Urban Inundation Prediction using High Density Gauge Network and Weather Radar in Seoul, Korea

Over the past few years in Seoul, locally heavy rain has been occurred occasionally. It causes flash floods and landslides in downtown. In order to prevent these disasters, the WISE (a next-generation Weather Information Service Engine) project is proceeding for improvements of technology and infrastructure of urban and rural meteorological information services in Korea. The main objective of this project is to develop the urban flood and inundation simulation system using weather information.

As shown in Fig. 1, the procedure to estimate and forecast inundation map are consisted of following four steps.

Fig.1. Range of the Gwangdeok radar station (a); and (b) the weather gauge station in Seoul

 

First, high-resolution rainfall fields are produced by quantitative precipitation estimates (QPEs) and quantitative precipitation forecasts (QPFs) process to use as an input data of urban hydrological model. The study data are collected by Mt. Gwangdeok weather radar, 34 Automatic Weather Stations (AWS) of Korea Meteorological Administration (KMA), and 262 meteorological sensors of SK Planet Telecommunication Company (SKP) to estimate the QPEs, as shown in Fig. 2. Z-R relationship, Kriging, and Conditional merging method are applied in QPE process. Four types of QPE product considered. QPE1 is estimated by the ordinary kriging method based on data from KMA's rain gauge networks; QPE2 is similar as QPE1 but it uses a combination of the KMA's and SKP's rainfall gauge networks; QPE3 is involved by Mt. Gwangdeok weather radar using the Marshall-Palmer equation; QPE4 employs the conditional merging method for combining QPE2 and QPE3. The QPEs has 250-m spatial resolution and 10 minutes temporal resolution. In addition, we used an extrapolation method to predict the forecasted rainfall fields and initial field is given by QPEs.

Fig.2. Range of the Gwangdeok radar station (a); and (b) the rain gauge station in Seoul

 

Second, Storm Water Management Model (SWMM) is applied to analysis the surface flow and sewer system of urban area. The RUNOFF module of SWMM simulates the route of surface runoff to subcatchments after rainfall. The outputs of the RUNOFF module of SWMM is an inputs of the EXTRAN module. Then, the EXTRAN module simulates the movement of water in the sewer system by solving Saint-Venant equations. We setup the SWMM for Samsung area in Seoul, which was damaged by locally heavy rain in September 21, 2010 as shown in Fig. 3.

 

Fig. 3. The locations of estimated surcharge using predicted radar rainfall and the drainage network in Samsung area for SWMM

 

Third, when surcharging occurs, two-dimensional diffusion model is adopted to simulate inundation on the surface by routing overland flow with the sources coming from surcharging flow. This model is based on two-dimensional shallow water equations. A diffusion wave can simulate the drainage effect and can be applied to water flow and surface flow on a rough surface. The flow can be analyzed by using a momentum and continuity equations for inundation analysis with topographical map and grid network.

Finally, the result will be providing and displaying through the integrated weather service of WISE platform for Seoul area.  It is expected that urban flood information of WISE platform will be used to reduce life and property of urban flood disaster.

The applicability and accuracy of this system is ongoing at this stage. The accuracy of QPE and QPF products will be evaluated using k-fold cross-validation method for storm events in July 2013. The surface flow, surcharge and inundation in Samsung area will be simulated.