In the previous post we went through some of the background to this post: the structure of the Earth, the change in minerals at various depth in the mantle which change the density, and a bit on seismology which can show how the temperature changes with depth. Now it is time to see what we can do with this knowledge, study the alien life of the Earth – and find the Earth’s heart.
Seismography has shown us how the conditions in the mantle change with depth. But there are deviations: not all places on Earth have the same mantle structure underneath them. There are locations below which the earthquake waves travel unexpectedly fast or slow. Some of these we understand. For instance, S-waves in subduction regions show high velocities, indicating low temperatures. The subducting slabs are cold and hard, and therefore resistant to deformation. The shear wave velocity is particularly sensitive to this: in harder material, it travels faster. Think of the students in the previous post: if they push back hard against the wave coming from behind, they push the person in the line equally hard. A hard push gives a fast response, and therefore a cold slab has a higher vs than a warmer, softer slab. Try kicking a deflated football and compare it to one at full pressure, if you need experimental evidence (always recommended). In these places of subduction, the high shear wave velocity reveals the departed but still cold ocean floor.
Other anomalies are not nearly as well understood. In particular, there are two very large regions in the lower mantle where the shear wave velocity decreases significantly. ‘Significant’ means by around 1-2%. This does not sound dramatic, but it is far more than expected. The two regions occupy around 8% of the volume of the mantle.
These two areas of tardiness have become known as ‘Large Low Shear Velocity Provinces’. The name kind of makes sense: the waves are shear waves, they are slow, and the regions of slowness are very large and therefore are called ‘provinces’ rather than say ‘spots’. (But ‘Large’ and ‘Provinces’ does seem a tautology, saying the same thing twice.) However, it is clearly a name designed by committee. Nowadays it is abbreviated as ‘LLSVP’ which to me still sounds like a quango! Wikipedia calls them ‘superplumes’, whilst the term ‘superswells’ is also in use; both names, however, are also used for other (perhaps related) phenomena. The name ‘thermo-chemical pile’ has been used in scientific papers but this sounds like a severe and socially inhibiting disease. The plethora of unhelpful names shows that the understanding of them is still lacking. Once we know what they are, a more palatable name (not involving piles) will be invented.
Source: Cottaar & Lekic, 2016, Morphology of seismically slow lower-mantle structures, Geophysical Journal International, Volume 207, Pages 1122–1136
One of the regions is located underneath Africa, and the other is almost (but not quite) opposite, underneath the Pacific. They sit on top of the core-mantle boundary, entombed within the lower mantle. The two regions have different shapes: the Pacific LLSVP is roundish and the African LLSVP has a more complex shape, elongated north–south with two extension east-west. The African LLSVP also reaches much higher than the Pacific one. The Pacific LLSVP extends 700 km above the core-mantle boundary, with the top about 2300 km below the surface. The African LLSVP is more than twice as high, with its top at 1600 km: it reaches halfway to the crust. They are still quite flat structures: the width of perhaps 5000 km or more is several times larger than the height.
The explanation of the LLSVPs has been discussed for decades. They may be (likely are) zones of higher temperature (because of the lower shear wave velocity), but they may (likely do) consist of rock with different composition. Perhaps they are both. The first suggestion are the so-called ‘thermal models’, and the second are the ‘chemical models’. Combining both models, the LLSVPs are known as the thermo-chemical piles (this is where that name comes from), where the words cover pretty much all possibilities while staying safely on the geological fence – even the term ‘plume’ is carefully avoided.
But are the LLSVPs mantle plumes? It is tempting to see them as the rising part of the mantle flow, countering the descending subducting plates. But if they are ascending, why don’t they reach higher? And models show that a hot rising plume in the mantle becomes narrow and tall, while these structures are very broad. Perhaps they consist of a large number of narrow plumes, which in the data appear as a large structure. But that would still not explain why they stop halfway. No, they do not appear to be mantle plumes.
