Abstract

MORVILLO, M. C et al. Prediction of regionalized variables with undetected data: two cases of study. Geoacta [online]. 2015, vol.40, n.1, pp.54-64. ISSN 1852-7744.

Non-detected data are usually found in sampling of georeferenced variables in contexts such as mineral deposit exploration, environmental monitoring, and groundwater reservoir among others. Such data are non-detected because their values are below a detection limit or above the maximum detection level of the instrument used to measure. These extreme data, when they are not properly imputed can produce significant distortions in the contour maps that are made with interpolation methods. In fact, if predicted values in locations with this kind of data are calculated before imputing, would intermediate values, in clear contrast to extreme condition. In this work, two samplings with non-detected data are shown. One, on gold exploration, with values below a detection level. Another, on the depth of groundwater layer with values above of the maximum detection level of measuring. Three criteria -Elimination, LDI and NCIP - are considered to impute it in places with no detected values and their effects on the prediction are compared. The results are more favorable the CNPI criterion.

Keywords : detected georeferenced data; Imputation CNPI; Kriging.

        · abstract in Spanish     · text in Spanish     · Spanish ( pdf )