AFTER THE TSUNAMI – HARNESSING AUSTRALIAN EXPERTISE FOR RECOVERY
Canberra, 31 March 2005

The Indian Ocean tsunamis – science and seismics
Dr Phil Cummins
Geoscience Australia

Powerpoint presentation (10,210KB)

As a scientist involved in monitoring earthquakes, there is a tendency for me to think that my work is done after the tsunami has passed, but as we have seen from the recent earthquake and also as demonstrated by this lithograph from the 1861 Sumatra tsunami, these earthquakes have happened in the past and they are going to happen in the future, so the work is actually never done.

My talk, unlike the previous talks, will be focused on the science behind tsunami. I will just talk briefly about the physics and why I think we perhaps could have better anticipated the Boxing Day tsunami. I will go on to describe the science behind what happened on Boxing Day and just talk briefly about comparing the very recent earthquake with the Boxing Day earthquake, and finally I will talk a bit about suggestions for future research.

Plate tectonics – subduction zones
(Click on image for a larger version)

This illustrates plate tectonics; I thought I should discuss this briefly. As most of us probably know, the Earth is expelling heat and its interior is convecting, much like a pot of boiling water. Rising material comes up and forms the rigid tectonic plates which cover the Earth’s surface. These move away from these ridges, which are called spreading centres, and eventually cool and densify enough to sink back into the Earth’s interior.

Where this happens you have what is called a subducting plate and an overriding plate. They join in an interplate interface, which is a massive thrust fault called the megathrust. That is where most tsunamis are generated. It is certainly the scene of the largest earthquakes in the world, much larger than any earthquakes that occur on the interior of the plates.

You can see these kinds of features are readily evident when you look at a plot of seismicity, earthquakes on the globe colour coded with depth. All these shallow earthquakes generally occur where these spreading ridges are, and the subduction zones are illustrated by the deep earthquakes. There is actually a progression from shallow to deep along the subducting plate.

Earthquake excitation of tsunami
(Click on image for a larger version)

So how does an earthquake actually excite a tsunami? That is illustrated in this figure. We have the subducting plate – in this case it is the Indian or Australian plate – sliding beneath Sumatra. If the interplate contact exhibits stick-slip friction, it drags the overriding plate down with it and it deforms the overriding plate. It will come down underneath the ocean, it will be pulled down, but it will also be pushed down more landward, and the entire overriding plate will be pushed landward in a horizontal sense.

Eventually the stress building up on this interplate contact will exceed the strength of the friction, and the plate will simply pop back up into position. It will move a mass of water vertically, and that is what causes a tsunami. You will notice also that on this [right-hand] side of the earthquake there will be a draw-down of water. That is seen as a receding wave, so there is actually a natural warning if you are on this side of the earthquake.

Eventually this tsunami will propagate out, and in some cases it can propagate and still cause damage at great distances.

Comparative tectonics: M>8 Earthquakes: Japan vs Indonesia
(Click on image for a larger version)

The really big tsunami are caused by magnitude-9 class earthquakes, and these tend to occur where young oceanic lithospheres are subducting. That is illustrated in this figure, where I am comparing Japan with Indonesia. What you see from this colour coding is that this orange and yellow colour indicates a young oceanic lithosphere in the south-west of Japan and off Sumatra, in the case of Indonesia, but if you go to north-east Japan or off Java you have relatively old lithospheres subducting. I have plotted here the magnitude-8-and-above earthquakes, and you will see that where old lithospheres subduct you don’t get many earthquakes above magnitude 8. You do get a few of these normal faulting earthquakes that are in a reverse sense of the thrust earthquakes that I have been talking about. They do occur out here [somewhere on slide!] but generally the earthquakes are smaller than you get where you have young lithospheres subducting.

By and large, the really massive earthquakes – the magnitude-9 earthquakes – occur where young oceanic lithosphere is subducting.

Tsunami generated by the great Sumatran earthquake of 1833 (Mw=9.2)
(Click on image for a larger version)

So I think, based on that kind of work, comparative tectonics, but also based on the earthquake history in Sumatra and on geodetic measurements of the strain buildup – the way the crust moves and accumulates strain energy – that it really should have been possible to anticipate that Indian-wide tsunami could occur. In fact, we were looking at that last year and did some simulations of the tsunami caused by the great 1833 Sumatra earthquake, and indeed came to the conclusion that it could affect the entire Indian Ocean basin.

