iSDAsoil: soil total Carbon for Africa predicted at 30 m resolution at 0-20 and 20-50 cm depths

iSDAsoil dataset soil total carbon in permilles (g/kg) log-transformed predicted at 30 m resolution for 0–20 and 20–50 cm depth intervals. Data has been projected in WGS84 coordinate system and compiled as COG. Predictions have been generated using multi-scale Ensemble Machine Learning with 250 m (MODIS, PROBA-V, climatic variables and similar) and 30 m (DTM derivatives, Landsat, Sentinel-2 and similar) resolution covariates. For model training we use a pan-African compilations of soil samples and profiles (iSDA points, AfSPDB, and other national and regional soil datasets). Cite as:

Hengl, T., Miller, M.A.E., Križan, J. et al. African soil properties and nutrients mapped at 30 m spatial resolution using two-scale ensemble machine learning. Sci Rep 11, 6130 (2021). https://doi.org/10.1038/s41…

To open the maps in QGIS and/or directly compute with them, please use the Cloud-Optimized GeoTIFF ver….

Layer description:

• sol_log.c_tot_mehlich3_m_30m_*..*cm_2001..2017_v0.13_wgs84.tif = predicted soil total Carbon mean value,
• sol_log.c_tot_mehlich3_md_30m_*..*cm_2001..2017_v0.13_wgs84.tif = predicted soil total Carbon model (prediction) errors,

Model errors were derived using bootstrapping: md is derived as standard deviation of individual learners from 5-fold cross-validation (using spatial blocking). The model 5-fold cross-validation (mlr::makeStackedLearner) for this variable indicates:

Variable: log.c_tot 
R-square: 0.794 
Fitted values sd: 0.571 
RMSE: 0.291 

Random forest model:
Call:
stats::lm(formula = f, data = d)

Residuals:
     Min       1Q   Median       3Q      Max 
-2.70312 -0.16714 -0.00549  0.15691  3.01116 

Coefficients:
               Estimate Std. Error t value Pr(>|t|)    
(Intercept)    0.025841   0.032713   0.790 0.429570    
regr.ranger    0.902240   0.008462 106.619  < 2e-16 ***
regr.xgboost   0.066535   0.008145   8.169 3.18e-16 ***
regr.cubist    0.145730   0.006927  21.039  < 2e-16 ***
regr.nnet     -0.048957   0.013466  -3.636 0.000278 ***
regr.cvglmnet -0.075212   0.005556 -13.537  < 2e-16 ***
---
Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1

Residual standard error: 0.291 on 50140 degrees of freedom
Multiple R-squared:  0.7938,	Adjusted R-squared:  0.7938 
F-statistic: 3.861e+04 on 5 and 50140 DF,  p-value: < 2.2e-16

To back-transform values (y) to g/kg use the following formula:

g/kg = expm1( y / 10 )

To submit an issue or request support please visit https://isda-africa.com/isd…

iSDA is a social enterprise with the mission to improve smallholder farmer profitability across Africa. iSDA builds on the legacy of the African Soils information service (AfSIS) project. We are grateful for the outputs generated by all former AfSIS project partners: Columbia University, Rothamsted Research, World Agroforestry (ICRAF), Quantitative Engineering Design (QED), ISRIC — World Soil Information, International Institute of Tropical Agriculture (IITA), Ethiopia Soil Information Service (EthioSIS), Ghana Soil Information Service (GhaSIS), Nigeria Soil Information Service (NiSIS) and Tanzania Soil Information Service (TanSIS). More details on AfSIS partners and data contributors can be found at isda-africa.com