ABSTRACT One of the most prominent sources of error and uncertainty in water quality modelling results is the input data. In this study, data from three meteorological databases were used to test the performance of a water temperature model of Lake Diefenbaker: the data from Environment and Climate Change Canada (ECCC) had long-term quality control history (>20 years); the data from the AccuWeather had short-term quality control history (<10 years), and the data from the MeteoBlue database were modelled values. The CE-QUAL-W2 hydrodynamic and water quality model was used for this study. The model was calibrated by adjusting model coefficients controlling the amounts of measured solar radiation and wind that reach the surface of the water. The sensitivity results showed very similar performances, with slightly better performances (root mean square root difference of ± 0.1) with the ECCC data followed by the MeteoBlue data and thereafter by the AccuWeather data.

Abstract Algal simulations in many water quality models perform poorly because of oversimplifications in the process descriptions of the algae growth mechanisms. In this study, algae simulations were improved by implementing variable chlorophyll a/algal biomass ratios in the CE-QUAL-W2 model, a sophisticated two-dimensional laterally-averaged water quality model. Originally a constant in the model, the chlorophyll a/algal biomass ratio was reprogrammed to vary according to the nutrient and light limiting conditions in the water column. The modified model was tested on Lake Diefenbaker, a prairie reservoir in Saskatchewan, Canada, where, similar to many other lakes in the world, field observations confirm variable spatiotemporal ratios between chlorophyll a and algal biomass. The modified version yielded more accurate simulations compared to the standard version and provides a promising algorithm to improve results for many lakes and reservoirs globally.

In this study, we built a two-dimensional sediment transport model of Lake Diefenbaker, Saskatchewan, Canada. It was calibrated by using measured turbidity data from stations along the reservoir and satellite images based on a flood event in 2013. In June 2013, there was heavy rainfall for two consecutive days on the frozen and snow-covered ground in the higher elevations of western Alberta, Canada. The runoff from the rainfall and the melted snow caused one of the largest recorded inflows to the headwaters of the South Saskatchewan River and Lake Diefenbaker downstream. An estimated discharge peak of over 5200 m3/s arrived at the reservoir inlet with a thick sediment front within a few days. The sediment plume moved quickly through the entire reservoir and remained visible from satellite images for over 2 weeks along most of the reservoir, leading to concerns regarding water quality. The aims of this study are to compare, quantitatively and qualitatively, the efficacy of using turbidity data and satellite images for sediment transport model calibration and to determine how accurately a sediment transport model can simulate sediment transport based on each of them. Both turbidity data and satellite images were very useful for calibrating the sediment transport model quantitatively and qualitatively. Model predictions and turbidity measurements show that the flood water and suspended sediments entered upstream fairly well mixed and moved downstream as overflow with a sharp gradient at the plume front. The model results suggest that the settling and resuspension rates of sediment are directly proportional to flow characteristics and that the use of constant coefficients leads to model underestimation or overestimation unless more data on sediment formation become available. Hence, this study reiterates the significance of the availability of data on sediment distribution and characteristics for building a robust and reliable sediment transport model.