Reevaluating Flood Risk Methodologies in Houston, Texas Using a 2D Hydraulic Flood Model

Abstract:

Floods are some of the most devastating natural disasters that occur every year and are expected to worsen under climate change. Here we show trends in extreme precipitation and an analysis of the validity of the rational method, used to estimate peak flows, in Houston, Texas. To accomplish this, comparisons of flooding between mid 20^th century to early 21^st century extreme rainfall estimates were completed using a 2D flood model, LISFLOOD-FP. We also compare flood estimates generated from two hyetographs, the frequency-based storm and a uniform rainfall distribution, an underlying assumption of the rational method. The LISFLOOD-FP source code was modified to allow for spatially varying infiltration rates as opposed to a single infiltration rate for the entire model domain. To illustrate the applicability of this analysis, the flood extents generated in this work were compared to United States Federal Emergency Management Agency (FEMA) flood hazard maps for Houston. We show that flood risk should be presented as non-stationary and identify sources of flood extent underestimation within current flood estimation methodologies: 1) assuming uniform daily rainfall rates; 2) omitting pluvial flood risk; 3) not accounting for a changing climate. We highlight the importance of assessing sub-daily, as well as daily, flood estimates to understand total flood risk. Assuming a constant rainfall rate, we find that the area flooded from the 100-year 6-hour storm is 18.5% greater than the area flooded from the 100-year 24-hour storm in Houston, Texas.