Dependence and spatial correlation of soil attributes as function of sample arrangement

Abstract

National agriculture has undergone intense changes, both in the way of managing the properties and in the technological level employed in them. Among the technologies adopted, emphasis is placed on precision agriculture, mainly used for the application of fertilizers at various rates, using the information from the analysis of soil samples, collected in georeferenced continuous grids. The present study was conducted in a plot of Fazenda Planalto, located in the municipality of Maracaju - MS, with the objective of understanding the different sampling arrangements adopted for soil collection and its influence on the geostatistical analysis of phosphorus, potassium, base saturation, pH, organic matter and clay content, as well as the reliability of the prediction maps elaborated from the geostatistical analysis. Different numbers of soil simple samples were used to form the soil composite sample (1, 5, 10, 17 and 25) combined with a thinning of the sample points to form different grid resolutions (187, 95, 47 and 23 points). It was concluded that eight simple samples are necessary for a smaller error of prediction of values in non-sampled locations. To estimate spatial dependence, sample grids with: 187 points for potassium phosphorus and base saturation, 95 points for pH and organic matter and 23 points for clay content are needed. The best estimates in unsampled locations are obtained for the clay.