The most likely model for the LLSVP is that they consist of slightly heavier minerals than the rest of the mantle. If the temperature is the same as that of the mantle, they will happy lie at the bottom. However, the difference in density is not large, perhaps only a few per cent, as is evident both from the fact that they don’t cover all of the bottom, and that some parts have managed to rise. A higher temperature can compensate for the higher density. The model assumes that the temperature varies throughout the LLSVP. All parts are warm, but not equally warm. Where the excess warmth is less, the LLSVP is denser and lies at the bottom. Where the temperature is a bit higher, the material expands a tiny bit, the density lowers, and the LLSVP begins to rise. But it does not rise very far, at most halfway up the mantle. This fact already indicates that they are different from the surrounding mantle, and they do not have exactly the same chemistry. The combination of lower and rising parts creates a dome-like structure, a bit like the plant at the top of this post (a yareta from the Atacama desert). Depending where you are, it could also be compared to the domes of the Sydney opera house or those of the Cornwall Eden project.
The two LLSVPs may have existed here for hundreds of millions of years. Deep in the upper mantle, they should leave no direct trace on the surface, 1500 km or more above. But perhaps surprisingly, there is one indication for their existence on the surface. It is visible in the terrain of Southern Africa. Don’t look for small things: as appropriate for such a large structure, go big. Step back and look from Afar. You can now see that the southern third of Africa is a high plateau, up to 2 km high and extending to almost the southern edge of the continent. (The low-lying strip between the high plateau and the ocean is some of the most beautiful land on Earth, in my opinion.) This six-nation plateau is much higher than a typical continental plate (500 meters). There is no clear explanation for this excess height: there is no mountain region (the Drakensberg are the edge of the plateau and are mountainous only from one side), no plate collision (not since before Gondwana formed), and no hot spot activity since the break-up of Gondwana. It has no known cause. There is only the LLSVP sitting below it.
Southern Africa is the only area where an LLSVP lies below a continent: all other regions above the two LLSVP are oceanic.
There is a second indication of a long-distance impact of LLSVPs on the surface, but we will come back to that.
The LLSVPs are old – old as the mountains (which actually are much younger – a case of our language lagging far behind our science). But where did this 8% of the mantle come from? How did it acquire a different composition compared to the rest of the mantle?
There have been many, always speculative, suggestions. One proposes an ancient origin, when the Earth was young and the mantle was a magma ocean. It solidified very quickly, but the magma ocean lasted long enough that the heavy iron was able to sink to the bottom and form the Earth’s core (that part is not speculative – it is the only way our core could have formed). It ended when the mantle solidified. Perhaps the bottom of the mantle stayed molten longer than the rest of the mantle, and the material from the LLSVP sank to the bottom of the mantle at the this time: it is a memory of when the world was young and life was still an unforeseen development. Other suggestions go off on more recent tangents. One proposal is that the LLSVPs are the remnant of the oldest crust of the young Earth, sinking to the bottom in the earliest subduction, a few hundred million years after the Earth’s birth. An even more speculative idea is that the LLSVP formed from the high density parts of subducted oceanic plates. An oceanic plate consists of the crust and the attached lithosphere. The idea is that when the subducted pate reaches the bottom, the crust detaches from the lithosphere. The crusty bit begins to rise as it warms up (think lava lamp) but the lithospheric bit stays behind. Both eventually dissolve into the surrounding mantle, but the region where the lithosphere was remains overdense. It is a rather complex, multistep model, and as with any complex model, it can explain a lot but is very hard to prove. It may have a problem with the clear separation between the LLSVPs and the rest of the mantle, and with the fact there are only two.
Combining the various proposals, the LLSVPs are sometimes described as a ‘melange’, a mix of all proposed components. The bottom of the mantle acts as the waste bin of the Earth: everything that isn’t wanted elsewhere ends up here. That makes the LLSVPs the Earth’s rubbish dump – although that is a term you will not find in any scientific papers.
