NSF terminates funding for its main meteorology focused AI institute

In weather, lots of potential threats in the West this week

Starting off this morning's newsletter talking about recent developments with regard to federal science agencies and funding. First, I wanted to highlight a recent forum I participated in here in Oklahoma that was organized by KGOU, our local National Public Radio affiliate. A number of national and regional meteorological institutions affiliated with the federal government and the University of Oklahoma call Norman home, and this forum was a live event to bring awareness to the community about potential cuts and risks to these entities. Myself, Keli Pirtle, the recently retired NOAA communications lead for the Norman NOAA community, and Dr. Renee McPherson, University of Oklahoma professor and Director of the South Central Climate Adaptation Science Center (CASC) were the panelists, answering questions both from the KGOU reporters and a live audience. You can listen to the entire forum (about 1 hour and 15 minutes) here.

While much of this forum focused on NOAA given its major role in most of the major weather related entities in this region, we did also talk about cuts and reductions to the US Geological Survey (USGS), which funds the CASC, and the National Science Foundation (NSF), which funds a tremendous amount of scientific research nationally, including in the earth and atmospheric sciences. A key institute that NSF funds here in Norman is AI2ES, a multi-institution research consortium led by the University of Oklahoma focused on developing "trustworthy artificial intelligence (AI) methods for Earth system and environmental science users that will revolutionize our understanding and prediction of high-impact atmospheric and ocean science phenomena and create new educational pathways to strengthen the AI and environmental science workforce." AI2ES was launched as an inaugural NSF AI institute in 2020, and in its first 5 years has already conducted research that has resulted in more than 130 peer reviewed publications and the mentoring of more than 75 graduate and undergraduate students.

It was recently confirmed to AI2ES director Dr. Amy McGovern that NSF will not be funding a second five year grant period for AI2ES. While Dr. McGovern and the leadership at AI2ES will try to keep AI2ES going with other grant funds, this NSF decision is obviously a huge blow and will have serious negative impacts on the development of AI focused R&D in the earth, atmospheric and oceanic sciences.

While I know that AI (particularly generative AI) is getting a lot of "mixed" attention in the public and media these days, as I have written in earlier posts I have been amazed by the rapid progress that AI has made in meteorological R&D. While I am by no means an AI expert, my sense is that the scientific challenges presented by meteorology -- and specifically forecasting -- are particularly amenable to AI approaches, and we have seen a number of significant advances in the development of AI based weather modeling and forecasting systems. Just last week, my colleague James Franklin, the retired forecast branch chief at the National Hurricane Center, posted analyses showing that the Google DeepMind AI system has produced the most accurate track forecasts for Atlantic tropical systems this season.

AI2ES has not only been working on specific AI related science projects, they have been doing the important work of developing research approaches to ensure that the AI techniques developed will be usable for and trusted by scientists. They have also been conducting social science research on how the public will utilize and trust (or not) AI developed output for applications like weather forecasts and warnings.

The list of scientific progress that is being decimated by the current presidential administration is of course long and growing, but the loss of this funding is to me a particular blow given the early stage of AI development and what I view as the potential for AI to revolutionize the field. Applying AI to meteorological research and operations is an area in which China is investing a tremendous amount of resources — so the more we reduce our investment, the more likely we are making it that China (or other nations) will lead in the development of this new technology rather than our own country.

In other federal government science news, a self-organized group of 85 scientists released a review this morning of the Department of Energy (DOE) Climate Working Group report that the Environmental Protection Agency (EPA) is using to reconsider its finding that climate change is endangering human health. You can find the review of the DOE report at this website, and these paragraphs from the site summarize their findings:

Our review reveals that the DOE report's key assertions—including claims of no trends in extreme weather and the supposed broad benefits of carbon dioxide—are either misleading or fundamentally incorrect. The authors reached these flawed conclusions through selective filtering of evidence ('cherry picking'), overemphasis of uncertainties, misquoting peer-reviewed research, and a general dismissal of the vast majority of decades of peer-reviewed research.

