Method for outlier detection: a tool to assess the consistency between laboratory data and ultraviolet-visible absorbance spectra in wastewater samples

Reliable estimations of the evolution of water quality parameters by using in situ technologies make it possible to follow the operation of a wastewater treatment plant (WWTP), as well as improving the understanding and control of the operation, especially in the detection of disturbances. However, ultraviolet (UV)-Vis sensors have to be calibrated by means of a local fingerprint laboratory reference concentration-value data-set. The detection of outliers in these data-sets is therefore important. This paper presents a method for detecting outliers in UV-Vis absorbances coupled to water quality reference laboratory concentrations for samples used for calibration purposes. Application to samples from the influent of the San Fernando WWTP (Medellin, Colombia) is shown. After the removal of outliers, improvements in the predictability of the influent concentrations using absorbance spectra were found.