3rd NOAA Workshop on Leveraging AI in Environmental Sciences

The hackathon is jointly organized with the Climate Informatics community (climateinformatics.org).

The 3rd NOAA AI Workshop and Climate Informatics community invite you to join this virtual joint hackathon event to use data science methods to address climate and environmental challenges.
This year we have three hackathon topics in collaboration with different research groups. The three topics include:

• Colorado wildfire risk estimation (w/ Earth Lab at University of Colorado, Boulder): this hackathon challenge invite data science solutions to estimate the risk of wildfire damages to human settlements/buildings in Colorado using weather information, topology, development density, and vegetation data;
• Marine heatwave prediction (w/ Columbia University): this hackathon challenge seeks innovative ways to understand the relationship between weather and surface mixed layer factors and the onset and end of marine heatwave events using historical climatology data;
• Climate model emulation (w/ Oxford University): this hackathon challenge aims to use data science methods to predict annual mean global distributions of temperature and precipitation given emissions and concentrations of key anthropogenic climate forcers: SO2, BC, CH4, and CO2.

All participants are invited to form interdisciplinary groups with fellow attendees to tackle one or more of these challenges. All teams are invited to submit their solutions to these challenges to the public competition at CodaLab (https://codalab.org/). Instructions for how to submit to CodaLab will be provided after the kickoff of the event.

The winning teams will be invited to share their hackathon results at the closing day of the workshop at the Hackathon Showcase session.

To attend the hackathon, simply register for the workshop event using this google form prior to the hackathon event.

Sign up for the hackathon and tutorial here: https://ncsu.zoom.us/meeting/register/tJIkcOChpzoiH9JX9M7qF9bkbfYBNhZ660-e

Large computer models are ubiquitous in the environmental sciences. These models often have tens or hundreds of tuneable parameters and can take thousands of core-hours to run to completion while outputting terabytes of output. It is becoming common practice to develop emulators as fast approximations, or surrogates, of these models in order to explore the relationships between these inputs and outputs, understand uncertainties and generate large ensembles datasets. While the purpose of these surrogates may differ, their development is often very similar. Here we introduce GCEm: an open-source tool providing a general workflow for emulating and validating a wide variety of environmental models and outputs. It includes efficient routines for sampling these emulators for the purpose of uncertainty quantification and model calibration. It is built on well established, high performance libraries to ensure robustness, extensibility and scalability. We demonstrate the flexibility of GCEm through three case-studies using GCEm to reduce parametric uncertainty in a general circulation model, explore precipitation sensitivity in a cloud resolving model and scenario uncertainty in the CMIP6 multi-model ensemble.

Virtual Help Desk for all attendees. Conference questions? We can help.

Read this participant handbook for detailed information about workshop technology platform!

Questions about technology, logistics, how to network, or any aspect of the conference can be answered during Help Desk time. We're also available to answer your questions now by several methods:

Our Help Desk email: ai.workshop@noaa.gov
Our Slack channel Help Desk: enviro-ai.slack.com, #help-desk
And during Virtual Coffee Breaks:
12:15 p.m. Monday
12:15 p.m. Tuesday
11:00 a.m. Wednesday
2:00 p.m. Thursday

See our Code of Conduct and Etiquette. We encourage a safe, productive, inclusive, and welcoming environment for all attendees.

Session Agenda (Mountain Time):
09:00 Workshop Highlights & Welcome - Drs. Yuhan “Douglas” Rao and Chris Slocum
09:10 NOAA Center for Artificial Intelligence - Dr. Rob Redmon
09:15 National Centers for Environmental Information - Mary Wohlgemuth
09:25 NOAA Leadership Address - Dr. Rick Spinrad
09:45 National AI Initiative - Dr. Lynne Parker
10:05 Panel Q&A Moderated by Dr. Eric Kihn
10:30 Close

Speakers and Panelists:

"NOAA's Perspective on AI" – Rick Spinrad
Dr. Richard (Rick) Spinrad is the Under Secretary of Commerce for Oceans and Atmosphere and the 11th NOAA Administrator. Prior to returning to NOAA, Dr. Spinrad served as a Professor of Oceanography at Oregon State University. He was also Vice President for Research at OSU from 2010-2014. Dr. Spinrad served as NOAA’s Chief Scientist under President Barack Obama from 2014 until 2016. He also led NOAA’s Office of Oceanic and Atmospheric Research and National Ocean Service from 2003-2010. While at NOAA, Dr. Spinrad co-led the White House Committee that developed the nation’s first set of ocean research priorities and oversaw the revamping of NOAA’s research enterprise, including the development of the agency’s Scientific Integrity policy. Prior to initially joining NOAA, Dr. Spinrad held leadership positions at the U.S. Office of Naval Research and Oceanographer of the Navy, where he was awarded the Distinguished Civilian Service Award — the highest award given by the U.S. Navy to a civilian.

“National AI Initiative” – Lynne Parker
Dr. Lynne E. Parker is the Founding Director of the National AI Initiative Office. She also serves as the Deputy United States Chief Technology Officer and the Assistant Director of Artificial Intelligence at the White House Office of Science and Technology Policy (OSTP). She has served in OSTP since 2018 and played an integral role in numerous landmark national AI policies. Dr. Parker is on assignment to OSTP from the University of Tennessee, Knoxville, where she is a professor of computer science and former interim dean of engineering. She has served as National Science Foundation’s Division Director for Information and Intelligent Systems, and as a Distinguished Research and Development Staff Member at Oak Ridge National Laboratory. She earned a Ph.D. from the Massachusetts Institute of Technology and is a Fellow of IEEE, a Fellow of AAAS, and a Distinguished Member of the Association for Computing Machinery.

Mary Wohlgemuth is the Director of NOAA National Centers for Environmental Information. She provides direction and oversight of the organization's operations and resources to develop robust, service-centric programs linking environmental information and data stewardship.

Robert Redmon is a space scientist and the lead of NOAA Center for Artificial Intelligence.

Jebb Stewart and Greg Dusek are the co-chairs of the NOAA Artificial Intelligence Executive Committee.