I think it is worth reflecting on why this wasn’t appreciated sufficiently, prior to the earthquake, because I think the evidence was there.

Seismic waves generated by the great Sumatra-Andaman earthquake of 2004
(Click on image for a larger version)

But I will leave that for the discussion groups and go on to discuss what actually happened on Boxing Day. This was a great earthquake that occurred off Sumatra. The first indication of it was the seismic waves, and that is what is illustrated here. These are the ground motion recorded at seismic stations distributed throuthout the globe. Time is going this way [y axis?], so you see the first arriving waves are the largest; distance goes this way [x axis?], so it arrives at close stations first. And these waves are propagating all the way to the other side of the globe – that is 0° and 180°, so this wave has gone all the way to the other side of the globe, it has wrapped back around and these waves, which are roughly 20 seconds period, are still, although you can’t really see the scale, of roughly half a centimetre to a centimetre amplitude as they travel around the Earth over hours. So these are really massive waves.

And yet they don’t really indicate the size of this earthquake. To see that you really have to look at very long-period energy that is not visible on these seismograms. But I just wanted to plot this to illustrate that these are massive waves. You couldn’t actually feel these, because of the long period, but they are of the scale we could actually see.

26 December Sumatra earthquake
(Click on image for a larger version)

The earthquake appears to have ruptured a very long segment of the Sumatran–Andaman subduction zone. The first indications were that there was a magnitude-9 earthquake that ruptured about a 500-kilometre segment. Some of the analysis of seismic waves and certainly tsunami waves indicated that it was a somewhat larger source, about 1,000 kilometres in length, along the subduction zone. But the aftershocks and crustal deformation indicate that it was 1,300 kilometres. So there has been some confusion about the magnitude of the earthquake, but the length of subduction zone which slipped is not in question.

Crustal movement in Andman Islands
(Click on image for a larger version)

That is evidenced not only by the aftershocks but also by the obvious uplift and subsidence that has occurred in the Andaman Islands.

Those are well to the north in the Andaman Islands, and you can see here coral reefs that have emerged from the water – they were obviously submerged prior to the earthquake. This is a lighthouse which prior to the earthquake was on land and now it is in roughly a metre of water. There are many observations of this kind of uplift and subsidence well to the north in the Andaman Islands.

This is from Roger Bilham’s web site series. He is a researcher in the US. You can actually use these kinds of observations to try and estimate how much fault slip occurred. So this [graph at right-hand side of slide] is showing where the uplift occurred, right here. This is vertical deformation, so here is uplift. This is the subsidence, and so you get negative vertical deformation. And you can try to fit the fault slip. What he found was that you needed about 10 metres of fault slip, even that far north in the Andaman Islands, to generate this kind of crustal motion.

Free oscillations – Mw=9.3!
(Click on image for a larger version)

That was borne out weeks later by these very long-period waves – it is in fact a standing wave, the Earth is ringing like a bell at very long periods, in excess of 1,000 seconds. This is a way of plotting the seismic energy as a function of those very low frequencies, and if you look at those very low frequencies you indeed find that this was even bigger than suggested by the original seismic analyses. This analysis of these free-oscillation data by Stein et al indicates that the actual magnitude of this earthquake was 9.3, which makes it the second largest earthquake ever recorded.

26 December Sumatra earthquake
(Click on image for a larger version)

And this 9.3 magnitude agrees with the kind of rupture zone that extends for 1,300 kilometres all the way up into the Andaman Islands.

Effects in Sumatra
(Click on image for a larger version)

As for the tsunami, it was a truly massive tsunami, as we all know. One of the first Japanese teams that went to Banda Aceh made measurements, looking at things like debris up in trees to estimate maximum wave height, and this is what they found. This is the north-western tip of Sumatra here, and these are measurements of 0–30 metres; that is what this scale goes to. So you can see that on this side of Sumatra there were 30-metre waves observed over a fairly broad range along the coast. Even on the lee side the waves were up to 10 metres. So this was a truly massive tsunami. Just reading the email reports from some of the survey teams, people just could not believe the size of the tsunamis seen in northern Sumatra. They hadn’t seen them before.

Tsunami waves
(Click on image for a larger version)

Similarly, some of the observations of the tsunami are…

[tape change]

…a yacht off Phuket – this is a fishfinder. The yacht had a fishfinder on it, and it simply rode the wave out. You can see that it experienced roughly five metres of wave. It just simply rode it out and its fishfinder faithfully recorded the wave form of the tsunami.