There are some things we do know. If the sunken material were too dense, it would form a stable bottom layer like the D″ (remember, D-double-prime). That is not what we see. This also rules out more exotic ideas such as extraterrestrial impacts or core dredge-up. If it were the same density as the surroundings, it would relatively quickly get stirred into the rest of the mantle. That too did not happen. Creating a structure akin to what we see, rising domes in places, lying low in others, only works if the density is only a little higher than the surroundings, by a few per cent. It is like a lava lamp that it isn’t quite working: the blobs are shuffling around in the bottom half but fail to rise to the top. Blobs that are warm enough rise a bit, but become stuck halfway up and remain there, until they have cooled a bit and sink again. Such a failed lava lamp is dynamic but disappointing (so much so that not even youtube has an example, while working lava lamps have many hours of footage), and moves sluggishly. (Our garden snails would take offence at that term.) The LLSVPs may behave like this. A part of the denser material heats up (for some reason), rises, cools, and eventually drops down, while another part in the mean time has been heated and begins to rise in turn. Speed up the movie to show a billion years in a minute, and the mantle comes alive, with the LLSVP moving around in the lower mantle like a massive, monster amoeba. Or like a pumping heart.
The depiction of the mantle above shows the different parts. Yellow indicates the subducting plates, descending into the deep where they partly dissolve. Black is the normal composition of the mantle. Blue is the LLSVP that fluctuates along the bottom but doesn’t easily mix – sometimes a subducting plate will penetrate into it. The cartoon does not separate the lower and upper mantle, so it doesn’t capture everything. But it gives the idea of the life of the mantle.
There is one more component which we haven’t mentioned yet. At the very bottom of the mantle there are a few smaller regions where the velocity of the shear waves drops by a massive 20-30%. This drop is ten times more than that of the LLSVPs! The primary or P-waves (remember them?) also slow down here, but ‘only’ by 10%. These regions are therefore called the ultralow velocity zones (ULVZs). Scientific names may be accurate, but they do lack in impact, memorability or poetic power! Geology in particular has invented an obscure nomenclature, also known as geo-babble:
A geo-babble prosody
The language of geology
a whimsical neology
The lexicon is chockablock
with complex names for sandy rock
Starting as a periodite
forshape into a ringwoodite
reincarnate as bridgmanite
transfigurate to perovskite
continuing ad infinite
A wordy sample
of the lower mantle
The layers too have epithets
A glossology of figureheads
with the asthenosphere
but nothing will rhyme
And the same is true, to a lesser degree
with the esoteric LLSVP
What you see is what you get
unless you name it ULV-zet
and if you read it ULV-cee
what it is you still don’t see
the discombobulated uninitiated
(Presumably professional geo-babble requires ownership of a Babble fish.)
(The story that the asthenosphere was made up by geologists, and that this is why it is called ‘as-the-no-sphere’, appears not to be true.)
Back to our story. All ULVZs lie at the very bottom of the mantle. They are typically a few hundred kilometers across, between 10 and 40 km thick, and are located directly on top of the D″ layer. They are tiny compared to the LLSVPs. ULVZs are very hard to detect, and only parts of the Earth have been investigated for them. On the map below, areas where they are known to exist are shown in bright red, and areas where they do not exist are blue. Note that the red areas are not the shape of the ULVZs: they indicate that one or more ULVZs exist somewhere within the red area but their shape or size are not known. The LLSVPs are indicated as the pinkish regions.
The known ULVZs are mainly seen around the edges of the LLSVPs (writing a sentence with two long acronyms feels really scientific). The non-detections are mainly further from the LLSVPs.