Chair: Kirstine Dale

Agenda:
10.45 - 10.48:    Welcome
10.50 - 11.10:    Stephen Haddad – Embedding Data Science at the Met Office through a Community of Practice
11.12 - 11.22:    Mesut Akdere – The Global Artificial Intelligence Transformation and Its Implications for Workforce Development & Education
11.24 - 11.34:    Amy McGovern  – Update from the NSF AI Institute for Research on Trustworthy AI in Weather, Climate, and Coastal Oceanography
11.36 - 11.46:    David John Gagne  – The NCAR Analytics and Integrative Machine Learning Group: Research Overview
11.48 - 11.58:    Andrew Neang – Assembling heterogenous data for machine learning: A communities of practice perspective
12.00 - 12.10:    Isaac Passmore  – Mission Operations Assistant (MOA) - Community Driven A.I. for Satellite Operations
12.10 - 12.15: Q+A

Slack channel: #05_community-of-practice-workforce-dev

Chairs: Kyle Hilburn, Yoonjin Lee

Presentations:
David Hall – "AI for the Environmental Sciences at NVIDIA" (10:45 am MDT - 11:05)
Chongxing Fan – "Use different machine-learning algorithms for clear-sky detections in infrared hyperspectral observations: assessment and physical interpretability" (11:07 - 11:19)
Daniel Lopez – "Generating CloudSat Reflectivity Using Passive Microwave Brightness Temperatures and cGANs" (11:21 - 11:33)
Janni Yuval – "Neural-network parameterization of subgrid momentum transport learned from a high-resolution simulation" (11:35 - 11:47)
Linsey Passarella – "Super Resolution Reconstruction of E3SM Data Using a FSRCNN" (11:49 - 12:01)
Spencer Clark – "Applying machine learning parameterization through coarse graining to improve the skill in simulating multiple climates in a full complexity GCM" (12:03 - 12:15)

Slack channel: #03_weather-and-climate

We're gathering on Wonder.me!

Whether you're a seasoned user or a newbie, let's share resources to improve our AI/ML learning curve. What tool, tutorial, instructional material, or expert helped you solve a problem or learn the AI ropes? Bring your answers and open mind to this virtual space to find resources that may help your work.

ACTIVITY
Participate in our data gather whether or not you can join us for virtual coffee!!
Questionnaire: Where have you gotten the best AI education? <LINK>

HOUSEKEEPING NOTES

• Help During Conference Presentations: If you need help during the conference, please reach out to our Help Desk email, ai.workshop@noaa.gov, or Slack channel, enviro-ai.slack.com, #help-desk.
• Help During Coffee Breaks: You can hover your avatar over the Help Area in the bottom left corner, and we’ll do our best to assist you. Look for the i for "Information" avatars to find people who can answer questions. Outside of Coffee Breaks, please use email or Slack for assistance.
• Wonder.me: You’ll notice Wonder.me is a very fluid environment in which you can join and leave conversation bubbles at any time. We encourage you to join a group, engage in discussions, listen, and learn what you can. 
• Note About Wonder.me Backgrounds: They cannot be blurred in a Wonder Space, so if you are concerned about privacy, be cognizant of your setting.  
• Code of Conduct and Etiquette: Please take a few minutes to read our Code of Conduct and Etiquette statement.

“Trustworthy and Responsible AI: Risk Management Framework, Standards, and Evaluations”
Dr. Elham Tabassi is the Chief of Staff in the Information Technology Laboratory (ITL) at NIST and leads the NIST Trustworthy AI program. ITL, one of six research laboratories within NIST, supports NIST’s mission to promote US innovation and industrial competitiveness by advancing measurement science, standards, and technology in ways that enhance economic security and improve our quality of life. ITL cultivates trust in information technology and metrology by conducting research, developing and disseminating standards, measurements, and testing for interoperability, security, usability, and reliability of information systems. As a scientist, she has been working on various computer vision research projects with applications in biometrics since 1999. She is a member of the National AI Research Resource Task Force, a Fellow of the Washington Academy of Sciences, and a senior member of IEEE.

Chairs: Amy McGovern, Julie Demuth


Presentations (Mountain time):
14:15 – Antonios Mamalakis – "Explainable Artificial Intelligence for Environmental Sciences: A benchmark to assess and compare neural network attribution methods"
14:27 – David Harrison – "Challenges and Benefits of Machine Learning in an Operational Environment: Survey Results from the 2021 Hazardous Weather Testbed Spring Forecasting Experiment"
14:39 – Cindy Kaiying Lin – "Coding Climate Justice "
14:51 – Arjan Durresi – "Trustworthy Acceptance of AI-based Collective Management of Food-Water-Energy Resources"
15:03 – Brian Hu – "Explainable Artificial Intelligence (XAI): From Fish Classification to an Open-Source Toolkit"
15:15 – Courtney Bowman – "Responsible AI: An Ethical Framework"
15:30 – Discussion

Slack channel: #09_trustworthy-ai

Chair: Nikunj Oza

Presentations:
14:15 – Hassan Moustahfid & Anass El Aouni – "An Unsupervised Machine Learning Algorithm for the Identification and Tracking of Lagrangian Mesoscale Eddies"
14:35 – Andrew Osborne – "Development of the Multi-Radar Multi-Sensor (MRMS) Machine Learning Quantitative Precipitation Estimate (QPE) for Complex Terrain"
14:47 – Ignacio Lopez-Gomez – "Predicting Extreme Heat using Physics-based AI"
14:59 – Zhongrun Xiang – "Fully distributed rainfall-runoff modeling using spatial-temporal graph neural network"
15:11 – Christina Kumler – "Improving and Implementing the Random Forest Fire Radiative Power Model with Additional Variables and Time"
15:23 – Kate Duffy – "A generalizable machine learning approach to predict land surface temperature"

