トピックス

【2025.1.31 GIR公開セミナー報告】Dr. Fang Zhao “Are We Overestimating Future Floods Under Climate Change? Evaluating Methodological Biases and Role of Model Calibration in Large-Scale Hydrological Simulations”

イベント報告

2025.2.7

◆講演者：Dr. Fang Zhao (中国、華東師範大学、教授)
◆講演タイトル：”Are We Overestimating Future Floods Under Climate Change? Evaluating Methodological Biases and Role of Model Calibration in Large-Scale Hydrological Simulations”
◆日時：2025年1月31日(金) （13：00～15：00）
◆会場：東京農工大学 府中キャンパス 第2講義棟 2階 21講義室、Zoom
◆言語：英語
◆開催担当者：グローバルイノベーション研究院・農学研究院　ジュリアン　ブランジェ（グローバルイノベーション研究院　食料分野 ジュリアン・ブランジェチーム）
◆開催案内
◆参加人数：8人

講演概要

Prof. Fang started his lecture by introducing key climate change impact flood analyses which evaluated future flood exposure and damage. However, recent advances in climate science strongly suggested that the methodology used to determine 100-year flood introduce some bias that may result in the overestimation of future flood impacts.

In the second part of his lecture, Prof. Fang concretely presented where and how that bias comes from using synthetic Monte Carlo simulations. Owing to the very technical nature of the topic, prof. Fang and the participants often engaged in discussions. Furthermore, although the lecture was centered on flood, the identified bias issue is present in any analysis relying on the estimation of return periods (i.e., droughts, cyclones, earthquake).

In the third section, hydrodynamic simulations were used to confirm the validity of the Monte Carlo approach. The methodology and forcing data used in the CaMa-Flood model (the hydrodynamic model) were presented in detail.

The last section of Prof. Fang lecture was less technical as he introduced recent flood studies in China (Yangtze River Basin). The studies implemented the bias correction methods previously presented. The CaMa-Flood model was used in three configurations: non-calibrated, calibrated for mean, high, and low streamflow. A key finding was that, overall, calibration does not strongly affect the estimation of 100-year return flood. This finding is due to large floods being largely precipitation driven with catchment characteristics playing a less critical role.