Site icon

Intense seismic swarm punctuated by a magnitude 7.5 Japan shock

A magnitude 7.5 earthquake on the western shores of Honshu, Japan, caused extreme shaking and felled a five-story building on New Year’s Day. The shaking was far more severe than was thought likely for an earthquake of this size and depth.

By Shinji Toda, Ph.D., Tohoku University, Ross S. Stein, Ph.D., Temblor, Inc.

Citation: Toda, S., and Stein, Ross S., 2024, Intense seismic swarm punctuated by a magnitude 7.5 Japan shock, Temblor,

On the first day of 2024, a magnitude 7.5 earthquake struck the western shore of Honshu, Japan’s main island. The New Year’s Day shock caused strong shaking along the entire Noto Peninsula, and took 84 lives as of this writing. An at least 1.2-meter tsunami inundated parts of the Noto Peninsula, causing extensive damage. The Japan Times reported landslides, liquefaction and several hundred destroyed buildings — including the collapse of a five-story building.

This earthquake is unusual in two key ways. First, it was preceded by a an extremely active three-year long seismic swarm accompanied by continuous ground uplift in the epicentral region (Amezawa et al., 2023). Some researchers suspect the swarm results from upward migration of crustal fluids through existing fault networks (Yoshida et al., 2023). Second, the shaking produced by the earthquake was much larger than expected for an earthquake of its size and depth. Why it shook so strongly is unclear, but it may be due to the concentrated burst of fault slip beneath the Noto Peninsula.

Japan’s surprisingly active west coast — far from subduction

Japan’s Noto Peninsula has hosted a series of large earthquakes over the past 20 years (Figure 1). These are compressional events along surface-rupturing and buried thrust faults located onshore and offshore. (Buried faults don’t cut the surface sediments and are sometimes called “blind” though we will refrain from using that term.) The source of compression is the diffuse plate boundary between the island of Honshu and the Eurasian Plate, converging at a rate of between 13 and 28 millimeters per year, depending on the model (Taira, 2001). Much of this convergence occurs along the Noto Peninsula and Toyama Bay shore, with a rate less than 10 millimeters per year.

Figure 1. The region surrounding the 2024 magnitude 7.5 earthquake has been very seismically active in the past 20 years (left panel); quakes are the product of compressional faulting along a series of onshore and offshore thrust faults (right panel). Credit: Data from USGS (left panel), Inoue and Okamura (2010), and JMA (Japan Meteorological Agency) (right panel)


How frequent are earthquakes of this size on the Noto Peninsula?

Given a few millimeter per year long-term slip rate and the calculated 1- to 5-meter fault slip in the 2024 event, rough inter-event times of quakes of this size should be several hundred to a few thousand years.

The Noto Peninsula itself has been uplifted over the past million years, with the northern coast near the 2024 epicenter rising at an average rate of about 1 millimeter per year (Ota and Hirakawa, 1979). Based on satellite radar imaging, the earthquake caused 4 meters of uplift of the northwestern Noto Peninsula (which incidentally resulted in a shoreline advance toward the sea of about 200 meters, according to GSI). This uplift implies that the quake has a much longer inter-event time — about 4,000 years.

In Temblor’s 50,000-year worldwide earthquake simulation (also called a “stochastic event set”), the inter-event time for shocks greater than or equal to magnitude 7.5 within a 50-kilometer radius around the 2024 epicenter is 1022±134 years. The event set also includes a magnitude 7.4 earthquake within 6 kilometers of the 2024 event.

So, while three different ways to look at inter-event times of earthquakes resembling the 2024 event cover a wide range, about a thousand years appears plausible.

Dangerously high shaking

An especially troubling element of the earthquake is its generation of very strong shaking within 40 to 50 kilometers of the rupture (Figure 2). The U.S. Geological Survey (USGS) ShakeMap model under-predicted the shaking within 40 kilometers of the rupture by factors of two to five.

Figure 2. The shaking from the magnitude 7.5 earthquake was recorded by the Japanese strong ground motion network, with observations over 15 to 600 kilometers from the fault rupture modeled by the USGS. The shaking within 40 kilometers of the rupture is extremely high — about four times higher than the ground motion model expected by the USGS ShakeMap algorithm. (See orange and red triangles on the left side of the figure. Some are well above the USGS model confidence interval.) Credit: USGS v.1 ShakeMap Analysis


Often, stronger shaking is associated with a more compact source, with higher slip over a smaller area than usual. But the rupture is about 100 kilometers long according to the USGS and Japan’s Geographical Survey Institute (GSI) and so is typical of its magnitude. Nevertheless, the highest slip is concentrated in the northern Noto Peninsula (Figure 3), and so this burst of slip along a southeast-dipping thrust fault beneath the coastal towns might explain the extreme shaking. The 2007 magnitude 6.7 earthquake off the Noto Peninsula (Figure 1) also shook harder than the ShakeMap model, but not nearly by so much. And another 2007 magnitude 6.6 quake that ruptured offshore to the east (Figure 1) produced shaking consistent with ShakeMap.