Slack channel: #03_weather-and-climate

Chairs: Brian Meyer; Yoonjin Lee

Posters:
Leilei Duan – "Training the Next Generation Geodesign Practitioners for Leveraging AI in Environmental Sciences"
Abhishek Gupta – "Carbon accounting as a way to build more sustainable AI systems: An analysis and roadmap for the community"
Jaleesia Amos – "Community led efforts to capture environmental nanomaterial transformations"
Brandon Smith – "Advancing Organizational Capabilities in AI/ML through a Deep Learning Academy"
Shreya Agrawal – "Interpreting Large-Context Multi-dimensional Predictions of Precipitation Neural Networks"
Eric Loken – "Understanding next-day random forest severe weather hazard forecasts using predictor sensitivity tests and tree interpreter analysis"
Zachary M. Labe – "Climate model evaluation with explainable neural networks"
Sally Wang – "Identifying key drivers of wildfires in the contiguous US using machine learning and game theory interpretation"
Dr. Jennifer Sleeman – "Deep Learning Autoencoders for Detecting and Predicting Extreme Air Quality Pollution Events"
Randy Chase – "Next hour tornado prediction dual-polarization radar signatures gleaned from deep learning and explainable artificial intelligence"

Chairs: Jebb Stewart; Yoonjin Lee

Posters:
Rackhun Son – "Machine Learning provides substantial improvements to county-level fire weather forecasting over the western United States"
Meghna Babbar-Sebens – "InterACTWEL Cyberinfrastructure: Enabling Longitudinal Stakeholder Decision Support for Adaptation Planning in Local Communities"
Carroll A. Hood – "Proof-of-concept Demonstrations of the Application of Emerging Technologies into the Environmental Data Stewardship Lifecycle"
Lakshmish Ramaswamy – "Engineering Robust Machine Learning Systems for Blue Carbon Assessment"
Katherine Wilson – "Development of automated video review tools and processing to evaluate the performance of bycatch reduction devices in Alaska’s walleye pollock (Gadus chalcogrammus) fishery "
Jordan A. Caraballo-Vega – "Applicability of Mask R-CNN for Object Classification in Very High-Resolution Satellite Imagery"
Devaki Raj – "Scaling Computer Vision (CV) for Environmental Monitoring through Code-Free Tools
Izzat Kamarudzaman – "Towards AI Adoption: Data-driven Machine Learning Systems"
Gabrielle Gantos – "4D Post-Processing of Holographic Images"
Achraf Tounsi – "The use of machine learning to account for reservoir management and predict downstream streamflow values"
Xingming Liang – "A Deep Learning Based Forward Microwave Radiative Transfer Model "
Fan Meng – "Forecasting Tropical Cyclones Wave Height using Bidirectional Gated Recurrent Unit"
Handan Sun – "Tropical cyclone frequency prediction method based on neural network"
Ruchika – "DATA-DRIVEN WEATHER FORECASTING USING ML METHODS ON IMDAA DATA"

Chair: Rob Redmon

Presentations:
09:30     Welcome
09:31     Shannon Rankin (Highlight) - "Combining the power of citizen science and artificial intelligence to make new discoveries in our ocean soundscapes"
09:50     Daniel Woodrich - "INSTINCT, and the dream of an end-to-end (and beyond) data pipeline for AI programs in Passive Acoustic Monitoring"
10:02     Thomas Vandal - "GeoNEX-ML: A Machine Learning System for Geostationary Satellite Imagery"
10:14     Matt Brown - "ADAPT: Drone Payload for Data Collection and Real-Time AI Processing in the Field"
10:26     Damien Sileo (Highlight) - "Unsupervised pre-training of neural networks on satellite imagery improves food insecurity forecasting"
10:44     Final Q&A pending remaining time.

Slack channel: #07_seamless-ai-system

Chair: John Williams

Schedule of Presentations:

09:30  Nicole D. Jackson – "Weather impacts on utility-scale photovoltaic plant performance"
09:45  Amy McGovern and Jason Hickey – "Lessons learned while creating cross-sector partnerships for AI for weather"
10:00  Julien Brajard – "Combining machine learning with data assimilation to improve numerical models and estimate the system’s state."
10:15  Rose Yu – " Physics-informed Deep Learning for Turbulent Flow Prediction"
10:30  Haonan Chen – "Deep learning for precipitation estimation and nowcasting using operational radar and multi-source in situ measurements"

These times are Mountain Daylight Time, UTC-6.

Slack channel: #03_weather-and-climate

"Designing for Effective Human-AI Teaming"

Dr. Emilie Roth is the owner and principal scientist of Roth Cognitive Engineering. She is a cognitive psychologist by training and has over 30 years of experience in cognitive analysis and design in a variety of domains including nuclear power plant operations, railroad operations, military command and control, and healthcare. Most recently she has participated in the design and evaluation of collaborative automated systems for planning and scheduling of air transport missions for the air force, which will form the basis of her talk today.

"Operationalizing AI for Disaster Response" – Pedro Rodriguez

Dr. Pedro Rodriguez is the group supervisor and a senior machine learning researcher of Humanitarian Assistance for Disaster Relief (HADR) program at JHU Applied Physics Laboratory. HADR is a machine learning and artificial intelligence-enabled program to collect and process overhead imagery into categories for analysis. In partnership with the Department of Defense's Joint Artificial Intelligence Center, HADR aids FEMA's recovery effort after hazards including Hurricane Florence, Dorian, and Laura. Pedro also leads various DoD projects bringing image processing and target recognition algorithms to the tactical edge.

We're gathering on Wonder.me!

Failures are simply attempts to succeed. Have you tried something in AI that just didn't work? Share your failure stories and lessons learned during this Coffee Break.

ACTIVITY
What are your epic failures and lessons learned? Take this short Q&A! The activity is open whether or not you can attend our virtual coffee hour!

HOUSEKEEPING NOTES

• Help During Conference Presentations: If you need help during the conference, please reach out to our Help Desk email, ai.workshop@noaa.gov, or Slack channel, enviro-ai.slack.com, #help-desk.
• Help During Coffee Breaks: You can hover your avatar over the Help Area in the bottom left corner, and we’ll do our best to assist you. Look for the i for "Information" avatars to find people who can answer questions. Outside of Coffee Breaks, please use email or Slack for assistance.
• Wonder.me: You’ll notice Wonder.me is a very fluid environment in which you can join and leave conversation bubbles at any time. We encourage you to join a group, engage in discussions, listen, and learn what you can. 
• Note About Wonder.me Backgrounds: They cannot be blurred in a Wonder Space, so if you are concerned about privacy, be cognizant of your setting.  
• Code of Conduct and Etiquette: Please take a few minutes to read our Code of Conduct and Etiquette statement.

