Mw5.8 earthquake strikes eastern Indonesia
We finally take on an earthquake in the Molucca Sea region
Earthquake Insights is an ad-free newsletter written by two earthquake scientists. Our posts are written for a general audience, with some advanced science thrown in! To get these posts delivered by email, become a free subscriber. If you would like to support our work here, please also consider a paid subscription.
On October 21 at 3:32 PM local time, a Mw5.8 earthquake struck offshore between the eastern Indonesian islands of Bacan and Obi. The earthquake was soon followed by a productive sequence of at least 20 aftershocks, including a Mw5.7 event. We are not aware of any reports of serious damage.
Earthquakes are common in the Molucca Sea region. (For those unfamiliar with this area, it is south of the Philippines, west of Papua New Guinea, east of Borneo and Sulawesi, and north of Australia.) This area is so tectonically complicated that we usually ignore moderate-sized earthquakes that happen there. This gives us a bit of a bad conscience. However, the recent earthquake sequence is particularly interesting, and so we’re giving it a shot. We’ll cover some big-picture ideas, and then talk about the October 21 earthquakes more specifically.
First, let’s look at the active tectonics of the Molucca Sea. Sandwiched between the long, arm-waving islands of Sulawesi and Halmahera, this small sea conceals a large tectonic conundrum. We will start with a zoomed-out view of the topography of eastern Indonesia and surrounding areas:
There’s a lot to take in here.
An unsuccessful tectonic comparison
One way to try to understand the tectonics of a region with relatively little data is to compare it to another, better-understood region with more data. So, just for the sake of argument, we will compare the Molucca Sea to another, better understood, region of Earth with some similar characteristics. Please note in advance that the comparison fails spectacularly.
So, what features should we look for, to find a comparison for the Molucca Sea? Here are a few basic facts.
The Molucca Sea is a shallow sea. While many of the ocean basins within the Indonesian archipelago are quite deep — reaching to 4–5 kilometers below sea level —, the Molucca Sea is only rarely deeper than 2 kilometers.
The Molucca Sea is also fairly rectangular: it is longer than it is wide, with kind-of-parallel coastlines on two sides.
There is a mostly underwater ridge located right in the center of the sea, which rises above water only in a few isolated islands.
When we look around Earth for a similar setup, we find only a few places to match these characteristics. One of them is the Gulf of Aden, between the Horn of Africa and the Arabian Peninsula. Let’s plot the topography of these two areas for comparison, at the same scale. We have rotated the maps to give a fresh perspective on both areas, for those who are already familiar with the geography.
Both seas are about the same length and width, have about the same water depth, have a prominent ridge down the middle, and separate mountainous areas with young volcanoes (shown as red triangles).
Let’s plot shallow earthquakes (<30 km deep): this should show us where the crust is actively deforming. We exclude deeper earthquakes, because the Molucca Sea is surrounded by all kinds of subducted slabs, and we want to look only at surface activity. Our map for the Molucca Sea below uses the earthquake catalog from BMKG because it has better locations than global catalogs.
Both seas have a band of shallow earthquakes beneath the central ridge. The land on either side is seismically quiet in comparison.
The Gulf of Aden is a great example of active oceanic spreading within a young rift. The continental crust of Africa and Arabia was once joined, but has broken apart as a new ocean basin formed. The water depth of the Gulf of Aden is typical for young oceanic crust, which usually forms at about 2.5 km water depth. The volcanic fields on the rift flank in Yemen represent melting of the upwelling mantle; some of these volcanic fields are still active. The higher topography on both sides of the rift is partly due to isostatic rebound of the crust. These features are all common in similar tectonic settings.
The Molucca Sea is similar at first glance — does that mean it is also a rift with an active spreading ridge? Let’s look at some more data. First, we can examine the Bouguer gravity anomaly (we use WGM2012 data). Without going into details, this anomaly gives information about the density distribution of the crust — a proxy for both composition and temperature.
Clearly, the areas are quite different.
At Aden, the land regions on both sides of the rift have low gravity anomaly values, close to 0, while the ocean area in between has higher values, about 200-300 mgal. This reflects the fact that the land at Aden is underlain by less dense material (continental crust) while the sea is underlain by more dense material (oceanic crust).
At Molucca, the picture is somewhat reversed. The land areas have higher gravity anomaly values (300-450 mgal), while the sea area in between has lower values (100-200 mgal). This indicates that the Molucca Sea is underlain by low-density material — not exactly continental crust, but certainly not oceanic crust, despite the fact that this is a sea.
We can also look at maps of crustal magnetization (we use EMAG_V2 data):
At Aden, there is a clear pattern of magnetic domains aligned along the axis of the sea. These are the famous “magnetic stripes” that are frozen into the cooling oceanic crust as seafloor spreading progresses; each stripe represents one iteration of the Earth’s magnetic field pointing either north or south.
