Evaluating Hydrologic Design under Stationary and Non-Stationary Assumptions

by Vasquez Juan Rivadeneira

Institution: University of California – Irvine
Year: 2016
Keywords: Hydrologic sciences; Climate change; Climate Change; Extremes; Freeboard; Non-Stationary; Return Period; Stationary
Posted: 02/05/2017
Record ID: 2097374
Full text PDF: http://www.escholarship.org/uc/item/456900hw


In a warming climate, the magnitude and/or frequency of precipitation is expected to change in some regions, leading to higher local flood risk. This issue has raised much needed attention in the engineering design and flood management communities, and has motivated discussion on guidelines for improving infrastructure design in a changing climate. The existing infrastructure design concept assumes a stationary climate indicating no significant change in statistics of extremes over time. In a changing climate, however, frequency and severity of extreme events might change - a notion known as non-stationarity. Typically, a freeboard is included in hydrologic design as safety factor to improve infrastructure resilience, which might fail to deliver the intended level of reliability in a changing climate. This study considers 5 locations in the Northeastern United States that have shown a non-stationary behavior over time in the annual streamflow maxima. First, a freeboard is estimated based on the stationary assumption, commonly used for engineering design. The freeboard assuming a non-stationary climate is then evaluated against the commonly used stationary assumption. The Non-stationary Extreme Value Analysis (NEVA) software, which utilizes Extreme Value Theory (EVT) framework and Bayesian inference for parameter estimation, is used to obtain Return Level-Return Period curves under both stationary and non-stationary assumptions. The 100-year streamflow estimates under both stationary and non-stationary assumptions are used as input into HEC-RAS software to simulate the flow and compare the change in freeboard. Results show an expected increase in the magnitude of streamflow for 100-year events when there is a positive trend in observations. Under a non-stationary assumption, the freeboard decreases approximately 16% on average in our 5 case studies. This indicates that for a bridge design case study, there will be a higher chance of overtopping because the freeboard is less than the initial design under a stationary assumption. Potential changes in freeboard and, hence, failure risk in a warming climate calls for improved design concepts that consider potential changes in extremes over time. However, incorporating non-stationarity in design concepts is a challenging task, and requires not only model development but also an adaptation strategy.