Uncertainty quantification of interpolated maps derived from observations with different accuracy levels

Most practical applications of spatial interpolation ignore that some measurements may be more accurate than others. As a result all measurements are treated equally important, while it is intuitively clear that more accurate measurements should carry more weight than less accurate measurements. Geostatistics provides the tools to perform spatial interpolation using measurements with different accuracy levels. In this short paper we use these tools to explore the sensitivity of interpolated maps to differences in measurement accuracy for a case study on mapping topsoil clay content in Namibia using kriging with external drift (KED). We also compare the kriging variance maps and show how incorporation of different measurement accuracy levels influences estimation of the KED model parameters. Most practical applications of spatial interpolation ignore that some measurements may be more accurate than others. As a result all measurements are treated equally important, while it is intuitively clear that more accurate measurements should carry more weight than less accurate measurements. Geostatistics provides the tools to perform spatial interpolation using measurements with different accuracy levels. In this short paper we use these tools to explore the sensitivity of interpolated maps to differences in measurement accuracy for a case study on mapping topsoil clay content in Namibia using kriging with external drift (KED). We also compare the kriging variance maps and show how incorporation of different measurement accuracy levels influences estimation of the KED model parameters.