No one should doubt that human-caused climate change is real, is already producing potentially dangerous impacts, and that humanity is on track for a geologically enormous amount of warming. No one knows what socioeconomic impacts of this warming will be. It should also be clear that the DOE report's approach to undermining scientific evidence mirrors tactics previously employed by the tobacco industry to create artificial doubt.

Though our review doesn't explicitly address the EPA's 2009 Endangerment Finding, the process of compiling our response made one thing clear: in the 16 years since that finding, evidence of human-caused climate change and its threats to public health and welfare has only strengthened.

The last sentence really hit me because it resonates with my own personal experience of being an early skeptic of how much climate change would actually impact weather and human risks in my lifetime to being convinced by evidence and experience of the growing threats to society we face right now. The funding cut to AI2ES is the latest example of how we are actively reducing our ability to use science to protect society from those weather and climate threats.

Turning to those imminent weather threats, this week the main focus for impact weather will be in the western United States and Canada. This morning, clusters of thunderstorms producing frequent cloud to ground lightning have been moving through central California as shown in this morning satellite loop with GOES Lightning Mapper lightning flashes overlaid. While these storms are not completely dry, they are moving through an area with very dry conditions and SF Chronicle meteorologist Anthony Edwards wrote this morning about wildfires already being ignited across the region by this thunderstorm activity. (Note that that in the satellite loop above you can see the hotspot of a large wildfire just east of the thunderstorm activity in the overnight infrared data, and when the sun comes up you can see the large smoke plume pushing north of it.)

As this thunderstorm activity builds north today, it will move into an even drier low level environment, meaning any wetting rainfall will become less common and lightning will be even more dangerous as far as igniting new wildfires. Red flag warnings are in effect for much of the Pacific Northwest the next couple of days.

This region will also be seeing record heat much of this week, with both record highs (today’s forecast record highs shown above) and record warm minimum temperatures expected. Conditions are expected to be even more anomalous north of the border in western Canada, where the national September record maximum of 40C could be eclipsed this week in interior British Columbia. Numerous large wildfires are already burning across the Pacific Northwest and western Canada, so these weather conditions are of serious concern for aggravating these issues.

Meanwhile, in the usually dry desert Southwest, the potential for thunderstorms and heavy rainfall will be the focus this week. Tropical Storm Lorena has formed in the eastern Pacific off the western Mexican coast, and as we have been talking about for a few days now, the track of this system looks to bring a surge of moisture into the Southwest over the next several days.

The threat for thunderstorms with heavy rainfall and localized significant flash flooding should increase through the week, but as the graphic from NWS Tucson notes the forecasts for the most impacted locations and magnitude of rainfall and impact will evolve as we monitor the evolution of Lorena. The monsoonal moisture feeding the thunderstorms over California this morning will bring a risk of localized flash flooding today to parts of the region, and a flash flood watch is in effect for the deserts of southern California.

As expected, Kiko has entered what looks to be a rapid intensification phase, and is now an 85 mph hurricane as it continues to move slowly west over the east-central Pacific. It is now forecast to become a major hurricane over the next 24-36 hours before gradually weakening as it turns more west-northwest. It is still at least a week away from any impacts to Hawaii, and the overall situation has not changed much since yesterday's tropical post, with the most likely scenario being a weakening system bringing some rain and wind to the state.

Not a lot has changed in the Atlantic either. NHC continues to monitor the area off the west coast of Africa near the Cabo Verde Islands for development, and has increased the likelihood of a tropical cyclone forming over the next week to 70%. The thunderstorm activity in the region is disorganized right now, and to me it looks like the most likely path to development will be a rather complex interaction of a couple of tropical waves. Where exactly a system congeals and how quickly it develops will play a huge role in an eventual track, but overall the system should move generally west (European ensemble envelope of day 6 forecasts in red circle above) under a large subtropical high pressure area.

We also continue to keep a little watch on the old front over the Gulf and southwest Atlantic. Old fronts over warm water are always a little concerning in September, and the models continue to sporadically try to generate a weak system as shown in the yellow area on the European ensemble above. Right now though there is nothing to focus on for any specific risk.