“…In the DSM field, machine-learning techniques are increasingly used to elucidate the spatial distribution of both soil type and soil properties across a large range of scales (Bui and Moran., 2001;Scull et al, 2005;Malone et al, 2009;Nelson and Odeh, 2009;Abdel-Kader, 2011;Lacoste et al, 2011;Lemercier et al, 2012;Kempen et al, 2012;Jafari et al, 2013;Nauman and Thompson, 2014;Brungard et al, 2015;Mosleh et al, 2016;Viloria et al, 2016;Nussbaum et al, 2018;Vaysse and Lagacherie, 2015;Ellili et al, 2019;Padarian et al, 2019;Arrouays et al, 2020).They were also applied to disaggregate superficial geology maps available at a 1 : 250 000 scale in Australia (Bui and Moran, 2001). The main advantage of these approaches is they allow the handling of both quantitative and categorical (ordinal or nominal) soil and environmental variables as explanatory covariates (Bui and Moran, 2001).…”