Using a Faster Modeling Framework To Advance Flood Prediction in Texas

As flooding becomes an ever-growing risk in Texas, researchers are working to improve its prediction with a faster, more efficient modeling framework.

When studying complex flood-prone areas, small changes in rainfall location, soil moisture or river runoff can lead to dramatically different outcomes. Capturing this uncertainty with traditional probabilistic models requires substantial time and computing power — which is an issue when time is precious.

In response, researchers developed an ensemble-based probabilistic forecasting framework centered on the Super-Fast INundation of CoastS (SFINCS) model. SFINCS is a hydrodynamic model that simulates interacting coastal, river, and rainfall-driven flooding, and can do so in as little as four hours under certain computing conditions — instead of the days traditional probabilistic models take.

The researchers tested the framework using Hurricane Ike, which wreaked havoc on the Texas Gulf Coast in 2008. Model results were evaluated against observed tidal water-level, stream gauge height, and high-water marks. The team was able to forecast flooding one to three days before the hurricane made landfall and do so 16% more accurately than deterministic models.

Although the study focused on the Texas Gulf Coast, the framework could also be extended to flash-flood-prone inland watersheds such as the Texas Hill Country.

Research by Wonhyun Lee, research assistant professor; Alex Sun, senior research scientist; Bridget Scanlon, research professor
Bureau of Economic Geology
Research published May 2025 in Coastal Engineering

This story is part of the 2026 Jackson School Newsletter. Read it in the Newsletter

Back to the Newsletter