We don’t know what the ULVZs are! Their relation to LLSVPs (if any) is hotly debated; the shear velocities are very different, suggesting they are not directly related but have different compositions. The location at the bottom of the mantle indicates a high density, and the low shear velocities suggests they could be partly melted. There are two main suggestions regarding their origin: blobs with enhanced iron-oxide, perhaps diffused from the core, or subducted ocean crust which has melted, with the densest material dropped to the bottom. The clustering around the LLSVPs can be a result of mantle circulation. If the LLSVPs have higher temperatures, then they can heat the mantle material around them. This begins to rise, and sucks in new material at the bottom. This causes a sideways flow at the bottom of the mantle, directed towards the LLSVP. The lateral flow takes the ULVZs along with it as a (perhaps unwanted) passenger, but once adjacent to the LLSVP where the mantle flow goes up, the ULVZs are unable to rise with because of their higher density and they are ejected from the flow. This leaves them stranded on the beach of the LLSVPs. Think clogged arteries.
There is one rather intriguing aspect to the ULVZs. Some (not all) seem to be related to surface volcanism. Both Hawai’i and Iceland sit above a nearby ULVZ. In the case of Hawai’i, the ULVZ is located just to the west, and for Iceland it is a little to the east. This is surprising, in view of the separation by three thousand of vertical kilometers and the fact that the ULVz are immobile and stuck at the bottom. However, a relation to the surface volcanism is far from proven. People may just have looked for ULVZs mainly in such areas of interest. The data is suggestive, but not conclusive.
We have seen that one of the LLSVP seems to be pushing up Southern Africa over an area 1500 km in size, whilst ULVZs may be related to volcanic island hot spots 3000 km above. Could LLSVPs also be related to hot spots or is this a monopoly of the ULVZs?
Hot spots are locations where volcanoes are active but which lack a clear reason for why there should be a volcano there. Volcanoes are expected at mid-oceanic ridges and above subduction zones, both at the edges of plates. In contrast, hot spot volcanoes such as Hawai’i and Yellowstone can be located in the middle of a plate. Their activity requires an underground heat source – hence the name ‘hot spot’. The source of that heat may differ between hot spots. In some cases there is a clear connection deeper into the mantle, in other cases there isn’t. Where there is a signature of a connection, it often ends midway in the lower mantle, as for instance in the Galapagos. Plumes from the core-mantle boundary are a rarity – perhaps even non-existent. But starting a plume in the middle of the mantle just moves the problem, for why would there be a heat source there?
The map above shows the locations of the known hot spots. The size of the orange circles showing the magma production rate, in the slightly funny unit of mega-grams (or tons) per second. The known hot spots tend to be located around the LLSVP regions, and are often near their edges. Can it be that some hot spots are somehow powered not by the ULVZs but by these LLSVPs? This would explain why hot spot plumes can often not be traced to the bottom of the mantle.
It gets even more interesting when looking at past flood basalts. (These are also known as large igneous provinces, or “LIP” in this acronym-flooded field (AFF)). The LIP is a sure sign that there was a hot spot there in the past. These (now extinct) hot spots have been added to the map above, and are shown as the green circles. The positions are corrected for plate motion. For instance, the Columbia flood basalt is plotted not at its current location but near Yellowstone, which is where the flood basalt actually occurred, before the moving continent carried the lava shield away. Almost all the LIPs are located above the LLSVPs! And just as for the current hot spots, they have a tendency to go for the edges of the LLSVPs, not their centre.
A mantle wind
The relation between hot spot volcanism and LLSVPs is intriguing. Both the hot spots and the ULVZs are found around the edges of the LLSVPs, but the ULVZs are at the bottom and the hot spots are much higher up.
The LLSVPs themselves are largely immobile, while the mantle flows around them at a leisurely speed of perhaps half a meter per year. The LLSVPs are warmer than the surrounding mantle and pass on some of their heat. Around the LLSVP you get a thin, warm mantle region. This begins to rise: there is now a mantle updraft around the LLSVP. The LLSVP stays where it is (it is too dense to rise any further) but it powers the updraft around it.
