Megumi Watanabe

JSPS Postdoc

The University of Tokyo

Dr. Megumi Watanabe is a Japan Society for the Promotion of Science (JSPS) Postdoctoral Fellow at Institute of Industrial Science of the University of Toyo. Her research develops advanced flood forecasting and applications for disaster prevention and mitigation and a water body map as fundamental earth science data. Before joining the University of Tokyo, she was a project researcher at Tokyo Institute of Technology, where she investigated the relationship between extreme precipitation and temperature rise using observation, models, and a theory. She completed a PhD in Information Science and Engineering at Tokyo Institute of Technology, investigating climate change impacts on sparsely observed High Mountain Asian glaciers and their uncertainty using glacio-hydrological model and remote sensing techniques. Throughout her research activities, she has contributed to revealing climate change impacts on mountainous hydrology and extreme hydrological phenomena and realizing social implementation.

Importance of Two-way Communication with Users for Effective Utilization of Ensemble Flood Forecasts

Ensemble flood forecasts have a long lead time and provide probabilistic information. They are expected to be used for early hazard warnings. However, know-how for effectively using the information delivered from ensemble flood forecasts has not been established because ensemble flood forecasts have different characteristics from those of conventional deterministic flood forecasts. This study analyzed how communication between forecasters and users improved the operation of ensemble flood forecasting systems by surveying antecedent cases in overseas countries. Effective visualization of forecasting is an important process for delivering actionable information to users. We classified visualization techniques of ensemble flood forecasts into four categories and discussed the characteristics and expected usage of the respective visualization techniques. Results indicated that ideas for handling forecast uncertainties were included in visualization techniques to support users’ decision-making by fully utilizing information taken from ensemble forecasts for issuing appropriate flood alerts. This can be done by converting a simulated discharge to an equivalent return period to avoid system bias issues or showing consistency of estimated risks with recent forecast results. Results also show that feedback from users collected through workshops can improve the operation of flood forecasting systems. This user feedback helped to issue “very low likelihood but severe impact" alerts effectively and to reveal users’ capacity limitations because of an insufficient legal framework. We concluded that two-way communication between forecasters and users is necessary for effective flood forecast operation and for technical development.