Mapping LUCAS topsoil chemical properties at European scale using Gaussian process regression

• Published in: Geoderma Vol. 355; p. 113912
• Main Authors: Ballabio, Cristiano, Lugato, Emanuele, Fernández-Ugalde, Oihane, Orgiazzi, Alberto, Jones, Arwyn, Borrelli, Pasquale, Montanarella, Luca, Panagos, Panos
• Format: Journal Article
• Language: English
• Published: Elsevier B.V 01.12.2019; Elsevier Scientific Pub. Co
• Subjects: Bulgaria; C:N ratio; calcium carbonate; carbon nitrogen ratio; Cation Exchange Capacity; covariance; European Union; Gaussian process regression; issues and policy; land cover; land use; Mapping; model uncertainty; Nitrogen; normal distribution; Phosphorus; physical properties; Potassium; Romania; soil quality; soil sampling; surveys; topsoil; Bulgaria; Romania; Cation Exchange Capacity; C:N ratio; pH; Gaussian process regression; Phosphorus; Mapping; Nitrogen; Potassium
• ISSN: 0016-7061; 1872-6259
• DOI: 10.1016/j.geoderma.2019.113912

This paper presents the second part of the mapping of topsoil properties based on the Land Use and Cover Area frame Survey (LUCAS). The first part described the physical properties (Ballabio et al., 2016) while this second part includes the following chemical properties: pH, Cation Exchange Capacity (CEC), calcium carbonates (CaCO3), C:N ratio, nitrogen (N), phosphorus (P) and potassium (K). The LUCAS survey collected harmonised data on changes in land cover and the state of land use for the European Union (EU). Among the 270,000 land use and cover observations selected for field visit, approximately 20,000 soil samples were collected in 24 EU Member States in 2009 together with more than 2000 samples from Bulgaria and Romania in 2012. The chemical properties maps for the European Union were produced using Gaussian process regression (GPR) models. GPR was selected for its capacity to assess model uncertainty and the possibility of adding prior knowledge in the form of covariance functions to the model. The derived maps will establish baselines that will help monitor soil quality and provide guidance to agro-environmental research and policy developments in the European Union. •The LUCAS harmonised soil survey comprising 20,000 observations was used in this study.•Soil chemical properties were mapped over the extent of Europe using Gaussian process regression (GPR).•GPR with Matérn kernel modelled soil properties with high accuracy.•GPR allows the production of uncertainty maps as it allows the estimation of prediction variance.•These maps complete the series of topsoil properties for Europe based on LUCAS.