Análise da variabilidade espaço-temporal de atributos químicos do solo em uma área agrícola produtora de grãos

Abstract

Assessing changes in soil attributes is essential in agronomy, as it directly impacts both productivity and soil health. With the advancement of precision agriculture, quantifying these variations has become increasingly important. The characterization of the spatial and temporal variability of soil attributes requires the application of various statistical techniques. Thus, the main objective of this thesis is to analyze the spatio-temporal distribution pattern of soil chemical attributes in a commercial agricultural area, applying a diverse set of statistical techniques. To achieve this objective, the thesis was structured into two scientific articles. The first study focused on analyzing the spatio-temporal variability of soil chemical attributes in a soybean cultivation area from 2010 to 2022, using spatio-temporal geostatistics techniques. Semivariance models were evaluated, particularly the metric-sum and simple metric-sum models. The results indicated that soil chemical attributes displayed moderate spatial dependence, with a limited range, while temporal dependence ranged from moderate to strong, with a span of one to six years, depending on the attribute. The joint spatio-temporal dependence was observed to be weak to moderate, suggesting limited interaction between spatial and temporal components. The study emphasized the sensitivity of chemical attributes to temporal factors and reinforced the significance of geostatistical methods for sustainable agricultural management. In the second article, the study analyzed the variation in calcium content, phosphorus content, organic matter content, and pH in a 167.35-hectare soybean area located in Cascavel, Paraná, between 2010 and 2022. By applying Empirical Orthogonal Functions (EOFs) and the Lomb-Scargle Discrete Fourier Transform (LSDFT) for irregular data, it was found that the variability of the attributes could be captured by a small number of EOFs: two for calcium content, three for pH and organic matter content, and four for phosphorus content. The patterns identified specific variability trends for each attribute and highlighted the influence of certain years on their variation. Power spectra were used to detect seasonal events, providing valuable insights. This technique proved effective in identifying relevant spatial and temporal patterns that are crucial for a deeper understanding of soil chemical attribute variability.