Suchergebnisse

In the spatial variability management of fields, the approach based on management zones (MZs) divides the area into sub-regions, which have spatially homogeneous topography and soil conditions. Such MZs should lead to the same potential yields. Farmers understand which areas of a field have high and low yields, and use of this knowledge may allow the identification of MZs in a field based on production history. The objective of the present study was to evaluate the application of farmer's experience to determine MZs. The study was conducted in three agricultural fields located in the west of the Paraná State in Brazil, and the MZs were generated considering three cases: a) without the use of the farmer's experience variable; b) with the variable of farmer's experience and the stable soil properties selected at the variable selection stage; and c) only with the farmer's experience variable. The generated MZs were evaluated using the Variance Reduction (VR) index, Fuzziness Performance Index (FPI), Modified Partition Entropy (MPE), Smooth Index (SI), and Analysis of Variance (ANOVA). The study showed that the use of farmer's experience to set MZs could be an efficient and simple tool, that it could reduce costs for the processes of setting MZs, compared to the traditional method of using stable soil variables and relief.

The application of precision agriculture considers the values of non-sampled places by the interpolation of sample data. The accuracy with which the maps of spatial distribution of yield and the soil attributes are produced in the interpolation process influences their application and utilization. This paper aimed to compare three interpolation methods (inverse of the distance, inverse of the square distance, and ordinary kriging) in the construction of thematic maps of soybean yield and soil chemical attributes. A set of data referred to 55 sampling units for the construction maps of soybean yield and of eight soil chemical attributes, by different interpolation methods. The comparison was made based on the error matrix, by calculating the Kappa and Tau indices, beyond the relative deviation coefficient (RDC). It was noticed that the inverse of the square distance was the interpolator that less influenced the data behavior, and the best interpolation method dependent of the variability of the studied attribute. The kriging and the inverse of the square distance were considered the methods that presented the best results in the interpolation of data.