Hyperspectral reflectance models for predicting soil organic matter content in coastal tidal land area, northern Jiangsu

Based on the advantage of rapid and non-destructive testing of hyperspectral reflectance compared with conventional methods, the hyperspectral models for predicting soil organic matter (SOM) content of different pedogenic time by partial least square regression (PLSR) was conducted in this study. Besides original spectra, several spectral indices were also calculated, including first derivative reflectance spectra (FDR), inverse-log spectra (lg(1/R)) and band depth (BD). The root mean square error (RMSE) and correlation coefficient (r) were used to validate the models. The results show that there is a difference in spectral indices for construct optimal model. REF is the optimal index of building PLSR model, which RMSE and r is 2.7231 and 0.8701, respectively, for predicting SOM content of overall samples. FDR and BD are better index for predicting SOM content form in millennium and century scales, r is 0.9436 and 0.8770, respectively. Correlation analysis shows that visible light region, near infrared that centered in 1 400 nm, and ranged in 1 900–2 450 nm is optimal band for predicting SOM content.