At Molucca, there is no such pattern. The data appear a lot more random, although there are some big anomalies associated with structures outside the Molucca Sea. Even if magnetic stripes exist here, they are probably obscured by the thick overlying sediment.
Finally, let’s plot focal mechanisms. These tell us directly about the type of earthquakes occurring in each area: normal (associated with spreading or stretching), strike-slip (associated with sideways sliding), or thrust (associated with subduction or collision).
At Aden, the focal mechanisms are either normal (white in the middle) or strike-slip (like an X with opposite quadrants colored white). This is the typical signal of oceanic spreading. The spreading ridges produce the normal events, and the transform faults between the ridge segments produce the strike-slip events.
At Molucca, the shallow earthquakes along the central ridge are obviously different: almost all are thrust events (the colored quadrant is in the middle). This indicates significant shortening within the Molucca Sea.
This quick exercise is just meant to demonstrate that superficial appearances can be deceiving in the Molucca Sea area. If we only had access to topography/bathymetry and earthquake epicenter data, and if we had to look at the maps from across the room, we would have a hard time telling the Molucca Sea apart from the Gulf of Aden. With more data, it is clear that they are completely different beasts.
Tectonics of the Molucca Sea
So, what is actually going on below the Molucca Sea? Fortunately, some other people have done a lot of work to try to puzzle this out, so we can lean on their work.
Basically, an oceanic slab is subducting in two directions; one side is sinking eastward under Halmahera, and the other side is sinking westward under Sulawesi. In other words, instead of representing the birth of a new ocean (as in the Gulf of Aden), this is the last gasp of an ancient ocean, being swallowed up on two sides.
In the not-too-distant past (geologically speaking), there were two real-deal subduction zones here, facing away from each other, each producing its own volcanic arc. Here’s a cross section from Hennig‐Breitfeld et al. (2024), which builds on decades of previous studies of this area. We will return to that paper eventually.
As each side of the oceanic plate subducted, the area of oceanic lithosphere at the surface grew smaller and smaller. The two volcanic arcs got closer and closer together.
Starting about 2.6 million years ago, give or take, the last oceanic crust disappeared. However, the plates on either side were still moving towards each other, and as a result, the sediments deposited on both sides of the ocean margin started to smush together. An arc-arc collision should eventually occur here, if the sedimentary stuff in between isn’t so strong that the collision simply stops.
Another cross section shows the present-day picture:
The subducting slabs are still sinking its way downward into the mantle, but their apex is now covered by a thick package of previous ocean floor sediments, sandwiched between two volcanic arcs. The ridge in the middle of the Molucca Sea marks a major thrust structure that is still stacking things up.
This tectonic pattern is almost unique in the world.
When we draw a cross section of seismicity across the Molucca Sea (along the same A-A’ profile as before), we can clear see the shape of the arched slab in the deep earthquakes. The slab seems to be still barely attached beneath the Molucca Sea, with intense seismicity that extends even below the crust, indicating that the slab itself is deforming. The background color for this profile is the estimated seismic P-wave velocity, which is slower in the crust and faster in the mantle, as taken from the Litho1.0 global model.
Zooming out, we can see that the Molucca Sea fits within a rapidly deforming mosaic of continental blocks, volcanic arcs, ocean basins, subducted slabs, and huge faults cutting through the crust. Global plate motion models aren’t reliable in this area, which is more like a tectonic “smush zone.” To show the complexity of the motions, we have plotted GPS velocities in an arbitrary reference frame. The convergence across the Molucca Sea is clear, at ~70 mm/yr. If the shortening continues at this rate, in about 3-4 million years the islands of Sulawesi and Halmahera be able to shake hands instead of just waving at each other.
This extreme degree of complexity is exacerbated by the remote and challenging conditions of this area, much of which is underwater and blanketed by volcanic deposits and sediments. This is why we tend to avoid talking about earthquakes around the Molucca Sea, unless they are particularly large or particularly interesting.
Fitting the October 21 earthquakes into a tectonic puzzle
Fortunately, the 21 October earthquake does have particularly interesting story. Let’s return to it.
The earthquake occurred between the islands of Bacan and Obi, on the east side of the Molucca Sea.
If you’re having trouble with your geography, here is a helpful map. The island of Bacan is the largest of the four Bacan Islands. The island of Obi, also called Obira, is the largest of the 42 Obi Islands. The October 21 earthquakes struck just west of two small Obi Islands, strongly felt onshore. If your hobby is learning the names of remote islands, this is the place for you.
The normal-type focal mechanisms indicate east-southeast to west-northwest crustal extension. Recall from our discussion above that the central part of the Molucca Sea produces mostly thrust-type earthquakes. Here, on the eastern side, the smooshed-up sediments are trying to ride up and over the eastern volcanic arc. So, why are we seeing extension?