This basic model is shown above. The subducted plate comes down to the core-mantle boundary and moves along the bottom. Where it reaches the region of the LLSVP a sharp boundary forms, the mantle winds moves upward, and from the heated edge of the LLVSP a thin plume begins to rise. This plume will form the hot spot where it reaches the surface. In the depiction, the plumes come from the LLSVP itself. In reality, it appears to come from the surrounding mantle.
So hot spots and related volcanism appear to be closely related to the LLSVPs.
To rise and fall
There are two circulations in the mantle. In the upper mantle, the circulation begins with inhalation of old oceanic crust in a subduction zone, and ends with exhalation at a mid-oceanic spreading zone. Most of the time this circulation remains within the upper mantle, although sometimes the subducting crust is cold enough to sink into the lower mantle while some hot spots break through the other way and disturb the breathing cycle of the upper mantle. This cycle of the upper mantle is powered by the subduction of the oceanic plate which pulls the rest of the plate sideways. This causes the plate to tear, creating the weakness of the mid-oceanic rift where the mantle material can find its way back to the surface, closing the cycle.
The lower mantle also has a circulation. Part of this is a slow convection that remains confined to the lower mantle. But there is some interaction with the upper mantle, especially when a cold subducting plate reaches the lower mantle and plummets to the bottom. Convection is driven by heat from the below and it doesn’t like a cold bottom. Instead the plummeting plate causes a downdraft behind it. This disturbs the circulation of the lower mantle and pushes the updraft away from the frigid plate. Over time, perhaps a few hundred million years, the plate on the bottom warms up. Now it becomes buoyant again and begins to rise. At this time it will also begin to dissolve into the lower mantle.
The role of the LLSVP in this process is absent from the youtube simulation (it just has a heat source below the centre), but appears to be quite important. The LLSVP causes local regions of updraft. The mantle material just around the LLSVP is heated and rises, and this pulls in new material from the sides. The excess heat triggers a narrow plume which may be hot enough to break through the barrier, enter the upper mantle and cause a hot spot. A volcano will now erupt. The sparsity of hot spots already shows that this does not happen everywhere around the LLSVP. Only a few regions become hot enough to do this. One suggestion is that regions where an old oceanic plate mixed into the mantle are especially susceptible to form such plumes, because they have slightly lower density (due to the mixed-in oceanic crust). The extra heat makes them more buoyant and they take off, leaving the rest of the mantle behind.
This predicts that mid-oceanic ridges have different magma from hot spot volcanoes. Both produce basalt, but mid-oceanic ridges are largely recycled oceanic crust, while hot spots are mostly unprocessed material from the lower mantle. (Whether they could contain any material from the LLSVP itself is disputed.) Indeed, the basalt from hot spot volcanoes is more primitive, differs somewhat in the ratio of chemical elements and contains more of the rare isotope 3He which is strongly depleted in the upper mantle (and is therefore very very expensive!). There is also more diversity in the magma from different hot spot volcanoes.
Whilst the upper mantle circulation is powered by descending plates, the circulation of the lower mantle is powered by the heat from the LLSVP. The upper mantle inhales – the lower mantle pumps.
This still begs the question why the hot spots are mainly found around the edges of the LLSVP. For if they get their heat from the LLSVP, why wouldn’t the central top of the LLSVP heat the mantle directly above it just as well? At first thought, the answer to this seems simple. Assume that every square meter of LLSVP surface heats the mantle with the same efficiency. Where are those surfaces? Imagine a cube sitting on the mantle floor. It has 6 sides, 5 of which face the mantle (the 6th is the bottom and faces the D″). One of those sides is the top, and four are around the edges. So the surface area around the edges is four times larger than the surface area on top. A quick count of hot spots on the map above shows that around 20 are on edge and perhaps 7 are more central. That is a ratio of 3 to 1, not far from 4 to 1 expected for a cube. This is the simple answer, but it falls flat because the LLSVP is not a cube and is much wider than it is high. The excess surface around the edges is now much less. Using the dimensions of the African LLSVP, it now predicts that perhaps half of the hot spots are at the edge and half on top. That is not quite what is seen.
