A turbidity compensation method for COD measurements by UV–vis spectroscopy

In water quality detection by ultraviolet-visible (UV–vis) spectroscopy, the detection equipment faces diverse water bodies. It is easy to be interfered by turbidity caused by suspended solid particles (SSP), resulting in a nonlinear rise of the whole spectrum and a significant decrease in measurement accuracy of chemical oxygen demand (COD). A turbidity compensation algorithm for UV–vis spectroscopy based on Lambert-Beer’s law for fitting the absorbance caused by turbidity in the whole spectrum region is studied. According to the additivity of absorbance, a few absorbance spectra of turbidity were obtained by filtering and difference method. They are then used as base spectra to linearly fit the absorbance spectrum of any turbidity value water. Finally, turbidity compensation of all samples in the whole region is realized by the subtraction method. Compared with 350 nm proportional compensation (350 nm PC) and multiplicative scatter correction (MSC), the compensated spectra of the method proposed in this paper are in the best coincidence with the filtered spectra by a 0.45 um filter membrane. Partial least squares (PLS) method is used to establish the COD prediction model of the raw spectra and compensated spectra of the three methods. The results show that the Rpred of the proposed method is the biggest and the RMSEP is the smallest. The experimental results show that the proposed turbidity compensation method can effectively correct the absorption spectrum without affecting the absorption characteristics of water samples. Furthermore, improving the accuracy of COD measurement by UV–vis spectroscopy.