4:30 PM - 4:45 PM
[AAS03-23] Extreme Flood Events in Japan: Impact-Based Hydrodynamic Forecasting Using Ensemble Precipitation Forcing
Keywords:Ensemble Flood Prediction, Impact-Based Forecasting, Integrated Land Simulator (ILS), Extreme Flood Events
Extreme weather events, such as Typhoon Hagibis (2019), have caused severe flooding in Japan, underscoring the need for advanced forecasting methods that integrate meteorological observations with hydrodynamic modeling. Accurately predicting flood extent and associated impacts remains challenging due to uncertainties in precipitation forecasts and complex flood dynamics in urban environments. This study develops an impact-based ensemble flood prediction framework, combining ensemble precipitation forecasts with high-resolution hydrodynamic modeling using the Integrated Land Simulator (ILS).
To assess flood simulation sensitivity to precipitation variability, 39-hour MEPS ensemble precipitation forecasts are incorporated as forcing data. A depth-damage curve, derived from insurance loss data from the 2015 Kanto-Tohoku flood, is applied to estimate building damages across different precipitation scenarios. An improved levee representation explicitly simulates hydraulic interventions influencing flood depth, inundation extent, and flow retention dynamics. AMeDAS precipitation observations serve as a baseline reference, ensuring realistic constraints on extreme event simulations.
ILS effectively reproduces large-scale inundation areas, demonstrating high consistency with GSI Japan’s observed flood extent maps. The simulation with levee protection improves accuracy in reproducing inundation patterns. In contrast, a comparison between observed and simulated affected buildings in the no-protection scenario yields an R² of -687.29, indicating significant overestimation of flood impacts. Incorporating levee adjustments improves correlation to R² = -16.82, particularly in the Kanto, Tohoku, and Chikuma River areas. However, biases in flood depth and urban inundation remain, especially in Tokyo, due to complex hydrodynamic interactions and high-density urban areas.
This study highlights the potential of ensemble precipitation forecasts in improving flood modeling under extreme weather conditions. Future work will focus on conducting ensemble hydrodynamic simulations, refining levee parameters, and integrating ensemble impact assessments to enhance predictive reliability and quantify flood damage uncertainties.
To assess flood simulation sensitivity to precipitation variability, 39-hour MEPS ensemble precipitation forecasts are incorporated as forcing data. A depth-damage curve, derived from insurance loss data from the 2015 Kanto-Tohoku flood, is applied to estimate building damages across different precipitation scenarios. An improved levee representation explicitly simulates hydraulic interventions influencing flood depth, inundation extent, and flow retention dynamics. AMeDAS precipitation observations serve as a baseline reference, ensuring realistic constraints on extreme event simulations.
ILS effectively reproduces large-scale inundation areas, demonstrating high consistency with GSI Japan’s observed flood extent maps. The simulation with levee protection improves accuracy in reproducing inundation patterns. In contrast, a comparison between observed and simulated affected buildings in the no-protection scenario yields an R² of -687.29, indicating significant overestimation of flood impacts. Incorporating levee adjustments improves correlation to R² = -16.82, particularly in the Kanto, Tohoku, and Chikuma River areas. However, biases in flood depth and urban inundation remain, especially in Tokyo, due to complex hydrodynamic interactions and high-density urban areas.
This study highlights the potential of ensemble precipitation forecasts in improving flood modeling under extreme weather conditions. Future work will focus on conducting ensemble hydrodynamic simulations, refining levee parameters, and integrating ensemble impact assessments to enhance predictive reliability and quantify flood damage uncertainties.