A more difficult answer could be that a mantle flow rising around the edges can pick up more heat that one rising from the top. This is because an edge flow remains close to the LLSVP during the rise, while one rising from the top loses contact with the heat reservoir almost immediately. A combination of the simple and more difficult answer may work.
The second begging question is where does the LLSVP gets it heat from? There are two obvious possibilities. It may just be better insulation. The mantle transports heat from the core to the surface through convection. The LLSVP does not take part in that, and it only loses heat by conduction into the rest of the mantle, and this is a much slower process. Compare the temperature in your greenhouse with that of the air outside. The outside air convects, the air inside only conducts. (Interesting, a greenhouse is not heated by the greenhouse effect!) Less heat loss (hold it – this part is difficult, at least to some who prefer not to grapple with global warming) means that you stay warmer.
The heart of the matter
We started with a search for the Earth’s heart. We found that there are two cycles, one in the upper mantle and one in the lower mantle. The upper mantle cycle is compared to breathing (of healthy rock rather than air). In contrast, the lower mantle cycle is pumped by heat from the LLSVPs. The pump acts the same as does a human heart.
Thus, the mysterious LLSVP is in fact our planet’s warm heart.
There are many open questions here. At what depth does the plume form?. Does it come from the top of the LLSVP or does it start lower down? The data remains inconclusive. What is the precise structure of the LLSVP? The images look nice but we may be missing a lot of detail. Do they perhaps consist of clusters of narrow plumes, or are they indeed more like domes?
And what is an LLSVP anyway? What is it made of and does it really date from the earliest days of the Earth? And what are the even more mysterious ULVZs?
There is a final question. The mantle is a windy place, where nothing is stationary. That means the LLSVPs should also move around. In that case, why do even LIPs from 300 million years ago line up with the modern edges of the LLSVPs? Perhaps it is because LLSVPs move only very slowly. Models suggest that their viscosity is ten time higher than that of the average mantle. The movement may be sluggish enough (with apologies to our rocket-like garden snails) that they don’t move far in a brief 300 million years.
But if they move, why should hot spots be stationary? And even if the LLSVP is stuck, the hot spot should still be able to move around the edges of the LLVSPs. There are in fact indications that hot spots do move around, and have not always been at the place where we see them now. The Hawai’i hot spot may have moved by 10 degrees in latitude over its life time.
Attempts have been made to model the mantle wind, taking into the effect of the changes in the location of subduction zones over time. Around 700 million years ago, there was a network of subduction zones around western Gondwana. 200 million later those cold subducting blocks reached the bottom of the mantle. Another subduction zone formed 400 million years ago around the northern edge of the Tethys ocean, and the process repeated here. The clusters of hot spots and LIPs occur above these locations: perhaps the plumes surrounding the African LLSVP are a memory of Gondwana, rather than the direct trace of the large low shear velocity province. The LLSVPs activated them but the fuel came from the Gondwana. Perhaps.
The life of the Earth is like the cicadas who live in the dark, unobserved, for 17 years, before making their brief but very loud appearance above ground. We only see 1% of their life, and have to guess about the other 99%. For the Earth, volcanoes are that 1% we can see. The other 99% of the life of the Earth is underground, much like the living rocks which Spock referred to. We hear the noise, but fail to see the party underneath to which our homely continents are not invited.
We started this series with Spock’s famous statement from Star Trek: ‘It is life, Jim, but not as we know it’. (The one he never said.) But now we have the answer, and it turns out the Earth’s life is very much as we know it. The clue is in the tautological LLSVP. For the Earth has two of these, and thereby has two hearts. The Earth is like the only alien more famous than Spock, the one with two hearts. The Earth, it turns out, is Doctor Who. We have been studying a Time Lord. We are Gallifrey.
Perhaps this is the reason for the unearthly language of geology.
Albert, June 2022