|Monday, May 31|
Validation of Canadian Arctic river temperature in a continental hydrologic model
* Michèle Rochette, University of Manitoba, Canada
Tricia Stadnyk, Canada
Masoud Asadzadeh, Canada
"Due to the climate change and a warming arctic (Government of Canada, 2019), surface water temperatures are expected to increase, having an unknown cumulative effect on the Canadian Arctic freshwater ecosystems. Since temperature is a major variable affecting fish spawning, this has serious implications on biodiversity, tying into the United Nations’ Sustainable Goal 14: Life Below Water (United Nations, 2015). As temperatures increase, biodiversity and ice cover will be affected in most rivers and streams, and therefore it is important to be able to properly predict stream temperatures. Using the third version of the HYPE model (HYdrologic Predictions for the Environment), stream temperature and river ice thickness are modelled in rivers contributing to the Nelson-Churchill and Mackenzie River basins that are parts of the Arctic basin. The model is currently being validated against historical water temperature and river ice data in these regions. Additionally, the performance of the routine is analyzed and compared against climatic and physiographic basin details. Long-term, there is uncertainty in the cumulative, interactive energy and mass-balance relationships between air temperature, runoff volume, and stream temperature. A non-linear relationship is expected between the surface water temperature and the air temperature. The routine is validated against existing stream temperature data, in order to improve our knowledge of stream temperature across the Canadian Arctic basins. This model is driven into the future under different climate scenarios to examine the effects of climate change stream temperature and river ice and to assess which sub-basins may be at higher risk for increases in river temperature and decreases in river ice formation. Government of Canada. (2019). Canada's Chaning Climate Report. Gatineau, QC: Government of Canada. United Nations. (2015). Sustainable Development Goals. Retrieved from Sustainable Development Goals Knowledge Platform: https://sustainabledevelopment.un.org/?menu=1300 "
Regional analysis of water temperature metrics for Atlantic salmon in Eastern Canada
* Olfa Abidi, INRS, Canada
A. St-Hilaire, Canada
T.B.M.J. Ouarda, Canada
C. Charron, Canada
C. Boyer, Canada
A. Daigle, Canada
"This study sheds light on the modeling of thermal regimes of rivers in 146 hydrometric stations located in Eastern Canada. To meet this objective, classical linear (MLR) and non-linear (GAM) regional estimation methods were tested and compared to estimate five thermal metrics related to maximum temperature and its date of occurrence, as well as the seasonal variability. Homogenous regions were also determined based on hierarchical clustering analysis (HCA), region of influence (ROI) as well as canonical correlation analysis (CCA) approaches. Then, the regional models MLR and GAM were applied within the delimited homogeneous regions. Also, they were compared with an application to all stations without delimitation of homogeneous regions. For each regional estimation model, a set of optimal explanatory variables were selected using a forward stepwise regression procedure. The results demonstrate that the non-linear GAM model was consistently better than the simpler multiple regression model (MLR) for estimating the five thermal metrics. Results also show that the best practice consists on delineating homogeneous regions before applying the regional GAM model. According to all performance criteria, delineation of regions with the fixed non-contiguous approach (HCA) is considered to be more flexible and to lead to better performances than neighborhood-based approaches (ROI and CCA). HCA benefits also the most from the coupling with the GAM model. "
ESTIMATION OF THE AREA OF POTENTIAL THERMAL REFUGIA USING THE GENERALIZED ADDITIVE MODEL AND MULTIVARIATE ADAPTIVE REGRESSION SPLINES: CASE STUDY OF THE STE-MARGUERITE RIVER
* André St-Hilaire, INRS, Canada
Thermal refugia in rivers are becoming critical habitat for ectotherm fish, including Atlantic salmon (Salmo salar). In this study, two statistical modelling approaches were used to estimate the areas of potential thermal refugia: the generalized additive models (GAM) and multivariate adaptive regression splines (MARS). GAM and MARS models were fitted independently for four sites on the Ste-Marguerite River, (Quebec, Canada). Model performances were evaluated using the leave-one-out cross validation (LOOCV) and the following criteria: the Akaike information criterion (AIC), root-mean-square error (RMSE), relative root-mean-square error (rRMSE), Nash-Sutcliffe efficiency coefficient (NASH), and finally the bias (BIAS). Using an array of thermographs deployed at the confluence of a cold tributary and the warmer main river stem, refugia were delineated at a daily time step. Model results indicate that the estimated areas are close to the surfaces interpolated using measurements, with both models and for all sites. Results suggest also that MARS provides a better performance than GAM in terms of forecasting and estimating the variability of the area of thermal refugia in all study-stations.
Comparing reference climatological datasets and their usefulness in water temperature modelling
* Philippe Gatien, ETS, Canada
Richard Arsenault, Canada
Jean-Luc Martel, Canada
André St-Hilaire, Canada
"Water temperature plays a major role on physical, chemical and biological processes within rivers. It is imperative that rivers downstream from dams be managed in a way that does not negatively impact existing flora and fauna, in order to preserve biodiversity. Proper water management that includes water temperatures must then be capable of predicting the thermal behavior a river, ideally with observed meteorological data. Unfortunately, this data is often costly to acquire or simply not readily available. This study aims to compare the capability of a the HEC-RAS hydraulic model’s thermal module by using multiple reference datasets to ascertain their ability to simulate water temperatures where observed data may not be available. The reference datasets used are both European Center for Medium Range Weather Forecast (ECMWF) reanalysis products: ERA5 and ERA5-Land. These are very similar datasets with the notable differences being grid density (0.25° for ERA and 0.1° for ERA5-Land). Data has been extracted from these datasets over the Nechako basin in British-Colombia, Canada, where the required observed data to run the thermal model was readily available from 2017 to 2020 and partially available from 2005 to 2014 allowing for an in-depth analysis of input data quality and sensitivity. Thermal model calibrations followed those of the hydraulic model, ensuring good mass balance throughout the analysis. Calibrations were performed on a yearly basis, and also for the 2017-2019 period with more complete data in an attempt to increase predictive capabilities for all datasets (observed and reanalyses). As expected, the observed data fared better when evaluating simulation accuracy using the mean absolute error (MAE). Interestingly, increasing the density of the grid did not seem to correlate with an increase in predictive capacity of the model. This may be due to the fact that ERA5-Land is a simulation driven by ERA5 directly. When analysing this data over the diurnal cycle, it is apparent that trends in the errors appear. Independently of the dataset used, the model seems to over/underestimate peak values on a regular basis. Therefore, different objective functions were tested during calibration in order to attribute more importance to peak values, and cross-validation on the 2005-2014 period were also performed to test the model’s robustness. Overall, the thermal model reacted well with all of the datasets. This study therefore demonstrates that reanalysis datasets, such as ERA5 and ERA5-Land, are an acceptable alternative to observed data when observations are not readily available. Furthermore, ERA5 seems to be a sufficiently dense grid, with ERA5-Land offering no real advantage during thermal simulations in the river of interest in this study. "