Informations et ressources scientifiques
sur le développement des zones arides et semi-arides

Accueil du site → Master → Canada → Probabilistic drought analysis

University of Manitoba Canada (1995)

Probabilistic drought analysis

DeWit, William J.

Titre : Probabilistic drought analysis

Auteur : DeWit, William J.

Université de soutenance : University of Manitoba,

Grade : M.Sc.1995

Drought conditions can have severe consequences for those enterprises that depend upon the availability of water. The operation of a water dependent system must, therefore, take into consideration the fact that these adverse conditions will certainly occur in the future. One method of dlought design is to use the drought of record as a design cliterion, as is done at Manitoba Hydro. A more theoretical approach would be to investigate the drought characteristics of the region in question. This, however, poses some difficulty since historic records are typically too short to adequately characterize droughts. This study uses synthetically generated data to investigate droughts in the Manitoba portion of the Nelson River basin. Droughts are censored from the historic and synthetic records using a theory of runs analysis which classifies droughts by their length, severity and magnitude. Exceedence probabilities and return periods are assigned to the events by applying the Weibull plotting position formula to the severity data. power generation during drought periods is also studied. The results of the analysis show that the historic record does not provide a good representation of drought events in the study area. The historic data ptoduces a return period of 79 years for the drought of record while the synthetic data gives a return period of 38 years. Using historic data alone produces return period estimates that are very conservative. Relationships between drought parameters and power generation levels showed a high degree of variability.


Version intégrale (7,33 MB)

Page publiée le 21 mars 2010, mise à jour le 28 janvier 2018