Mapping rootable depth and root zone plant-available water holding capacity of the soil of sub-Saharan Africa.

In rainfed crop production, root zone plant-available water holding capacity (RZ-PAWHC) of the soil has a large influence on crop growth and the yield response to management inputs such as improved seeds and fertilisers. However, data are lacking for this parameter in sub-Saharan Africa (SSA). This study produced the first spatially explicit, coherent and complete maps of the rootable depth and RZ-PAWHC of soil in SSA. We compiled geo-referenced data from 28,000 soil profiles from SSA, which were used as input for digital soil mapping (DSM) techniques to produce soil property maps of SSA. Based on these soil properties, we developed and parameterised (pedotransfer) functions, rules and criteria to evaluate soil water retention at field capacity and wilting point, the soil fine earth fraction from coarse fragments content and, for maize, the soil rootability (relative to threshold values) and rootable depth. Maps of these secondary soil properties were derived using the primary soil property maps as input for the evaluation rules and the results were aggregated over the rootable depth to obtain a map of RZ-PAWHC, with a spatial resolution of 1 km2. The mean RZ-PAWHC for SSA is 74 mm and the associated average root zone depth is 96 cm. Pearson correlation between the two is 0.95. RZ-PAWHC proves most limited by the rootable depth but is also highly sensitive to the definition of field capacity. The total soil volume of SSA potentially rootable by maize is reduced by one third (over 10,500 km3) due to soil conditions restricting root zone depth. Of these, 4800 km3 are due to limited depth of aeration, which is the factor most severely limiting in terms of extent (km2), and 2500 km3 due to sodicity which is most severely limiting in terms of degree (depth in cm). Depth of soil to bedrock reduces the rootable soil volume by 2500 km3, aluminium toxicity by 600 km3, porosity by 120 km3 and alkalinity by 20 km3. The accuracy of the map of rootable depth and thus of RZ-PAWHC could not be validated quantitatively due to absent data on rootability and rootable depth but is limited by the accuracy of the primary soil property maps. The methodological framework is robust and has been operationalised such that the maps can easily be updated as additional data become available.

SoilGrids250m: Global gridded soil information based on machine learning.

This paper describes the technical development and accuracy assessment of the most recent and improved version of the SoilGrids system at 250m resolution (June 2016 update). SoilGrids provides global predictions for standard numeric soil properties (organic carbon, bulk density, Cation Exchange Capacity (CEC), pH, soil texture fractions and coarse fragments) at seven standard depths (0, 5, 15, 30, 60, 100 and 200 cm), in addition to predictions of depth to bedrock and distribution of soil classes based on the World Reference Base (WRB) and USDA classification systems (ca. 280 raster layers in total). Predictions were based on ca. 150,000 soil profiles used for training and a stack of 158 remote sensing-based soil covariates (primarily derived from MODIS land products, SRTM DEM derivatives, climatic images and global landform and lithology maps), which were used to fit an ensemble of machine learning methods-random forest and gradient boosting and/or multinomial logistic regression-as implemented in the R packages ranger, xgboost, nnet and caret. The results of 10-fold cross-validation show that the ensemble models explain between 56% (coarse fragments) and 83% (pH) of variation with an overall average of 61%. Improvements in the relative accuracy considering the amount of variation explained, in comparison to the previous version of SoilGrids at 1 km spatial resolution, range from 60 to 230%. Improvements can be attributed to: (1) the use of machine learning instead of linear regression, (2) to considerable investments in preparing finer resolution covariate layers and (3) to insertion of additional soil profiles. Further development of SoilGrids could include refinement of methods to incorporate input uncertainties and derivation of posterior probability distributions (per pixel), and further automation of spatial modeling so that soil maps can be generated for potentially hundreds of soil variables. Another area of future research is the development of methods for multiscale merging of SoilGrids predictions with local and/or national gridded soil products (e.g. up to 50 m spatial resolution) so that increasingly more accurate, complete and consistent global soil information can be produced. SoilGrids are available under the Open Data Base License.