Magnitude 6 earthquake shakes Taiwan

Thrusting within a complicated and mysterious tectonic collision

若要閱讀該文章的中文版（由Google自動翻譯），請點擊這裡。

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.

A magnitude 6 earthquake struck southwestern Taiwan on January 21, 2025 just after midnight local time (2025-01-20 16:17 UTC). This is a seismically active region that has produced many shallow, large, and dangerous earthquakes. Let’s take a look at the earthquake effects, and why this area is so seismically active. First, a plot of M5+ earthquakes in and around Taiwan:

Figure 1: Map of M5+ earthquakes in and around Taiwan. Earthquakes are colored by depth.

The map above only plots the earthquakes larger than magnitude 5, because the entire area gets covered up when smaller earthquakes are included! Even a zoomed-in view of the source region has thousands of instrumentally recorded events:

Figure 2: Recorded earthquakes around the location of the 2025-01-20 M6 earthquake, colored by depth. The timeline below shows seismicity since 1940.

The timeline of seismicity (for only the map area above) shows that M6+ earthquakes occur in this particular region about every 5 years or so. The catalog improves dramatically around 1990, not because more earthquakes happened, but because more seismometers were deployed in and around Taiwan. We will discuss the large earthquake that occurred in 1941 in a moment.

While the USGS and EMSC are reporting today’s event as moment magnitude 6 (Mw), the Taiwan Central Weather Administration reports local magnitude 6.4 (ML). The moment magnitude is typically considered more authoritative, and more capable of describing the energy released in the earthquake, but the local magnitude can often be more useful to understand the intensity local shaking.

Indeed, this earthquake caused severe shaking near the epicenter, reaching intensity 6- on the Taiwan seismic intensity scale (this value is roughly equivalent to MMI VIII). A seismometer at Dapu, five kilometers from the epicenter, recorded the waveforms below, showing that accelerations in the east-west direction apparently exceeded twice g, the acceleration due to gravity. That’s a pretty intense number! However, previous earthquakes have taught us that isolated strong acceleration readings can be non-representative of the actual shaking due to site effects, so we shouldn’t draw too many conclusions from this single observation without further investigation.

Figure 3: Seismic waveforms in the 2025-01-20 M6 earthquake in Taiwan, recorded at Dapu. Source: https://www.cwa.gov.tw/V8/E/E/EQ/EQ114007-0121-001727-WaveAcc.html?v=20250121002813

The Japan Meteorological Agency provides this useful visual description of what that kind of shaking can do — including causing non-resilient buildings to collapse (the “6 Lower” panel). Indeed, news reports indicate that at least one house did collapse in this event, and a number of others were damaged. (Edit: as noted by Dr. Wald in the comment below, the intensity scale used in Taiwan does not exactly match that used by the JMA — you can find a written description of the scale used by the Taiwan Central Weather Administration here. In general, the numbers match up pretty well, but the values are calculated in a different way. For more details, please see Dr. Wald’s comment below, and the paper that he links: Wu et al., 2003. Sorry for the confusion! We learn something new every day, and really appreciate the contributions of experts to that effect.)

Figure 4: Visual representation of the impact of shaking at different JMA intensity levels. https://upload.wikimedia.org/wikipedia/commons/d/de/Summary_of_Tables_explaining_the_JMA_Seismic_Intensity_Scale.pdf

Fortunately, the spatial area of these strong intensities was relatively limited because the earthquake was quite shallow, although shaking was probably felt all across Taiwan. In the capital city of Taipei, at the northern end of the island, many people may have slept through the shaking (Taiwan intensity 2).

Figure 5: Shaking intensities reported for the 2025-01-20 M6 earthquake. Source: Taiwan Central Weather Administration. https://www.cwa.gov.tw/Data/earthquake/zip/EEQ114007-2025-0121-001727/2025007ie.png?v=20250121002813

Taiwan has a topographic backbone: a tall mountain range that runs north-south. The rugged terrain is beautiful, but its steepness discourages habitation. Instead, most people live within the low, fertile plains on the western side of the island, with a few communities in the various mountainous valleys within the island and along its eastern edge. The recent earthquake occurred within the southwestern foothills of the mountains, about 15 kilometers east of the densely populated plains.

