Replicating Hurricane Harvey’s flows and water levels was a challenge, but the 2D models will help prepare us for future extreme weather events
Hurricane Harvey was a record-setting, unprecedented storm with more than half of the average annual rainfall for the Houston region occurring in less than five days. And I happened to be living in Houston at the time. As a stormwater engineer with a passion for water resources, it was a devastating, yet fascinating, event to experience. Now I’m able to study and learn from Hurricane Harvey to help plan for future major flooding events.
The massive amount of flooding prompted studies of Hurricane Harvey in the 2D modeling environment. Replicating Hurricane Harvey’s flows and water levels was a challenge. Many of the stream gauges in the Houston area have existed for 30 years or less, and most of the gauges only record rainfall and water-surface elevation but not flow. This meant we had to use rainfall-runoff methods to reproduce flows and achieve accurate results.
Usually after a large storm, high-water marks are surveyed, but Hurricane Harvey produced so much rain that the surveyors couldn’t get to flooded areas to gather data—sometimes even after the water had gone down by several feet. Producing hydrologic and hydraulic models of Harvey in an unsteady model—with the added component of 2D areas—required creative thinking and problem-solving.
The intersection of Shepherd Drive and Memorial Parkway near Buffalo Bayou in Houston, Texas, during Hurricane Harvey.
How to calibrate models without all the data points
Choosing the proper rainfall-runoff methodology, approximating surface storage outside of 2D flow areas, and modifying hydrologic and hydraulic parameters while meeting applicable criteria and remaining within industry standards while calibrating to very limited data was especially challenging for Hurricane Harvey.
So how did we find the necessary sources of data?
For rainfall, we mainly used the existing rain gauges for smaller areas or gauge-adjusted radar rainfall for larger areas with more variability in precipitation. However, sourcing data points as calibration targets was an entirely different story. In some cases, we had only one high-water mark, or the nearest stream gauge was too far away to accurately match the data. So, we did what millennials do best: we went to the internet. The National Oceanic and Atmospheric Administration and National Aeronautics and Space Administration both had timestamped satellite images posted online, but the most unique data we sourced was drone videos. YouTube had several drone videos that were flown during the storm when brief respites from the rain occurred. Unknown to the drone pilots, these videos had very valuable data showing times and dates of flood levels. Relating the aerial views of the flooding to known landmarks and topographic data allowed us to obtain points for calibrating our models.
Drone videos taken during Hurricane Harvey provided valuable information; here is a shot from the 49-second mark of a drone video used for model calibration.
The model calibration cycle.
To build these 2D models, we used HEC-RAS (Hydraulic Engineering Center’s River Analysis System) and XP-SWMM (Stormwater Management Model) as the hydrologic and hydraulic modeling programs and LiDAR (Light Detection and Ranging) data to create virtual land surfaces in the models. Generally, when calibrating models to an actual storm, you begin by making changes to the hydraulics part of the model—changing roughness values, including minor energy losses, adjusting 1D/2D connections, etc., and then you move into hydrologic calibration, if necessary.
The struggle in calibrating to something as big as Hurricane Harvey was that it was hard to get the models to represent the physical amount of water that was on the ground. Harvey was unique because it rained irregularly for almost five days, which meant we needed to reproduce both the timing of the flooding on multiple days and the corresponding water levels. Using our understanding of the earth’s response to rainfall and the dynamics of water flowing in pipes, streams, and overland areas, coupled with creative thinking and collaboration with peers, we were able produce accurate models calibrated to Hurricane Harvey.
Modeling beyond the typical design storm and planning for the future
There are certain levels of storms that most municipalities and government agencies require for modeling related to infrastructure and development. These events are usually storms that have high chances of occurrence in any given year, with considerations for storms that, due to their magnitude, occur less often. But it’s important for cities and agencies to do extreme event modeling. We need to start looking beyond what we previously considered an extreme event and think about what could happen if there’s another Hurricane Harvey or a storm of such intensity that drainage infrastructure can’t respond quickly enough to prevent structure flooding (like Houston’s tax day flood in April 2016).
_q_tweetable:Although we can never fully mitigate flood risk, it’s time to invest in our future and protect our livelihoods to the best of our ability._q_We should be thinking about modeling these storms before an extreme weather event rather than after, like we typically do now. Since we have no way of knowing the shape and form of the next catastrophic storm, we can use recent extreme events to plan—especially since these events are happening more often. Hurricane Harvey and the tax day flood were different types of storms, but each resulted in thousands of flooded homes and loss of mobility. Models recreating these events can be used to evaluate existing drainage infrastructure and assist in planning and design to reduce risk of flooding during the next disastrous storm.
Statistically, another Hurricane Harvey-sized storm is unlikely, but with climate change seeming to increase the number of extreme events, it could happen again, and it could be worse.
While many are creating models for the Houston area due to the recent flooding events there, it’s important that municipalities and agencies around the nation begin evaluating extreme events for design and planning of infrastructure. Flooding causes more damage in the US than any other type of natural disaster—another example: the devastating floods and hundreds of levee breaches in the Midwest this year—and yet we are almost never prepared for it. Although we can never fully mitigate flood risk, it’s time to invest in our future and protect our livelihoods to the best of our ability. And we need to stop pretending that it won’t happen to us, because I can tell you personally: it will.