SEPTEMBER 30, 2024
"Croucher Advanced Study Institute (ASI): Revolutionizing Weather Forecasting and Climate Prediction Through Artificial Intelligence" held successfully from 19 to 21 September 2024 at the Hong Kong University of Science and Technology (HKUST)
<aside> 🧑🏫 Organizing Committee
Chair: Prof. Hui Su (HKUST)
Co-Chairs
Prof. Zhizhao Liu (The Hong Kong Polytechic University (PolyU))
Prof. Xiaoming Shi (HKUST)
Prof. Eun-Soon IM(HKUST)
Prof. Chengxing Zhai (HKUST)
Prof. Lin Liu (The Chinese University of Hong Kong (CUHK))
Prof. Xiaomeng Li (HKUST)
</aside>
The Croucher Advanced Study Institute (ASI) hosted a series of keynote lectures featuring renowned experts in artificial intelligence (AI) applied to weather forecasting and climate prediction. The distinguished speakers included **Prof. Christopher Bretherton (Allen Institute for Artificial Intelligence),** Prof. Tapio Schneider (California Institute of Technology), Prof. Tom Beucler (University of Lausanne), Prof. Aditya Grover (University of California, Los Angeles (UCLA)), Dr. Wai Kin Wong (Hong Kong Observatory (HKO)), and Dr. Xiaohui Zhong (Fudan University).
During their presentations, the keynote speakers articulated the advancements and applications of AI in weather forecasting and climate prediction from various perspectives. They addressed the limitations of traditional numerical models and highlighted the potential of AI models to enhance forecast accuracy. The discussions encompassed both overarching themes and intricate technical details related to AI model development and implementation.
The speakers acknowledged that the rise of extreme weather events, exacerbated by global warming, complicates weather forecasting. They noted that our limited understanding of factors such as clouds and aerosols introduces significant uncertainty in climate predictions. Relying on traditional numerical models not only results in systematic biases but also demands excessive computational resources, especially for high-resolution simulations. In contrast, AI models can deliver forecasts thousands of times faster post-training, often yielding results that are equal to or surpass the accuracy of traditional models. AI models specifically tailored for particular problems show even greater promise, enhancing expectations for their future contributions to forecasting.
While expressing optimism for the future, the keynote speakers also addressed the current limitations of AI models, including their struggles to predict extreme events like heavy precipitation, challenges related to interpretability due to their data-driven nature, and inadequate performance in sub-seasonal to seasonal predictions. They proposed solutions such as developing improved observation and reanalysis datasets, designing physics-based hybrid AI models, and refining model architectures. Many of these solutions have already been implemented, leading to significant progress and bolstering confidence in the future efficacy of AI models.
Prof. Yike Guo (Provosr of HKUST ) gave Welcome Speech on 19 Sept
Dr. Wai Kin Wong gave keynote lecture on 19 Sept
Prof. Christopher Bretherton gave keynote lecture on 19 Sept
Prof. Tom Beucler gave keynote lecture on 20 Sept
Dr. Xiaohui Zhong gave keynote lecture on 20 Sept
Prof. Tapio Schneider gave keynote lecture on 21 Sept
Prof. Hui Su (first on the left), Dr. Wai Kin Wong (second on the left), Mr. Chi-ming Shun (second on the right), and Prof. Shuaqi Tang (first on the right) in the panel discussion
Prof. Zhizhao Liu (first on the left), Prof. Christopher Bretherton (second on the left), Prof. David Neelin (second on the right), and Prof. Shuyi Chen (first on the right) in the panel discussion
Prof. Xiaoming Shi (first on the left), Prof. Dit-Yan Yeung (second on the left), Prof. Tom Beucler (second on the right), and Dr. Sirod Sirisup (first on the right) in the panel discussion
Prof. Eun-Soon IM (first on the left), Dr. Xiaohui Zhong (second on the left), Prof. Ming Shangguan (second on the right), and Prof. Alexis Lau (first on the right) in the panel discussion
Prof. David Neelin (first on the left), Prof. Tapio Schneider (second on the left), Dr. Chen Chen(second on the right), and Dr. Jeff Lo (first on the right) in the panel discussion
Following each keynote lecture, speakers participated in panel discussions alongside other distinguished panelists. These panelists included prominent experts from around the world, such as Prof. David Neelin (UCLA), Prof. Shuyi Chen (University of Washington), Mr. Chi-ming Shun (former Director of HKO), Prof. Dit-Yan Yeung (HKUST), Prof. Alexis Lau (HKUST), Dr. Jeff Lo (Centre for Climate Research Singapore, i.e., CCRS), Dr. Chen Chen (CCRS), Dr. Sirod Sirisup (National Electronics and Computer Technology Center), Prof. Shuaiqi Tang (Nanjing University), and Prof. Ming Shangguan (China University of Geosciences, Wuhan). During these discussions, experts shared their insights and experiences with AI models, while participants eagerly engaged with questions related to the keynote lectures and their own research projects, fostering stimulating conversations. Additionally, Dr. Chen Chen and Mr. Tao Han (HKUST) presented practical examples of AI applications in weather forecasting and climate prediction.
To encourage early-career scholars to explore the advancements and applications of AI models in weather forecasting and climate prediction, the organizing committee hosted a Precipitation Forecast Model Intercomparison Project (PFMIP). Participants were invited to utilize any AI model and dataset to forecast precipitation in Hong Kong. Five teams from various universities and commercial companies presented their model results during the event. On the afternoon of September 21, Prof. Chengxing Zhai assessed all the submitted results from multiple perspectives, providing valuable insights.
Additionally, Croucher ASI showcased 17 posters from various research groups, igniting active discussions and fostering in-depth interactions beyond the keynote lectures and panel discussions.
The Croucher ASI concluded on the evening of September 21, wrapping up three days filled with engaging lectures and stimulating discussions. Prof. Hui Su reflected on the keynote presentations and panel discussions, noting that the event achieved significant outcomes by providing an open and inclusive platform for global AI researchers. She highlighted how the ASI stimulated active dialogue on future directions and technical details of AI models while offering early-career scholars a valuable opportunity to learn about the latest advancements in AI.
Prof. Su expressed her gratitude to all ASI participants, sponsors, co-organizers, co-chairs, and volunteers. She conveyed her hope to organize more events of this nature in the future and looked forward to welcoming additional researchers from around the world to participate.
Prof. Chengxing Zhai assessed PFMIP results
Prof. Hui Su delivered the round-up session
Group photo
Poster Viewing session
Four distinguished speakers are discussing
Guests and volunteers group photo