Figure 6: Screenshot from Google Earth, showing the earthquake location. Note the dark greens around the epicenter and to the east, in contrast to the lighter tans representing populated regions to the west.

Taiwan has a long history of damaging earthquakes in its western ranges. The most important of these was the September 21, 1999 Chi-Chi earthquake, a Mw7.7 event. That earthquake triggered rockfalls, damaged schools, destroyed dams, and ultimately killed 2,415 people. Closer to today’s epicenter, the December 17, 1941 Mw7.2 Chungpu Earthquake was also highly destructive.

Much of the earthquake planning and resilience in Taiwan today is a consequence of the public and government responses to the 1999 earthquake. Indeed, that resilience was put on display in May 2024, when a M7.4 earthquake struck the city of Hualien, in eastern Taiwan. While 18 people died in that event, the damage could have been much worse without significant preparation and improvements in building construction over the last decades.

Tectonic context

So, why did this earthquake happen? Well, there is an easy answer, and a more difficult answer.

In general terms, it is pretty easy to understand why earthquakes occur in Taiwan: it is a nexus of deformation. Northeast of Taiwan, the Ryukyu Trench, which extends southwest from Japan, curves to the west and terminates against the northeastern part of Taiwan: here, the Philippine Sea Plate is sinking northward beneath the Eurasian Plate. Rollback of the Philippine Sea Plate is opening up the intervening Okinawa Trough, which incidentally hosted a cluster of M5 extensional events a few days ago.

And yet, the Eurasian Plate to the west of Taiwan is moving east, and sinking beneath the Philippine Sea Plate. Taiwan itself has been caught between these opposing subduction systems for millions of years. The earthquakes we see are just the ongoing negotiation between the plates in this tremendous oblique collision zone.

If that didn’t make sense to you, take a look at the cartoon below.

Figure 7: Schematic 3D diagram showing the plate interactions below Taiwan. Source: Song (1993).

If the cartoon still boggles your mind, don’t worry. You have joined the ranks of countless geologists who have struggled to visualize this particular 3D geometry.

The epicenters of the 1999 and 1941 earthquakes, along with today’s magnitude 6 earthquake to the south, both fall within the fold-and-thrust belt running down the center of the island. There, a thick pile of sediments is being scraped off of the sinking Eurasian Plate, lifted up above sea level, and piled up in great slivers to form the foothills of the mountains. This is a classic example of an on-land accretionary wedge. A great deal of effort went into mapping the faults responsible for the Chi-Chi earthquake in particular — and that’s where our story gets tricky.

The faults at depth can be mapped in remarkable detail using a combination of data from oil and gas wells, surface geology, and seismic reflection imaging.

Older cross sections show stacked ramp faults that extend downward to about 10-14 kilometers depth. If these sections were correct, then events like the Chi-Chi earthquake and today’s earthquake might just be interpretable as slip on ramp faults. In that case, the scraping-off of Eurasian sediments is the whole story. For instance, see these two older cross-sections from around the location of today’s magnitude 6 earthquake.

Figure 8: Two older geological cross-sections in the region of the 2025-01-20 M6 earthquake. Red star marks the approximate location of the earthquake.

The focal mechanism for today’s earthquake would have be a great match for the Lunhou Thrust, or a similar ramp fault dipping 34°, as shown on these older cross sections. However, we now know that this is likely not the case.

Figure 9: Focal mechanism of the 2025-01-20 M6 earthquake, from the USGS. https://earthquake.usgs.gov/earthquakes/eventpage/us6000pl8h/moment-tensor

An important branch of structural geology that relates faults and folds can help us make better cross sections. Using these techniques, combined with clues from wells and subsurface imaging, a clearer picture emerges: the steep thrust faults rise upward from a relatively flat fault that sits only about 5-6 kilometers below the surface. That fault is known as the Chinshui detachment. The mountain ranges we see today are the eroded remnants of the rocks that have first slid easily along the detachment and then been forcefully shoved upward on the ramp faults.

