Analyse des grilles de précipitation à haute résolution et des critères de calage pour la modélisation hydrologique au Québec

Abstract

The research presented in this master’s degree thesis is centered on the analysis of three highresolution gridded precipitation datasets that cover the entire province of Quebec. These gridded datasets are produced by Natural Resources Canada (NRCan), by Quebec’s Ministère du Développement durable, de l'Environnement et de la Lutte contre les changements climatiques (MDDELCC), and by Environment and Climate Change Canada (the Canadian Precipitation Analysis, CaPA). Operational precipitation gridded dataset from Rio Tinto that covers the Saguenay-Lac-Saint-Jean region is also considered. This study aims at determining the spatial and temporal differences in gridded precipitation datasets and their implications for hydrological modeling over the province of Quebec and over a specific region, here the Lac-Saint-Jean watershed. Three calibration criteria, namely the Nash-Sutcliffe Efficiency (NSE), the Kling-Gupta Efficiency (KGE), as well as a combination of criteria such as bias, root mean square error (RMSE) and the NSE are evaluated for calibrating hydrological models on various watersheds in the province of Quebec. Indicators such as the relative differences, correlation, ratio of variances, cumulative precipitation, etc., are used to determine the similarities and distinctions between precipitation datasets. The conceptual global model HSAMI is used to perform simulations on 181 watersheds in Quebec. The semi-distributed conceptual model CEQUEAU is also used for hydrological simulations, but only on the Lac-Saint- Jean watershed. Different calibrations of the HSAMI model are carried out with each of the three above-mentioned criteria on the 181 watersheds. Results indicate strong similarities between gridded precipitation datasets in the southern portion of the province and disparities in the center and northern portions. The analysis of hydrological simulations performed with HSAMI indicates that the CaPA dataset provides the best results, especially for the watersheds located in the center and north of the province. The MDDELCC dataset shows the best performance in watersheds located on the South shore of the St. Lawrence River and comes out as the overall second best option. The analysis of the simulations with CEQUEAU on the Lac-St-Jean watershed leads to the same conclusion. Comparing the simulated streamflow series obtained after calibration of the HSAMI model using the different calibration criteria mentioned above reveals that calibrating the model according to the KGE criterion provides a better match between the variances of simulated and observed streamflows.

Date	9 Nov 2016
Original language	French
Awarding Institution	École de technologie supérieure
Supervisor	Annie Poulin (Supervisor) & Marie Amélie Boucher (Co-supervisor)