Research Article
Estimating Salt Concentrations Based on Optimized Spectral Indices in Soils with Regional Heterogeneity
Table 5
Accuracy of cross-validation dataset.
| Predictive models | Calibration/validation dataset | n | R2Cal | RMSECal (g·kg−1) | R2Pre | RMSEPre (g·kg−1) |
| Cross validation of Ebinur Lake Basin | E-model (NDSIs) | Ebinur/Ugan-Kuqa | 15 | 0.82 | 3.82 | 0.40 | 10.23 | E-model (RSIs) | Ebinur/Ugan-Kuqa | 15 | 0.82 | 3.84 | 0.41 | 10.12 | E-model(Com) | Ebinur/Ugan-Kuqa | 15 | 0.83 | 3.79 | 0.43 | 9.68 | E-model (NDSIs) | Ebinur/Keriya | 15 | 0.82 | 3.82 | 0.37 | 11.96 | E-model (RSIs) | Ebinur/Keriya | 15 | 0.82 | 3.84 | 0.39 | 11.54 | E-model(Com) | Ebinur/Keriya | 15 | 0.83 | 3.79 | 0.41 | 10.89 | Cross validation of Ugan-Kuqa Oasis | U-model (NDSIs) | Ugan-Kuqa/Ebinur | 15 | 0.74 | 7.38 | 0.33 | 12.35 | U-model (RSIs) | Ugan-Kuqa/Ebinur | 15 | 0.77 | 10.50 | 0.34 | 12.78 | U-model(Com) | Ugan-Kuqa/Ebinur | 15 | 0.78 | 4.96 | 0.38 | 11.97 | U-model (NDSIs) | Ugan-Kuqa/Keriya | 15 | 0.74 | 7.38 | 0.30 | 11.45 | U-model (RSIs) | Ugan-Kuqa/Keriya | 15 | 0.77 | 10.50 | 0.40 | 11.35 | U-model(Com) | Ugan-Kuqa/Keriya | 15 | 0.78 | 4.96 | 0.42 | 11.14 | Cross validation of Keriya Oasis | K-model (NDSIs) | Keriya/Ebinur | 15 | 0.62 | 4.53 | 0.27 | 11.96 | K-model (RSIs) | Keriya/Ebinur | 15 | 0.63 | 4.22 | 0.29 | 11.84 | K-model(Com) | Keriya/Ebinur | 15 | 0.67 | 3.18 | 0.31 | 11.02 | K-model (NDSIs) | Keriya/Ugan-Kuqa | 15 | 0.62 | 4.53 | 0.28 | 12.57 | K-model (RSIs) | Keriya/Ugan-Kuqa | 15 | 0.63 | 4.22 | 0.29 | 12.58 | K-model(Com) | Keriya/Ugan-Kuqa | 15 | 0.67 | 3.18 | 0.29 | 12.31 | Revalidation of global model | Global model | Ebinur/Ugan-Kuqa/Keriya | 45 | 0.77 | 5.32 | 0.69 | 8.45 |
|
|