This transcript was generated automatically and lightly edited for readability.

Oliver Strimpel: This is Geology Bites with Oliver Strimpel. The Earth's magnetic field changes over time. In fact, there have been frequent reversals in the direction of the field, with the north and south magnetic poles switching places. We tend to think of these changes as applying to the whole globe. But in addition to this temporal element to the magnetic field's variability, there is also a spatial element, with the field changing very differently in different places across the planet.

Cathy Constable is a professor at the Scripps Institution of Oceanography. She measures the remanent magnetism in rocks and human artifacts around the globe to reconstruct not just the variations of the Earth's magnetic field as a whole, but also its spatial variations across the globe over geological time. Cathy Constable, welcome to Geology Bites.

Cathy Constable: Hello. It's nice to be here.

Oliver Strimpel: It seems hard enough to reconstruct the history of the Earth's magnetic field in one dimension, the temporal one, but uncovering the spatial distribution of the field over time seems really challenging. Are you really able to create a spatial map of the field at different times over the past?

Cathy Constable: Yes, but the map is a little bit crude. The temporal dimension, as you describe it, really only reflects the field that we might expect from a magnetic dipole, if we imagine the familiar permanent bar magnet at the center of the Earth. But we really know that the field is more complicated, because it's driven by convection in the liquid outer core. So having better mapping in both space and time is actually essential to getting better acquainted with what's going on in the core. And we've been successful in building more detailed time-varying spatial maps for the past 10,000 and the past 100,000 years. The temporal resolution is better for the more recent times, and in both cases, as I said, the spatial resolution is a bit crude. It's still only on rather large scales.

Oliver Strimpel: So what are the materials you analyze to do this work?

Cathy Constable: There's a broad range of materials, actually. Human artifacts that have been heated to a high enough temperature to reset the magnetization as they cool in place: that would be things like fired pottery, bricks, hearths where there have been high-temperature fires, kilns. And there's a decreasing number of these human artifacts as one goes back in time. Beyond about 3,000 years ago, there aren't that many of them, and the spatial distribution is limited by human activity. Rocks, in fact, cover a much broader temporal range. If we think about lava flows, they solidify and cool, and then they preserve an instantaneous, geologically speaking, spot record of the field. Lake and marine sediments give a more continuous record, one that tends to be smoothed in time by an amount that's dependent on the sedimentation rate. So all of these things contribute to our spatial reconstructions.

Oliver Strimpel: So are such suitable materials conveniently distributed across the globe for you to collect?

Cathy Constable: No, not at all conveniently. As you know, about 70% of the Earth's surface is ocean. There's a lot more ocean in the Southern Hemisphere than in the Northern Hemisphere, so there's a huge challenge in the distribution, especially in the Southern Hemisphere. And it's also true that there's less human activity at high latitudes, and fewer human artifacts there. And if we look at marine sediments, which can be recovered from the ocean, they tend to have lower accumulation rates than the things that you get on land, like lake sediments, so the temporal resolution isn't as good.

Oliver Strimpel: It seems as if we can get a pretty good record of reversals from many, many magnetic stripes that are frozen into the ocean crust as it's formed and spreads on either side of the mid-ocean ridge. Are these some of your best sources?

Cathy Constable: They're very good on some timescales. What happens is that the magnetic stripes are limited by the processes forming the ocean crust, and also by the fact that measurements are usually made at the ocean surface, several kilometers from the source. So the temporal resolution is actually a lot better in lava flows and in sediments with high accumulation rates; they do a better job. And so there are some fortunate geological circumstances where, for example, at Steens Mountain in western North America, 15 million years ago, there were a series of lava flow eruptions that provided a wonderful record of a magnetic reversal, and you can actually see the evolution of this magnetic reversal through a record preserved in these lava flows. And there are also records in sediments, an example from Italian sediments. And these two records are very interesting, because even though they are somewhat controversial, they preserve very high rates of change, or appear to preserve very high rates of change, in the magnetic field, by a matter of degrees per year in terms of the direction. And that's something that is not necessarily completely expected, and we're exploring it a lot further as part of the work that we're doing, where we build a global view of what's going on. In more recent times, these things can really give us very good temporal resolution.

Oliver Strimpel: Tell us a bit more about why we are so interested in reconstructing the spatial distribution of the field.

Cathy Constable: So my interest is purely in the geomagnetic record and getting a better understanding of how the magnetic field is produced in the Earth's core and what controls its variability. If we think about the limitations of the seafloor stripes, they tell us about normal or reverse polarity directions rather than the details of field strength variations, and the resolution is typically no better than about 20,000 to 40,000 years. They also only give us the polarity of the field. They don't give us the details of directional variability, or even field strength variability, around the globe. And the spatial structure of the field does vary with time, and the rates of change in the field also depend on location and strength. So we have to go to a broader understanding of the spatial scales to get this. Now, numerical simulations of the core field provide another way to look at how the field might vary in time and space, but they also have limitations. We know the equations to use, but it's not currently possible to do the computations with the right values for core properties.