We know that at least one of these shallow ramp thrusts slipped in the 1999 Chi-Chi earthquake — the Chelungpu thrust — because impressive offsets were mapped at the surface. We also know that the detachment fault slipped, because sites farther into the mountains translated moved a few meters horizontally, as recorded by GPS stations. These motions suggest that the earthquake was just an incremental growth of the accretionary wedge, as expected.

However, seismic records show that the earthquake started much deeper, at ~8-10 kilometers: below the base of the Chelungpu thrust, and below the Chinshui detachment. That’s where the complex part of the story begins.

Figure 10: Two detailed cross-sections showing fault structures in the region of the 1999 Chi-Chi earthquake. The two focal mechanisms show hypocenters that fall below the detachment. Figure 5 of Yue et al. (2005).

When data are well constrained but don’t match the expectations from our simple models, we have to keep our minds open to more complex options. This scenario is extremely common in geology. In this case, the deep origin of large earthquakes beneath western Taiwan ultimately forced geologists to draw the following (somewhat mind-bending) cross section:

Figure 11: A geological cross-section in the area of the 1999 Chi-Chi earthquake, illustrating the connection of a deep fault with the shallow fold-and-thrust belt. Figure 16a of Yue et al. (2005).

The data show that the lower plate (the leading edge of Eurasia) is breaking up as it is forced downward beneath the high mountains of Taiwan. We can’t map these deep faults at the surface because they are covered by an inconveniently enormous mountain range, but we can infer their presence from their seismicity. Sometimes (although we don’t know how commonly), rupture of a deep fault can trigger a huge rupture on the shallower fault system — that’s what happened in 1999.

So, while older geological cross-sections across southern Taiwan show thrust faults dipping east at roughly the depth of the recent earthquake, more recent cross-sections have revised the detachment upward — placing it at a depth of 5-6 kilometers. Meanwhile, Taiwan’s Central Weather Administration places the hypocenter of today’s M6 earthquake at 9.7 km. So you can see the problem.

We have collected a few published cross-sections to further illustrate the conundrum. Please forgive minor inaccuracies in how we have projected the earthquakes onto the sections. The key point is: if we stick to the old cross-sections (left), we could easily match the earthquake to a fault: the Lunhou Thrust is right next door. The earthquake would be good evidence that the fault is still active. However, more recent and much more reliable cross sections (right) show that the earthquake hypocenter — like most of the hypocenters in the entire region — occurred well below the detachment.

This isn’t actually that surprising: in other accretionary prisms, we often see that much of the seismicity occurs in enigmatic structures below the fold-and-thrust-belt — for instance, in Bangladesh. The real question is whether and how those small earthquakes might be related to the larger, much more hazardous faults above.

Figure 12: Four geological cross-sections in the region of the 2025-01-20. M6 earthquake. Red stars mark the approximate projection of the hypocenter onto the cross-sections. Left: older cross-sections showing more deeply-rooted faults, which seem to match up with the hypocenter. Right: more recent cross-sections, showing the hypocenter below the detachment.

So, while the simple geo-story that the earthquakes in western Taiwan are due to collision is certainly true, exactly how that happens is complicated.

If we accept the shallow detachment model, some important questions spring to mind, .

1. How exactly do deep ruptures jump onto the detachment fault? Are there special characteristics of the occasional ruptures that do trigger the shallow system?

2. Can the detachment fault ever start its own large earthquakes, or does it need to be kick-started by a deeper event? Stein and Bird (2024) recently suggested that great strike-slip events may always start on smaller, connected splay faults — maybe the same is true for some detachment systems?

3. How many of the ramp faults in western Taiwan are actually active today?

Fortunately, people are working on answering these kinds of questions.

Only about three weeks ago, Le Béon et al. (2024) published a new study that carefully mapped river terraces in the region, to determine how much they had been offset along the Lunhou Thrust (among other faults). By dating terraces of different ages and offsets, they were able to estimate slip rates for different faults in the area. Their results indicate that the Lunhou Thrust isn’t really even active anymore, and hasn’t been for the last 2,000 years. They write: “In terms of active deformation, the main players appear to be the Kouhsiaoli Fault and Wushantou Anticline, while the geologically most significant Lunhou Thrust seems inactive.” A cross section from their paper is shown above. Given the preceding discussion, their observations aren’t that surprising. However, they again highlight the ongoing problem of the deep earthquakes — deeper than the mountains themselves — beneath western Taiwan.