A more standard type of tsunami recording is this tide gauge on the other side of the Bay of Bengal, in India, where you can see it reached ±2 metres in amplitude. This was actually very important for understanding the source zone, and it was one of the first indications of how big this tsunami really was.

Finally, I was monitoring the tsunami bulletin board during this time, and you could see that these people who monitor the tsunami data kept sending in reports, ‘Now it’s reached the Pacific Ocean,’ ‘Now it’s going up the Pacific Ocean.’ It reached all the way up into the north-west Pacific Ocean; finally it got into the Atlantic Ocean and was recorded in France, the northern Atlantic. So this is the first truly global tsunami. I know that it is said that the tsunami due to the eruption of Krakatoa generated disturbances in the ocean seen throughout the globe, but many people think that was actually due to the atmospheric disturbances, that they were exciting water motion, and that that was not a true tsunami. Clearly this tsunami has affected every ocean on the globe.

28 March 2005 Sumatra earthquake
(Click on image for a larger version)

Now I will just talk quickly about the earthquake that recently occurred, which didn’t generate much of a tsunami. That was certainly surprising to me, and it is still a bit puzzling. This [on slide] is from the USGS’s web site, illustrating the major earthquakes that have occurred off Sumatra. This [towards bottom of map] is the massive 1833 earthquake that we did the tsunami simulation for. There was also an 1861 earthquake, slightly up the trench, and then the 2004 earthquake. You can see that the 2004 earthquake rupture began here [position not noted] and then propagated north for 1,300 kilometres. The more recent earthquake seems to have initiated at the northern end of the 1861 rupture zone, and most people presume that it propagated south-eastward. That is basically true. This is the actual fault slip that was imaged from seismic data, and there is the epicentre. Most of the rupture did propagate to the south-east, but a fair bit of it was concentrated right around the epicentre and some of it also propagated to the north-west. So I am wondering if it really is a repeat of the 1861, or if this is a different kind of earthquake.

2004 and 2005 Sumatra earthquakes
(Click on image for a larger version)

I just wanted to compare and try and explain why this might have been so different. You can see that the Boxing Day tsunami initiated at depth – that is 30 kilometres depth, if you follow the plate interface down – but most of the slip occurred at shallower depths and it was well distributed over the fault plane (this is fault slip here). What the vertical movement of the sea floor results in is pronounced concentration of sea floor deformation near the trench access. That is very deep water so the earthquake could put a lot of energy into the water column, whereas if you look at the more recent earthquake you see that most of the slip was concentrated near the epicentre at roughly 30 kilometres depth. Being at that depth, it doesn’t excite as much sea floor deformation, and this is also much shallower water. That is just a guess, but it could be an explanation for why this did not excite a very large tsunami, which was quite puzzling to myself and a lot of other seismologists.

I won’t spend much time on the conclusions. I do think that there was evidence that we could get Indian Ocean-wide tsunami, even prior to the Boxing Day event, and I think it is worth reflecting on why it was not fully appreciated early.

Finally, the Great Sumatra–Andaman earthquake ruptured an entire 1,300-kilometre long section of the Sumatran subduction zone, making it the second largest earthquake ever recorded. And comparison of that with the 2005 earthquake suggests that the latter may have generated a small tsunami due to concentration of slip at considerable depth.

Suggestions for future research on Indian Ocean tsunami: I think this is what really the forum is about today. One of the highest priorities is to collect and analyse both historical data and palaeo-tsunami data to characterise this hazard, to understand what the recurrence rates are and how big these events can be.

We need to do more modelling and analysis to quantify what the risk really is, how often we can actually expect these things and what our vulnerable areas are.

Third, geodetic monitoring of strain accumulation in the subduction zones is crucial for trying to understand where these earthquakes are most likely – and where the really big earthquakes are most likely.

We need further work on the tectonic history of the Indian Ocean. I found that there really wasn’t that much information on plate reconstructions, which are crucial for trying to determine the character of the subduction zone interface, and so more work could be done there.

We should not forget, with all this attention focused on the Sumatran subduction zone, that it is not the only subduction zone in the Indian Ocean. There is also the Makran subduction zone, off Iran-Pakistan, which can generate magnitude-8 earthquakes and large tsunami.

Finally, we need better collaboration among Indian Ocean and Pacific countries, and also among the earthquake researchers and the tsunami researchers. I think that was one reason why the Boxing Day tsunami was not fully anticipated.