If we zoom out a bit, and plot only the shallow (<30 km) earthquakes, we can see that a broad band of seismicity crosses Bacan Island, connecting a cluster of earthquakes in the Molucca Sea to earthquakes along the southern coast of Halmahera island, on the right-hand side of the map.
Zooming around, we can find some extremely interesting geomorphic features:
Shaded relief maps of the nearby islands, like Mandioli Island in the figure above, sometimes look like they have topographic contours already drawn. These lines are flights of marine terraces, which form at sea level. When the area is tectonically uplifted, those terraces are progressively raised, preserving the older ones high up as the newest terrace is formed at present sea level. In the tropics, these terraces are often decorated with, or even built out of, coral reefs. The huge flights of terraces on many islands in this area indicate rapid tectonic uplift.
Bacan Island itself exhibits some remarkable topography. There is an active volcano (elevation 927 m), but it is dwarfed by a tall mountain called Baku Sibela (elevation 2,085 m). This extremely rugged mountain is anomalous: it is the tallest landform between Sulawesi and West Papua, towering high above all of the surrounding islands. Only recently have guided multi-day tours become possible. The view from the top is apparently amazing (photo by Dan Quinn, retrieved from the Gunung Bagging website, a place for people who climb Indonesian mountains):
Because this ridge is completely non-volcanic, its steep, tall relief must reflect tectonic uplift as well.
At the same time, many of the islands are separated by small but deep marine basins. That suggests that major down-dropping has also occurred, also due to tectonics. This is an unusual type of basin-and-range topography, on a small scale.
Luckily, a recent study has already looked at this area in detail — recall the paper we touched on in the earlier section. Hennig‐Breitfeld et al. (2024) studied the tectonic ups-and-downs of Bacan Island and surrounds (a PDF of the open-access paper can be found here). Using better quality bathymetry than we have access to, they showed that the uplifted mountains are matched by downthrown valleys of similar area, giving a polygonal character to the landscape. They also mapped submarine reef complexes that have actually sunk downward beneath the waves, from their starting point near sea level down to depths of 2 kilometers. (Reefs can only grow in shallow water, where sunlight can still filter through the water.)
By applying thermochronology techniques to samples collected from Baku Sibela, Hennig‐Breitfeld et al. also determined that the rocks presently exposed along the tall ridge have been exhumed from significant depth in the crust to the surface only within the last ~1.4 million years. (As the name suggests, thermochronology dates how much time has passed since a rock cooled below a certain temperature, so we can back out an exhumation rate by assuming a temperature vs. depth curve). This implies pretty extraordinary average uplift rates of about 1 centimeter per year, for at least a million years (so, at least ~10 km of uplift).
So, some areas are moving up very fast, while other are moving down. What’s going on?
Hennig‐Breitfeld et al. discuss two possible reasons for this rapid vertical motion.
There are two plausible explanations of the observed rapid and very young exhumation of the Sibela Mountains metamorphic complex, one related to vertical movement along the Sorong strike-slip system and the other to eastward thrusting of the collision complex. - Hennig‐Breitfeld et al. (2024)
Let’s start with the second option.
Basically, the idea is that as the thick pile of sediments in the Molucca Sea drives eastward, it rests its weight on the crust of Halmahera Island. This causes the western edge of the island to sink, carrying down the reef complex into deep water. Because the crust is elastic, the downward flexure also causes an upward bulge further to the east, which increases the shallow tensional stress near Bacan Island. This could explain the normal-mechanism earthquakes that occurred on October 21. The following figure shows an elevation profile (black line), along with a model of this flexure (red line). This idea is similar to what happens in a subduction zone, where the downward flexure of a subducting slab induces an upward bulge outboard of the trench, with associated normal-mechanism earthquakes.
The reality of large-scale flexure is pretty clear from the downwarped coral reef complex, and it makes sense that a resulting flexural bulge could affect the crust near Bacan. However, we aren’t entirely convinced that this is the only factor at play. First, the bulge seems to be limited to the southern part of the Molucca Sea: it isn’t apparent much farther north than Bacan (to us at least), even though the collision is presumably also happening there. It’s also not clear to us how this flexural bulge would drive such rapid and local exhumation of deep-seated metamorphic rocks.
The other option is that the rapid uplift of Baku Sibela is due to slip on a nearby strike-slip fault system, called the Sorong Fault Zone. In the words of Watkinson and Hall (2017), that fault is the “wildcard of eastern Indonesian active tectonics.” That is because the topographic expression of the fault is extremely impressive, but existing observations of the fault aren’t sufficient to say much about its slip rate or seismic potential. One of the reasons that it is difficult to characterize this system is because it is actually many different faults: the AFEAD fault map (plotted below) shows many fault strands in this area that together compose the fault zone, including some that cut east-west near the Bacan and Obi Islands.
