|Title||Impressions of digital soil maps: The good, the not so good, and making them ever better|
|Author(s)||Arrouays, Dominique; McBratney, Alex; Bouma, Johan; Libohova, Zamir; Richer-de-Forges, Anne C.; Morgan, Cristine L.S.; Roudier, Pierre; Poggio, Laura; Mulder, Vera Leatitia|
|Source||Geoderma Regional 20 (2020). - ISSN 2352-0094|
ISRIC - World Soil Information
Soil Geography and Landscape
|Publication type||Refereed Article in a scientific journal|
|Keyword(s)||Digital soil assessment - Digital soil mapping - Machine learning - Pedology - Soil survey|
Since the turn of the millennium, digital soil mapping (DSM) has revolutionized the production of fine resolution gridded soil data with associated uncertainty. However, the link to conventional soil maps has not been sufficiently explained nor are the approaches complementary and synergistic. Further training on the digital soil mapping approaches, and associated strengths and weaknesses is required. The user community requires training in, and experience with, the new digital soil map products, especially about the use of uncertainties for risk modelling and policy development. Standards are required for public and private sector digital soil map products to prevent the production of poor-quality information which will become misleading and counter-productive. Machine-learning methods are to be used with caution with respect to their interpretability and parsimony. The use of DSM products for improved pedological understanding and soil survey interpretations requires urgent investigation.