To make water quality monitoring networks useful for practice, the automation of data collection and data validation still represents an important challenge. Efficient monitoring depends on careful quality control and quality assessment. With a practical orientation a data quality assurance procedure is presented that combines univariate off-line and on-line methods to assess water quality sensors and to detect and replace doubtful data. While the off-line concept uses control charts for quality control, the on-line methods aim at outlier and fault detection by using autoregressive models. The proposed tools were successfully tested with data sets collected at the inlet of a primary clarifier,where probably the toughest measurement conditions are found in wastewater treatment plants.
Data quality assessment; On-line wastewater monitoring; Univariate methods
Main Research Area:
11th IWA conference on instrumentation control and automation, 2013