Explore the vast range of titles available.

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.

Soils are widely recognized as a non-renewable natural resource and as biophysical carbon sinks. As such, there is a growing requirement for global soil information. Although several global soil information systems already exist, these tend to suffer from inconsistencies and limited spatial detail. We present SoilGrids1km--a global 3D soil information system at 1 km resolution--containing spatial predictions for a selection of soil properties (at six standard depths): soil organic carbon (g kg-1), soil pH, sand, silt and clay fractions (%), bulk density (kg m-3), cation-exchange capacity (cmol+/kg), coarse fragments (%), soil organic carbon stock (t ha-1), depth to bedrock (cm), World Reference Base soil groups, and USDA Soil Taxonomy suborders. Our predictions are based on global spatial prediction models which we fitted, per soil variable, using a compilation of major international soil profile databases (ca. 110,000 soil profiles), and a selection of ca. 75 global environmental covariates representing soil forming factors. Results of regression modeling indicate that the most useful covariates for modeling soils at the global scale are climatic and biomass indices (based on MODIS images), lithology, and taxonomic mapping units derived from conventional soil survey (Harmonized World Soil Database). Prediction accuracies assessed using 5-fold cross-validation were between 23-51%. SoilGrids1km provide an initial set of examples of soil spatial data for input into global models at a resolution and consistency not previously available. Some of the main limitations of the current version of SoilGrids1km are: (1) weak relationships between soil properties/classes and explanatory variables due to scale mismatches, (2) difficulty to obtain covariates that capture soil forming factors, (3) low sampling density and spatial clustering of soil profile locations. However, as the SoilGrids system is highly automated and flexible, increasingly accurate predictions can be generated as new input data become available. SoilGrids1km are available for download via http://soilgrids.org under a Creative Commons Non Commercial license.

80% of arable land in Africa has low soil fertility and suffers from physical soil problems. Additionally, significant amounts of nutrients are lost every year due to unsustainable soil management practices. This is partially the result of insufficient use of soil management knowledge. To help bridge the soil information gap in Africa, the Africa Soil Information Service (AfSIS) project was established in 2008. Over the period 2008-2014, the AfSIS project compiled two point data sets: the Africa Soil Profiles (legacy) database and the AfSIS Sentinel Site database. These data sets contain over 28 thousand sampling locations and represent the most comprehensive soil sample data sets of the African continent to date. Utilizing these point data sets in combination with a large number of covariates, we have generated a series of spatial predictions of soil properties relevant to the agricultural management--organic carbon, pH, sand, silt and clay fractions, bulk density, cation-exchange capacity, total nitrogen, exchangeable acidity, Al content and exchangeable bases (Ca, K, Mg, Na). We specifically investigate differences between two predictive approaches: random forests and linear regression. Results of 5-fold cross-validation demonstrate that the random forests algorithm consistently outperforms the linear regression algorithm, with average decreases of 15-75% in Root Mean Squared Error (RMSE) across soil properties and depths. Fitting and running random forests models takes an order of magnitude more time and the modelling success is sensitive to artifacts in the input data, but as long as quality-controlled point data are provided, an increase in soil mapping accuracy can be expected. Results also indicate that globally predicted soil classes (USDA Soil Taxonomy, especially Alfisols and Mollisols) help improve continental scale soil property mapping, and are among the most important predictors. This indicates a promising potential for transferring pedological knowledge from data rich countries to countries with limited soil data.

This study compared the efficiency of geostatistical digital soil mapping (DSM) with conventional soil mapping (CSM) for updating soil class and property maps of a cultivated peatland in the Netherlands. For digital soil class mapping, the generalized linear geostatistical model was used. Digital mapping of the soil organic matter (SOM) content and peat thickness was done by universal kriging. The conventional soil class map was created by free survey, while the property maps were created with the representative profile description (RPD) and map unit means (MUM) methods. For each method, we computed the effort invested in the mapping in terms of the sampling and cost densities. The accuracies of the created soil maps were estimated from independent probability sample data. The results showed that for DSM, the cost density could be reduced by a factor of three compared with CSM without compromising accuracy. The map purity of both maps was around 55%. For conventional soil property mapping, the MUM maps were more accurate than the RPD maps. For SOM, CSM-MUM (RMSE 7.5%) performed better than DSM (RMSE 12.1%), although accuracy differences were not significant. For peat thickness, DSM (RMSE 23.3 cm) performed slightly better than CSM-MUM (RMSE 24.9 cm). Despite the differences in accuracy being small, the digital soil property maps were produced more efficiently. The cost density was a factor of 3.5 smaller. We conclude that for updating conventional soil maps in the Dutch peatlands, geostatistical DSM can be more efficient, although not necessarily more accurate, than CSM.

