Research Article
Quick Determination of Soil Quality Using Portable Spectroscopy and Efficient Multivariate Techniques
Table 6
Modelling results from different variable selection models for soil quality parameters.
| Parameters | Models | Var | Wavelengths (nm), selected | Calibration (77) | Prediction (33) | R2 | RMSEC | Bias | R2 | RMSEP | Bias |
| N (%) | PLS | 331 | 740–1070 | 0.560 | 0.033 | 0.0038 | 0.777 | 0.026 | 0.004 | IPLS | 17 | 842–858 | 0.441 | 0.036 | −0.0002 | 0.413 | 0.038 | −0.005 | Si-PLS | 43 | 770–784 945–958 973–986 | 0.571 | 0.033 | −0.0005 | 0.699 | 0.033 | −0.013 |
| P (mg·kg−1) | PLS | 331 | 740–1070 | 0.763 | 0.789 | 0.0900 | 0.803 | 0.684 | 0.119 | IPLS | 17 | 1039–1054 | 0.698 | 0.882 | 0.0144 | 0.758 | 0.817 | 0.093 | Si-PLS | 45 | 768–781 894–907 950–963 1058–1070 | 0.779 | 0.765 | 0.0284 | 0.804 | 0.738 | 0.048 |
| K (cmolc·kg−1) | PLS | 331 | 740–1070 | 0.878 | 0.096 | 0.0110 | 0.878 | 0.107 | 0.018 | IPLS | 17 | 842–858 | 0.798 | 0.121 | −0.0019 | 0.732 | 0.134 | 0.016 | Si-PLS | 60 | 846–860 876–890 921–935 996–1010 | 0.910 | 0.083 | 0.0005 | 0.898 | 0.117 | −0.047 |
| Mg2+ (cmolc·kg−1) | PLS | 331 | 740–1070 | 0.711 | 0.550 | 0.0627 | 0.807 | 0.568 | 0.098 | IPLS | 17 | 1055–1070 | 0.770 | 0.502 | 0.0008 | 0.776 | 0.608 | −0.025 | Si-PLS | 53 | 768–781 824–836 967–979 1019–1030 | 0.778 | 0.492 | 0.0160 | 0.761 | 0.654 | 0.109 |
| Ca2+ (cmolc·kg−1) | PLS | 331 | 740–1070 | 0.783 | 2.439 | 0.2780 | 0.787 | 2.470 | 0.430 | IPLS | 17 | 774–790 | 0.736 | 2.652 | 0.0119 | 0.734 | 2.8916 | −0.9689 | Si-PLS | 60 | 756–770 801–815 936–950 981–995 | 0.826 | 2.228 | 0.0642 | 0.666 | 3.0219 | −0.4531 |
| pH | PLS | 331 | 740–1070 | 0.891 | 0.351 | 0.0401 | 0.785 | 0.3655 | 0.0636 | IPLS | 17 | 876–890 | 0.688 | 0.564 | 0.0407 | 0.500 | 0.5301 | 0.1165 | Si-PLS | 53 | 810–823 824–837 922–935 1019–1031 | 0.904 | 0.331 | −0.0004 | 0.646 | 0.4760 | −0.0369 |
|
|
Par: quality parameter, Var: selected variable. Bold values represents best options. |