W.O.M.B.A.T. | Stats + Short Stories Episode 324 / by Stats Stories

Noel Cressie is Distinguished Professor at the University of Wollongong, Australia, and Director of its Centre for Environmental Informatics, which is a vibrant interdisciplinary group doing research in spatio-temporal statistics, satellite remote sensing, and broader fields of environmental science; he is also Adjunct Professor at the University of Missouri and Affiliate at NASA’s Jet Propulsion Laboratory in the USA. Noel grew up in Western Australia, received a PhD from Princeton University, and shared a career between the US and Australia. He is author and co-author of four books, three of them on spatial and spatio-temporal statistics, and of more than 300 peer-reviewed publications. His recent research involves hunting for atmospheric-carbon-dioxide sources around the world and focusing on Antarctica’s environmental future. He has won a number of awards, including the Fisher Award and Lectureship from the Committee of Presidents of Statistical Societies (COPSS), the Pitman Medal from the Statistical Society of Australia, the Barnett Award from the Royal Statistical Society, and the Matheron Award and Lecture from the International Association for Mathematical Geosciences. Noel is a Fellow of the Australian Academy of Science, of the Royal Society of New South Wales, and of a number of other learned societies.

Episode Description

Would you be surprised if a wombat won a statistical achievement award? well our guest Noel Cressie is here to talk about the WOllongong Methodology for Bayesian Assimilation of Trace-gases and how it can impact the environmental landscape.

+Full Transcript

John Bailer
Would you be surprised to learn that a wombat received the Outstanding Statistical Application Award, the Statistics and Physical Engineering Sciences Award, both from the American Statistical Association, and the Mitchell Prize from the International Society for Bayesian Analysis last year? Well, maybe not such a surprise if you learned that wombat is the methodology for Bayesian assimilation of trace gases. And they're working on a framework for estimating carbon dioxide co2 sources and sink or flux as well at the Earth's surface on a global scale. Today's episode of Stats and Short Stories will feature a conversation about why understanding co2 sources and sinks matter for evaluating global climate change. I'm John Bailer. Stats and Stories is a production of Miami University's departments of statistics and media, journalism and film, as well as the American Statistical Association. Rosemary Pennington, professor of journalism, is away. Our guest is Noel Cressie, distinguished professor at University of Wollongong, Australia and director of its Center for Environmental Informatics in an interdisciplinary group doing research in spatial temporal statistics, satellite remote sensing and broader fields of environmental science. He's also an adjunct professor at the University of Missouri and an affiliate at NASA's Jet Propulsion Lab in the US. His recent research involves hunting for atmospheric carbon dioxide sources around the world and focusing on Antarctica's environmental future. Noel, first, congratulations on your team receiving this outstanding affirmation of the work that you're doing. Can you tell us a little bit about wombats?

Noel Cressie
Yes, thanks, John. Really good to be here. And to be talking about my favorite Australian animal. It's a cute little cube shaped animal with a very large nose and it has a pouch. But it has a reverse pouch because it does a lot of digging and burrowing. And if the pouch faced forward, there would be a lot of dirt in the pouch, and the baby would actually be spitting out dirt all the time. So back to wombat. Now, wombat, the wall and gold methodology that we developed. It really started when a colleague of mine, came as a postdoc, to the University of Wollongong in 2014. And we're talking about all sorts of things, he was my postdoc. And we got involved in a lot of spatial statistics. And at some point over a cup of coffee, I think we said we were both interested in what we call flux inversion. Now, flux inversion is to try to work out where the carbon dioxide comes from, at the birth surface. All you can measure though, is the carbon dioxide in the atmosphere. So what you have is an instrument like The Orbiting Carbon Observatory from NASA, you have that instrument that's measuring the concentration of carbon dioxide in the atmosphere, from a height of 700 kilometers. And what you want to do is trace those carbon molecules back to where they come from the surface. And they come from photosynthesis of trees, and they come from respiration. You know, when the trees die off, then carbon dioxide is actually released. And that's the source. And of course, the photosynthesis is a sink when they're taken in. And so there's something called a carbon cycle. And that carbon cycle in the atmosphere is what we need to understand in order to mitigate climate change. And there is this satellite referred to as the Auburn cabinet observatory to satellites, of which I'm a team member. And now most of my team are part of that membership as well. So Andrews and I haven't discovered coffee, we're both interested in flux and version. And we both agree that it's an impossible problem that can be made possible by being a Bayesian. And so what a Bayesian does is actually put a prior distribution on what this transport might be like, that takes the flux forward into the atmosphere. And we want to invert that transport and take it back to where it actually came from. And we do it globally over a period of time on the order of six years. It was the actual paper that won the award we did over six years.