So, it’s not clear if very high shaking is a feature of the Noto peninsular region (e.g., perhaps due to thick local alluvium amplifying the shaking) or if the extreme shaking is more unique to the 2024 event. Regardless of the cause of strong shaking, the 2024 shaking appears to have triggered landslides and liquefaction, and likely played a role in the collapse of a five-story building. Typical two-story wood-framed houses in Japan, particularly those built before World War II, suffered the greatest damage in this earthquake, similar to what happened in the 1995 magnitude 6.9 Kobe earthquake.

The measured shaking, with five stations recording over 1 g (this is the force of gravity acting sideways), exceeds even the most stringent building codes worldwide, and so would likely have caused catastrophic damage almost anywhere outside of well-prepared and resilient Japan. Nonetheless, it’s a reminder of what could happen in the earthquake-prone region.

Figure 3. The epicenter of the magnitude 7.5 shock lies in a “saddle” in the calculated fault slip (USGS, 2024). Over a 30-second period, the rupture propagated both to the southwest, where the slip peaked at about 3 meters, and also to the northeast, where it reached 1.5 meters. Aftershocks (from the USGS ANSS catalog) are seen both near the site of high slip in the southwest and in the epicentral slip saddle. JMA (Japan Meteorological Agency) plans to release the full catalog of aftershocks greater than or equal to magnitude 1.0 soon, which will provide a richer and perhaps more accurate picture. Credit: Modified from USGS


Massive seismic swarm preceded quake

In December 2020, a powerful seismic swarm began in the area that eventually hosted the epicenter of the magnitude 7.5 quake. The swarm included 48 shocks greater than magnitude 4.0 and seven quakes greater than magnitude 5.0 (Figure 4), including a magnitude 6.5 earthquake that occurred in May of 2023. A magnitude 7.5 shock following a seismic swarm is highly unusual. In fact, if swarm precursors were common, we could predict some earthquakes — and we cannot.

But by the same token, it is hard to imagine that the swarm and the magnitude 7.5 epicenter coincided by chance. Instead, we view the swarm as a clue that conditions for failure at this site dramatically changed three years ago, culminating — in a manner we do not fully understand — with the nucleation of a large earthquake.

Figure 4. The intense swarm beginning in December 2020 is remarkable. The earthquake frequency of magnitude greater than 1.0 shocks increased by about 400-fold, and so the probability of large shocks most likely also increased by the same factor. Note that the March 2007 magnitude 6.7 Noto-Hanto quake, far west of this area, did not increase the seismicity rate in the future 2024 epicentral site, and did not initiate a swarm. Credit: Toda and Stein, 2024, CC BY-NC-ND 4.0


What we can say is that the swarm raised the seismicity rate by a factor of about 400. Often, swarms have a higher ratio of small to large shocks than is expected; this is typically the case for volcanic swarms in Japan (Toda et al., 2002) and elsewhere. But in this case, the ratio of small to large shocks (the “b value” in seismic parlance) did not change. So, if the rate of small shocks increased by a factor of 400, the simplest interpretation is that the probability of large ones did too. Regarding the area in the inset of Figure 4, we calculated that the chance of a shock greater than or equal to magnitude 7.0 grew from about 0.08% per year before the swarm to 3% per year in the past three years, which is extraordinarily high.

Because the swarm’s seismicity has grown shallower with time, it has been interpreted to accompany the rise of crustal fluids through the network of faults and fractures at the head of the Noto Peninsula (Yoshida et al., 2023; Amezawa et al., 2023). Fluid pressures are not actually observed since the earthquakes are so deep, and so this remains a hypothesis. Unlike far more common volcanic swarms, there does not appear to be magmatic processes (with molten rock at depth) occurring at Noto, making it all the more unusual.

Effects of the 2007 magnitude 6.7 Noto-Hanto earthquake

So, is it coincidental or causative that the epicenter of the magnitude 7.5 shock, the western extent of the 100-kilometer-long 2024 rupture, and the site of the preceding seismic swarm are all located in an area of increased earthquake probability (a “trigger zone”) due to the 2007 magnitude 6.7 event? This earlier Noto Peninsula earthquake produced a pattern of four regions, or lobes, rich in aftershocks. Toda (2008) explained this by Coulomb stress transfer, in which faults brought closer to failure produced aftershocks (the red “trigger zones” in Figure 5), and faults brought further from failure inhibited aftershocks (the blue “stress shadows” in Figure 5).