And if that weren’t enough to wonder about, there is (at least) one more process at work: some of the fault related structures above the detachment are apparently growing aseismically — creeping little by little, rather than (or in addition to) the sudden slip of earthquakes. This creeping behavior is thought to be related to the materials that make up the cores of the anticlines: clayey rocks with high fluid pressures.

This might seem like a good thing: stresses that are released aseismically can’t be released by earthquakes. And indeed, one study of the Houchiali fault, located in southern Taiwan, highlighted that this fault has produced fewer M~6 earthquakes than you might expect given the shortening rate. However, we also know that faults can experience both kinds of behavior over time. Given the high deformation rates, complex fault systems, and history of large earthquakes in this region, it is clear that while some stress may be released aseismically, there is plenty left over to create large earthquakes.

We welcome input or corrections from anyone who has more experience with the complex geology in this area! Please leave a comment below.

References:

Brown, D., Alvarez-Marrón, J., Camanni, G., Biete, C., Kuo-Chen, H. and Wu, Y.M., 2022. Structure of the south-central Taiwan fold-and-thrust belt: Testing the viability of the model. Earth-Science Reviews, 231, p.104094. https://doi.org/10.1016/j.earscirev.2022.104094

Bürgi, P., Hubbard, J., Akhter, S.H. and Peterson, D.E., 2021. Geometry of the décollement below eastern Bangladesh and implications for seismic hazard. Journal of Geophysical Research: Solid Earth, 126(8), p.e2020JB021519. https://doi.org/10.1029/2020JB021519

Hubbard, J. and Bradley, K., 2024. Deadly M7.4 earthquake strikes Taiwan. Earthquake Insights, https://doi.org/10.62481/c4a3297f

Hung, J.-H., Wiltschko, D. V., Lin, H.-C., Hickman, J. B., Fang, P., Bock, Y., 1999. Structure and motion of the Southwestern Taiwan Fold and Thrust Belt. Terrestrial, Atmospheric, and Oceanic Sciences, 10 (3), p. 543-568. https://doi.org/10.3319/TAO.1999.10.3.543(T)

Le Béon, M., Chen, C.C., Huang, W.J., Ching, K.E., Shih, J.W., Tseng, Y.C., Chiou, Y.W., Liu, Y.C., Hsieh, M.L., Pathier, E. and Lu, C.H., 2024. Aseismic deformation within fold-and-thrust belts: example from the Tsengwen River section of southwest Taiwan. Geoscience Letters, 11(1), p.57. https://doi.org/10.1186/s40562-024-00373-3

Le Béon, M., Marc, O., Suppe, J., Huang, M.H., Huang, S.T. and Chen, W.S., 2019. Structure and deformation history of the rapidly growing Tainan anticline at the deformation front of the Taiwan mountain belt. Tectonics, 38(9), pp.3311-3334. https://doi.org/10.1029/2019TC005510

Mouthereau, F., Lacombe, O., Deffontaines, B., Angelier, J. and Brusset, S., 2001. Deformation history of the southwestern Taiwan foreland thrust belt: insights from tectono-sedimentary analyses and balanced cross-sections. Tectonophysics, 333(1-2), pp.293-322. https://doi.org/10.1016/S0040-1951(00)00280-8

Song, G.-S., 1993. Numerical simulations of collision process in Eastern Taiwan. Terrestrial, Atmospheric, and Oceanic Sciences, 4(1), http://dx.doi.org/10.3319/TAO.1993.4.1.141(O)

Stein, R.S. and Bird, P., 2024. Why do great continental transform earthquakes nucleate on branch faults?. Seismological Research Letters, 95(6), pp.3406-3415. https://doi.org/10.1785/0220240175

Yue, L.F., Suppe, J. and Hung, J.H., 2005. Structural geology of a classic thrust belt earthquake: the 1999 Chi-Chi earthquake Taiwan (Mw= 7.6). Journal of Structural Geology, 27(11), pp.2058-2083. https://doi.org/10.1016/j.jsg.2005.05.020