Contact for the resource
Extractable aluminium content (Al measured by Mehlich 3) in mg/kg (fine earth) at 2 depth intervals (0-20 cm and 20-50 cm) predicted using two sets of Africa soil profiles data. For details see published paper here below (Hengl T., G.B.M. Heuvelink, B. Kempen, J.G.B. Leenaars, M.G. Walsh, K.D. Shepherd, A. Sila, R.A. MacMillan, J. Mendes de Jesus, L.T. Desta, J.E. Tondoh, 2015. Mapping Soil Properties of Africa at 250 m Resolution: Random Forests Significantly Improve Current Predictions. PLoS ONE 10(6)
Extractable Aluminium (Al) content of the soil fine earth fraction in mg/kg (ppm) as measured according to the soil analytical procedure of Mehlich 3 and spatially predicted for 0-30 cm depth interval at 250 m spatial resolution across sub-Saharan Africa using Machine Learning (ensemble between random forest and gradient boosting) using soil data from the Africa Soil Profiles database (AfSP) compiled by AfSIS and recent soil data newly collected by AfSIS in partnership with EthioSIS (Ethiopia), GhaSIS (Ghana) and NiSIS (Nigeria as made possible by OCP Africa and IITA), combined with soil data as made available by Wageningen University and Research, IFDC, VitalSigns, University of California and the OneAcreFund. [Values M = mean value predicted]. For details see below for peer reviewed paper (T. Hengl, J.G.B. Leenaars, K.D. Shepherd, M.G. Walsh, G.B.M. Heuvelink, Tekalign Mamo, H. Tilahun, E. Berkhout, M. Cooper, E. Fegraus, I. Wheeler, N.A. Kwabena, 2017. Soil nutrient maps of Sub-Saharan Africa: assessment of soil nutrient content at 250 m spatial resolution using machine learning. Nutriënt Cycling in Agroecosystems 109(1): 77-102). Maps produced for the Environmental Assessment Agency (PBL), funded by the Netherlands government, in collaboration with the AfSIS and the Vital Signs projects.
Exchangeable aluminium (Al3+ measured in 1M KCl solution) in cmolc/kg (fine earth) at two depth intervals (0-20 cm and 20-50 cm) predicted using two sets of Africa soil profiles data. For details see published paper here below (Hengl T., G.B.M. Heuvelink, B. Kempen, J.G.B. Leenaars, M.G. Walsh, K.D. Shepherd, A. Sila, R.A. MacMillan, J. Mendes de Jesus, L.T. Desta, J.E. Tondoh, 2015. Mapping Soil Properties of Africa at 250 m Resolution: Random Forests Significantly Improve Current Predictions. PLoS ONE 10(6)
This harmonized dataset of derived soil properties for the world (WISE30sec) is comprised of a soil-geographical and a soil attribute component. The GIS dataset was created using the soil map unit delineations of the broad scale Harmonised World Soil Database, version 1.21, with minor corrections, overlaid by a climate zones map (Köppen-Geiger) as co-variate, and soil property estimates derived from analyses of the ISRIC-WISE soil profile database for the respective mapped ‘soil/climate’ combinations. The dataset considers 20 soil properties that are commonly required for global agro-ecological zoning, land evaluation, crop growth simulation, modelling of soil gaseous emissions, and analyses of global environmental change. It presents ‘best’ estimates for: organic carbon content, total nitrogen, C/N ratio, pH(H2O), CECsoil, CECclay, effective CEC, total exchangeable bases (TEB), base saturation, aluminium saturation, calcium carbonate content, gypsum content, exchangeable sodium percentage (ESP), electrical conductivity, particle size distribution (content of sand, silt and clay), proportion of coarse fragments (less than 2 mm), bulk density, and available water capacity (-33 to -1500 kPa); also the dominant soil drainage class. Soil property estimates are presented for fixed depth intervals of 20 cm up to a depth of 100 cm, respectively of 50 cm between 100 cm to 200 cm (or less when appropriate) for so-called ‘synthetic’ profiles’ (as defined by their ‘soil/climate’ class). The respective soil property estimates were derived from statistical analyses of data for some 21,000 soil profiles managed in a working copy of the ISRIC-WISE database; this was done using an elaborate scheme of taxonomy-based transfer rules complemented with expert-rules that consider the ‘in-pedon’ consistency of the predictions. The type of rules used was flagged to provide an indication of the possible confidence (i.e. lineage) in the derived data. Best estimates for each attribute are given as means and standard deviations (STD), as calculated for the sample populations that remained upon application of a robust data outlier detection scheme. Results of the analyses can be linked to the spatial data through the unique map unit (grid cell) identifier, which is a combination of the soil unit and climate class code. Most map units are comprised of up to ten different components; each of these with their own range of derived soil properties and associated statistical uncertainties. Estimates of global soil organic carbon (SOC) stocks to 200 cm are presented in the technical documentation as an example of possible application.