Fig. 1. Distribution of sampling points in the study area

Fig. 2. Comparison of soil organic matter contents of two duplicate sets of samples

Fig. 3. Soil spectral curves of soil horizons with different organic matter contents
(a): Horizon with the lowest soil organic matter content; (b): Horizon with the organic matter content close to average value;(c): Horizon with the highasu soil organic matter content

Fig. 4. Soil organic matter content and the differences between soil spectra
(a): The angle between spectrum of each test point and spectrum of each sample; (b): The angle between spectrum of each sample and spectrum of each horizon

Fig. 5. Correlogram of soil organic matter content to spectral reflectance of each sample

Fig. 6. Modeling results of PLSR
(a): Validation: RMSEP (Cross-validated using 236 leave-one-out segments); (b): Training: % variance explained

Fig. 7. Statistics of the independent random validation in PLSR prediction model
The red dotted line represents the evaluation result of LOOCV

Fig. 8. Estimated soil organic matter content (black) by the spectrum of each test point (a) and sample (b), and their mean value (green), standard deviation (blue) against the measured. In (b), the estimated are compared with the measured soil organic matter contents, while the other values are compared with the average soil organic matter content of two samples in each horizon

Table 1. Statistics of soil organic matter content(g·kg^-1)

Table 2. Statistics of differences in degree between the sample spectrum and reference spectrum

Table 4. Statistics of the error of soil organic matter content estimated by in situ soil spectra (by standard deviation)