Venezuela Earthquakes
Jeff Z
Angel Rodriguez
--
You received this message because you are subscribed to the Google Groups "Public Seismic Network Mailing List" group.
To unsubscribe from this group and stop receiving emails from it, send an email to psnlist+u...@googlegroups.com.
To view this discussion visit https://groups.google.com/d/msgid/psnlist/CAM%3DZhQ1bRw-zc41dikvKXv6KbmYAM_F_hYEm5c%2Bv-MUNwJ-c3w%40mail.gmail.com.
Bob Hammond
Early scientific picture of the deadly Venezuela earthquake emergesEastward rupture from the Boconó fault onto the San Sebastián fault
Can you help? We are working to make earthquake science accessible to the public, but this effort is entirely supported by readers like you. If you can, please consider supporting our work by 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! A day and a half have passed since a magnitude 7.5 earthquake struck northern Venezuela, causing widespread damage from San Felipe in the west to Caracas in the east. The toll of the known dead has climbed to 589. The number of unknown dead remains possibly much higher — those within collapsed buildings, in more remote areas where official rescue efforts have not reached, or simply not reported to officials. The USGS PAGER still estimates that the true fatalities likely exceed one thousand, and may well be in the tens of thousands. These first few days are the best chance for rescue, and the focus must remain there.
The human stories of the earthquake will continue to emerge. Over time, we will also learn about the engineering stories, and the political stories — the explanations for why buildings collapsed. There will be much to learn from these, as they can tell us how much of the tragedy was inevitably caused by the physical earthquake, and how much was avoidable, if those structures had been built differently. Here, our goal is to tell the science story — what we can of it, so far. These early scientific results are important, because they can help guide rescue and recovery efforts, helping organizations figure out where the areas of heaviest damage are likely to be and focusing their resources. They also will play an important result in the field of earthquake science more broadly, helping researchers better understand the behavior of fault systems across the world. The ruptureOne of the questions we left open in our last post was: what was the length and location of the earthquake rupture? Many maps of earthquakes show these events as points, or stars. That is a reasonable approximation for a small earthquake. But for an event like this — magnitude 7.5 — we can expect that the earthquake will actually be caused by slip on a section of fault that is ~80-180 kilometers long. This graph from the classic paper by Wells and Coppersmith shows a collection of earthquakes, comparing their magnitudes to their lengths. Note that the X-axis is logarithmic.
When the earthquake happened in Venezuela, earthquake scientists knew right away that the epicenters plotted on the map were misleading, or at least very incomplete: the rupture, and area of strong shaking, would necessarily follow one fault or another for at least ~80 kilometers, and probably longer. What we didn’t know was: in what direction? Did the rupture center around the epicenter, or propagate toward the east, towards Caracas, or towards the west? We now have the answer to that question. Actually, we have it four times over:
Together, all of these methods and data point to the same thing: the earthquake started near San Felipe, and propagated nearly uniformly eastward, toward Caracas. The rupture was probably ~160-180 kilometers long, and much of it occurred on a fault that lies offshore, paralleling the coast. Below, we discuss these different datasets. We also address the question of whether this was one large earthquake or two distinct earthquakes, and what to expect going forward. Felt reportsMost of what we currently know about the strength of shaking is based on reports from people in the area. So far, 525 people have filled out these reports. These have trickled in over time, and they provide an invaluable window into the earthquake. 525 is simultaneously enough to learn a lot, and a very small percentage of the people impacted. The USGS estimates that about 15 million people experienced at least moderate shaking.
More reports would be extremely useful; if you are in the area, please consider adding your own. There is an option at the top of the report for Spanish, and the questions are straightforward. For instance:
Answers to these questions are then used to calculate the MMI shaking intensity at your location, and added to the map. Shaking at all levels is important — not just the violent shaking. The data points with weaker shaking help to outline the earthquake area. Here is the current map. It shows orange to red squares (intensities VIII-IX, severe to violent) from San Felipe to Caracas. There are still a lot of blind spots. A map like this will naturally highlight cities and areas with higher population densities. The strongest shaking is right along the coast.
This map is a clear indication that the earthquake ruptured eastward, towards Caracas. However, shaking intensities are an imperfect estimate of rupture. Local geological conditions can dramatically affect shaking — a process known as “site effects.” Sedimentary basins commonly increase shaking, because the seismic waves slow down and pile up in soft sedimentary rocks, amplifying in the process. In addition, the directivity of the rupture can change the shaking pattern: as the earthquake ruptures in one direction, the seismic waves can “pile up” in front of the rupture, increasing shaking intensities. Both of these effects may have played a role for Caracas. We note that FUNVISIS (the Fundación Venezolana de Investigaciones Sismológicas) also has a web portal for reporting earthquake shaking. However, it does not seem possible to access earthquakes that occurred earlier than today, and we are not sure how to find the resulting maps of reports. If any of our readers have insight into this, please let us know in the comments! AftershocksThe USGS is currently only reporting three aftershocks (M4.4-M4.5). However, that is a dramatic undercount. We know that many more aftershocks must be occurring, given the size of this sequence. Fortunately, FUNVISIS has a much denser seismic network in the country, and is able to detect many more events — more than 120, so far. Here is a map from their website showing seismicity since June 24, 2026. This catalog includes events down to the low magnitude-2s.
