Soil functions modelling at global scale using SoilGrids 2.0

Soils provide a variety of goods and services and are a key natural resource, non-renewable on the time scale of a human life-span, to realise several UN-Sustainable-Development–Goals, including zero hunger (SDG2), climate action (SDG13) and life on land (SDG15). Consistent soil information is required to underpin a large range of global assessments, such as soil and land degradation, sustainable land management, and environmental conservation. In this study, we present an application addressing the modelling of soil functions at global scale, using erosivity risk and soil carbon sequestration potential as examples. We used SoilGridsv2.0, as set of soil property maps at 250m resolution, derived from a large set of standardized soil profile observations (WoSIS), an updated set of environmental covariates, and improved machine learning models. It provides global assessments of prediction uncertainty, quantified with a 90% prediction interval, and considers an evaluation procedure that provides more realistic metrics of map accuracy. We used simplified models to derive soil functions (e.g. erosivity or carbon sequestration potential) from basic soil properties, meaningful for different pedo-climatic regions. We provide an indication of areas of low/high risk of soil degradation to support sustainable soil management planning. The uncertainty limits of the input soil layers were used to provide a preliminary assessment of the uncertainty of the derived layers. The present modelling framework, using soil properties maps to derive soil functions, offers great flexibility and may be applied to a diverse set of models to generate soil information products tailored to specific applications. We highlight some of the challenges of assessing soil functions at global scale.