We need your help! We are dedicated to providing rapid scientific analysis following earthquakes, but to make this effort sustainable, we need financial support. If you can, please consider upgrading to a paid subscription. If you use our Substack as part of a group, or for teaching, consider a teaching or institutional subscription. Every paid subscription makes a real difference — and unlocks access to the more than three hundred posts in our archive!
Free subscribers still get all of our new posts by email, and all of our posts are free to read for one month. If you have recently lost your job or are a student or researcher without ability to pay, let us know and we will upgrade your subscription to “paid” at no cost.
Residents of San Ramon, California have been feeling on edge as a swarm of earthquakes has rattled the ground beneath their feet over the last month. The earthquakes began on November 9th with a magnitude 3.8 event at 9:38 AM local time. The largest event so far was a magnitude 4.0 earthquake at 7:56 PM on December 19th (2025-12-20 UTC time), shown on the map and timeline below as the targeted event.
Because these earthquakes are located close to one of the USA’s largest cities, many news reports have appeared over the last month. Most of these articles tend to follow the same basic script: Another earthquake has happened. Is this a sign of the coming “Big One”? A geophysicist was interviewed: they said “No.”
Here, we will try to dig a little deeper and look at some of the reasoning behind that “No” answer (usually qualified by pointing out that large earthquakes are nevertheless inevitable in this region, which is riddled with large and dangerous faults).
None of the earthquakes in the active swarm have been strong enough to cause significant damage. Based on seismometer readings and felt reports to the USGS, the maximum shaking in the largest-magnitude probably reached intensity IV (light). Shaking at that level can be disconcerting — like a truck bumping into a building, causing dishes to rattle and walls to creak. Based on reports to the USGS DYFI program, many people across the Bay Area have sensed shaking from some of these earthquakes.
One person was quoted in the news as having been warned by their cat that one of the earthquakes was coming.
He made “weird little noises” and hid under a table. “Within seconds of him yelling this big loud meow, there was a big earthquake,” Heys said. “It really shook me. It sounded like a dresser hit the wall. It was like this one big bang.”
People have posited for centuries that animals might be able to detect earthquakes in advance, citing the odd behavior of snakes, frogs, pets, and domestic animals in the minutes, hours, or days before large earthquakes. However, there is no evidence to support this hypothesis, and no observations of anomalous signals during those time periods. What is possible is that some animals may react to the first ground shaking — the P wave — which is much smaller amplitude than the shear and surface waves that follow.
As an example, here is one seismogram for the 2025-12-20 M4.0 earthquake. The first several seconds of shaking after the P arrival might pass undetected, except perhaps by an especially sensitive cat, creating the impression that the cat had special advance warning.
(On the topic of cats, we would like to share with our readers this video of our newest kitten being welcomed into the family by an already resident cat. We suspect that these particular orange cats are too lazy to give us any early warnings, but in the unlikely scenario that a relevant earthquake does cause shaking in Ithaca, NY, we will be sure to mention their reaction.)
Ok, enough clickbait. What’s going on in California?
Since we started Earthquake Insights in 2023, we have written over two dozen posts about earthquakes in California. It might seem surprising that few of those posts were about earthquakes on the San Andreas Fault. While the San Andreas is certainly the most prominent strike-slip fault in California, it actually only takes up about half of the five centimeters per year of relative motion along this plate boundary. The rest is accommodated by a complex network of faults, many of them concentrated in a ~100-kilometer-wide band along the western side of California. That being said, active faulting extends as far east as New Mexico!
Most earthquakes in California occur on these other faults, including this latest seismicity. The current swarm has been occurring below the San Ramon Valley, about 45 kilometers northeast of the San Andreas Fault.
So, what fault is responsible for this current misbehavior? The San Ramon Valley is flanked by, and was presumably formed due to the action of, two fault systems: to the southwest, the Calaveras Fault System, and to the northeast, the Pleasanton Fault. The recent earthquakes are clustered near the trace of the Pleasanton Fault.
Cross-sections show that the seismicity can be traced almost vertically down from the fault trace to depth, perhaps with a slight dip towards the southwest. This suggests that the seismicity is associated with the Pleasanton Fault, although (as we shall see in a moment) past seismicity in this area has illuminated all kinds of fault structures at depth, some cutting perpendicular to the main fault traces. It is likely that more accurate relocation of these events in the future will provide a much more accurate view of the activated fault geometry.