Oliver Strimpel: And why is that?

Cathy Constable: The reason is that the core properties are just very challenging for numerical simulations. The viscosity in the core turns out to be very low. What that means is that the numerical simulations would ideally have very high spatial resolution and take very short time steps in order to resolve what's going on on very small spatial scales, and that's very numerically challenging. The viscosity is basically expressed in the dimensionless Ekman number, and it's effectively orders of magnitude too high in the numerical simulations that we're currently using. It's too hard to compute if we use the correct numbers. And what happens then is that there are issues in converting the nondimensional numbers that are used in the equations in these simulations to actual dimensional timescales for the Earth. Usually we have to choose between scaling according to an advection or a diffusion timescale, depending on what we're interested in and the relative importance of advection versus diffusion in the processes we're looking at.

So diffusion is basically what happens as a result of ohmic heat losses in the core: electrical losses lose energy in the magnetic field because of the finite conductivity of the core. The advection basically is a product of the convection that drives the magnetic field in the core. And with advection, what happens is that on short timescales and large length scales, the advection carries the field around with the fluid. And we can see this actually in maps of the modern-day field, where you can see some features of the field that are basically drifting westward or moving around the polar region. And the advection is dominant on short timescales, whereas on longer timescales the diffusion becomes important, and that's one of the things that has to play into the loss of energy in the dipole part of the field when you get an excursion or a reversal.

Oliver Strimpel: Just to come back to the problem of getting the right viscosity into the models: roughly, what is the viscosity of the core? Can you compare it to something we're familiar with?

Cathy Constable: The viscosity of the core is really like that of water. And you might contrast that with the silicate mantle of the Earth, which has a viscosity that is orders of magnitude higher, and therefore the timescales for convection in the mantle are much, much longer than they are in the core. If you think about the convection, advection timescales in the core, they're of the order of hundreds of years, versus, we believe, millions of years for most of the processes that are going on in the mantle. And so in trying to mimic the viscosity in the core, the best numerical simulations right now are using this dimensionless Ekman number that I mentioned of the order of 10 to the minus 6 or 10 to the minus 7, and we believe that the correct number for the core is somewhere between 10 to the minus 10 and 10 to the minus 15.

Oliver Strimpel: Wow, not even close.

Cathy Constable: Not even close. And that's why we have to be very careful about drawing reliable physical conclusions about what's going on in the core from these numerical simulations. It's not that they're wrong in what they do; it's that they're just not necessarily completely appropriate for what we want to use them for when we're comparing with the geomagnetic field.

Oliver Strimpel: So we're mapping the field here on the Earth's surface with the goal of improving our understanding of what's going on in the core, but between us and the core lie 2,900 kilometers of mantle. Is the mantle completely transparent to the magnetic field?

Cathy Constable: No, it's not, but it is to a good approximation for what we do. The mantle is basically made of silicates, and it has a finite but low electrical conductivity. The conductivity in the core is much, much higher. In fact, studying the conductivity of the mantle is another field entirely in its own right, and that's done using magnetic and electrical data as well. So the effect of the mantle, when we're looking at the signal that's coming out of the core, is basically to smooth things out. That's what we could call a temporal damping in the signal, so that everything that comes to us is smoothed over a period of a few months. So we can't detect very short-term variations that might be going on in the core. And that's the main effect, and it doesn't matter for the purpose of looking at paleofield structure, because the temporal resolution that we have in the best of circumstances might be a decade to a hundred years.

Oliver Strimpel: What kind of things are we looking to pin down better using your spatial measurements? Can you constrain the values, for example, of the various parameters that you mentioned, the viscosity in the Ekman number? And there's also the Rayleigh number, which defines the amount of energy in the convecting core.

Cathy Constable: Not directly. The best estimates that we have for these parameters basically come from mineral physics and petrology, and our understanding of the materials in the core and the temperature of the core. But in the spatial representations we have, we can look at representative properties of the Earth's field, and we can compare them with the results from simulations to decide if the simulations are doing a good job. And one issue that we address, which is an interesting one, is whether there are long-term geographic variations in the spatial structure that reflect boundary conditions at the core-mantle boundary. For example, there's some evidence that the field appears to be more active in the Southern Hemisphere, and also more active in the Atlantic than in the Pacific region. And this builds on a bigger picture that we get from seismic tomography, which supports differences in the region around the core-mantle boundary. So that's something that we can actually see the consequence of in the magnetic measurements, and we can then go back to the numerical simulations and say, well, actually, if you want to get a realistic representation of the geomagnetic field, you need to include these spatially varying boundary conditions in the numerical simulations. And that's one part of it. And then there are other things that are harder to deal with in terms of the constraints, which we don't really have a systematic way of addressing.

