(1) Create awareness of capabilities at GSFC and UMD in the use of AI/ML (Artificial Intelligence/Machine Learning) in Earth Science.
(2) Strengthen or establish new connections among researchers from different disciplines.
(3) Identify opportunities for collaboration and acceleration of innovative applications.
Format: Hybrid = In-person + Virtual
- September 22, 2021: Main Sessions at Stamp Student Union, UMD
- September 23, 2021: Poster session via Topia
Final Report: Released November 1, 2021
- Workshop Booklet: Participants bios and abstracts
- Presentation Files: Only those with author’s consent, access limited to the registered attendees with email address
September 22, 2021: In-Person at Stamp Union
- Opening remarks:
- Dr. James Irons (Director of NASA Earth Science Division)
- Professor Amitabh Varshney (Dean of the College of Computer, Mathematical and Natural Sciences)
- Keynote Speaker: Dr. Kevin Murphy (NASA HQ, Chief Science Data Officer), "Grand Challenges for AI in Earth Science"
- Plenary talk 1: Dr. Tom Goldstein (UMD CS/UMIACS): "Making Neural Network Systems Robust"
- Session 1: Lightning talks and discussion in three parallel breakout sessions
- Panel Discussion, including:
- Dr. Christa Peters-Lidard (Associate Director ESD – GSFC)
- Prof. X.-Z. Liang (Head of DAWN project)
- Dr. Nargess Memarsadeghi (Lead of NASA AI Center of Excellence)
- Prof. Margaret Palmer (Head of SESYNC)
- Dr. William Regli (Head of ARLIS)
- On-site lunch and Poster Viewing
- Plenary talk 2: Dr. Matthew Hansen (Professor, Department of Geographical Sciences, UMD) "Global Monitoring of Land Cover and Land Use"
- Session 2: Lightning talks and discussion in three parallel breakout sessions
- Reports from breakout sessions
- Closing session
- Social hour (5:00 - 6:00pm)
September 23, 2021: Virtual Poster Session via Topia
- Meet-the-Presenter: Interactive Virtual Poster Session (1pm - 3pm)
- Open access through October 21, 2021
- Please be engaged! The workshop aims to bring the diverse areas of expertise from NASA GSFC and UMD, and foster active interaction at the workshop and in the future.
New and innovative Earth observations, analysis and modeling are fast expanding our understanding of climate change and the Earth System. The aspirational goal to create a ‘digital twin’ of the Earth* represents the benefits that could arise at the nexus of these fields. Developing the path to a Digital Earth will deliver many intermediate goals in our ability to predict the impact of the changing climate on the environment and human needs, and conversely the impact of our energy and sustainability choices on the Earth System.
Emerging developments in AI (Artificial Intelligence) and ML (Machine Learning) enable novel approaches to Earth Science. For example, spectroscopic instrument design and data retrievals can be informed by ML-based spectral fitting methods; Earth System models can substitute ML models for prohibitively expensive full-physics simulations; and flood, landslide, and famine risk can be assessed using remote sensing images with ML approaches for classification, detection, and regression. Recent work applying AI/ML techniques to Earth science problems has demonstrated its promise, and there are still many challenges and opportunities to be addressed through multi-disciplinary teams of AI/ML and Earth science researchers.
NASA Goddard Space Flight Center (GSFC) and the University of Maryland (UMD) will host a workshop on September 22, 2021, in College Park, MD, to bring together expertise at NASA GSFC and UMD in the Earth Sciences, AI/ML, and applied sciences. The goal of the workshop is to create awareness of current capabilities at GSFC and UMD in the use of AI/ML for Earth Sciences, create connections among researchers from different disciplines, and identify opportunities for collaboration and acceleration of innovative applications.
* The digital revolution of Earth-system science” by Peter Bauer, Peter D. Dueben, Torsten Hoefler, Tiago Quintino, Thomas C. Schulthess and Nils P. Wedi, 22 February 2021, Nature Computational Science. DOI: 10.1038/s43588-021-00023-0'
Targeting strength and directions of GSFC and UMD researchers who work on earth sciences, discussion themes of this workshop include but not limited to:
- Data bottlenecks in ML/AI
- Explainable, interpretable, trustworthy AI
- ML/AI algorithms applied to earth science
- Physical modeling, simulation, prediction
- Applications to ocean and atmosphere
- Applications to Land, environment, agriculture, food, society
- Collaborative research encompassing both earth science and data science
- GSFC: Mark Carroll (CISTO), Nargess Memarsadeghi (CISTO), Sabrina Delgado Arias (SSAI)
- UMD: Ellen D. Williams (ESSIC Director), Kayo Ide (Scientific lead: AOSC/ESSIC/IPST/MATH), Julie Nicely (ESSIC/GSFC), Hannah Kerner (GEOG SCI), David Jacobs (CS/UMIACS), Ramani Duraiswamy (CS/UMIACS), Mihai Pop (CS/UMIACS)
The workshop will follow University of Maryland policies. See: https://umd.edu/4maryland/health-plan