Table 1 List of variables provided via iSDAsoil and average accuracy performance based on the fivefold spatial cross-validation (R-square, RMSE and Concordance Correlation Coefficient). Extractable elements are based on Mehlich-3 method. *Statistics based on the ln-transformed values. See also Supplementary material for detailed summary statistics per variable.

From: African soil properties and nutrients mapped at 30 m spatial resolution using two-scale ensemble machine learning

Variable

Unit

Training samples

R-square

RMSE

CCC

Sand content

%

122,261

0.736

13.7

0.848

Silt content

%

122,223

0.640

8.92

0.780

Clay content

%

122,269

0.746

9.6

0.854

Bulk density, <2 mm fraction

g/cc

13,565

0.819

126

0.901

Carbon, organic

g/kg

122,457

*0.791

*0.369

0.883

Carbon, total

g/kg

50,140

*0.794

*0.291

0.820

pH in \(\hbox {H}_2\)O

133,378

0.818

0.459

0.900

Stone content

%

92,785

*0.709

*0.803

0.701

Effective Cation Exchange Capacity

cmol(+)/kg

66,380

*0.754

*0.417

0.860

Calcium, extractable

mg/kg

144,593

*0.840

*0.543

0.913

Iron, extractable

mg/kg

57,526

*0.817

*0.235

0.899

Potassium, extractable

mg/kg

139,122

*0.773

*0.509

0.872

Magnesium, extractable

mg/kg

136,681

*0.815

*0.497

0.898

Nitrogen, total

g/kg

99,249

*0.732

*0.197

0.845

Phosphorus, extractable

mg/kg

53,493

*0.486

*0.707

0.654

Sulphur, extractable

mg/kg

37,530

*0.548

*0.384

0.708

Zinc, extractable

mg/kg

39,344

*0.711

*0.375

0.831

Aluminium, extractable

mg/kg

63,551

*0.881

*0.321

0.937

Depth to Bedrock

cm

28,054

0.429

41.3

0.725