Anais do XXXIV Congresso Brasileiro de Ciência do Solo

ORDINARY KRIGING AND REGRESSION KRIGING FOR PREDICTING SOIL PHYSICAL-CHEMICAL PROPERTIES IN TWO WATERSHEDS AT MINAS GERAIS STATE

MICHELE DUARTE DE MENEZES⁽¹⁾; PHILLIP R. OWENS⁽²⁾; CARLOS ROGÉRIO DE MELLO⁽³⁾; NILTON CURI⁽³⁾;
1 - UNIVERSIDADE FEDERAL RURAL DO RIO DE JANEIRO; 2 - PURDUE UNIVERSITY; 3 - UFLA;

The estimation of soil properties at non-sampled areas can be source of valuable information for land management. This study compared the performance of ordinary kriging (OK) and regression kriging (RK) for predicting soil physical-chemical properties in two watersheds with contrasting soil-landscape features. Mean prediction of error and root mean square of prediction error were used for assessing the prediction methods. A better linear correlation was found at Lavrinha Creek Watershed, suggesting some relationship between contemporaneous landforms and soil properties, and RK outperformed OK. In the most cases, RK were not performed at Marcela Creek Watershed due to lack of linear correlation between covariates and soil properties. Since alternatives of simple OK kriging have been sought, another prediction method should be tested, considering the systematic pattern of soil properties distribution over that landscape

[baixar arquivo pdf]