This study compares the performance of five popular equal-area projections supported by Free and Open Source Software for Geo-spatial (FOSS4G)—Sinusoidal, Mollweide, Hammer, Eckert IV and Homolosine. A set of 21,872 discrete distortion vindicatrices were positioned on the ellipsoid surface, centred on the cells of a Snyder icosahedral equal-area grid. These indicatrices were projected on the plane and the resulting angular and distance distortions computed, all using FOSS4G. The Homolosine is the only projection that manages to minimise angular and distance distortions simultaneously. It yields the lowest distortions among this set of projections and clearly outclasses when only land masses are considered. These results also indicate the Sinusoidal and Hammer projections to be largely outdated, imposing too large distortions to be useful. In contrast, the Mollweide and Eckert IV projections present trade-offs between visual expression and accuracy that are worth considering. However, for the purposes of storing and analysing big spatial data with FOSS4G the superior performance of the Homolosine projection makes its choice difficult to avoid.

SoilGrids produces maps of soil properties for the entire globe at medium spatial resolution (250 m cell size) using state-of-the-art machine learning methods to generate the necessary models. It takes as inputs soil observations from about 240 000 locations worldwide and over 400 global environmental covariates describing vegetation, terrain morphology, climate, geology and hydrology. The aim of this work was the production of global maps of soil properties, with cross-validation, hyper-parameter selection and quantification of spatially explicit uncertainty, as implemented in the SoilGrids version 2.0 product incorporating state-of-the-art practices and adapting them for global digital soil mapping with legacy data. The paper presents the evaluation of the global predictions produced for soil organic carbon content, total nitrogen, coarse fragments, pH (water), cation exchange capacity, bulk density and texture fractions at six standard depths (up to 200 cm). The quantitative evaluation showed metrics in line with previous global, continental and large-region studies. The qualitative evaluation showed that coarse-scale patterns are well reproduced. The spatial uncertainty at global scale highlighted the need for more soil observations, especially in high-latitude regions.

Reducing emissions from deforestation and forest degradation plus the conservation of forest carbon stocks, sustainable management of forests and enhancement of forest carbon stocks in developing countries (REDD+) requires information on land-use and land-cover changes (LULCCs) and carbon emission trends from the past to the present and into the future. Here, we use the results of participatory scenario development in Tanzania to assess the potential interacting impacts on carbon stock, biodiversity and water yield of alternative scenarios where REDD+ is or is not effectively implemented by 2025, a green economy (GE) scenario and a business as usual (BAU) scenario, respectively. Under the BAU scenario, LULCCs will cause 296 million tonnes of carbon (MtC) national stock loss by 2025, reduce the extent of suitable habitats for endemic and rare species (mainly in encroached protected mountain forests) and change water yields. In the GE scenario, national stock loss decreases to 133 MtC. In this scenario, consistent LULCC impacts occur within small forest patches with high carbon density, water catchment capacity and biodiversity richness. Opportunities for maximizing carbon emission reductions nationally are largely related to sustainable woodland management, but also contain trade-offs with biodiversity conservation and changes in water availability.

Recent estimates show that one third of the world’s land and water resources are highly or moderately degraded. Global economic losses from land degradation (LD) are as high as USD $10.6 trillion annually. These trends catalyzed a call for avoiding future LD, reducing ongoing LD, and reversing past LD, which has culminated in the adoption of Sustainable Development Goal (SDG) Target 15.3 which aims to achieve global land degradation neutrality (LDN) by 2030. The political momentum and increased body of scientific literature have led to calls for a ‘new science of LDN’ and highlighted the practical challenges of implementing LDN. The aim of the present study was to derive LDN soil organic carbon (SOC) stock baseline maps by comparing different digital soil mapping (DSM) methods and sampling densities in a case study (Otjozondjupa, Namibia) and evaluate each approach with respect to complexity, cost, and map accuracy. The mean absolute error (MAE) leveled off after 100 samples were included in the DSM models resulting in a cost tradeoff for additional soil sample collection. If capacity is sufficient, the random forest DSM method out-performed other methods, but the improvement from using this more complex method compared to interpolating the soil sample data by ordinary kriging was minimal. The lessons learned while developing the Otjozondjupa LDN SOC baseline provide valuable insights for others who are responsible for developing LDN baselines elsewhere.

