Soil spatial variability in relation to soil formation and agricultural management has been an appealing area of research recently. When the distance between samples causes increased variation of a property, this variation can in part be explained by spatial dependency. The analysis of spatial variability is an important support for the planning of experiments and to evaluate effects of treatments, aiming to reduce experimental variations attributed to random error. The infiltration rate, normally of high spatial variability, is agronomically important due to its role in runoff and to determine irrigation intensities. In this paper we estimated the spatial dependence degree and its range and influence on the variability of infiltration rates. Another aim was to determine the number of samples needed to obtain a certain level of uncertainty/accuracy. An infiltration experiment was set up in a medium textured Alfisol under no-tillage with 40 observation points in a straight transect, spaced one meter apart. The infiltration rate was determined with a cylinder infiltrometer, and the data submitted to descriptive and exploratory statistics as well as to geostatistics. The infiltration rate was highly variable, especially in the initial measurements. Spatial dependence was found in a range of 3.5 m. The number of observations needed for a certain level of uncertainty increased when only independent spatial data were used.
infiltration rate; spatial variability; zero tillage