Oliver Strimpel: You said the field actually doesn't generally conform to a simple dipole. But can you describe what your spatial-temporal map of the Earth's field actually does look like?

Cathy Constable: Yes. It depends on the time interval that we look at. If we look at long-term averages over, say, a million years, it does look a lot like a simple dipole, with direction varying around the Earth as you might expect for a simple bar magnet at the center of the Earth. But over time, the direction and the magnitude can vary a lot, and it doesn't always look like a dipole. So as I mentioned, on average there are indications that you see more activity in the field variations in the Southern Hemisphere than in the north, and also in the Atlantic hemisphere (the piece of the Earth that includes the Atlantic and the Indian Ocean) versus the Pacific hemisphere. And the Laschamp excursion, which occurred about 41,000 years ago, is one example where the dipole and the overall field decay in strength, and structures really get quite complicated. Rates of change in the field can also be quite variable over time, and also with position, and we can actually estimate these from our time-varying representations. So we find the fastest directional changes at equatorial to mid-latitudes; we find the fastest field strength changes at high latitudes; and the most rapid directional changes occur when the field strength is lower.

Oliver Strimpel: You mentioned the Laschamp excursion. What exactly is that? Is that a very special occasion, or was it just the most recent such event?

Cathy Constable: So we think that excursions are basically failed reversals: that they occur when the field strength drops, and in principle the magnetic field could reverse, but something prevents that from happening. And one possible reason might be that the reversed field isn't able to establish a hold in Earth's solid inner core, and so it's just kind of moving around in the outer core, and then the inner core, which retains the original polarity, allows the original dipole polarity to reassert itself over time. So there are lots of excursions that have been found in the paleomagnetic record. The Laschamp excursion is the best documented one. It's also one of the more recent ones. There are suggestions of a couple more that occurred since then, but we don't find a global extent for these, so maybe those are things that didn't quite even make it to excursion status. I should have said that the overall strength decays, and once the dipole strength decays, then the pieces of the field that are not represented through the dipole become more dominant. You could think of them as being background noise, and they basically make the field look much more complicated during an excursion.

Oliver Strimpel: Do you have a qualitative understanding of what's going on in the core that makes this happen?

Cathy Constable: The global picture that we have now of excursions suggests that they're mostly related to the loss of dipole strength. The highest field strength occurs at high latitudes, so we get rapid changes there as the dipole strength drops. And the other thing that we think is that excursions may be initiated at low latitudes by the growth of reverse flux patches, or reverse magnetic field, at or near the core surface, and these may migrate poleward and result in a full-scale polarity reversal.

Oliver Strimpel: Are there any other qualitative findings you see in your paleo-spatial map?

Cathy Constable: Our time-varying paleofield models, I think, is what we call them. But you know, one person's model is somebody else's idea of reality. So if we look over the past two million years, one of the things that we see is that on average the dipole decays more slowly, basically on diffusion timescales, which are of the order of 30,000 years, than it bounces back. And this is a fairly subtle result, in the sense that there's a lot of variability in the dipole moment, where it goes up and down anyway. But when you look at it on long timescales, it's decaying more slowly than it comes back, which suggests that in bouncing back, the convection, or the advection of the field to produce new field, is a more important process. The other thing that we see is that if we look at the long-term decay of field strength going into excursions or reversals, those decay timescales are of the order of 10,000 to 15,000 years, but the directional changes can be much more rapid, and they can be as large as one, two, maybe even 10 degrees per year.

Oliver Strimpel: These are things that we would expect to be able to find in the core models, the numerical models that people make about the core. Do these observations sort out the models? You said it wasn't specifically good enough to tell you what the Ekman number should be or what the Rayleigh number should be, but does it nonetheless sift out some models from others in a fruitful way?

Cathy Constable: It does, yes. The simulations are numerically precise, and they can be used to predict anything that we could observe and make into our paleomagnetic time-varying models. But even though they're numerically precise, they're not yet universally successful. And of course, paleofield data are messy, so we need to be careful about demanding precise agreement. But many simulations do not show this asymmetric growth and decay rate for the dipole. We don't yet understand a systematic reason for this, in fact, because we would like to be able to find a family of simulations that showed it, and be able to say, ah, yes, these simulations have an Ekman number that's small enough, or an energy distribution that's the right size. We don't yet have a systematic reason for this. In the numerical simulations, reversals are sometimes produced by just turning up the Rayleigh number, which is effectively a mechanism for increasing the amount of energy in the system and tipping the balance away from dipole dominance. What we see in the real field at the Earth's surface is a decrease in the dipole strength, but not necessarily any increase in energy in the rest of the field. I think the short answer to your question is that this is an area where we can expect interesting developments as the numerical modelers find a path to more Earth-like conditions.