Most tectonic earthquakes follow a general pattern: there is a mainshock (with or without smaller foreshocks), followed by smaller aftershocks that decay away with time. For earthquakes of this type, we generally attribute the aftershock sequence to stress triggering, where slip on one fault changes the stresses on other faults in the same region, causing them to also break.
An earthquake swarm does not follow that (foreshock)-mainshock-aftershock pattern. Instead, we typically see multiple earthquakes occurring within a small region, with little rhyme or reason to the magnitudes.
Earthquake swarms are often attributed to physical processes happening down inside the crust, like fluids moving along faults, or in volcanic regions, movement of magma. The seismicity is a response to these other forces at work, rather than just triggering from an earlier earthquake.
However, swarms can sometimes also take on characteristics of typical fault rupture, with mainshock-aftershock sequences rising up out of the swarm. This is basically what happened at Santorini, Greece at the beginning of this year, where injection of magmatic dykes into the crust also activated shallow tectonic faults.
The most recent earthquakes near San Ramon share some features of both patterns. There have been several “mainshocks” of similar magnitude (~M3.7-M4). Those earthquakes have each been followed by smaller aftershocks. For instance, this seismic sequence kicked off with a “mainshock” of M3.8. In the weeks that followed, the events were all much smaller (maximum M3.3). On December 8th, there was an M3.6 — plausibly triggered by the M3.8.
It wasn’t really until just recently that the seismicity developed a more swarmy character: on December 20-21, we saw an M3.8, followed by an M4.0, followed by an M3.9, all within 24 hours.
The main reason that people have been relatively quick to label this sequence as a swarm is because there is a long history of swarm-type seismicity around San Ramon.
Here is a map of earthquakes in the region, colored by time. Earthquakes before 1970 are colored black. Clusters of similar color events can either be mainshock-aftershock sequences, or swarms. For this and the following figures, we are using data from the Northern California Seismic Network, conveniently available via Github, along with data from the USGS ComCat database covering the last month.
As you can see, this map looks a bit like the wall of a paintball arena. Instead of a nice line of earthquakes, like we can see along the Hayward Fault along the left edge of the figure, the hilly upland region has a lot of scattered earthquake clusters. Many of the earthquake clusters do not occur along the mapped active faults, most of which are vertical strike-slip structures oriented northwest-southeast. Instead, the clusters tend to form tight groupings, or sometimes line up in various directions that seem unrelated to the mapped faults.
If we zoom in to the area between Alamo and Dublin (an odd association of otherwise familiar names), and plot a timeline, we can see just how scatter-shot the earthquakes have been, in both space and time.
The long timeline above compresses the individual clusters, which usually only last for a few months. Let’s examine some more zoomed-in maps and timelines of selected clusters, as indicated on the figure above. Note that each following figure has its own color scale, to allow us to look at the time evolution within that cluster.
The earliest major swarm for which we have good data was in 1970. The swarm began with a single small event followed by a smattering of even smaller earthquakes, waited a few months, and then burst into more serious activity. Although it is difficult to define when a swarm actually ends, it looks like this one lasted for a little less than one year, with the main action taking place over about three months.
Our second swarm occurred in 1976. It started out with a sudden burst on August 15th, had a few larger events around the 21st, and lasted for only two more weeks.
In 1990, a very interesting swarm began with a burst of events on a NE-SW oriented plane, cutting across the mapped fault trace and the chains of hills. The first earthquake in the sequence was not a mainshock; it took about 10 hours to reach the highest magnitude events. That initial burst died down, only to be followed by a second, similar burst a few weeks later, which apparently re-activated the same fault structure. A baby burst also happened just to the south, and the swarm faded away over a period of months.
In 2003, the area where we are currently seeing earthquakes experienced a different swarm. As with the current sequence, the early earthquakes occurred near the trace of the Pleasanton Fault, along the east side of the valley. That swarm died down over the course of about two months, and then a second burst with several large events occurred to the south.
In 2015, we saw a particularly interesting swarm over the course of about one month. Beginning with a small burst, the earthquakes propagated both toward the north and toward the south, eventually leaving a quieter area between the two active branches. We will discuss this event further in a moment, by referring to a published study.
Starting in February of 2018, another swarm occurred, with periodic bursts migrating both northward and southward, separated by weeks of quiescence. The largest magnitude event was quite small at M~3.4. That swarm lasted about four months.