To produce a large earthquake, a large fault is necessary; to promote an earthquake, stress must accumulate. The faults certainly exist (as shown in the right panel of Figure 1), and the 0.25-bar stress change imparted by the 2007 shock is modest, but enough to trigger seismicity. But what’s clear is that the 2007 quake did not initiate the swarm (as shown in Figure 4). Perhaps more important is that the highest-slip region of the 2024 earthquake (as shown in Figure 3) occurred where the greatest stress was imparted by the 2007 event (as you can see in Figure 5 below), just to the northeast of the 2007 rupture.

Figure 5. Toda (2008) calculated the stress transfer to surrounding faults with a geometry similar to the 2024 magnitude 7.5 shock (which had a strike of 49°, dip 42°, rake of 102° according to the USGS). The epicenter of the magnitude 7.5 shock, and the locus of the 2020-2023 seismic swarm, lie in a region of calculated 0.25-bar stress increase. Credit: Toda (2008), with the Jan. 1, 2024, epicenter added


What next?

We calculate that the 2024 shock increased stress by several bars on nearby active faults beyond the ends of the rupture, some of which are experiencing aftershocks. This observation might indicate that our calculations have some reliability (see Figure 6). Whether the stressed faults will fail in a progressive earthquake sequence, and whether the seismic swarm will continue, are unknown.

Figure 6. Stress transferred by the 2024 magnitude 7.5 earthquake to the surrounding faults indicates that several faults beyond the ends of the rupture were likely brought closer to failure (NT9, NT10, KZ1, KZ2, SD05, SD06). In contrast, most faults in Toyama Bay should be inhibited from failing as a result of the mainshock. Credit: Toda and Stein, 2024, CC BY-NC-ND 4.0


New observations lead to unanswered questions

This large inland quake has raised some troubling and scientifically interesting questions about the strength of seismic shaking and the role of crustal movement, seismic swarms and Coulomb stress transfer in promoting the earthquake.

What this event has in common with the 2023 magnitude 7.8 and magnitude 7.7 Kahramanmaraş, Turkey, earthquakes is excellent seismic monitoring networks and extensive preceding research investigations — a gift to seismology and to public safety. Both cases show shaking near the rupture much stronger than heretofore thought possible. Each earthquake sequence also shows how we are just beginning to decipher fundamental issues in seismology like progressive earthquakes and seismic swarms.

Science editor: Dr. Alka Tripathy-Lang, Ph.D.
Reviewer: Dr. Wendy Bohon, Ph.D.


Amezawa, Y., Hiramatsu, Y., Miyakawa, A., Imanishi, K., & Otsubo, M. (2023), Long-living earthquake swarm and intermittent seismicity in the northeastern tip of the Noto Peninsula, Japan. Geophys. Res. Letts., 50, e2022GL102670.

GSI (Geographical Survey Institute) (2024),

Inoue, T., Okamura, Y. (2010). Digital geological map. Bull. Geol. Surv. Jpn. AIST S-1.

Ota, Y. and Hirakawa, K. (1979) Marine terraces and their deformation in Noto Peninsula, Japan Sea side of central Japan, Geographical Review of Japan, 52, 169-189, (in Japanese with English abstract).

Taira, A. (2001), Tectonic evolution of the Japanese island arc system, Ann. Rev. Earth and Planet. Sci. 14, 109-34,

Toda, S. (2008), Coulomb stresses imparted by the 25 March 2007 Mw=6.6 Noto-Hanto, Japan, earthquake explain its ‘butterfly’ distribution of aftershocks and suggest a heightened seismic hazard. Earth Planet Sp 60, 1041–1046.

Toda, S., Awata, Y. (2008), Does the 2007 Noto Hanto earthquake reveal a weakness in the Japanese national seismic hazard map that could be remedied with geological data?. Earth Planet Sp 60, 1047–1052.

Toda, S., Stein, R. and Sagiya, T. (2002), Evidence from the AD 2000 Izu islands earthquake swarm that stressing rate governs seismicity. Nature 419, 58–61.

USGS (2024), Finite Fault model for 1 Jan 2024 M 7.5 earthquake,

Yoshida, K., Uchida, N., Matsumoto, Y., Orimo, M., Okada, T., Hirahara, S., et al. (2023). Updip fluid flow in the crust of the northeastern Noto Peninsula, Japan, triggered the 2023 Mw 6.2 Suzu earthquake during swarm activity. Geophysical Research Letters, 50, e2023GL106023.


Text © 2023 Temblor. CC BY-NC-ND 4.0

We publish our work — articles and maps made by Temblor — under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International (CC BY-NC-ND 4.0) license.

For more information, please see our Republishing Guidelines or reach out to with any questions.