Successful modelling of river dissolved oxygen dynamics requires knowledge of stream channel environments

Water quality in lowland rivers is sensitive to changes in flow during summer dry periods, when high temperatures and low pollutant dilution are problematic and may reduce oxygen concentrations to levels of ecological concern. A 10-year period of monitoring data was collated for a typical small lowland UK river. Two hourly-resolution applications of a process-based water quality model (QUESTOR) were made, with and without local knowledge, to establish whether specific information on stream channel hydraulics is an essential precursor to successful simulation. Results showed this information to be necessary, with considerably better goodness-of-fit statistics obtained when the local knowledge was used. In this regard, mean improvements in Nash-Sutcliffe Efficiency across all monitoring sites were from -0.33 to 0.18 and from 0.24 to 0.78 for dissolved oxygen and water temperature respectively. Percent bias was within 10% for the local model. The 10-year record also allowed a detailed characterisation of how changes in flow, as described by a comprehensive range of Indicators of Hydrological Alteration, relate to the water quality determinants. Analysis revealed these dynamics were also captured more realistically when the model was driven by local knowledge. The research concludes that river dissolved oxygen simulations driven by national-level information are of some value as screening tools, but model refinement supported by sufficient provision of local information is necessary when detailed simulations are required to support specific decision-making.

Automated summary

The study found that local hydrological information is essential for successful simulation of river water quality. The use of local knowledge significantly improves goodness-of-fit statistics in simulating dissolved oxygen and water temperature. This highlights the importance of incorporating local information for accurate water quality modeling.