John Bailer
So when you are building these kinds of models, did you do it for sort of a window of time and then use it to predict into the future? What are some of the things that you did in terms of validation of this?

Noel Cressie
Right, so we started out with something called an OSSE and that's not an Aussie like from Australia. That's an OSSE, which is an observing system simulation experiment. And so we validated by actually building if you like a digital twin in a computer that was a satellite moving over the computer generating satellite sampling concentration, but where we knew what the flux was, that was being output, you know, from Earth's surface, so we knew the answer. And so in that assay, we generated data that would look like what a satellite would retrieve. And we did the fluxion version, we used the Bayesian methods, and we looked at the accuracy and validity of what we obtained. And we didn't get the answer right the first time, there were a lot of dials to turn, as you can imagine, it's a fully global system, over a period of actually, the whole carbon cycle works in a period of 20 minutes in terms of the actual transport. And so there's a certain amount of aggregation to do spatially and temporally. And we worked out the units of that. And we built something called a bamboo, basic aerial, temporal unit. And we work within those batters to come up with the flux inversion.

John Bailer
So after doing all this, what are some of the insights that you've gained about co2 production sources and sinks?

Noel Cressie
First of all, when we actually ran one bad, we got rid of the fossil fuel component. And that is a huge component. And that tends to dominate. So we were interested in the natural carbon cycle. And since we were interested in whether carbon cycles were changing or not, I might sort of make a call out to a postdoc who joined us on this research, which was supported by the Australian Research Council. His name is Michael Bertolacci. And the three of us, Andrew, Michael, and I represent the core statistics team. We had interdisciplinary input from atmospheric chemistry, and from IT support and various other places. So basically, what did we learn? Well, in the work with Andrew and Michael, we're learning that the cycle is changing, that in the northern hemisphere, you know, the onset of photosynthesis, that is causing the carbon dioxide to be absorbed into Earth's surface is moving by the order of one or two days. Now, that might not seem a lot. But that phase change is actually significant. If I could put it that way, in the broad sense of the term, it is real from a signal processing point of view. And the Bayesian inversion that we get gives us those prediction intervals that tell us that the carbon cycle is changing, because of the excess carbon dioxide in the atmosphere.

John Bailer
That's a pretty scary result. So will these models make it possible also to think about if we change certain characteristics of the carbon production? How might that change? Ultimately, the models? Could the model be used to help predict changes given interventions?

Noel Cressie
Yes. So currently, we're not doing any forecasting. But what the model does is allows us to then make conjectures or hypothesis tests, or sorry, or this generation of what might be causing these changes. Unfortunately, that particular data set that we have from The Orbiting Carbon Observatory does not do polar regions very well. And that's because the satellite doesn't have a good enough signal to noise ratio to do the retrieval. But generally speaking, there is now moves afoot. Because we believe that there's going to be a certain amount of outgassing of carbon dioxide in the polar regions, both methane and carbon dioxide coming from, you know, permafrost melting. And so, you know, there's this evolution of science, if you like, we're seeing some of these things happening in the northern latitudes. And there's an attempt now by NASA to come up with retrieval mechanisms that would allow them to get sort of better information in the polar regions.

John Bailer
This is fascinating work, but I'm afraid that's all the time we have for this episode of Stats and Short Stories. And thank you so much for joining us today. And, and thank you for this excellent work.

Noel Cressie
Thank you John, and signing off on behalf of the wombat team.

John Bailer
You’re welcome. Stats and Stories is a partnership between Miami University’s Departments of Statistics, and Media, Journalism and Film, and the American Statistical Association. You can follow us on Twitter, Apple podcasts, or other places you can find podcasts. If you’d like to share your thoughts on the program send your email to statsandstories@miamioh.edu or check us out at statsandstories.net, and be sure to listen for future editions of Stats and Stories, where we discuss the statistics behind the stories and the stories behind the statistics.