“…As a result, r has been successfully employed to predict a variety of related soil properties including clay [Triantafilis et al, 2013a], salinity [Zara et al, 2016], water content [Huang et al, 2016a]. A number of different inversion approaches have also been attempted over the past decade, including using the cumulative function under the low induction number (LIN) [e.g., Smiarowski et al, 2011;Triantafilis et al, 2013b;Viganotti et al, 2013;Jadoon et al, 2015], local and global optimizations [e.g., Mester et al, 2011;Kamm et al, 2013;von Hebel et al, 2014], regularizations [e.g., Borchers et al, 1997;Vervoort and Annen, 2006;P erez-Flores et al, 2012;Li et al, 2013], conditional simulations [e.g., Minsley, 2011;Dafflon et al, 2013], and joint-inversion in combination with other geophysical data sets [e.g., Farzamian et al, 2015]. However, compared with the well-established geostatistical space-time models used in other fields of geoscience [Kyriakidis and Journel, 1999], the continuity and the spatiotemporal variations in EMI data have little been explored.…”