Comparação de três métodos de interpolação na construção de mapas temáticos

Abstract

Samples of soil and plant properties are collected and analyzed to apply at precision agriculture (PA) and obtain information that can help on a decision making in order to correct problems that have been detected in the studied areas. Thus, computer systems are used to generate thematic maps by interpolation of sample data to measure information for nonsampled locations. The choice of interpolation method to be used is an important aspect, since most of the interpolators do not preserve the original data, so the thematic map is affected. The accuracy with which the spatial distribution maps of yield and soil properties are produced during the interpolation process influences the implementation and use of PA. A set of data from 55 samples of chemical properties of a soil and 130 samples of soybean yield, 2010/2011 harvest, were analyzed to compare three interpolation methods (inverse distance, inverse distance squared and ordinary kriging) on the thematic maps drawing. Maps of ten chemical properties of soil and two samples concerning yield were compared, using Kappa, Tau and CDR. It was observed some variation with the results and that the interpolation influence depends on the attribute and data frequency. However, kriging and inverse square distance were considered the methods that generated maps with lower dispersion of data