Some good stuff in Quanta magazine

I'd like to highlight a couple of excellent physics articles that appeared in Quanta magazine last month.

First, Thomas Lewton profiles Jonathan Oppenheim's work on hybrid classical-quantum theories of quantum gravity.  The idea seems to be that instead of attempting to quantize the gravitational field, let it remain classical.  To reconcile quantum uncertainty with a classical spacetime, gravity must be stochastic; it must be noisy.  

Second, Katie McCormick discusses a topological insulator analogy that has been used to explain atmospheric motions such as Kelvin waves in the Earth's atmosphere.  Taruh Matsuno's successful prediction of equatorial Kelvin waves in the 1960s was, evidently, one of the only times theoretical work in geophysical fluid dynamics was predictive of phenomena later discovered in nature.  The article focuses on Brad Marston and collaborators' theoretical and observational work demonstrating that Matsuno's waves can be understood using a topological insulator analogy (think quantum Hall effect).  Once again, a theoretical prediction (Poincare gravity waves in the stratosphere) was subsequently confirmed observationally.  Finally the article discusses David Tong's quantum field theoretical framing of coastal Kelvin waves.

I had attended a talk by Marston at the APS March Meeting earlier this year (see my earlier post), but did not quite follow it.  I am grateful to Quanta magazine for distilling the story into a form that can be consumed by a wider public (including me).

More on machine learning in weather and climate modeling

 My last post commented on the use of machine learning in Earth science modeling.  The Philosophical Transactions of the Royal Society, Series A, has just published a special issue on this topic, based on a workshop held at Oxford University in September 2019.  Here is a link to the introductory paper, and a link to one of the contributions, that deals with physically-aware machine learning models.  Ten case studies are discussed in the latter.

Physically aware machine learning models

My last post resurrected some Eos articles from a few years ago, but today I'd like to discuss a piece in this month's issue.  Maskey et al. write about "A Data Systems Perspective on Advancing AI", reporting on a NASA-sponsored workshop held in January.  They describe "traditional" Earth science modeling as "top down", starting with first principles (laws of physics), while the machine learning approach is "bottom up", having algorithms that learn relationships empirically from historical data.  An inherent limitation of empirical modeling, they recognize correctly, is the inability for a model trained on historical data to extrapolate into regimes never before seen in the training data.  Yet this is precisely what Earth science is called to do, when dealing with extreme weather events or climate change, for example.  The writers propose that "physically aware machine learning models" could overcome this limitation, suggesting a melding of the "top down" and "bottom up" approaches.  The authors mainly write about using physics to constrain the machine learning models or their cost functions during training, claiming promising results already.  It is less clear to me that placing constraints on an empirical model would allow it to creditably extrapolate, only that such constraints should improve interpolation capability.

On this blog, it was noted previously that there have been demonstrated cases of deep neural networks actually being capable of generalizing beyond the training data, though such cases are not well understood, and are not convincing unless validated in independent data.  From the context, it did not seem like these deep learning models were of the "physically aware" variety that Maskey et al. describe.

These are early days in the efforts to apply machine learning to physical problems.  We still have much to learn about what is possible, and what remains limited, with such efforts.

Congratulations to Alvin

Last month, the Deep Submergence Vessel (DSV) Alvin celebrated its 50th anniversary of service to oceanographic research.  As Humphris et al. (2014) note, Alvin is "the world's first deep-diving submarine and the only one dedicated to scientific research in the United States."  Named after geophysicist Allyn Vine, the half-century old submarine is returning to service after a major upgrade this year.  I recommend the article by Humphris et al. (2014) for readers interested in learning about the history of this unique vessel.

Reference

S. E. Humphris, C. R. German, and J. P. Hickey, 2014:  Fifty years of deep ocean exploration with the DSV Alvin.  Eos, Transactions, American Geophysical Union, 95 (22):  181-182.

Congratulations to Geophysical Research Letters

The American Geophysical Union (AGU) is celebrating the 40th anniversary of its letters journal, Geophysical Research Letters.  They've posted a collection of 40 papers published in the journal in past decades, a sort of 'greatest hits' list, and made them open access.  You can find it here.

I only recently joined the AGU and have not been an avid reader of this particular journal, yet I congratulate AGU on this milestone.

Congratulations to the winners of the International Data Rescue Award in the Geosciences!

Last month at the American Geophysical Union fall meeting, the above award was given to the Nimbus Data Rescue Project of the National Snow and Ice Data Center, an organization at the University of Colorado, Boulder, funded by NOAA, NASA, and NSF.  According to Showstack (2014), the Nimbus project is "recovering, reprocessing, and digitizing infrared and visible data from the NASA-funded Nimbus 1, 2, and 3 weather satellites, the first of which launched in 1964.  None of the early Nimbus data had been available for 4 decades because of archaic data formats and difficulty in accessing film rolls."  Three runners up for the prize were also chosen: you can read more about them here as well as in Showstack (2014).

The award is sponsored by the Integrated Earth Data Applications group at Columbia University, and Elsevier, who has a business interest in data stewardship services. Elsevier might be considered a controversial player, as many researchers are unhappy with their allegedly predatory journal pricing policies. Nonetheless, we should praise them when they do something right, an this appears to be a rare example.

I've written previously about data stewardship issues here and here. I am pleased that the scientific community is giving more and more attention to such issues, as exemplified by this new award program. I hope that it continues and that similar awards emerge for other sciences.  I am also more than relieved to begin a new year of blogging with some positive news.

Reference

Randy Showstack, 2014: Award program recognizes efforts to protect geoscience data. EOS, Transactions of the American Geophysical Union, 95 (1): 2.