Soils are back on the global political agenda. Renewed interest of the soil resource is fuelled by an increasing awareness about the importance of sustainable soil management to secure production of food and fiber for a quickly growing world population, and about the major role of soils in the global carbon cycle. With this has come great demand for accurate, up-to-date and detailed geographical soil information. The current generation soil information systems typically store data from conventional surveys. Besides soil data at points, these systems contain soil maps that are often restricted to soil type; thematic soil maps are mostly missing. The maps are frequently outdated, lack detail and quantitative information on accuracy, or have no full spatial coverage. Consequently, these data are of limited use in today´s soil data applications.The Dutch soil information system BIS is no exception to this situation. The main source of soil information in the Netherlands, the nationwide 1:50 000 soil map, is becoming outdated, particularly for the areas with peat soils, and needs to be updated. Furthermore, maps of basic soil properties with quantified accuracy are lacking. Such maps are essential input for environmental process models that predict the effect of policy measures on for example soil acidification, pesticide leaching and greenhouse gas emission. Now the urgent need is felt to update the national soil map and to extend BIS with full-coverage thematic maps of all major soil properties with quantified accuracy. Efficient, quantitative methods for (geo)statistical modelling of soil maps, referred to as digital soil mapping (DSM), might be very useful for this purpose. Yet, despite growing global popularity DSM has not been applied in an operational way in the Netherlands so far. The main objective of this thesis is therefore to investigate and evaluate the merit of DSM for updating soil information in the Netherlands. Research focuses on DSM methods for updating soil type maps as well as maps of continuous soil properties. The province of Drenthe with large areas of peat soils is selected as case study area to illustrate and evaluate the developed methods.After the general introduction in Chapter 1, Chapter 2 describes a study on the possibility of updating the 1:50 000 soil map using a simple generalized linear regression Summary model and legacy soil point data from BIS. Map unit-specific multinomial logit models (MLM) were used to predict probability distributions of soil types within ten map units of the simplified soil map 1:50 000. For this purpose a framework for selecting an MLM was taken from the literature and adapted for soil mapping. Updating not only focused on peat soils but also on mineral soils to investigate if the purity of these map units could be increased through disaggregation with high-resolution covariates. Validation showed a modest 6% improvement in map purity compared to the existing, outdated soil map. This improvement was mainly attributed to better representation of soil distribution within the peat map units of the simplified map. However, map unit purities and class representations of the four peat soils as depicted on the updated map were still small.Digital soil type maps offer new possibilities for mapping individual soil properties. Chapter 3 describes the development of a model that exploits the information from such soil type map for spatial prediction of continuous soil properties. This model has important advantages compared to the conventional geostatistical model. First, actual (observed) soil type at sampling locations can be used as covariate instead of the mapped soil type. This has the advantage that the relationship between soil property and soil type is not confounded by impurities in the map units. Second, using actual soil type as covariate in the model makes it possible to quantify the proportion of the prediction variance that arises from uncertainty of the actual soil type at prediction locations. The developed model is applied to map the soil organic matter (SOM) content using the digital soil type map created in Chapter 2. Validation showed that the prediction performance of the proposed model was slightly better than that of the conventional geostatistical model.In Chapter 4 a method is proposed for three-dimensional mapping of SOM that combines general pedological knowledge with geostatistical modelling. A conceptual SOM depth profile was constructed by stacking building blocks (model horizons) for each soil type depicted on the updated digital map from Chapter 2. The vertical distribution of SOM within each building block was described by a function. The combination of building blocks—stacked in pre-defined order—with their associated parameters (thickness, average SOM content, exponential decay parameters) describes a soil type-specific depth profile. The parameters of each of these depth profiles were spatially predicted by geostatistical interpolation with covariates. A probability distribution of soil type-specific depth functions was then obtained by combining these predictions with the digital soil type map from Chapter 2. The depth functions and their associated probabilities were used to map the SOM stock for four depth intervals using the methodology described in Chapter 3. Validation of the predicted stocks with an independent probability sample showed accurate results for the topsoil. Results for deeper soil layers, however, were modest. Prediction performance of pedometric depth functions was comparable to that of conventional depth functions.The main drawback of the MLM, which was applied for soil type mapping in Chapter 2, is that spatial dependency in the data is not exploited for spatial prediction. Chapter 5 addresses this issue and investigates if a soil type map predicted by a spatial model is more accurate than one predicted by a non-spatial model. As spatial model the generalized linear geostatistical model (GLGM) was chosen. The GLGM is central to the methodological framework of model-based geostatistics, which is considered state-of-the-art in DSM. A pragmatic approach was adopted in which each of the five soil types in the case study area in the cultivated peatlands was modelled separately with a binomial logit-linear GLGM. Predictions with the soil type-specific GLGMs resulted in five binomial probabilities at each prediction location, which were scaled to multinomial probabilities so that they sum to one. Validation showed that use of a spatial model for digital soil type mapping did not result in more accurate predictions of soil type than those with the non-spatial MLM.Chapter 6 compares the efficiency of DSM methods with that of conventional soil mapping (CSM) methods for updating soil type and property maps. In addition, the effect of mapping effort (expressed in a monetary unit per ha) on accuracy is assessed for digital soil type and property maps. For digital soil type mapping the GLGM was used. For soil property mapping (SOM content en peat thickness) two methods are considered for both DSM and CSM. For DSM these are the method from Chapter 3 and the conventional geostatistical method (universal kriging). For CSM these are the representative profile descriptions (RPD) and map-unit-means (MUM) methods. For DSM both methods gave similar results in terms of accuracy. The MUM method gave better results than the RPD. For CSM the MUM method gave better results than the RPD. Validation results further showed that DSM produced soil type and property maps that were of similar accuracy as those produced by CSM. Furthermore, DSM maps were produced much more efficiently than the CSM maps: costs per hectare were a factor three to four smaller without compromising accuracy. This shows that for future updating of soil information DSM can be an attractive alternative to CSM. Finally, Chapter 7 presents a synthesis of the results and the main findings of Chapters 2 to 6. Implications of the results for the soil information system BIS and future updating of soil information in the Netherlands are discussed and an outlook on future research is given. It is argued that soil survey is shifting from conventional, qualitative soil survey to quantitative soil survey. This means that a toolbox with quantitative, state-of-the-art methods for soil mapping is not sufficient for effective and successful operational use of DSM. It requires the development of a next-generation soil information system based on new strategies and methods for collecting, storing, processing, visualizing and disseminating soil information. This thesis presents a first step on the road towards such system.

