Evaluating river water quality modelling uncertainties at multiple time and space scales

Maintaining healthy river ecosystems is crucial for sustaining human needs and biodiversity. Therefore, accurately assessing the ecological status of river systems and their response to short and long-term pollution events is paramount. Water quality modelling is a useful tool for gaining a better understanding of the river system and for simulating conditions that may not be obtained by field monitoring. Environmental models can be highly unreliable due to our limited knowledge of environmental systems, the difficulty of mathematically and physically representing these systems, and limitations to the data used to develop, calibrate and run these models. The extensive range of physical, biochemical and ecological processes within river systems is represented by a wide variety of models: from simpler one-dimensional advection dispersion equation (1D ADE) models to complex eutrophication models. Gaining an understanding of uncertainties within catchment water quality models across different spatial and temporal scales for the evaluation and regulation of water compliance is still required. Thus, this thesis work 1) evaluates the impact of parameter uncertainty from the longitudinal dispersion coefficient on the one-dimensional advection-dispersion model and water quality compliance at the reach scale and sub-hourly scale, 2) evaluates the impact of input data uncertainty and the representation of ecological processes on an integrated catchment water quality model, and 3) evaluates the impact of one-dimensional model structures on water quality regulation. Findings from this thesis stress the importance of longitudinal mixing specifically in the sub daily time scales and in-between 10s of meters to 100s of meters. After the sub daily time scale, other biological and ecological processes become more important than longitudinal mixing for representing the seasonal dynamics of dissolved oxygen (DO). The thorough representation of the dominant ecological processes assists in obtaining accurate seasonal patterns even under input data variability. Furthermore, the use of incorrect model structures for water quality evaluation and regulation leads to considerable sources of uncertainty when applying duration over threshold regulation within the first 100s of meters and sub hourly time scale.