报告题目： Does AI Offer Solutions for Hydroclimatic Extremes?
报 告 人： Dr John Xiaogang Shi
时    间： 2024年11月20日（星期三）上午10:00-11:30
地    点： 地理资源所A901会议室

专家简介：

Dr John Xiaogang Shi，现任英国格拉斯哥大学长聘副教授，博士生导师，社会与环境可持续发展学院研究主任及格拉斯哥大学-南开大学联合研究生院环境科学项目负责人。师从世界知名水文学家、 美国工程院院士Dennis P. Lettenmaier教授，2013年获美国University of Washington博士学位。2013-2015年获加拿大自然科学和工程研究委员会（NSERC）博士后奖学金。自2013年先后任职于加拿大联邦环境部国家水文研究中心, 澳大利亚联邦科学与工业组织（CSIRO), 英国兰卡斯特大学以及利物浦大学。现兼任加拿大约克大学客座教授和Environmental Impact Assessment Review (IF 9.8) 副主编， 并担任英国UKRI 自然环境委员会(NERC)和未来领袖奖学金(Future Leaders Fellowships)评审专家组成员。Dr Shi主要致力于大尺度水文模拟、极端气候及水资源灾害预测和预报、以及气候变化对水资源影响等研究领域。近5年，主持英国皇家学会、英国皇家工程学会、英国文化协会、英国自然科学基金、苏格兰全球挑战基金等科研项目20余项。

报告摘要：

Hydroclimatic extremes (e.g. floods and droughts) are destructive natural disasters resulting in severe impacts on vulnerable individuals and communities with massive social, economic, and environmental consequences. Globally, flood risk already affects 1.81 billion people. The hydrologic forecasting is therefore essential to allow evidence-informed actions to reduce disaster risk and increase flood/drought resilience for growing populations. Traditional hydrologic forecasting systems include physics-based or conceptual hydrology models to forecast streamflow and hydrodynamic models to translate forecasted streamflow. However, this approach involves great uncertainties. As a facet of Artificial Intelligence (AI), purely data-driven deep learning models have successfully produced very high-accuracy simulations for many hydrologic variables with large data sets, opening up many new opportunities to progress research in hydrology. However, it has very generic internal structures that do not resolve hydrologic dynamics. Therefore, the question remains whether AI-based hydrologic models could improve flood/drought resilience.