 

Chairs: John Williams, Rob Redmon

Presentation Schedule (All in Mountain Time/UTC-6):
1:00 pm  Mayank Kejriwal – "Domain-Specific Knowledge Graph Construction for Enabling Seamless AI Applications in the Environmental Sciences"
1:15 pm  Vidya Samadi – "A tour of end-to-end deep learning for flood image classification and floodwater level estimation, are we there yet?"
1:30 pm  Enrico Camporeale – "ACCRUE: Accurate and Reliable Uncertainty Estimate in Deterministic Models"
1:45 pm  Matt Dawkins – "Enhancements to VIAME, an Open-Source, Do-It-Yourself AI Platform for Environmental Science" (Highlight talk)
2:05 pm  Speaker panel Q&A
Times are in Mountain Daylight Time (UTC-6).

Slack channel: #07_seamless-ai-system

Chair: Christina Kumler

Presentations:
13:00 – Tse-Chun Chen – "Learning UFS State-Dependent Systematic Errors from the Analysis Increments"
13:20 – John S Schreck – "Accelerating Mechanistic Simulation of Organic Chemistry in Weather and Climate Models with Machine Learning"
13:32 – Marko Orescanin – "Bayesian Deep Learning for Addressing Uncertainty - Precipitation Type Detection"
13:44 – Steve Penny – "Integrating AI/ML with Data Assimilation for data-driven state estimation"
13:56 – Matthew Brothers – "Random Forest Approach for Improving Non-Convective High Wind Forecasting Across Southeast Wyoming"

Slack channel: #03_weather-and-climate

Chair: Chris Slocum

Presentations:
2:30. Welcome
2:32. Imme Ebert-Uphoff (highlighted talk) – "Guide to Custom Loss Functions for Neural Networks in Environmental Sciences"
2:55. Jason Platt – "Robust Forecasting through Generalized Synchronization in Reservoir Computing"
3:10. Yoonjin Lee – "Exploring ways to effectively use temporal satellite images in detecting convection from GOES-16"
3:25. Weng-Keen Wong – "Structural Change Point Detection for Detecting Change in Weather and Climate Data"
3:40. Panel Q+A

These times are Mountain Daylight Time, UTC-6.

Slack channel: #08_tools-and-datasets

Chair: V. Ramaswamy

Presentations:
14:30 – William Chapman (highlited talk) – "Forecasting Seasonal Western US Precipitation with Interpretable Machine Learning Trained on Large Climate Model Simulations"
14:50 – Jason Stock – "Using Machine Learning to Improve Vertical Profiles of Temperature and Moisture for Severe Weather Forecasting"
15:05 – Randy Chase – "Near real-time hail forecasts using machine learning and convective allowing models"
15:20 – Rocky Talchabhadel – "Spatial downscaling of coarse resolution satellite-based precipitation estimates to 1 km using Machine Learning"
15:35 – Linsey Passarella – "Using Multi-Task Learning to Isolate the Response of Western US Precipitation During ENSO Events"

Slack channel: #03_weather-and-climate

Chair: Nikunj Oza

Posters:
Mariam Hussain – "Prediction of Winter Diurnal Variability in Bangladesh Utilizing Supervised ML techniques"
Yuan-Heng Wang – "Hypothesis Testing Using Long Short-Term Memory Networks (LSTMs) Applied to Large Pixel-Scale Snowpack Dataset"
Anthony Wimmers – "Image-to-image processing of satellite imagery to estimate 2D winds of tropical cyclones"
Deepank Kumar Singh – "Mapping Flood Recovery Times across the United States using Machine Learning "
Feng He – "Prediction of Monthly Arctic Sea Ice Concentration with Convolutional Neural Networks: a Case Study with APP-x Climate Data Records"
John Cintineo – "Evaluating and exploring lightning predictions from NOAA/CIMSS ProbLightning"
Ryan Lagerquist – "U-nets for nowcasting the timing and location of thunderstorms based on satellite data"
William Chapman – "Probabilistic Weather Predictions from Deterministic Forecasts with Machine Learning Methods"
Yusuf Sermet – "Next-Generation Environmental Information Systems using Immersive Technologies and Artificial Intelligence"

"Fusing Simulation and Data sciences for Weather and Climate Services" - Stephen Belcher

Prof. Stephen Belcher is the Met Office Chief Scientist. As Chief Scientist, Stephen has overall responsibility for the leadership and management of the Met Office’s scientific programme, by providing strategic direction, ensuring high quality delivery and nurturing scientific excellence. He represents the Met Office on science and research technology to UK Government, ensuring that the Met Office science programme fits properly into the wider UK environmental science landscape. In 2010 he became the Joint Met Office Chair in Weather Systems. This role gave him a taster of working closely with the Met Office, and in 2012 he joined the Met Office as Director of the Met Office Hadley Centre.

Prof. Belcher led the evolution of the Met Office Hadley Centre to focus on climate science and services: motivated by the need to provide governments, industry and society with actionable advice, i.e. ‘climate services’. He was a driving force behind the initiation of the Newton Fund Climate Science for Service Partnership China (CSSP China), in which scientists from both China and the UK are now working together to develop fundamental climate science and climate services.

Prof. Belcher will be joined by Jebb Stewart, a co-chair of NOAA AI Executive Committee for a discussion afer his talk.