Because they slide mostly sideways, strike-slip faults are not very efficient at bringing deep rocks to the surface.
The Sorong fault zone cannot directly explain the October 21 earthquakes: the faults are strike-slip, oriented roughly east-west or northwest-southeast, and the earthquakes occurred on normal faults, oriented northeast-southwest.
However, relatively recent events have given us some clues about this fault and its influence on Bacan.
On July 14, 2019, a Mw7.2 North Maluku earthquake ruptured a significant part of the Sorong Fault along the southern arm of Halmahera Island — i.e., very close to the recent M5.8. This earthquake is incredibly understudied for a shallow M7.2 event, especially considering that it caused significant damage as well as injuries and loss of life (some of the casualties were apparently due to bad environmental conditions in temporary camps). According to the ISC Earthquake Bibliography entry, only two papers have been published that reference this large earthquake, and they are both concerned with electromagnetic disturbances possibly associated with earthquakes. ( the literature could use a study of this earthquake!… hint… hint).
We plotted up the seismicity from a one month period following the earthquake:
The focal mechanism of the mainshock indicates left-lateral strike-slip motion on a northwest-southeast trending, almost vertical fault, i.e. one of the strands of the Sorong Fault Zone.
While most of the aftershocks align with the southern coast of Halmahera Island, the earthquake also caused many aftershocks around Bacan Island. These events trace an alignment almost perpendicular to the Sorong Fault: running straight through Baku Sibela (the big ridge on Bacan Island). The aftershock trend basically takes a 90° turn.
Focal mechanisms for some of the aftershocks near Baku Sibela prove that this 90° turn is real, and that northeast-southwest trending normal faults were indeed activated. While the fault orientation of these normal faults is different than the Sorong Fault, the extension direction is basically parallel to the strike-slip motion. The normal faults that were activated are presumably the faults that are still lifting up the metamorphic ridge — perhaps as the footwall of a horst-and-graben structure, although no detailed structural models exist here.
If the long-term pattern of slip (over 100,000 years) is similar to the pattern from this one earthquake, that would suggest that the Sorong Fault here transfers into an extensional stepover, cutting across Bacan Island and heading west toward the Molucca Sea. The extension rate on the normal faulting system would be the same as the strike-slip rate. If that rate were high enough, then it could potentially drive rapid exhumation of about 1 cm/year. However, the slip rate of this part of the Sorong Fault is simply not well enough known to tell.
In this model, the earthquakes on October 21 also represent slip on the extensional system associated with the apparent termination of slip on the Sorong Fault.
So, which model is right? Maybe both.
Hennig‐Breitfeld et al. (2024) clearly demonstrated that tectonic flexure is important at this margin, and provided absolutely critical new constraints on how fast this unusual metamorphic complex has been unroofed.
However, the Sorong Fault also clearly had something to say in 2019. That points toward strike-slip faulting and a linked extensional stepover as another important factor. Where the stepover connects to is another important issue.
Is there a connection between the closure of the Molucca Sea and this strand of the Sorong Fault? Connecting old and cold geology to currently active structures is always a great challenge.
In any case, the October 21 earthquakes certainly picked an interesting place to happen. And having gotten to this end of this post, you may more deeply appreciate our hesitance to dive into earthquakes in this part of the world — truly the deep end of the tectonic pool. As usual, we welcome corrections or elaborations from those who have better knowledge of this tricky area!
References
Hennig-Breitfeld, J., Hall, R., White, L.T., Breitfeld, H.T., Forster, M.A., Armstrong, R.A. and Kohn, B.P., 2024. Age, origin and tectonic controls on rapid recent exhumation of the Sibela Mountains, Bacan, Indonesia. International Journal of Earth Sciences, 113(3), pp.501-521. https://doi.org/10.1007/s00531-024-02390-1
Pasyanos, M.E., Masters, T.G., Laske, G. and Ma, Z., 2014. LITHO1. 0: An updated crust and lithospheric model of the Earth. Journal of Geophysical Research: Solid Earth, 119(3), pp.2153-2173. https://doi.org/10.1002/2013JB010626
Watkinson, I.M. and Hall, R., 2017. Fault systems of the eastern Indonesian triple junction: evaluation of Quaternary activity and implications for seismic hazards. https://doi.org/10.1144/SP441
Zelenin, E., Bachmanov, D., Garipova, S., Trifonov, V. and Kozhurin, A., 2021. The database of the active faults of Eurasia (AFEAD): Ontology and design behind the continental-scale dataset. Earth System Science Data Discussions, 2021, pp.1-20. https://doi.org/10.5194/essd-14-4489-2022