80% of arable land in Africa has low soil fertility and suffers from physical soil problems. Additionally, significant amounts of nutrients are lost every year due to unsustainable soil management practices. This is partially the result of insufficient use of soil management knowledge. To help bridge the soil information gap in Africa, the Africa Soil Information Service (AfSIS) project was established in 2008. Over the period 2008-2014, the AfSIS project compiled two point data sets: the Africa Soil Profiles (legacy) database and the AfSIS Sentinel Site database. These data sets contain over 28 thousand sampling locations and represent the most comprehensive soil sample data sets of the African continent to date. Utilizing these point data sets in combination with a large number of covariates, we have generated a series of spatial predictions of soil properties relevant to the agricultural management-organic carbon, pH, sand, silt and clay fractions, bulk density, cation-exchange capacity, total nitrogen, exchangeable acidity, Al content and exchangeable bases (Ca, K, Mg, Na). We specifically investigate differences between two predictive approaches: random forests and linear regression. Results of 5-fold cross-validation demonstrate that the random forests algorithm consistently outperforms the linear regression algorithm, with average decreases of 15-75% in Root Mean Squared Error (RMSE) across soil properties and depths. Fitting and running random forests models takes an order of magnitude more time and the modelling success is sensitive to artifacts in the input data, but as long as quality-controlled point data are provided, an increase in soil mapping accuracy can be expected. Results also indicate that globally predicted soil classes (USDA Soil Taxonomy, especially Alfisols and Mollisols) help improve continental scale soil property mapping, and are among the most important predictors. This indicates a promising potential for transferring pedological knowledge from data rich countries to countries with limited soil data.