Chair: Greg Dusek

Presentations:
09:45 – Zhongrun Xiang (highlighted talk) – "WaterBench: A Large-scale Benchmark Dataset for Data-Driven Streamflow Forecasting"
10:05 – Tyler Christensen – "AI Data Readiness: Designing Community-Driven Data Standards and Tools"
10:17 – Benedikt Schulz – "Machine learning methods for postprocessing ensemble forecasts of wind gusts: A systematic comparison"
10:29 – Chris Slocum – "Beyond Dvorak: creating feature-based labels in geostationary imagery for tropical cyclones"
10:41 – Muhammed Sit – "IowaRain: A Statewide Rain Event Dataset Based on Weather Radars and Quantitative Precipitation Estimation"

Slack channel: #08_tools-and-datasets

Chair: Philippe Tissot

Presentations:
09:46 – Evgenia Galytska – "Identifying causal relationships between Arctic and midlatitude weather and climate in CMIP6 models"
09:59 – Emily Gordon – "Oceanic harbingers of Pacific Decadal Oscillation transitions detected by neural networks"
10:12 – Kyle Hilburn – "Improving GREMLIN: A Case Study in AI Application Development"
10:25 – Zane K Martin – "Using simple, explainable neural networks to predict the Madden-Julian oscillation"
10:38 – Christine Piatko – "AI and Multimodal Imagery for Sea Ice Forecasting"

Slack channel: #03_weather-and-climate

We're gathering on Wonder.me!

Money and time don't grow on trees, but if they did, what specifically could help you get to the AI finish line? This session will encourage you to think pie-in-the-sky about your AI needs.

ACTIVITY
Short Answer: "If money were no object for AI/ML, I'd . . ."

HOUSEKEEPING NOTES

• Help During Conference Presentations: If you need help during the conference, please reach out to our Help Desk email, ai.workshop@noaa.gov, or Slack channel, enviro-ai.slack.com, #help-desk.
• Help During Coffee Breaks: You can hover your avatar over the Help Area in the bottom left corner, and we’ll do our best to assist you. Look for the i for "Information" avatars to find people who can answer questions. Outside of Coffee Breaks, please use email or Slack for assistance.
• Wonder.me: You’ll notice Wonder.me is a very fluid environment in which you can join and leave conversation bubbles at any time. We encourage you to join a group, engage in discussions, listen, and learn what you can. 
• Note About Wonder.me Backgrounds: They cannot be blurred in a Wonder Space, so if you are concerned about privacy, be cognizant of your setting.  
• Code of Conduct and Etiquette: Please take a few minutes to read our Code of Conduct and Etiquette statement.

Chair: Jebb Q. Stewart

Presentations:
12:00 – Peter Dueben (highlighted talk) – "Machine learning at ECMWF"
12:20 – Montgomery Flora – "Developing Machine Learning-based Severe Weather Hazard Guidance for the Warn-on-Forecast System (WoFS) "
12:40 – Auroop R. Ganguly & Evan Kodra – "Climate resilience from academia to startup "
13:00 – Nate Herold – "Integrating Deep Learning Methods into a National High-Resolution Land Cover Mapping Program"

Slack channel: #06_research-to-operation

Chair: Sue Haupt

Presentations:
12:00  Sebastian Lerch – "Incorporating large-scale spatial information into ensemble post-processing via autoencoder neural networks"
12:15  Charlie Becker – "Evaluation of two machine learning approaches for the objective identification of convective storm mode using convection-allowing model output"
12:30  John M. Haynes – "Low Cloud Detection in Multilayer Scenes from GOES ABI using Machine Learning Methods"
12:45  Jonathan Frame – "Data-driven streamflow simulations of extreme (high flow) events."
13:00 Yingkai Sha – "A hybrid analog-ensemble, convolutional-neural-network method for post-processing precipitation forecasts"

Slack channel: #03_weather-and-climate

Dr. Laure Zanna is a Professor in Mathematics & Atmosphere/Ocean Science at the Courant Institute, New York University. Her research focuses on the dynamics of the climate system and the main emphasis of her work is to study the influence of the ocean on local and global scales. Prior to NYU, she was a faculty member at the University of Oxford until 2019, and obtained her PhD in 2009 in Climate Dynamics from Harvard University. She was the recipient of the 2020 Nicholas P. Fofonoff Award from the American Meteorological Society “For exceptional creativity in the development and application of new concepts in ocean and climate dynamics”. She is the lead principal investigator of the NSF-NOAA Climate Process Team on Ocean Transport and Eddy Energy, and M2LInES – an international effort to improve climate models with scientific machine learning. She currently serves as an editor for the Journal of Climate, a member on the International CLIVAR Ocean Model Development Panel, and on the CESM Advisory Board.

Chair: Jebb Q. Stewart

Presentations:
14:30 - 14:45 - Grey Nearing – "Google’s Flood Forecasting Initiative: Globally-Scalable Models for Hyper-Local Information"
14:45 - 15:00 - Maryam Esfarjani – "Operationalization of AI/ML-Enabled Automated Detection of Fish and Habitats in Optical Surveys "
15:00 - 15:15 - Caleb White – "Wildfire Hazard Mapping: Combining Fire Weather with AI-based Vegetation Fuels Data in the Cloud"
15:15 - 15:30 - Ryan Sobash – "Using machine learning to advance next-day probabilistic convective hazard prediction with convection-allowing models"
15:30 - 15:45 - Manoj Nair – "MagNet: An open data-science competition for space-weather forecasting"
(all times Mountain Daylight Time)

Slack channel: #06_research-to-operation

Chair: V. Krasnopolsky

Presentations:
14:30 – Sue Ellen Haupt (highlighted)– "Using Deep Learning to Downscale from Mesoscale to Microscale in Complex Terrain"
14:45 – Ryan Lagerquist – "U-net++ for emulation and acceleration of a radiative-transfer model"
15:00 – Katherine Haynes – "Using Machine Learning to Simulate 89-GHz Imagery from Geostationary Satellites"
15:15 – Timothy Higgins – "Using Deep Learning for a High-Precision Analysis of Atmospheric Rivers in a High-Resolution Large Ensemble Climate Dataset "
15:30 – Nachiketa Acharya & Kamban Parasuraman – "Generative Adversarial Networks based multisite weather generator (GAN-MulWG)"

