Performance Evaluation of Flood Level Estimation Method Using State-Space Model with Time-Series Monitoring Data

This paper proposes a method for estimating flood levels using a state-space model in order to figure out the time-series changes at locations without observation devices. The proposed method integrates water level observations and flood analysis simulation results from physical models to estimate accurate flood level and its expansion process in channels. The method uses observation time-series data from observation devices as monitoring data for flood levels, and compensates for the accuracy of conventional methods by correcting a flood analysis simulation with a state-space model to provide a highly accurate estimate of flood levels for the entire urban area. We apply this method to several flood events in Aichi Prefecture, Japan, to investigate its performance.