5. CONCLUSIONS
This study focuses on the influence of the re-infiltration process for 100-year and 500-year flood events, which has so far not been well-recognized by the hydrologic/hydraulic modeling community. The sensitivity experiment and a 500-year Hurricane Harvey example both highlight the discernable differences between the with and without re-infiltration scheme. The major conclusions are summarized as follows:
1. In the 100-year design rainfall event, re-infiltration is found to make discernible differences with less flood extent (~1.56 km2), depth (~3 m), and dynamics (~4-hour delayed flooding and ~4-hour shorter inundation duration), compared to without re-infiltration. The 500-year Hurricane Harvey event shows a magnified difference in inundation duration up to 15 hours because of the longer event duration. However, the flood depth difference is less in the Harvey event due to the rapid saturation of the soils.
2. The hydraulic conductivity and antecedent soil condition from the designed sensitivity test are found to be the prime contributors to the difference between with and without re-infiltration, and comparatively, the antecedent soil moisture condition is the most sensitive among the three tested factors.
3. For the Harvey event, the differences are verified with stream gauge observations. On average, a 139.9% increase in NSE scores is found for re-infiltration with respect to without it. The improvements are mostly tied to better characterization of the recession limb after peak flow while the peak flows are well-captured by both. The proxy data – USGS High Water Marks – also indicate better performance with the inclusion of the re-infiltration scheme, as the re-infiltration scheme presents a 17.2% less flood depth difference than the case without the re-infiltration. The differences are further expected to enlarge for less intensive events and regions with a higher percentage of permeable soil media.
This study aims to raise attention to the important re-infiltration process in coupled H&H flood modeling to provide more accurate flood information, e.g., depth and timings. For future work, we will continue improving the current CREST-iMAP model framework by incorporating flood mitigation measures such as levees and dams into the system. Also, it is critical to couple with the NWP model to advance flood prediction lead time, which ensures more time for residents at risk to evacuate.