| dc.description.abstract | Patterns in concentration-discharge (c-q) relationships in rivers can be used to describe the complex interactions and combined effects of catchment processes affecting sources, mobilisation, and the transport of contaminants. Long-term monitoring, routine sampling, and targeted storm sampling have demonstrated the complexity of temporal variability in c-q relationships on diurnal, event, seasonal and annual timescales. Furthermore, seasonal and interannual controls on catchment functioning are seen to result in pronounced differences in the behaviour of water quality parameters on an event-by-event basis, both spatially and temporally.
The use of high-frequency water quality monitoring in rivers is an integral part of intake protection for the management of drinking water treatment works. Due to their operational focus, these routinely collected data are not commonly used beyond real-time decision making. The purpose of this research was to test whether these data have additional value, beyond immediate operational use.
This thesis aims to demonstrate the potential to increase the value of routinely collected, high-frequency, water quality data. It first examines data across multiple parameters and from multiple sites, aiming to identify tools to aid in the detection and removal of errors and anomalies. It then documents the development and testing of a rule-based approach for identifying rainfall-runoff events from sub-hourly flow data. This approach aims to provide a repeatable method for generating metrics for use in further analysis. The potential to unlock value in routinely collected, high-frequency, water quality data is demonstrated through the analysis of both low-frequency and high-frequency archived data. This analysis investigates how antecedent conditions and seasonal variability affect water quality during rainfall-runoff events.
After utilising the methods developed for processing these water quality and flow data, it is evident that routinely collected high-frequency data display the dominant patterns in water quality behaviour, particularly for pH, conductivity, turbidity, and, to a lesser extent, colour (and so by proxy, DOC). Analysis of these data identifies catchment process-driven diel cycles; concentration and dilution effects during rainfall-runoff events; and hysteresis in the c-q relationship. Additionally, turbidity data identifies antecedent and event conditions influencing the lagged c-q relationship, including the effects of seasonal changes in the catchment response to a range of catchment and meteorological drivers. These data also represented changing interactions and effects due to extremes in seasonal patterns across different years.
A key recommendation of this thesis is that, building on the insights gained from low-frequency data collected for regulatory purposes, routinely collected high-frequency data should be used to support the investigation of previously unexplored or unidentified processes and pathways affecting water quality, and treatment costs. The added value these data can bring to targeted research should also be considered. It is acknowledged that these archived high-frequency data can be challenging to use. These challenges are not easy to overcome; they result from the limited metadata available, the variability in observations which arise from both real-world errors and artefacts, and the complex interactions of drivers and functions within river catchments. However, in catchments where routinely collected data are the only source of multi-annual high-frequency water quality observations, these data may be crucial (alongside low-frequency regulatory sampling) for investigating the drivers affecting short-term and long-term water quality dynamics. | en_GB |
