Analyse des apports horizon 2050 pour les bassins Churchill Falls, Manic-5 et Caniapiscau

Lorrain, Karine (2007). Analyse des apports horizon 2050 pour les bassins Churchill Falls, Manic-5 et Caniapiscau. Mémoire de maîtrise électronique, Montréal, École de technologie supérieure.

Prévisualisation

PDF Télécharger (6MB) | Prévisualisation

Résumé

Cette étude porte sur la représentation des apports d'eau futurs pour trois réservoirs du Québec. La recherche consiste à simuler les apports futurs des bassins par l'emploi du modèle hydrologique HSAMI. La méthode des deltas est utilisée. Seize scénarios de cinq modèles climatiques globaux pour l'horizon 2050 sont ajoutés aux données climatiques des bassins. La seconde étape consiste à intégrer le générateur de climat WeaGETS dans la modélisation des apports prévisionnels des bassins à l'étude. Son utilisation a permis d'établir que les résidus des températures maximales et minimales ne suivent pas une loi normale et qu'une transformation Box-Cox améliore leur simulation. Enfin, cette étude consiste à comparer les apports futurs des séries synthétiques de WeaGETS avec ceux générés par la méthode des deltas. Selon les résultats obtenus, WeaGETS amplifie la quantité d'eau maximale de la crue printanière et semble avoir du mal à reproduire les extrêmes pour les températures minimales.

Titre traduit

Analysis of horizon 2050 supplies to the churchill falls, Manic-5 and Caniapiscau basins

Résumé traduit

Climate changes have been the subject of a particular scrutiny by the international community. Indeed, research are conducted to assess the probability they are the cause of determining repercussions on the future of generations to come. As a research example, certain studies consist in establishing impacts that could generate drastic disturbances on natural resources, especially water. This study specifically establishes a general representation of future supplies to three reservoirs owned by Hydro-Quebec and located in Northem Quebec. They largely influence the production of hydroelectric power and the planning of their supply will greatly help improve water reserve management in order to prevent extreme seasonal events provoked by meteorological variations.

This research is divided into three phases. The goal aimed by the first experiment is to simulate future supplies to these reservoirs by the use of Hydro-Quebec's HSAMI hydrological model. The process first consists in adding to the basins' climate data (precipitations, maximum and minimum temperatures) a monthly variation produced by global climate models. The use of 16 scenarios from 5 climate models has allowed the analysis of seasonal flow trends, such as spring flood or summer low-water period.
Furthermore, these scenarios help to visualize the occurrence of these phenomena in comparison with historical data. Tests have mainly concentrated on the interpretation of future supplies during horizon 2050, that is between 2040 and 2069. Simulation results first demonstrate a tendency to precocious spring flood. Furthermore, its supply is on average inferior to the observed supply. This situation is explained, among others, by a strong increase of potential evapotranspiration during the spring season, provoked by warmer temperatures. We also notice a more heightened low-water period during the summer and fall and a superior quantity of water during the winter season.

The second phase of the study consists in verifying the efficiency of the climate generator in the modelling of planned reservoir supplies. Indeed, this tool allows the production of synthetic series of climate variables over long periods. By adding monthly variations generated by a global climate model scenario to observed data, it is possible to simulate several series of statistically identical data, but with different daily variations. The use of this experimental climate generator has allowed to notice a more or less accurate speculation, that is, to confirm that maximum and minimum temperature residues follow a normal distribution. In fact, the issue is that the generation of these variables lies on this theory and that calculations have been designed as such. To improve their simulation and make the starting hypothesis true, a Box-Cox transformation has been used. By comparing synthetic series produced by this approach and those generated according to the original technique, we notice an improvement of the representativeness of these variables, especially for minimum temperatures. The use of the Kolmogorov-Smimov test as a comparison tool has allowed these observations. There are still research to be conducted on this new procedure, specifically concerning the simulation of maximum temperature inferior to minimum temperature for a given period.

The last phase of this study has been to compare supplies coming from the original climate generator's synthetic series to those generated by data observed for the Churchill Falls basin. According to results, the climate generator contributes to considerably amplifying the maximum quantity of water emerging during the spring snow melting period. Differences, of a value exceeding 1 000 m3 , are effectively produced by the use of the WeaGETS generator. To analyze this statistical difference, various tests have been conducted. The first experiment has been to compare each annual hydrograph in its ensemble, both for real, reconstructed and simulated supplies. This test has allowed the interpretation of their statistical tendency and thus deduce that simulated hydrographs present less dispersion between them year after year. The second analysis was to evaluate the source of the synthetic flood reproduction problem. After comparison of standard differences for the three climate variables used (precipitations, maximum and minimum temperatures), we notice a less significant dispersion of simulated minimum temperature values compared to observed data. Therefore, we can assert that the climate generator seems to have difficulty to reproduce extremes for minimum temperatures. In order to prove this assertion, two hybrid series have been produced. The first is constituted of observed temperature data and simulated precipitation values. The hydrograph created by the use of this series presents great similarities with the hydrograph produced with observed data. The second test has been conducted using a combination of simulated temperatures and observed precipitations. The hydrograph generated after the use of the HSAMI model effectively demonstrates that the synthetic series of temperatures greatly influence the amplitude of spring floods.

As another experiment, the Box-Cox modified generator has been used. Once more, the hydrograph produced with the synthetic series presents a far more important flood than the one presented with reconstructed supplies. However, after having conducted the same analyses performed for the original climate generator, we notice that results seem to improve the representativeness of reconstructed series. Indeed, even though values are still high, the dispersion of variables is better reproduced.

Finally, climate data representing horizon 2050, created with WeaGETS/Box-Cox, have been utilized to compare the hydrograph following the one resulting from the addition of the HadCM3a model's monthly deltas to observed data. We notice that the hydrograph simulated with WeaGETS respects the precociousness of the reconstructed horizon 2050 flood, but not the planned quantity of water. By using the same analyses as before, we note that the non- representativeness of minimum temperatures has a considerable impact on results. Thus, these experiments have allowed understanding the source of the problem and asserting that the Box-Cox theory cannot solve it alone.

Type de document:	Mémoire ou thèse (Mémoire de maîtrise électronique)
Renseignements supplémentaires:	"Mémoire présenté à l'École de technologie supérieure comme exigence partielle à l'obtention de la maîtrise." "par Lorrain, Karine" -- p. de t. Bibliogr. : f. [158]-176.
Mots-clés libres:	2050, Analyse, Apport, Bassin, Box, Box-Cox, Caniapiscau, Churchill, Climat, Cox, Falls, Generateur, Horizon, Hydrologie, Manic-5, Procedure, Stochastique, Temperature, Versant
Directeur de mémoire/thèse:	Directeur de mémoire/thèse Leconte, Robert
Codirecteur:	Codirecteur Brissette, François
Programme:	Maîtrise en ingénierie > Génie
Date de dépôt:	11 avr. 2011 14:55
Dernière modification:	30 nov. 2016 22:42
URI:	https://espace.etsmtl.ca/id/eprint/585

Gestion Actions (Identification requise)

Dernière vérification avant le dépôt