This brings us up to present day. As in 2003, the presently active swarm again seems to be expanding slowly southward and westward. This swarm has already lasted for two months, with periods of quiet followed by renewed earthquake bursts. We will have to wait to see how it evolves over time!
Earthquake swarms can be disconcerting because sensible shaking can go on for months, and there are no guarantees about how the swarm will evolve or when it will end. This is unlike the more typical mainshock-aftershock sequences, which tend to decay away on a (fairly) predictable timescale.
The causes of swarm seismicity can also be difficult to interpret, partly because swarms in different regions might involve entirely different processes. For example, the anxiety-inducing earthquake swarms in the central Aegean Sea region (near Santorini) at the start of this year were caused by deep magmatic intrusions that also activated shallow tectonic faults. That type of mechanism is highly unlikely in the San Ramon area, where no magmatic activity is known or expected.
Fortunately, researchers have already looked into the dynamics of some of the previous swarms in this area, so we can lean heavily on their results.
The most recent exploration that we are aware of was Xue et al. (2018), who studied the space-time evolution of the 2015 earthquake swarm, and also interpreted several earlier swarms. The image below, from that paper, shows the seismic clusters they examined, colored by when they occurred.
Xue et al. used more advanced seismic techniques to accurately locate the events in 3D, and also calculated focal mechanisms for the larger events to constrain both slip sense and fault orientations. They found that the earthquakes illuminated three different faults, all of them dipping towards the northwest, rather than aligning with the main Calaveras Fault. Panel b in the figure below is a map view, showing how the relocated events align on northeast-southwest oriented faults.
Unlike the recent swarm which started with a (relative) bang, the 2015 swarm started with small earthquakes that progressively grew in size to the mid-M3 range. There were then several bursts of seismicity over a period of several weeks before the activity died down.
This more detailed view of the swarm, compared to the standard catalog data we plotted above, is necessary for any mechanistic interpretation. So, what did they see?
First, Xue et al. saw evidence for fluid migration during some parts of the swarm. This inference can be made when earthquakes move progressively from one location toward another, at a velocity that makes sense for the diffusion of fluid through the crust, typically along networks of faults and fractures. As the fluids travel from areas of higher to lower pressure, they can force open fault fractures, allowing the pre-existing shear stresses to drive sudden slip.
Speaking broadly, this type of process is pretty slow — the triggered earthquakes should migrate at about 0.001 to 0.05 kilometers per hour, give or take. That is about the same speed as the flow of a slow glacier — literally, moving at a glacial pace.
The southward movement of earthquakes in the current (2025) swarm is perhaps an example of this type of motion over time. Eyeballing the data, the southern limit of earthquakes has shifted by about 1.5 kilometers over about 41 days. That gives an average velocity of 0.0015 kilometers per hour, which is pretty close to the expected rates from fluid diffusion. So, we could at least speculate that fluid diffusion might be involved in the active swarm as well.
However, Xue et al. also saw earthquake progressions in 2015 that were much faster than can be explained by diffusing fluids. That requires some other mechanism. One candidate is fault creep, which is a slow sliding of a fault without radiation of seismic waves. As area of the quietly sliding fault grows or shifts over time, the surrounding crustal rocks experience stress changes that can cause other earthquakes. In continental settings, slow slip usually propagates much (1000x) faster than fluid diffusion: up to about 0.5 kilometers per hour. That’s about as fast as the shuffle of a motivated tortoise.
Finally, every earthquake, no matter why it happens, redistributes stresses around the rupture, exactly when that rupture happens. That can change the state of stress on other faults, which can then be pushed to failure, usually after a delay — this is how we typically think of aftershocks. So, some of the swarm earthquakes are presumably aftershocks of the larger swarm events.
All of this interpretation is still a bit speculative, mainly because we lack more direct measurements of how and when the crust is deforming. As more and more events like this occur and are studied carefully, scientists should be able to more confidently interpret the earthquakes.
The northern end of the Calaveras Fault attracts a lot less attention than its southern end, where it merges with the more well-known Hayward Fault (which we last wrote about in September). That southern junction has been the subject of much interest because of the potential for a multi-fault large earthquake, activating both the Hayward and Calaveras Faults during a single event.
