The use of artificial intelligence (AI) to help collect, understand and analyze large sets of information has the potential to revolutionize our ability to observe, understand and predict processes in Earth’s systems.
Researchers and scientists are working together to advance the Earth and environmental sciences using AI and modeling techniques such as machine learning (ML). In particular, the group of scientists and experts aims to integrate modern technology into Earth system models, observations and theoretical work—as well as computational capabilities that provide speed, accuracy and more-informed, agile decision-making capabilities.
The Artificial Intelligence for Earth System Predictability (AI4ESP) workshop was organized in a collaborative effort between the US Department of Energy’s (DOE’s) Office of Biological and Environmental Research (BER) and DOE’s Advanced Scientific Computing Research Program, as well as experts from the community. From October to December 2021. The five-week virtual workshop explored the challenge and development of infrastructure that best integrates the combination of technological capabilities and human activities in the field and laboratories with computational resources. BER developed the process as the “Model-Experiment” paradigm, or ModEx.
“Effective improvements to Earth system prediction require radical advances in the ModEx environment. This workshop provided a cross-discipline and cross-mission opportunity for the scientific and application communities to contribute to the understanding of the necessary advances,” said AI4ESP Chief Nicky Hickman, DOE’s Argonne National Laboratory. Associate Director for Operations of the Office of Atmospheric Radiation Measurements of the Science User Facility.
According to a recently released report summarizing the AI4ESP workshop, the event brought together more than 700 participants from both the private and public sectors, along with representatives from earth and environmental sciences, computing and AI. The workshop was designed by around 100 experts based on 156 white papers provided by 640 authors from 112 organizations worldwide.
Information was condensed into 17 topics related to the integrated water cycle and extreme weather events within that cycle. Experts discussed nine focal points related to Earth System Prediction, with sessions including hydrology, watershed science and coastal dynamics; atmosphere, land, oceans and ice; and climate variability and extremes. During the sessions, participants explored the potential of AI to unlock scientific discoveries using tools such as neural networks, knowledge-aware machine learning, AI architectures and co-design.
In each session, researchers identified challenges that support the need for revolutionary AI technologies and infrastructure that can be applied to manage complex tasks in the environmental science field.
“We need new AI methods that understand the processes and respect physical laws to make predictions of Earth system behavior that are scalable, reliable and applicable under future climate regimes,” said Charu Varadharajan, a research scientist at DOE’s Lawrence Berkeley National Laboratory who leads the lab’s Earth. AI and data programming domain. “This workshop is unique in discussing how AI can improve models, observations and theory incorporating DOE’s ModEx approach.”
“The workshop and report allowed us to develop 2-, 5-, and 10-year goals for developing an integrated framework for each focal point. We also identified priorities for earth science, computational science, and programmatic and cultural changes that encompass AI4ESP’s mission. Varadharajan added.
Experts developed a comprehensive list of opportunities where AI research and development could help tackle some of the biggest challenges facing the Earth sciences. These challenges include the management and analysis of large sets of data to enhance the ability to observe and predict extreme events, and to promote the integration of human activities into theories and models.
“One of the most exciting opportunities in modeling is the development of new hybrid models that incorporate both process-based and ML-based modules,” said Forrest Hoffman, lead for the Computational Earth Sciences Group at DOE’s Oak Ridge National Laboratory. “These modeling frameworks enable the inclusion of data about poorly understood processes that can improve accuracy and often result in improved computational performance for Earth system models, enabling more simulations and analysis to be conducted within given resource constraints.”
Workshop participants also identified several priorities for addressing computational challenges, including advances in AI and ML, algorithms, data management and more. The results of those priorities can help develop a technology infrastructure that is efficient, accurate, strategic and convenient, and more accessible to resources.
There is also a need for programmatic and cultural changes across various scientific and government agencies to support a more cohesive mission, as well as a trained workforce that can successfully integrate technology into their humanitarian research and activities. Experts identified solutions that would include specific AI research centers for environmental sciences, frameworks to enable shared services across different communities, and ongoing training and support missions.
Participants of the 2021 AI4ESP workshop continue to discuss community computational activities, including those of the American Geophysical Union and the American Meteorological Society. Stay tuned for more workshops and meetings in the near future — further collaboration, engagement and framework development will continue to advance the AI4ESP mission.
Nicki Hickmon et al, Artificial Intelligence for Earth System Predictability (AI4ESP) Workshop Report, (2022). DOI: 10.2172/1888810
Provided by Argonne National Laboratory
Quote: New Report Details AI Infrastructure for Earth System Prediction (2023, January 24) Retrieved January 25, 2023 from https://phys.org/news/2023-01-ai-infrastructure-earth.html
This document is subject to copyright. No part may be reproduced without written permission, except for any fair dealing for purposes of private study or research. The content is provided for informational purposes only.