http://hdl.handle.net/10388/13260

The need to provide robust integrated models for defensible decision making is now critical for sustainable management of aquatic ecosystems. Uncertainties associated with results of integrated modelling studies render their implementation difficult for resource managers. Evidence suggests that use-inspired modelling studies can support sustainable resource management. The Qu’Appelle River Basin is currently experiencing water scarcity and quality issues due to growing agricultural, industrial, and domestic water demand. Furthermore, climate variability is anticipated to worsen the current state of the river. Future water allocation and apportionment across the basin are uncertain due to these issues. In this research, an integrated catchment-instream water quality model is developed and identified for the sustainable management of the Qu’Appelle River. Four research goals were pursued to aid the development of a sound modelling tool: a) to investigate dissolved oxygen variation under open water and ice-covered conditions, b) to assess parametric sensitivity of riverine, river-lake system, and lake configurations in summer and winter, c) to identify parameters and processes regulating different degrees of eutrophication and d) to identify the dominant processes controlling eutrophication: instream versus catchment. A new stepwise model development and identification framework was utilized in the development of the sound integrated tool to meet the overarching objective of this research. The approach couples top-down model development and Bayesian inference techniques in structuring and identifying internal mechanisms of aquatic systems. Model development progressed from a simple modified Streeter and Phelps model structure (Chapter 2) to an advanced eutrophication structure (Chapter 3). In Chapter 3, a novel instream modelling strategy was used to carefully integrate nutrients transport and transformation in the Qu’Appelle River. The approach was applied and validated for two stretches of the Qu’Appelle River. Mechanisms underlying different trophic states of the Qu’Appelle River were explored using a newly developed global sensitivity analysis tool (Chapter 4). Phased screening techniques incorporated into multiple Monte Carlo simulations, global sensitivity analyses, and Kolmogorov–Smirnov statistic tests were further employed to con-strain uncertain parameters and processes in the identifiability analysis of the integrated model (Chapter 5). Linked river-lake transport and reaeration processes that influence sediment oxygen demand (SOD) estimations were constrained and calibrated in Chapter 2. Study results showed that SOD20 rates gradually decreased from 1.9 to 0.79 g/m2/day in the riverine section of the system as urban effluent travelled through the river while a SOD20 rate of 2.2 g/m2/day was observed in the lakes. Seasonally, the modelled SOD rate increased three-fold in the lakes and the river in summer, as compared to winter periods. A global sensitivity analysis that bench-marked linked transport and nutrient transformations in river-lake systems with traditional river/lake eutrophication modelling showed that the influence of parameters on eutrophication mechanisms were more significant in the linked system than the river (up to 25-fold). Sensitive transformation parameters included the dissolved organic nitrogen mineralization rate, phytoplankton nitrogen to carbon ratio, phosphorus-to-carbon ratio, maximum phytoplankton growth rate, and phytoplankton death rate. The ability to properly capture river-lake connection, lake and backflow dynamics in the river-lake model may have accounted for this. The performance of eutrophication models for two stretches of the Qu’Appelle River measured using Variogram Analysis of Response Surface showed that diffuse loading has a significant influence on the middle Qu’Appelle River compared to the upper Qu’Appelle River. Some of the prevailing processes governing the eutrophic state in the upper Qu’Appelle River include organic matter oxidation, microbial respiration, phytoplankton productivity, mineralization, nitrification, nutrient uptake, denitrification, seasonality, dissolution, primary productivity, phytoplankton stoichiometry, and recycling. In the middle Qu’Appelle River, many processes including diffuse loading, mineralization, nitrification, primary productivity, phyto-plankton stoichiometry, nutrient uptake, and dissolution together sustain its hypereutrophic state. A multistage global sensitivity analysis of coupled catchment-scale nutrient export model and in-stream eutrophication model showed that catchment processes were, overall, more significant to the river’s water quality. This finding from this research provides a guide to managers of the basin as to where to invest money in terms of water quality monitoring and nutrient reduction. This study has demonstrated how top-down model development and Bayesian inference scheme can be used to limit error propagation in the development of integrated water quality models. The framework developed for this research provides an easy platform to continuously improve model structure and predictions based on available data and emerging mechanisms. With the co-evolution of the components of human-water systems in the Anthropocene, the proposed frame-work can be useful in handling emerging features of integrated systems.

http://hdl.handle.net/10388/13260