Abstract: NIR-Visible spectral reflectance of soil samples were measured using ASD2500 hyperspectral meter. The results indicated that the spectral reflectance curve shape of north-east black soil of Heilongjiang is different from fluvo-aquic soil of Henan and Shandong province. The former shows ascending in a slow curve but the latter in a steep curve in 600-900nm region. The correlation analysis indicated that organic matter (OM) content has negative correlation with spectral reflectance in 545-830nm at 0.05 significant level and in 580-738nm at 0.01 significant level. OM has positive correlation with first derivative spectral reflectance in 458-1069nm and 1166-1306nm at 0.01 significant level. Furthermore, the degree of correlation is stronger than original spectral reflectance. Different spectral characteristic parameters are selected including reflectance and its first derivation, reflected height and absorbed depth. The analysis results show that OM can be predicted using first derivative reflectance in 570-590nm and reflected height in 1280nm (P_Depth1280). By validating, the prediction models are practical and feasible. Different models can be made reference to each other in practice.

Key words: Spectral, Organic matter, Model