The main advantage of continuous water quality measurement systems is the ability to capture dynamics in water and wastewater systems, which allows for the identification of critical events, the evaluation of impacts on receiving water bodies, the identification of cause and effect relationships and the ability to discern trends. However, the challenge associated with automatic monitoring systems is the collection of data with sufficient quality for the intended application. That is, useful monitoring is dependent on cautious data quality assessment. With particular attention to its practical implementation, this paper presents a method for data quality assessment that attempts to extract useful information from individual water quality measurement time series. Based on forecasting techniques that make use of the historical behavior of the data, raw measurements are evaluated for the detection of doubtful data and outliers. Posterior treatment is then applied to remove noise and detect potential sensor faults. The proposed tool has been successfully tested on water quality time series collected from different water and wastewater systems.
Alferes, Janelcy; Copp, John; and Vanrolleghem, Peter A., "Forecasting Techniques Applied To Water Quality Time Series In View Of Data Quality Assessment" (2014). CUNY Academic Works.