes para detectar pontos que influenciam na escolha do modelo geoestatístico, nas estimativas dos parâmetros e na construção do mapa temático.

The use of geostatistical tools combined with precision agriculture, allow the monitoring of agricultural soybean producing areas, establishing relationships of spatial dependence between the sampled points. The modeling of spatial variability structure enables the construction of thematic maps of the attributes studied, using the kriging as the interpolation method. However, the presence of outliers among the elements sampling can influence the construction and interpretation of these maps. The distribution of t-Student probability has been used in attempts to reduce the influence of atypical points in the estimation of parameters of spatial dependence, having heavier tails than the normal distribution of probabilities. The detection of influential points in the study area, through the analysis of local influence diagnostics, provides greater reliability in the use of maps, providing an efficient use of inputs. Then, the objective was to apply the techniques of local influence on spatially referenced data with models of disturbance and using the matrix additive scale, considering the distribution t-Student n-variate. It was used a linear spatial model for the study of soybean yield data as a function of average plant height and number of pods per plant. The local influence techniques were effective to detect points that influence in the geostatistical model selection, estimation of parameters and construction of thematic maps.

A modelagem e estimação dos parâmetros que definem a estrutura de dependência espacial de uma variável regionalizada, utilizando métodos geoestatísticos, é de fundamental importância, pois a partir desses parâmetros é realizada a krigagem dos pontos não amostrados para a construção dos mapas temáticos. A presença de uma ou mais observações atípicas nos dados amostrais podem influenciar a estimação desses parâmetros. Assim, torna-se importante a avaliação da influência conjunta destas observações pela análise de Influência Local. Este trabalho tem por objetivo apresentar métodos de análise de influência local para variável regionalizada considerando que ela tenha distribuição t-Student n-variada e comparar com a análise de influência local considerando que esta mesma variável regionalizada tenha distribuição normal n-variada. Esses métodos de análise de influência local foram aplicados a atributos físicos do solo e produtividade da soja obtidos a partir de um experimento realizado em uma área comercial de 56,68 ha da região Oeste do Paraná, Brasil. O estudo mostrou que os valores influentes são identificados eficientemente com a distribuição t-Student n-variada.

The modeling and estimation of the parameters that define the spatial dependence structure of a regionalized variable by geostatistical methods are fundamental, since these parameters, underlying the kriging of unsampled points, allow the construction of thematic maps. One or more atypical observations in the sample data can affect the estimation of these parameters. Thus, the assessment of the combined influence of these observations by the analysis of Local Influence is essential. The purpose of this paper was to propose local influence analysis methods for the regionalized variable, given that it has n-variate Student's t-distribution, and compare it with the analysis of local influence when the same regionalized variable has n-variate normal distribution. These local influence analysis methods were applied to soil physical properties and soybean yield data of an experiment carried out in a 56.68 ha commercial field in western Paraná, Brazil. Results showed that influential values are efficiently determined with n-variate Student's t-distribution.

A modelagem da estrutura de dependência espacial pela abordagem da geoestatística é fundamental para a definição de parâmetros que definem esta estrutura, e que são utilizados na interpolação de valores em locais não amostrados pela técnica de krigagem. Entretanto, a estimação de parâmetros pode ser muito afetada pela presença de observações atípicas nos dados amostrados. O desenvolvimento deste trabalho teve por objetivo utilizar técnicas de diagnóstico de influência local em modelos espaciais lineares gaussianos, utilizados em geoestatística, para avaliar a sensibilidade dos estimadores de máxima verossimilhança e máxima verossimilhança restrita na presença de dados discrepantes. Estudos com dados experimentais mostraram que tanto a presença de valores atípicos como de valores considerados influentes, pela análise de diagnóstico, pode exercer forte influência nos mapas temáticos, alterando, assim, a estrutura de dependência espacial. As aplicações de técnicas de diagnóstico de influência local devem fazer parte de toda análise geoestatística a fim de garantir que as informações contidas nos mapas temáticos tenham maior qualidade e possam ser utilizadas com maior segurança pelo agricultor.

Modeling of spatial dependence structure, concerning geoestatistics approach, is an indispensable tool for fixing parameters that define this structure, applied on interpolation of values in places that are not sampled, by kriging techniques. However, the estimation of parameters can be greatly affected by the presence of atypical observations on sampled data. Thus, this trial aimed at using diagnostics techniques of local influence in spatial linear Gaussians models, applied at geoestatistics in order to evaluate sensitivity of maximum likelihood estimators and restrict maximum likelihood to small perturbations in these data. So, studies with simulated and experimental data were performed. Those results, obtained from the study of real data, allowed us to conclude that the presence of atypical values among the sampled data can have a strong influence on thematic maps, changing, therefore, the spatial dependence. The application of diagnostics techniques of local influence should be part of any geoestatistic analysis, ensuring that the information from thematic maps has better quality and can be used with greater security by farmers.