2019 CSCE Annual Conference - Laval (Greater Montreal) Conference
Dr. Van-Thanh-Van Nguyen, McGill University
Information on the variations of extreme rainfall events in space and time is essential for the design and management of different water resources systems. However, it is difficult in practice to obtain this information simply based on the available historical precipitation records due to the random behavior of these phenomena, especially in the context of climate change. Hence, statistical and stochastic approaches have been commonly used for describing more accurately the spatio-temporal variability of the precipitation process. In particular, in the context of climate change the statistical approach such as the popular SDSM method has been often relied on the physically unrealistic assumption that the statistical model parameters remain the same for current and future climates. Consequently, the use of a stochastic approach should be considered as more suitable in order to overcome this limitation of statistical methods. The main objective of the present study is therefore to develop an original stochastic model to represent the daily precipitation process in the context of climate change. The proposed model (referred herein as MCME- Markov Chain Mixed Exponential) consists of two components: (i) the first component representing the occurrences of daily rainfalls based on the first-order Markov Chain; and (ii) the second component describing the daily rainfall intensities using the Mixed Exponential distribution. The MCME model can generate synthetic daily rainfall series having the same statistical properties of the observed data. A comparative study was then carried out to assess the performance of the MCME as compared to the popular LARS-WG stochastic model, using NCEP re-analysis data and observed daily precipitation data available in stations across the province of Quebec, Canada. Both models are calibrated and validated for the period between 1961 and 1990 in consideration of different climate change scenarios. Results of this assessment have indicated the feasibility, accuracy, and robustness of the proposed MCME model as compared to the LARS-WG model using a set of common graphical and numerical performance criteria.