Anais do XXXV Congresso Brasileiro de Ciência do Solo

KNOWLEDGE-BASED DIGITAL SOIL MAPPING TECHNIQUE FOR PREDICTING SATURATED HYDRAULIC CONDUCTIVITY

MICHELE DUARTE DE MENEZES⁽¹⁾; PHILLIP RAY OWENS⁽²⁾; CARLOS ROGÉRIO DE MELLO⁽¹⁾; SÉRGIO HENRIQUE GODINHO SILVA⁽¹⁾; NILTON CURI⁽¹⁾;
1 - UFLA; 2 - PURDUE UNIVERSITY;

The estimation of saturated hydraulic conductivity (Ksat) at non-sampled areas in a continuous way can be source of valuable information for land and water management. The objective of this work was assessing the spatial prediction of Ksat in two watersheds by means of knowledge-based digital soil mapping technique. The way of choose typical values as well as the role of land use as a covariate was tested. An independent validation data set was used for assessing the accuracy of predictions methods. The knowledge-based digital soil mapping technique is an accurate option for spatial prediction of Ksat considering: 1) non-linearity distribution of data, 2) the high cost of an intense sampling scheme; and 3) the scarce resources for that in Brazil. The use of a fuzzy membership maps as a guide for sampling Ksat at Lavrinha Creek Watershed (LCW) is an accurate alternative. At Marcela Creek Watershed (MCW) the mean of values inside the polygons provide better accuracy of spatial prediction, since the land use does not influence Ksat statistically.

[baixar arquivo pdf]