2021 ESIP Winter Meeting

Deriving scientific insights from artificial intelligence methods requires adhering to best practices and moving beyond off-the-shelf approaches” (Imme Ebert-Uphoff et al 2019). Artificial intelligence (AI) has been showing promises to address many challenges associated with Earth sciences, such as remote mapping, prediction, anomaly detection, event classification, and potentially provide high-speed, effortless alternatives for representing vague non-observable processes in Earth system models. However, due to AI's uncertainty and black box nature, there is no consensus on a universal way to correctly use AI. This session calls for best practices of AI utilization and invites the current AI practitioners to present their experiences and workflows on preparing AI-ready data, training AI models, or applying AI in real scenarios, as examples for the community to learn from. The successful use of AI in any domain of Earth and Space Sciences is welcomed for this session.

How to prepare for this session: Please refer to this repository to find out the existing efforts on AI utilization in Earth science: https://github.com/ESIPFed/Awesome-Earth-Artificial-Intelligence

TALKS

Mike Giordano
AfriqAir; Observatoire de Sciences de l'UNIVERS EFLUVE, LISA/IPSL, UMR CNRS 758; Université Paris Est Créteil et Université Paris
Title: Low-Cost Air Quality Monitoring and Machine Learning: Challenges and Lessons Learned from the AfriqAir Network
S. Mostafa Mousavi
Stanford University
Title: Earthquake Monitoring in Artificial Intelligence Era

Aji John and Nicoleta Cristea
University of Washington
Title: High-resolution snow-covered area mapping in mountain ecosystems using PlanetScope imagery
Kevin Booth
Radiant Earth
Title: Radiant MLHub: An Open Library for Geospatial Training Data

Ryan McGranaghan
ASTRA LLC
Title: The opportunities and challenges of ML: Trends from the space weather perspective
Slides: https://doi.org/10.6084/m9.figshare.13728070.v1

Ziheng Sun
George Mason University
Title: Earth AI: Formulating ESIP ML Community Effort
Slides: https://doi.org/10.6084/m9.figshare.13721521.v1

View Recording
View Notes