2019 CSCE Annual Conference - Laval (Greater Montreal)

2019 CSCE Annual Conference - Laval (Greater Montreal) Conference

A Fuzzy clustering approach to diagnosis of change in natural streamflow regime

Return to Session

Mr. Masoud Zaerpour, Concordia Univeristy (Presenter)
Mrs. Shadi Hatami, Concordia University
Dr. Javad Sadri, Concordia University
Dr. Ali Nazemi, Concordia University

Climate change have significantly changed natural streamflow regime in Canada. Understanding the regime shifts in natural streamflow is important in terms of effective water management and preparedness for facing potential threats to water, food and energy security in Canada. Conventional approaches to assess the change in natural flow regime consider one or more streamflow characteristics and apply statistical tests to extract the form and magnitude of change individually in the considered characteristics. Here we propose an alternative approach to diagnose the change in natural streamflow regime by characterizing the gradual shift in shape and variability around the long-term annual streamflow hydrographs, which inherently include multiple streamflow characteristics within. The proposed methodology is based on (1) considering a wide range of streamflow characteristics that together represent the long-term annual hydrographs and their associated natural variability; (2) clustering streamflow series based on these features into a set of physically-relevant streamflow classes; and (3) monitoring the gradual shift from one flow regime to others using a systematic approach. To test this framework, we consider the streamflow data from 106 natural Canadian streams during the common period of 1966 to 2010. Fuzzy clustering is used to classify the streamflow series during this period, in which all streamflow series are associated to all clusters with certain degrees of belongingness. We then focus on natural streams in the province of Quebec and measure the gradual departure in degrees of belongingness using moving windows and take this information as an indicator for regime shift in natural streamflow regime. Based on this methodology, we diagnose some significant changes in shape and variability of annual expected hydrograph in Quebec, which provides a holistic understanding of recent changes in natural streamflow regime throughout the province.