Estimating the likelihood of roadway pluvial flood based on crowdsourced traffic data and depression-based DEM analysis

Abstract. Water ponding and pluvial flash flooding (PFF) on roadways can pose a significant risk to drivers. Furthermore, climate change, growing urbanization, increasing imperviousness, aging stormwater infrastructure have increased the frequency of these events. Using physics-based models predict at road segment scale requires notable terrain simplifications detailed information that is often not available fine scales (e.g., blockage inlets). This brings uncertainty into results, especially in highly urbanized areas where micro-topographic features typically govern actual flow dynamics. study evaluates potential for flood observations collected from Waze – community-based navigation app estimate likelihood PFF scale. We investigated correlation reports with well-known maps, including National Flood Hazard Layer (NFHL), high watermarks, low water crossings data inventories. In addition, localized surface depressions their catchments are derived 1 m resolution bare-earth digital elevation model (BE-DEM) investigate spatial association reports. analysis showed highest exists local rather than river flooding, indicating they potentially useful indicators PFF. Accordingly, two data-driven models, empirical Bayes (EB) random forest (RF) regression, were developed proxy susceptibility, three classes historical storm events (light, moderate, severe) every depressions. Applying 150 storms city Dallas depression catchment drainage area imperviousness most important predictive features. The EB performed reasonable precision estimating number out 92 light, 41 17 severe 0.84, 0.85, 1.09 mean absolute errors, respectively. shows provide prediction. superior performance compared RF included approach more accurate