Estimating model of soil total nitrogen content based on near-infrared spectroscopy analysis

Soil total nitrogen content is regarded as an important index for soil fertility diagnosis and crop growth management direction, the rapid monitoring model and establishing optimum bands on soil nitrogen content is the key to swiftly obtain soil nutrient information and develop precision agriculture. Five types of soil (Paddy soil, Fluvo-aquic soil, Salinized fluvo-aquic soil, Saline soil, Dark soil with lime concretion) collected from the middle and eastern of China were measured in the near-infrared region (4 000-10 000 cm-1). According to molecular vibration characteristics of full spectrum in the near infrared region, 8 wave bands (the whole region, combination region, the first overtone region, the second overtone region including the combination of bands referred above, and the protein functional group bands) were designed for establishing calibration models adopting multiplicative scatter correction(MSC) method. The calibration models for nitrogen content in soil samples were established by Partial Least Square (PLS) regression. The results showed that the prediction effect using MSC spectra in the combination region of 4 000-5 500 cm-1 was the best, the coefficient of determination and the root mean square error of cross-validation (RMSECV) for the calibration model were 0.90 and 0.16, respectively. Testing of the monitoring models with independent data of different types of soil samples indicated that determination coefficient and RMSE of validation were 0.91 and 0.15, respectively; RPD was 3.40. So, near-infrared diffuse reflection spectroscopy analysis technology can swiftly estimate total nitrogen content, and taking the combination band (4 000-5 500 cm-1) as modeling regional can get better calibration effect.