Figure 1.
Sampling density of the LUCAS SOIL database per European territorial units, level 1 (NUTS 1).
Map labels give the total number of samples per country.
Table 1.
Number of samples (n), frequency (in% of the total number of samples) and surface (in % of the total surface) occupied by land cover type as defined in the LUCAS survey [31].
Table 2.
Number of samples (n), frequency (in% of the total number of samples) and surface (in % of the total surface) occupied by World Reference Base (WRB) major soil groups [30].
Table 3.
Summary statistics of soil properties available in the LUCAS database, for mineral and organic soil materials.
Figure 2.
Eigenvectors and eigenvalues of the first three principal components of continuum-removed spectra.
The principal component analysis has been realized separately for mineral (top panel) and organic (bottom panel) soil materials.
Table 4.
Performance of the best spectroscopic models as measured against the test set.
Figure 3.
Predicted SOC content as a function of observed SOC content in test sets.
Model predictions are shown for models with (rfe+aux, right panels) and without auxiliary predictors (spc, left panels).
Figure 4.
Mean reflectance (left scale) and continuum-removed reflectance (right scale) spectra of LUCAS mineral soil samples, computed for arbitrary sand and SOC classes.
The sand classes are 0–25%, 25–50%, 50–75%, 75–100% and the SOC classes are 0–25 g C kg−1, 25–50 g C kg−1, 50–75 g C kg−1, 75–200 g C kg−1. Each panel regroups samples of a given SOC interval.
Figure 5.
Mean reflectance (left scale) and continuum-removed reflectance (right scale) spectra of LUCAS mineral soil samples, computed for arbitrary clay and SOC classes.
The clay classes are 0–20%, 20–40%, 40–60%, 60–80% and the SOC classes are 0–25 g C kg−1, 25–50 g C kg−1, 50–75 g C kg−1, 75–200 g C kg−1. Each panel regroups samples of a given SOC interval.
Figure 6.
Relative Root Mean Square Error of Prediction (RMSEP) per land cover, for arbitrary classes of SOC and sand content.
The sand classes are 0–25%, 25–50%, 50–75%, 75–100% and the SOC classes are 0–25 g C kg−1, 25–50 g C kg−1, 50–200 g C kg−1. The relative RMSEP is the RMSEP divided by the mean of observed SOC values of models developed with (red bars) and without auxiliary predictors (blue bars). Each panel regroups mineral samples of a given SOC interval and land cover type. The number of training samples (n) for each class of SOC content is given in each panel.
Table 5.
Reproducibility of SOC estimates (g C kg−1; Eq. 6) of the reference method and the spectroscopic models with (rfe+aux) and without (spc) the use of auxiliary predictors.