The symposium agenda included plenary sessions, panel discussions, breakout sessions and tutorials over three days. These sessions provided a comprehensive platform for engaging discussions, interactive sessions and presentations on various topics related to AI/ML in hydrology and Earth system science.
The symposium began with a warm welcome from Dipankar Dwivedi, Co-Chair of the HydroML Symposium and EESA Staff Scientist, expressing gratitude to the sponsors, the ExaSheds Project, EESA and Berkeley Lab. The Executive Committee and the Organizing Committee were recognized for their dedicated efforts in organizing and ensuring the success of the symposium.
During the opening remarks of the HydroML Symposium, Dwivedi provided an insightful overview of the event’s history. The inaugural HydroML 2022 (HydroML 1.0), chaired by Professor Shen at Pennsylvania State University (PSU), marked a significant milestone. Dwivedi summarized findings from HydroML 1.0, emphasizing the need for further exploration and collaboration in applying ML techniques in Earth system science. The attendees identified the limited availability of ML skills within the Earth science community as an area for growth and development. This symposium resulted in notable achievements, such as the collaborative research paper titled “Differentiable Modeling in Hydrology” which involved the active participation of over 30 researchers. These accomplishments underscore the impact and collaborative spirit fostered by the HydroML Symposium.
HydroML 2.0 commenced with a clear vision and ambitious objectives. The goals of HydroML 2.0 aimed to enhance the application of AI/ML in Earth System Science, examine explainability, interpretability and mechanistic understanding, promote integration and standardization, identify future research opportunities and enable networking and collaboration among researchers.
As the symposium came to a close, the focus shifted toward the key takeaways and the path forward; these included the challenges of data availability and locations, the integration of mechanistic understanding with ML and the future prospects of AI/ML techniques in the field.
The symposium’s success was attributed to the collective efforts of the organizing committee, sponsors and participants who actively engaged in discussions and shared their research findings. Moving forward, researchers aim to develop an event to showcase the symposium research, gather feedback through surveys, organize future events and maintain connectivity through social media channels and email.
The HydroML 2.0 Symposium has paved the way for further advancements in AI/ML applications in hydrology and Earth System Science; this symposium could not have been possible without the efforts of executive committee members Hoshin Gupta (University of Arizona), Carl Steefel (LBNL), Scott Painter (ORNL), Utkarsh Mital (LBNL), Reed Helgens (LBNL), and Kathryn Lawson (PSU), and several other dedicated members of the organizing committee. The valuable insights and collaborative efforts of participants have undoubtedly contributed to the progression of this exciting field.