RTI uses cookies to offer you the best experience online. By and clicking “accept” on this website, you opt in and you agree to the use of cookies. If you would like to know more about how RTI uses cookies and how to manage them please view our Privacy Policy here. You can “opt out” or change your mind by visiting: http://optout.aboutads.info/. Click “accept” to agree.

Accept
Insights

Appreciating Hydrologic Uncertainty to Support Risk-Informed Decisions

Aerial View Flooding

Nearly a decade ago, I was trying to explain the concepts of hydrologic uncertainty and risk quantification to a hydropower manager - I don't remember the exact words, but his response was something close to, "… why would I want to add uncertainty to an answer?! Just give me a single answer!" At the time, I didn't have a good response, but the conversation has stuck with me over the years, especially as we enter a new paradigm for greater acceptance of probabilistic risk methods. As Earth Day 2021 approaches, I want to emphasize that the uncertainty associated with water management, which is further amplified by global climate change, should be embraced for risk-informed decision-making.

Let us start by setting the stage for risk-informed decision-making. Let's say that you want to build a bridge over a river, and you want to design the bridge so that it isn’t impacted by any events less than the 100-year flood event. Being the diligent engineer, you find the 100-year flood elevation, and set the bottom of the deck a fraction of a millimeter higher than this elevation (to be conservative, of course) - is your bridge safe? You followed the directions and designed it precisely, so there shouldn't be any issues, right?

We all recognize that this doesn’t make sense because our estimation of the 100-year flood event may be incorrect. The design elevation is uncertain, thus it actually becomes a question of risk tolerance. How important is it that the bridge not be impacted at the 100-year flood? What are the consequences of it not passing the flow? For these and other reasons, additional factors of safety, such as a freeboard allowance, may be necessary.

The bridge design may be set to the expected (in a statistical definition) level, which means it has approximately a 50% chance of being flooded! Not only should the bridge be designed for an exceedance probability, but we actually need to understand both the uncertainty around the estimate, and your level of risk tolerance in order to make better decisions. In fact, the real problem actually may need to consider not just a single design frequency level, but instead the uncertainty around MANY flood frequencies (a 50-year flood may actually impact your bridge)!

This example is simplistic but highlights the importance of appreciating hydrologic uncertainty to support risk-informed decisions. It is better to estimate the uncertainty during the design process than to make a foolish and overly confident decision, or in the other direction, to be excessively conservative resulting in undue costs and an overly designed system. Risk-informed decisions are aimed at striking a balance between these extremes.

Estimation of hydrologic hazards annual exceedance probability (AEP), such as a river flow or stage, may involve the use of available streamgage records, hydrologic modeling, paleo-flood studies, or other methods, all aimed at estimating the return period of an extreme event. Methods such as in Bulletin 17C (England et al., 2018) and deployed by the Corps of Engineers (COE) HEC-SSP application, can help estimate frequency curves by integrating some of these data sources, and can estimate uncertainty in the frequency. A newer tool developed by the COE Risk Management Center (RMC BestFit) uses Bayesian methods for frequency estimation with an added benefit of including expert knowledge into the prior distributions when available. In many cases, due to lack of data, new storage regulation, the need to estimate more extreme events, or even the impacts of climate change, historical records may be insufficient to reliably estimate the contemporary AEP of extreme events. Simulation modeling, such as through Stochastic Event Flood Modeling (SEFM), can support hydrologic hazards estimation for risk-informed decision-making.

Regardless of the method used, appreciate and embrace the uncertainty in hydrologic processes, consider the potential impacts to the expected return frequency of contemporary (and future) expectations of extreme events within the design life, and make better data-driven, risk-informed decisions around hydrologic hazards.

What would I say to my colleague now about wanting a single design value? I might start by simply asking what he wants to accomplish and how much risk he is comfortable accepting.

Learn more about our Rapid Risk Suite of tools for estimating key components of risk.

References: England, John F. / Cohn, Timothy A. / Faber, Beth A. / Stedinger, Jery R. / Thomas, Wilbert O. / Veilleux, Andrea G. / Kiang, Julie E. / Mason, Robert R. 

Guidelines for Determining Flood Flow Frequency - Bulletin 17C 

2018, No. U.S. Geological Survey Techniques and Methods, book 4, chap. B5 United States Geological Survey, United States Geological Survey, 

 

Sign up for the CWR Monthly Email to stay up-to-date on our latest insights and project work.