Slack channel: #03_weather-and-climate

Chairs: Brian Meyer; Yoonjin Lee

Posters:
Akash Jain, Palantir USG and CTO – "Operationalizing AI: Lessons From the Field"
Aaron Hill – "Medium-range Forecasts of Hazardous Weather with Machine Learning"
Marouane Temimi – "A deep learning-based automated system to monitor hydrological and hydraulic processes in northern watersheds in the US using JPSS VIIRS imagery"
Ouri Wolfson – "MY-AIR: A Personalized Air-quality Information Service"
Nicole Haffke – "Machine Learning for Tornado Detection and Warnings"
Andrew Justin – "Toward Operational Real-time Identification of Frontal Boundaries Using Machine Learning"
Manmeet Singh – "Deep learning-based urban precipitation downscaling for smart city applications"
Alison Jarvis – "Automated solar feature identification using a random forest algorithm"
Garrett Graham – "A Feasibility Study on the Use of Machine Learning as a Quality Control Strategy"

Chairs: Jordan Watson; Yoonjin Lee

Posters:
Ali Muhamed Ali – "Prediction of Loop Current Frontal Expansion Using SSH Contour Lines with a Machine Learning Approach"
Sanjeev Koppal – "A.I.-enabled Underwater Lighthouses"
Marina Chaji – "Expanded standardized spatial data and development of site choice models to support economic impact analysis of offshore energy and aquaculture siting in the Greater Atlantic Region."
Peter Tueller – "FishSense: 3D Imaging for Fish Detection, Measurement, and Assessment"
SUZANNE ROMAIN – "Machine vision species identification for species groupings hard for humans to identify accurately"
Laura McGee – "Cloud-Free Reconstruction of Physical and Biochemical Variables Using a Machine Learning Method "
Erin Moreland – "Artificial intelligence applications in aerial surveys for ice-associated seals and polar bears"
KUNAL BHAVESH MEHTA – "OrcaAL: Active listening and learning for orca sound detection"
Liese Siemann – "Improving automated detection of scallops and flounder in optical surveys with stereo detection methods"
Ashley Woods – "Analysis on American Saltwater Marine Vertebrates’ behavior using a deep neural network to predict upcoming environmental instabilities"
Yawen Zhang – "Automated identification of Atlantic herring schools from decades of multifrequency echosounder data using weakly supervised machine learning"
Wu-Jung Lee – "Data-driven decomposition of echosounder time series for ecological insights"
William Tarpeh – "AI for automated marine data acquisition using novel nitrogen sensors"
Taylor Lee – "Geospatial machine learning to predict seafloor hard grounds/habitats"
Shlomi Ziskin Ziv – "Flash floods prediction using precipitable water vapor derived from tropospheric path delays over the Eastern Mediterranean"
Kshitij Dahal – "Machine learning to estimate precipitation with multiple satellite-based precipitation and gauge observations in Karnali River Basin, Nepal"

Chair: Nikunj Oza

Presentations:
09:30 – Alexei Belochitski (highlighted talk) – "Stable, Accurate, and Fast NN Emulator of Full Suite of Model Physics in NCEP GFS."
09:50 – Yu Huang – "3D-STNN: a Spatio-temporal neural networks for long-term volumetric velocity forecasting of loop current systems"
10:02 – Sébastien Barthélémy & Julien Brajard – "Super-resolution data assimilation"
10:14 – Duncan Watson-Parris – "Model calibration using GCEm v1.0.0 – an open, scalable General Climate Emulator"
10:26 – Ziheng Sun – "Monitoring Power Plant Emission Using EmissionAI"
Q&A 

Slack channel: #08_tools-and-datasets

Chair: Philippe Tissot

Presentations:
09:30 – Igor Granado – "Towards a framework for fishing route optimization decision support systems: state-of-the-art and challenges"
09:45 – Ali Muhamed Ali – "A Transform Model for Expanding In-Situ Velocity Measurement Corrections of Ocean Model to Data Absent Neighboring Model Areas"
10:00 – Megan Carr – "Assessing watershed and meteorological variable importance for predictive water quality models for shellfish waters"
10:15 – Dimitrios Politikos – "Automatic reading of fish age using deep learning"
10:30 – Q&A

Slack channel: #04_blue-economy

Amy McGovern, Jason Hickey - "Looking to the Future of AI & Weather Research"

Christin Khan - "Deep Blue Learning: How AI and Collaborative Innovation is Transforming NOAA Fisheries"

Enrico Camporeale - “Current status and possible path forward for ML-enhanced Space Weather predictions”

Chair: Greg Dusek

Presentations:
12:45 – Maryam Rahnemoonfar (highlighted talk) – "A comparison of Machine Learning and Physics Methods for Simulating Radar Echograms and Ice Thickness in Arctic and Antarctic"
13:05 – Kara Lamb – "Zero-Shot Learning of Aerosol Optical Properties with Graph Neural Networks"
13:17 – Tom Narock – "The Urban Flooding Open Knowledge Network (UF-OKN): AI for flood prediction and risk communication"
13:29 – Manzhu Yu – "Producing high-resolution nitrogen dioxide measurements on an hourly basis using neural networks"
13:41 – Sanjay Agravat – "Enabling Large Scale Enhancement of Weather Research at NOAA via Google Cloud and Open Source Technologies"

Slack channel: #08_tools-and-datasets

Chairs: Benjamin Richards, Jordan Watson

Presentations:
Océane Boulais & Aotian Zhang – "SEAMAPD21: a large-scale reef fish dataset for fine-grained categorization"
Hakob Avjyan – "Detecting coastline change from high resolution satellite imagery utilizing edge detection algorithm"
Vyacheslav Lyubchich – "Machine learning of factors related to production crashes at an oyster hatchery"
Mu-Chieh Ko – "The Development of a Consensus Machine Learning Model for Hurricane Rapid Intensification Probabilistic Forecast"
Caitlin Allen Akselrud – "Issues and best practices in machine learning with ecological time series data"

Slack channel: #04_blue-economy

We're gathering on Wonder.me!

Take a break to process with new and old friends what you've learned this week or an idea that will help you in the future.

