Flood hazard of a rural-urban catchment under spatiotemporally varying rainstorms

Abstract

<p>1. Introduction</p><p>Rural-urban catchments are susceptive to flood in the context of climate change and urbanization. Accurate and efficient assessment of flood hazard in rural-urban catchments are thus of scientific and practical significance. While a great amount of research attention has been paid to estimate catchment-scale flood responses from hydrologic point of view (typically the runoff at the basin outlet), those applying hydrodynamic models to derive the flood hazards distributions with time are relatively scarce. Yet, the spatiotemporal flood hazard distributions convey crucial information to both public safety as well as the understanding for catchment responses. Given these being aware, however, the role of <a>spatiotemporal variability</a> of rainstorms, a common but complex feature that controls the local flood hazards, is rarely explored under the framework of hydrodynamic models. This issue particularly arises when the model domain increases to a scale comparable with rainstorms.</p><p>2. Objectives</p><p>This study aims to explore the role of spatiotemporal variability mesoscale convective storms on catchment scale flood hazard by conducting a series of numerical experiments particularly for extreme events (larger than 2-year recurrence level).</p><p>3. Methods</p><p>A typical rural-urban catchment, the Upper Shenzhen River Basin, with an area of about 19 km² (comparable with mesoscale convective systems) is chosen as the study site. The historical storms are identified from the radar reflectivity data in the recent decade and are characterized in terms of mean and max core intensity, moving direction and speed, and the statistical information of these characteristics are employed to generate a pool of stochastic rain fields by a novel spatiotemporal rain field generator. A hydrodynamic model based on 2D Shallow water equations is developed and calibrated with streamflow in a gauge station against typical flood events.</p><p>4. Results</p><p>The model simulates the dynamic inundation processes for extensive rainstorm events sampled from the stochastic event pool. Preliminary results show that different temporal distribution of rainstorm events may lead to contrasting results of peak runoff and inundation area, the difference of which can be even larger than that between idealized uniform events with 10-year and 50-year return levels. The spatial variability yields smaller variability of maximum flood hazard compared to temporal variability. In the upcoming studies, we will depict the flood hazard distributions using a simple hazard index and carry out systematic sensitivity analysis on the contribution of various aspects of rainstorm characteristics on the hazard pattern. The effect of shape, area, as well as the slope of sub-watersheds on the sensitivity investigated above will also be explored.</p><p>5. Conclusions</p><p>This study highlights the importance of spatiotemporal variability in flood inundation modeling, and provides essential insights on how spatiotemporal patterns of flood hazards echo back to that of extreme rainstorms.</p>