RAPID DETERMINATION OF NITROGEN IN OVERLYING RIVER WATER DURING INTERMITTENT AERATION USING INTERVAL PARTIAL LEAST SQUARES AND NEAR-INFRARED SPECTROSCOPY

Near-infrared spectroscopy analysis was used in combination with interval partial least squares method to rapidly determine the total nitrogen, ammonia nitrogen, nitrite nitrogen in overlying water during intermittent aeration. Near-infrared spectroscopy was used to obtain absorbance data of water samples. Interval partial least squares was used to choose spectral region and establish calibration models (interval partial least squares calibration model) of total nitrogen, ammonia nitrogen and nitrite nitrogen. The full-spectrum was 4000-12500cm(-1) divided into 20 sub-intervals with interval partial least squares method. The best spectral region of total nitrogen, ammonia nitrogen and nitrite nitrogen were respectively 7432-7824 cm(-1), 10465-10860 cm(-1) and 7854-8250cm(-1). The interval partial least squares calibration models were established in the selected sub-interval. The established calibration models are capable of better predictions. The interval partial least squares calibration model of total nitrogen indicated that prediction coefficient of the prediction value and actual value is 0.9554, with the root mean square error of prediction being 0.1147. The interval partial least squares calibration model of ammonia nitrogen showed that correlation coefficient of the prediction value and actual value is 0.9651, with the root mean square error of prediction being 0.05437. The interval partial least squares calibration model of nitrite nitrogen showed that correlation coefficient of the prediction value and actual value is 0.9616, with the root mean square error of prediction being 0.03688. The results indicated that the proposed method was rapid, accurate, non-polluted, and reliable in determining the concentration of total nitrogen, ammonia nitrogen and nitrite nitrogen in overlying river water during biological nitrogen removal.