ACTIVITY
Slido Poll (Word Cloud): What do you want more of from conferences like this one? Answer the word cloud activity, whether or not you can come to our virtual coffee break.

• Still need help? Please reach out to our Help Desk email, ai.workshop@noaa.gov, or Slack channel, enviro-ai.slack.com, #help-desk.
• Code of Conduct and Etiquette: Please take a few minutes to read our Code of Conduct and Etiquette statement.

 

Chair: Philippe Tissot

Presentations:
14:45 – Elizabeth Barnes (highlighted talk) – "Controlled Abstention Networks: neural networks that say "I Don't Know" to learn better"
15:05 – Sudha Ram – "NEPACCESS: Leveraging AI methods and Tools for Environmental Impact Assessment"
15:18 – Lander Ver Hoef – "An Introduction to Topological Data Analysis for Remote Sensing"
15:31 – Delores J. Knipp – "Into the Storm: Machine Learning Approaches for Solar-Geomagnetic Superstorm Investigation"
15:44 – Eduardo Rodrigues – "Residual Dynamic Mode Decomposition - ResDMD"

Slack channel: #08_tools-and-datasets

Chairs: Benjamin Richards, Jordan Watson

Presentations:
Irina Benson – "Fourier transform near infrared spectroscopy of otoliths coupled with artificial intelligence can predict fish age more rapidly "
Kimberly Sparling, Cory Schillaci, and Jack Tisdale – "Autonomous Vehicles, AI & the Blue Economy: A Cost-Effective Solution to Protect Marine Resources"
Ali Ibrahim – "Fish Acoustic Detection Algorithm for Research "
Stacie Hardy – "Development of an Annotation Workflow and Dataset of Arctic Seals in Multispectral Imagery"
Maike Sonnewald – "Using interpretable and explainable machine learning to reveal the changing North Atlantic circulation under climate change"

Slack channel: #04_blue-economy

Chairs: Rob Redmon; Yoonjin Lee

Posters:
Linsey Passarella – "The Use of a Deep Neural Network to Represent Radiation Transfer Calculations in the E3SM"
Muhammed Sit – "Short-term Hourly Streamflow Prediction with Graph Convolutional GRU Networks"
Marwa Majdi – "Automatic Fog Detection and Visibility Estimation From Camera Images Using Deep Learning Features"
Micheal Simpson – "Deep Learning and Radar-Based Quantitative Precipitation Estimates"
Shiheng Duan – "Predicting daily snow water equivalent with machine learning models"
Stephen A. Shield – "Diagnosing Supercell Environments: A Machine Learning Approach"
Amy Mueller – "Multisensor and data science approaches to enable affordable, scaleable nutrient sensing in marine environments for coastal monitoring and aquaculture management"
Li Xu – "Improve the sub-seasonal forecast skills by the deep learning: a preliminary study by the GEFSv12 extended forecast"

Chairs: Philippe Tissot; Yoonjin Lee

Posters:
Ali Muhamed Ali – "A Transform Model for Expanding Velocity Measurement Corrections by Observed Time Sequences to Future and Past Ocean Model Predictions"
Maryam Rahnemoonfar – "Semantic Scene Understanding after Natural Disasters using Autonomous Aerial Robots"
Nachiketa Acharya, Kyle Joseph Chen Hall – "PyMME: A Python Platform For Multi-Model Ensemble for Climate Predictions"
Tianle Yuan – "The First Global Map of Ship-tracks Reveals a Nexus of Cloud Physics, Environmental Policy, and Maritime Shipping"
Weiming Hu – "Using Machine Learning to Find Better Weather Analogues: Analog Ensemble with Deep Networks"
Hsin-Yi Lin – "Fourier Reservoir Computing for data-driven prediction of multi-scale coupled quasi-geostrophic dynamics"
Alex Ayala – "Conditioned Emulation of Global Climate Models With Generative Adversarial Networks"
Arthur Grundner – "Machine learning-based parametrizations for the ICON model"
Matthew Chantry – "Parametrisation emulators for operational weather forecasting"
Burlen Loring – "Applied Machine Learning Based Extreme Event Detection on Large Scale Supercomputing Facilities with TECA"
Stephan Hoyer – "Machine learning accelerated computational fluid dynamics"
Fang Wang – "On deep learning bias correction for GCMs daily temperature outputs"
Nal Kalchbrenner – "Neural Precipitation Model for Longer Term Forecasting"
Samuel Olumide AKANDE – "Modelling Earth-Observation Assessments and Adaptation of North-Atlantic Coast to Extreme Climate and Ocean Events"

Dr. Rose Hendrix is a Senior Research Software Engineer at the Allen Institute for Artificial intelligence on the Applied Science & Technology team. There, she is the primary AI practitioner for Skylight, a maritime domain awareness tool. Prior to joining AI2, she completed her Ph.D. in Mechanical Engineering from the University of Washington. She completed her undergraduate education at the California Maritime Academy and sailed commercially with an unlimited 3AE steam/motor/GT license before beginning graduate work.

Dr. Ann Bostrom is the Weyerhaeuser endowed Professor of Environmental Policy at the Daniel J. Evans School of Public Policy and Governance, University of Washington. She researches risk perception, risk communication, and decision-making under uncertainty, with a focus on mental models of hazardous processes. Projects include interview, survey, and experimental research on perceptions, communication, and decision-making about climate change, earthquake early warning, and extreme weather forecasts and warnings. Dr. Bostrom earned her PhD. in Policy Analysis from Carnegie Mellon University, her M.B.A from Western Washington University, and her B.A. in English from the University of Washington.

Dr. Julie Demuth is a Project Scientist III at the National Center for Atmospheric Research (NCAR) in the Mesoscale and Microscale Meteorology (MMM) Lab with the Weather Risks and Decisions in Society (WRaDS) research group. Julie has a background in both atmospheric science (BS and MS) and communication (Ph.D.). She conducts research on hazardous weather risk communication, risk perceptions, and responses, and her research includes work with experts—including weather forecasters, emergency managers, and other government officials—and members of the public. Julie is co-leading the risk communication research effort with Ann Bostrom for the NSF AI Institute for Research on Trustworthy AI for Weather, Climate, and Coastal Oceanography, or what is referred to as AI2ES for short.

