The Mars Rover's driver

This weekend's Wall Street Journal features an interview with Vandi Verma, a naturalized US citizen from Indian, whom the author Tunku Varadarajan describes as "arguably the world's most experienced Martian robot operator".  Verma worked with Curiosity previously, and serves as chief engineer of robotic operations for Perseverance and Ingenuity.  DTLR commends the piece to all its readers.

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.

A couple of gems from EOS

I would like to direct readers to a couple of old Eos articles from 2016 that I found particularly engrossing.

The first is an article by JoAnna Wendel on dwarf planets.  I vaguely remember Pluto being dethroned from the status of planets some 15 years ago, and placed into the category of dwarf planets.  I remember agreeing with the reasoning, at the time, but didn't pay much attention to dwarf planets.  Wendel's article made me realize that dwarf planets constitute a rich set of objects, integral to our solar system, and as worthy of attention as their big siblings.

The second article by Jeffrey J. Love and Pierdavide Coisson, is about a cluster of geomagnetic storms triggered by solar activity in September 1941, and their effects on the earth during WWII.  There hasn't been such a cluster of geomagnetic within a single 14 hour period since then.  Our civilization would be much more vulnerable now to such an event than we were then, but the article tells of several examples of how life on Earth was disrupted.  It shows the importance of studying, and perhaps forecasting, space weather.