The more neglected northern area is geologically complicated. Here’s an overview geological map from a 2002 USGS-commissioned study on the fault zones of that region (from Unruh and Kelson, 2002). As you try to follow the Calaveras Fault northward, you end up in a geological tangle, from which emerge multiple other faults and folds (the synclines and anticlines marked on the map). That study goes into great detail about the structural complexity of the area and why it is difficult to point out the active faults in the area.
Put simply, as the crustal blocks in this area have slid past each other over time, they have also sometimes pushed slightly together or pulled slightly apart. As the location and style of the strain has evolved, so have the major faults that take up most of the motion. New faults have formed, old faults have died out, and these dead faults or other contacts have even been reactivated, taking on a new geological role. The system still hasn’t worked out all of its kinks — the main faults are not yet connected by throughgoing, simple structures. Instead, the deformation is distributed across many faults.
And just for fun, here are the earthquakes from our first figure, plotted over that geological map:
The complex geological setting might explain why the earthquakes in this region are so distributed and swarmy, and only occasionally occur along the main fault traces. Whether there are other geological factors at play is presumably an avenue for interesting future research!
What does all of this mean for seismic hazard? Should we worry that a large earthquake is more imminent now than before this current swarm began?
First, many of these smaller swarms appear to be occurring on faults that are probably quite small. That means that these structures, acting alone, won’t cause a large earthquake.
However, there are plenty of other, larger faults around — notably, the Calaveras and Pleasanton Faults. The largest known earthquake on the Calaveras Fault was a M6.5 event in 1911, and one study suggests that a multi-segment rupture of the fault could produce up to a M7.1 earthquake. But even if these swarms have somehow affected those faults, the change in probability of a large rupture will itself be extremely small. Yes, there is hazard — but it already existed before the swarm began.
It is important to remember that (as we have seen over the last few decades) seismic swarms like this have occurred a number of times without triggering a larger and more damaging event. That is the most common outcome of these kinds of swarms, both here and elsewhere. Occasionally, we see swarms that trigger, or at least are punctuated by, larger events. For instance, in July there was a M6.4 earthquake offshore Vancouver Island, in an area where a swarm was occurring. The 2025-01-01 M7.5 earthquake in western Japan also originated within a long-lived swarm. These examples are notable because they are the exception rather than the rule.
So, this current swarm seems (as usual) like a good reminder to update earthquake preparedness — but more because of the awareness of the background hazard, rather than concerns about elevated risk due to the swarm. We encourage our readers in seismically active areas to head over to the Earthquake Country Alliance and review their preparation advice.
And if anyone has any further insight into this swarm, or other earthquakes/faults in this fascinating and complicated area, please let us know in the comments!
Bradley, K., Hubbard, J., 2025. M4.3 earthquake rattles Berkeley, CA. Earthquake Insights, https://doi.org/10.62481/ac4d2661
Chaussard, E., Bürgmann, R., Fattahi, H., Johnson, C.W., Nadeau, R., Taira, T. and Johanson, I., 2015. Interseismic coupling and refined earthquake potential on the Hayward‐Calaveras fault zone. Journal of Geophysical Research: Solid Earth, 120(12), pp.8570-8590. https://doi.org/10.1002/2015JB012230
Hubbard, J. and Bradley, K., 2025. Earthquake swarm beneath the Aegean Sea. Earthquake Insights, https://doi.org/10.62481/69a6636a
Hubbard, J. and Bradley, K., 2024. Seismicity patterns around the Jan 1 earthquake in Japan. Earthquake Insights, https://doi.org/10.62481/72ea1b55
Hubbard, J. and Bradley, K., 2024. Swarm offshore Vancouver Island punctuated by M6.4 earthquake. Earthquake Insights, https://doi.org/10.62481/a23a9f0a
Unruh, J.R., and Kelson, K. I., 2002. Critical evaluation of the northern termination of the Calaveras fault, eastern San Francisco Bay area, California. William Lettis & Associates. https://earthquake.usgs.gov/cfusion/external_grants/reports/00HQGR0082.pdf
U.S. Geological Survey, 2020, Quaternary Fault and Fold Database for the Nation, accessed [12/21/2025], at https://doi.org/10.5066/P9BCVRCK
Xue, L., Bürgmann, R., Shelly, D.R., Johnson, C.W. and Taira, T.A., 2018. Kinematics of the 2015 San Ramon, California earthquake swarm: Implications for fault zone structure and driving mechanisms. Earth and Planetary Science Letters, 489, pp.135-144. https://doi.org/10.1016/j.epsl.2018.02.018
