Topsoil pH (0-20 cm depth) at 100 m resolution, 5% and 95% quantile maps, covering the territory of 14 EU-MS countries, produced under the research of the task6.2 of the EJPSOIL project and related the scientific paper entitled "Digital Soil Mapping at European scale: comparing national vs. continental inputs"

Soil property mapping at the European scale currently relies on the Land Use/Cover Area (LUCAS) soil survey managed by the Joint Research Centre (JRC). Many Member States (MS) also maintain national soil monitoring systems (N-SIMS) that typically provide higher sampling densities. Technical issues such as harmonization of soil information and legal issues such as constraints in sharing coordinates have hindered the integration of national and European soil monitoring systems in Eu- wide mapping. 16 topsoil pH (0-20 cm depht) 5% and 95% quantile maps 100 m resolution are published here, covering the territory of 17 European Member States: Austria (not published), Belgium (Flanders not published), Czech Republic, Denmark, Estonia, France, Germany, Hungary, Italy, Latvia (not published), Lithuania, Poland, Portugal, Slovakia, Spain, Sweden and The Netherlands. The 8 maps were produced to evaluate the influence of alternative digital soil mapping approaches on soil pH predictions, using different combinations of European-wide and national datasets in terms of both soil point observations and covariates, and with different modelling approaches country-driven, continental top-down, and continental with federated learing. The mapping procedure used was Quantile Random Forest for all the 8 maps. The study was porformed under the activities of the task 6.2 of EJPSOIL (European Joint Programme Soil) H2020 project. In the table below are reported the file names with a description of the input data and modeling approaches used. Modelling approach Data Sites Auxiliary covariates Quartile File name National model (Country Driven) LUCAS 2009* Eu-wide covariates 5% quartile threshold PHW_EUCOV_LUCAS_COUNTRY_DRIVEN_Q05 National model (Country Driven) LUCAS 2009* Eu-wide covariates 95% quartile threshold PHW_EUCOV_LUCAS_COUNTRY_DRIVEN_Q95 National model (Country Driven) N-SIMS Eu-wide covariates 5% quartile threshold PHW_EUCOV_NSIMS_COUNTRY_DRIVEN_Q05 National model (Country Driven) N-SIMS Eu-wide covariates 95% quartile threshold PHW_EUCOV_NSIMS_COUNTRY_DRIVEN_Q95 National model (Country Driven) LUCAS 2009* & N-SIMS Eu-wide covariates 5% quartile threshold PHW_EUCOV_COMB_COUNTRY_DRIVEN_Q05 National model (Country Driven) LUCAS 2009* & N-SIMS Eu-wide covariates 95% quartile threshold PHW_EUCOV_COMB_COUNTRY_DRIVEN_Q95 National model (Country Driven) LUCAS 2009* Eu-wide and National covariates together 5% quartile threshold PHW_EUNATCOV_LUCAS_COUNTRY_DRIVEN_Q05 National model (Country Driven) LUCAS 2009* Eu-wide and National covariates together 95% quartile threshold PHW_EUNATCOV_LUCAS_COUNTRY_DRIVEN_Q95 National model (Country Driven) N-SIMS Eu-wide and National covariates together 5% quartile threshold PHW_EUNATCOV_NSIMS_COUNTRY_DRIVEN_Q05 National model (Country Driven) N-SIMS Eu-wide and National covariates together 95% quartile threshold PHW_EUNATCOV_NSIMS_COUNTRY_DRIVEN_Q95 National model (Country Driven) LUCAS 2009* & N-SIMS Eu-wide and National covariates together 5% quartile threshold PHW_EUNATCOV_COMB_COUNTRY_DRIVEN_Q05 National model (Country Driven) LUCAS 2009* & N-SIMS Eu-wide and National covariates together 95% quartile threshold PHW_EUNATCOV_COMB_COUNTRY_DRIVEN_Q95 Continental model (Top down) LUCAS 2009* Eu-wide covariates 5% quartile threshold PHW_EUCOV_LUCAS_CONTINENTAL_Q05 Continental model (Top down) LUCAS 2009* Eu-wide covariates 95% quartile threshold PHW_EUCOV_LUCAS_CONTINENTAL_Q95 Continental model (Federated learning) LUCAS 2009* & N-SIMS Eu-wide covariates 5% quartile threshold PHW_EUCOV_COMB_FEDERATED_Q05 Continental model (Federated learning) LUCAS 2009* & N-SIMS Eu-wide covariates 95% quartile threshold PHW_EUCOV_COMB_FEDERATED_Q95 * Flanders, Czech Republic, Poland, Portugal used LUCAS 2018