IAHR Document Library

« Back to Library Homepage « Proceedings of the 35th IAHR World Congress (Chengdu, 2013)

Assessing the Uncertainties of Radar Rainfall Prediction and Runoff Simulation Parameter: Toward an Operational Ensemble Flood Forecasting in Urban River Basin

Author(s): Ratih Indri Hapsari; Satoru Oishi

Linked Author(s): Ratih Indri Hapsari, Satoru Oishi

Keywords: Flood prediction; Uncertainty; Ensemble; Urban River; Short-Term Prediction

Abstract: Urban river basins are vulnerable to floods due to intense rainfall, and hence posing a danger to communities. Non-structural flood measures, including forecasting and warning are indispensable to complement structural measures. Nonetheless, predictability of flash flood events is considered limited because of high spatial and temporal variability. In this research, uncertainties in the short torrential rainfall and its corresponding flood in urban river basin are explored. Uncertainty in rainfall prediction is assessed by developing ensemble prediction system. Initial condition of the translation model is perturbed based on advection vector incorrectness. An updating scheme is introduced to provide more reliable rainfall short-term prediction. Parametric uncertainty of the CDRMV3 distributed hydrological model is analyzed by perturbing the model parameters with scalar multiplier. Two sensitive parameters of the rainfall-runoff model are each perturbed five times within their physical bounds. The range of uncertainty generated by ensemble prediction and their potential for obtaining flood risk estimates is demonstrated. Understanding sources of uncertainties of complex dynamic systems and communicating the predictive uncertainties remain challenges in using ensemble operationally. Yet, the proposed system employing a fix number of members allows for a simple system, which is essential in the flash flood warning context. The approach is demonstrated through case studies in Sumiyoshi River Basin Japan, which passes through a highly urbanized area in Kobe City, Japan. Verified with observed data, the combined input-parameter ensemble could serve as a reliable and effective system for operational flood disaster prevention in urban area. The advantages of ensemble prediction over the single deterministic forecast are demonstrated. Uncertainty assessment in rainfall-runoff model parameters gives important information regarding the potential of flooding. The ensemble technique is less sophisticated and easier to be communicated to non-expert users in the operational systems. It could be a useful approach to manage the uncertainty in the flood short-term prediction in urban river basin and elevate the public security from the flash flood threat.


Year: 2013

Copyright © 2024 International Association for Hydro-Environment Engineering and Research. All rights reserved. | Terms and Conditions