The Santa Maria dataset contains \(n=\) 410 point soil observations made between 2004 and 2011 as part of different projects. These projects aimed at producing semi-detailed soil and land use maps (scale of 1:) \citep{Pedron2005, Miguel2010, SamuelRosaEtAl2011a, MiguelEtAl2012}, and predicting the vulnerability of the topsoil to erosion \citep{MouraBueno2012, Miguel2013}. As such, the dataset is composed of three subsets. The first subset contains \(n=\) 340 observations, the second \(n=\) 60, and the third \(n=\) 10 observations (Figure \ref{fig:points}).