From EU-SoilHydroGrids to HU-SoilHydroGrids: A leap forward in soil hydraulic mapping

• Published in: The Science of the total environment Vol. 921; p. 171258
• Main Authors: Szabó, Brigitta, Mészáros, János, Laborczi, Annamária, Takács, Katalin, Szatmári, Gábor, Bakacsi, Zsófia, Makó, András, Pásztor, László
• Format: Journal Article
• Language: English
• Published: Netherlands Elsevier B.V 15.04.2024
• Subjects: 3D soil hydraulic maps; Machine learning; Multi-layered gridded information; Pedotransfer functions; Soil hydraulic conductivity; Soil water retention; Van Genuchten parameters; Soil water retention; Multi-layered gridded information; Van Genuchten parameters; Machine learning; 3D soil hydraulic maps; Soil hydraulic conductivity; Pedotransfer functions
• ISSN: 0048-9697; 1879-1026
• DOI: 10.1016/j.scitotenv.2024.171258

Spatially explicit, quantitative information on soil hydraulic properties is required in various modelling schemes. At European scale, EU-SoilHydroGrids proved its applicability in a number of studies, in ecological predictions, geological and hydrological hazard assessment, agri-environmental models, among others. Inspired by its continental antecedent, an analogous, but larger scale, national, 3D soil hydraulic database was elaborated for the territory of Hungary (HU-SoilHydroGrids) supported by various improvements (i–iv) in the computation process. Pedotransfer functions (PTFs) were built in the form of i) advanced machine learning methods and ensemble models, and trained on the ii) national soil hydrophysical dataset. The set of predictors used in PTFs was supplemented by iii) additional environmental auxiliary variables. Spatial layers of the soil hydraulic parameters were generated using iv) 100 m resolution information on primary soil properties, namely DOSoReMI.hu. HU-SoilHydroGrids provides information on the most frequently required soil hydraulic properties (water content at saturation, field capacity and wilting point, saturated hydraulic conductivity and van Genuchten parameters for the description of the moisture retention curve) with national coverage at 100 m spatial resolution down to 2 m depth for six GSM standard depth layers. The HU-SoilHydroGrids has significantly lower squared error in the case of describing the moisture retention curve and hydraulic conductivity than the EU-SoilHydroGrids. The derived 3D soil hydraulic database (ver1.0) is presently available in National Laboratory for Water Science and Water Safety for project partners in order to test its functional performance in describing hydrological and ecological processes. [Display omitted] •3D soil hydraulic database was derived for the territory of Hungary up to 2 m depth.•HU-SoilHydroGrids performs significantly better than EU-SoilHydroGrids.•PTFs were derived with four machine learning methods and four ensemble models.•Ensemble model performed better if only the best performing MLs were included.•The PTFs were applied on available environmental data to compute the maps.