Jebb Q. Stewart is the lead of the Informatics and Visualization Branch with the NOAA Global System Laboratory in Boulder, Colorado. With a unique background in both Meteorology and Computer Science, he has over 20 years of experience in software development for visualizing, processing, distributing, and interacting with geophysical data. Over the last couple of years, his work has expanded to machine learning applications for object identification and improving data processing capabilities along with leveraging commercial cloud capabilities to provide tools and services to efficiently explore the ever-growing volumes of data.

Lak Lakshmanan is the Director for Data Analytics and AI Solutions on Google Cloud. He founded Google's Advanced Solutions Lab and is the author of several books on machine learning. Before Google, Lak was a Director of Data Science at Climate Corporation where his team developed a rainfall estimation system for precision agriculture. Before selling out to private industry, Lak was a Research Scientist at an NOAA Cooperative Institute at the University of Oklahoma where he developed WDSSII/MRMS, a suite of severe weather algorithms, and wrote numerous, rarely read journal papers. He was recently elected a fellow of the American Meteorological Society.

Hackathon has been increasingly used by AI and environmental science communities as learning opportunitites or to form collaboration cross groups.

This workshop hosted a jointed hackathon in collaboration with Climate Informatics community with support from researchers from the University of Oxford, University of Colorado, Boulder, and Columbia University on three challenging environmental research topics – climate model emulation, wildfire risk prediction, and marine heatwave classification. The hackathon attracted more than 150 registered participants. 

Additionally, NOAA High Performance Computing Program also hosts regular hackathon events in collaboration with NVIDIA to help NOAA scientists to improve their project performance. 

This session invites five teams from the joint hackathon and NOAA HPCC supported GPU hackathon to showcase their results and demonstrating the benefits of hackathon events for workforce development and adoption of emerging technologies. The session will conclude by a discussion with NOAA Center for AI and HPCC program on the benefits of hackathon events.

Session Agenda: 
1. Joint hackathon overview & showcase
    - ClimateBench (Team: bench cloudies, Emulateam)
    - Wildfire Risk Prediction (Team: SEEKingTreesOnFire)
    - Marine Heatwave Classification (Team: JAM & G)
2. NOAA GPU Hackathon showcase - Plankton Automation: Bringing AI and Machine Learning to Zooplankton Ecology (Deana Crouser, NOAA Fisheries AFSC)
3. Panel discussion on hackathon - Bill Cushman (NOAA HPCC), Frank Indiviglo (NOAA HPCC), Julia Levites (NVIDIA), Rob Redmon (NOAA NCAI)

While AI has the potential to address pressing environmental challenges, a recent survey (AI Index Report 2021) highlighted the lack of diversity in the people who are building AI systems. The diversity of the future AI workforce is crucial to solving grand environmental challenges for society. NOAA’s AI strategy outlined promoting AI proficiency in the workforce as one of the primary goals. Meanwhile, NOAA is also committed to workforce diversity by recruiting and attracting diverse and highly-capable talent. This interactive panel discussion brings together champions from NOAA, academic institutions, and professional organizations to highlight the needs, current efforts, and future pathways to diversify AI talent for environmental sciences.

Panelists:
Dr. Shirley M. Malcom is Senior Advisor and Director of SEA Change at the American Association for the Advancement of Science. Dr. Malcom, a behavioral ecologist by training, led AAAS programming in diversity, equity, inclusion, education, and literacy for over four decades. She was a member of the National Science Board from 1993-1998 and served on President Clinton’s PCAST from 1994 to 2001. Dr. Malcom is a Regent of Morgan State University and a member of the board of Caltech. In 2003 she received the Public Welfare Medal of the National Academy of Sciences, the highest award given by the Academy.

Dr. Alyson Wilson is the Associate Vice Chancellor for National Security and Special Research Initiatives at North Carolina State University. She is also a professor in the Department of Statistics and principal investigator for the Laboratory for Analytic Sciences. She is a Fellow of the American Statistical Association and the American Association for the Advancement of Science. Her research interests include statistical reliability, Bayesian methods, and the application of statistics to problems in defense and national security. Prior to joining NC State, she was a research staff member at the IDA Science and Technology Policy Institute (2011-2013); an associate professor in the Department of Statistics at Iowa State University (2008-2011); a technical staff member in the Statistical Sciences Group at Los Alamos National Laboratory (1999-2008); and a senior statistician and operations research analyst with Cowboy Programming Resources (1995-1999).

Dr. DaNa L. Carlis is an award-winning meteorologist and serves as the Deputy Director at NOAA’s Global Systems Laboratory (GSL). At GSL, he is responsible for leading the scientific and information technology efforts of the laboratory. Along with the GSL Director, he leads a laboratory of almost 200 scientists, engineers, and administrators. Prior to GSL, DaNa worked at the Weather Program Office (WPO) in Washington, DC where he was the founding program manager of the Earth Prediction Innovation Center (EPIC). DaNa enjoys the fact that he’s able to work between science, policy, and society to ensure better products and services to the American people.

Dr. Benjamin L. Richards is a Research Fishery Biologist at NOAA’s Pacific Islands Fisheries Science Center, where he leads the bottomfish fish survey team. His work focuses on development of fishery-independent surveys to continually improve the data used in fisheries stock assessment, and on the development of advanced technologies and operational machine learning solutions to optimize data processing and analysis. Dr. Richards currently serves on NOAA’s Artificial Intelligence Executive Committee and provides advice and service to the NOAA Center for Artificial Intelligence. He served as the chair of the NOAA Fisheries Automated Image Analysis Strategic Initiative, which received the 2019 US Department of Commerce Gold Medal for Scientific and Technical Achievement. In 2020, he was awarded the US Department of Commerce Silver Medal for enhancing stock assessment methodologies using camera-based surveys. In 2021, he received the NOAA Fisheries Silver Sherman award for safe and successful implementation of the bottomfish survey under COVID.