Determination of Total Nitrogen, Ammonia, and Nitrite in River Water by Near-infrared Spectroscopy and Chemometrics

ABSTRACT

Total nitrogen, ammonia nitrogen, and nitrite nitrogen were determined in river water during intermittent aeration by near-infrared spectroscopy with a back-propagation neural network. Near-infrared spectra were obtained for 138 samples. A total of 116 samples were used as the calibration set and the remainder as the test set. The spectral region was from 4000 to 12,500 cm⁻¹. Principal component analysis was used as a preprocessing method of the near-infrared spectra to eliminate redundant information. The six principal components were extracted through principal component analysis. Back-propagation neural network models of total nitrogen, ammonia nitrogen, and nitrite nitrogen showed that the correlation coefficients were 0.9816, 0.9783, and 0.9562, respectively, with root-mean-square error of cross validation values of 0.04735, 0.03689, and 0.03766. The results of the back-propagation neural network models of total nitrogen, ammonia nitrogen, and nitrite nitrogen indicated that the correlation coefficients were 0.9752, 0.9690, and 0.9524, respectively, with root-mean-square errors of prediction equal to 0.05763, 0.04537, and 0.04157. This study showed that the described approach accurately determined total nitrogen, ammonia nitrogen, and nitrite nitrogen. The concentrations of total nitrogen, ammonia nitrogen, and nitrite nitrogen were 0.64–54.25, 0.57–45.04, and 0.05–31.40 mg · L⁻¹ in river water, respectively.

KEYWORDS:

• Back-propagation neural network
• biological nitrogen removal
• intermittent aeration
• near-infrared spectroscopy
• principal component analysis