Establishment of Real-time Flood Prediction System by Data-Driven Model with X-band Radar
Abstract
The prediction of flood inundation is essential for urban safety with consideration of water-related disasters, and it is important to accurately predict inundation area and to disseminate the information to the residents for rapid evacuation. Since localized heavy rainfall is increasing due to the changes with climate patterns, a methodology of rapid urban flood prediction is required to reduce human and property damage. Furthermore, it will be helpful to enhance the ability for making flood defense strategies more efficiently.
In this study, the Nonlinear Auto-Regressive with eXogenous inputs(NARX) and Self-Organizing Map(SOM) model for flood prediction were applied to drainage districts in Seoul metropolitan city where the actual rainfall event that caused severe flooding in 2002, 2003 and 2010, repectively. The flood prediction based on data-driven model was trained and validated with the database of 1D and 2D flood analysis results. The NARX was applied for the prediction of flood volume, and this volume was linked with the expected inundation maps for real-time flood prediction. With the rainfall data which observed X-band radar was applied to NARX-SOM model for performing real-time flood prediction. The results of 2D hydraulic analysis were compared with the results of the flood prediction model in each drainage districts. The goodness of fit on the flood area was calculated for the result of prediction model, and the optimal expected inundation map for each basin was evaluated by clustering methods with SOM technique. The accuracy of flood prediction was analyzed according to the drainage system characteristics and the output dimensions of SOM. The results of this study are expected to provide basic process helpful in constructing rapid and accurate prediction models for urban flood.- Publication:
-
AGU Fall Meeting Abstracts
- Pub Date:
- December 2019
- Bibcode:
- 2019AGUFMNH31E0888K
- Keywords:
-
- 1821 Floods;
- HYDROLOGY;
- 1880 Water management;
- HYDROLOGY;
- 4313 Extreme events;
- NATURAL HAZARDS;
- 4328 Risk;
- NATURAL HAZARDS