Water Quality Modelling of Buffalo Pound Lake

The highly variable climate of the Canadian Prairies causes high economic losses from floods and droughts. Prairie waterbodies can be ice covered over half of the year impacting water transfer capacities and influencing aquatic processes and water quality. Available winter studies show ice cover has a wide ranging influence on physical, chemical and biological processes. Water quality models are an emerging tool in the Prairies for understanding complex ecosystem responses. Water quality modelling has traditionally been focussed on open water periods with under-ice processes largely ignored during calibration and model simulations. Management plans based on model results applicable to just four or five months of the year will overlook water quality issues under ice that can be informed by modelling. This thesis presents the first application of a complex hydrological-ecological model CE-QUAL-W2 to Buffalo Pound Lake an impounded, cold polymictic, natural lake supplying the water needs of approximately 25% of the Saskatchewan population. Three research themes investigate if 1) water quality is driven more by the lake’s catchment area or by internal lake processes, 2) how future flow management and climate change will affect the water quality of the lake, and 3) how under-ice processes can be successfully represented in the CE-QUAL-W2 model. Five water quality variables are simulated: Chlorophyll-a, ammonia (ammonium, NH4+-N), nitrate (NO3-N), dissolved oxygen, and phosphate (PO4-P). This thesis is written in manuscript format. The first manuscript improves the predictive capabilities of the zero-order sediment compartment and adapts the model code to read a variable sediment oxygen demand rate in place of the existing fixed coefficient. A semi-automated calibration method finds an annual pattern between chlorophyll-a, summer oxygen demand and rate of winter decay. The second manuscript looks to improve the under-ice heat and light environment in the model by modifying the ice algorithm to incorporate a variable albedo rate. Simulated ice-off dates are found to be highly sensitive to the ending albedo value. Improvements to water quality predictions are limited by the connection of the ice and eutrophication modules in CE-QUAL-W2. A targeted monitoring program is suggested to reduce uncertainty with boundary data. The third manuscript tests the sensitivity of the model to catchment and in-lake boundary conditions. All five water quality variables are found to be most sensitive to modelled inflow discharge. The final chapter summarises the findings of the three manuscripts and presents a scenario based flow management analysis for discussion. This research finds the Buffalo Pound Lake model is most sensitive to catchment boundary data. Water quality in the lake may be impacted by changing inflows resulting from lake management decisions and climate change