Oliver Strimpel: So the record that you found shows that there were periods, maybe even extended periods, when the Earth's field more or less vanished. Does that have an impact on life?

Cathy Constable: We might imagine that it would, because it's widely believed that a magnetic field is an important ingredient for life. But actually, it seems that we're OK even during reversals. Development of the human species has actually occurred over the past, you know, few million years, and it doesn't seem to have been detrimental to that evolution. What happens is that the Earth's field creates the magnetosphere, which holds off the solar wind and cosmic rays, which would normally be considered damaging and potentially carcinogenic, and that's good. The fact that the field existed was good for the original development and existence of the atmosphere, which is essential to life as we know it. If we turn down the field today, as we would during a reversal, the current atmosphere would still be present, even if the magnetosphere was very much compressed and smaller. And the atmosphere actually also plays a strong role now in protecting us from cosmic rays. So that would survive, certainly for long enough that it would be OK. It's not like on Mars, for example, where the dynamo died out long ago and there is essentially no remaining atmosphere.

Oliver Strimpel: So you're saying the impact of a disappearing field might have been more severe at the outset, when life was just emerging, than it would be now.

Cathy Constable: Yes, I think that's true.

Oliver Strimpel: Do we know if the Earth already had a magnetic field billions of years ago, when in fact life was just getting going?

Cathy Constable: So the earliest widely accepted paleomagnetic field records are from about three and a half billion years ago, and the field may have been in existence very early in Earth's four-and-a-half-billion-year history. There are ongoing attempts to recover records from the world's oldest rocks. It's widely believed that a magnetic field may be an important ingredient for life. An interesting thing that is emerging just recently is that the sequencing of some modern-day magnetic bacteria (bacteria that grow magnetite within their bodies and use the magnetic field for navigation and things like this) suggests that their genes go back to the Archean, and the Archean is two and a half to three and a half billion years ago. So if they were using magnetite for the same purpose at that time, that suggests that studying bacteria could actually also provide a mechanism for studying the existence of the magnetic field in the past, or at least provide supporting evidence for it.

Oliver Strimpel: For my last question: if you had unlimited funds for your research, what would you use them for?

Cathy Constable: OK, I'd like to mention three important issues that really limit understanding of long-term field variations. The number one thing I would like to deal with would be improving chronological constraints. How could we get a better measure of the age of these paleomagnetic samples when they're recording the field, so that we could make that more accurate? If we could do that, representations of the field would be better. We wouldn't be putting a measurement of field strength or direction at the wrong time when we reconstruct the field. So I would like to make everybody have to date every paleomagnetic sample they take before they record what's going on, and more importantly, I would like them to focus on samples that can be dated very accurately. What would that mean? In typical dating, we're relying on radiocarbon dating, and for that to be a good measure of age, we need to know what the production rate of radiocarbon is in the atmosphere, and that's a limiting factor. One way of getting around that might be to use something like speleothems or varve records, where in principle you can count the years and establish a record going back in time, in much the same way as one would for a tree-ring chronology.

Oliver Strimpel: What do speleothems involve?

Cathy Constable: Speleothems are basically stalagmites or stalactites that grow in caves, and if you've ever been inside a cave, you might have seen that they have these annual layers related to the seasonal feeding of the carbonate that forms the structure. And that record, if you can establish the age and magnetic record within the stalagmite, could allow you to get a very fine-scale temporal record of what's going on in the magnetic field. In principle, it could be annual resolution. Now, varve records come from sediments with annual layering due to changing seasonal input to the sediment, and they're especially prevalent in glacial, or sometimes in tropical or subtropical, regions. Basically, you can count the layers back in time and perhaps get a very accurate record of the time evolution of the field, and pair that with the magnetic record, which makes for a much better kind of data than data that are not accurately dated. So that would be my first thing.

My second thing, and I'm being quite greedy here, actually: I also want to have better spatial and temporal sampling, and uncertainty estimates for the records that people get, so that needs more field programs. So I want to send people out to collect more stuff, so that we have more records and better global and temporal coverage than we have at the moment. And the third thing I would put my money into would be a more concentrated effort in the numerical modeling, to understand whether we can get close enough to the right parameter regime to believe that we understand what's going on with the dominant physical processes in the core. First of all, we need to improve the paleofield records, so that we can build a better view of what we want to reproduce within numerical simulations, and then the numerical simulations have to figure out how to get across several orders of magnitude of difficulty in not being able to simulate the Ekman number.

Oliver Strimpel: Cathy Constable, thank you very much.

Cathy Constable: Thank you. It was a pleasure.

Oliver Strimpel: For more about Geology Bites, as well as pictures and diagrams that illustrate this podcast, you can go to geologybites.com.