Aftershocks stretch from San Felipe in the west to just past Caracas in the east. There is a section in the middle that seems relatively quiet. Typically, aftershocks concentrate in areas where the stress changes caused by the mainshock are highest — so around the edges of the rupture, or where there are discontinuities in fault geometry or slip amounts. Aftershocks can occur beyond the rupture itself, so this map does not tell us the actual length of the fault that slipped, but it gives us a good first sense. We note that the sizes of the circles on the map do not represent magnitude. The epicenters of the M7.2 and M7.5 are in the west, near San Felipe. We are guessing that the circle size may be scaled by how recent the events were, but we are not sure. Seismic basics and beyondIf you ever took a geology class in school, you probably learned about P and S waves. The P-wave is compressional, and it travels fastest from the rupture. The S-wave lags behind. Because these waves travel at different speeds, it is possible to use the time between the waves to estimate how far away from the earthquake was when it started. That is seismology 101. However, there are vast realms of seismology beyond that simple description. Those realms of seismology use all the other parts of the shaking, to extract much more information. Here is a seismometer reading from Barbados for the Venezuela earthquake. Note that there are actually three time series, representing the north, east, and vertical recordings.
The red and dark blue bars are basic guesses about when the P and S waves were expected to arrive. The PP and SS arrivals (orange and light blue bars) are estimates about similar waves that took a different path — bouncing once off of the Earth’s surface before being recorded. And within all of the other shaking are many more waves — ones that traveled through the core, or bounced off the core, or that started as S waves but then converted to P waves as a reflection, or that traveled along boundaries inside the Earth for awhile. You can see a comprehensive list here. There are dozens.
And that’s only the body waves. Much of the strongest shaking travels along the Earth’s surface (surface waves). Because the Earth’s surface is irregular, these waves can be very complex. Finally, because the earthquake rupture did not happen all at once, those same waves were all generated continuously over the rupture period (tens of seconds). One single seismograph shows the sum of all of the waves, generated across the entire rupture period. That is why those first P and S waves are so useful — they arrive before all the other ones, so they are much easier to read and interpret. But the richness of all of the other waves is a remarkable stash of information that only people who have dedicated years of their lives are able to read. Although people sometimes call us seismologists, we are not, and this is why. Two earthquakes or one?The USGS is reporting two M7+ earthquakes in this event: a magnitude 7.2 earthquake that started on June 24th at 18:04:33 local time, and then a magnitude 7.5 earthquake that started 38 seconds later, at 18:05:11. FUNVISIS also reports two events, also magnitudes 7.2 and 7.5. Because these two earthquakes were so close together in time, the second, larger earthquake fell within the shaking of the first earthquake, making it difficult to detect. (No clean P and S waves!) This raises questions about what the boundaries of an earthquake really are. An earthquake is not instantaneous. It does begin at a specific time, but the actual rupture happens as a pulse that travels along the fault. Think about it like ripping a piece of paper. You start the tear in one location. Then, as you continue to pull the edges, the tear travels across the paper. In an earthquake, this rupture pulse travels at ~1-2 kilometers per second along the fault surface. Only in a very few rare cases has this rupture process been observed by humans directly (although seismometers have certainly recorded it); one woman who witnessed the rupture of the M7.3 Borah Peak earthquake in Idaho described it as: “just as though one took a paint brush and painted a line along the hill.” The 39 seconds between the two earthquakes is the time between the start of the two events. But the gap is actually shorter than that, because the first earthquake took some time to complete. How long? We don’t know at this point, but we can guess based on past events. The 2019 M7.1 Ridgecrest earthquake in California was extremely well documented; the USGS slip model for Ridgecrest show that most of the slip occurred within the first 15 seconds, but that the full event took about 40 seconds. The 2025 M7.3 Sand Point, Alaska earthquake took ~25-35 seconds. The 2024 M7.4 Hualien earthquake in Taiwan took ~35-45 seconds. So, it is clear that the “gap” between the two earthquake might be only 20 seconds, or less. Why does the USGS think there were two earthquakes rather than one? We don’t know for sure, but we can guess. It is possible that the first earthquake did actually stop before the second one started, or at least that it slowed down a lot before accelerating again. That second acceleration would look just like a new earthquake. In fact, it is fairly normal in seismology to define large earthquakes as several sub-events, representing exactly these kinds of pulses. The distinction gets a little murky. The IRIS webpage does now list one seismometer within Venezuela, from Santo Domingo, about 300 kilometers southwest of the epicenter. The phase arrivals on the plot below are estimated for the second earthquake — the M7.5. It is clear that something else was happening in the ~40 seconds before those phases arrived. What is also clear is that the P-wave of the M7.5 must have arrived at just about the same time as the S-waves of the first earthquake, making them very difficult to see. Although the time gap between the P and S waves would have increased with distance from the earthquakes, the gap would have remained similar for the two events, because they occurred very close together. No wonder the second event was difficult to tease out. Please remember that we are not seismologists, but rather earthquake geologists! We encourage our seismologist readers to correct any mistakes in our discussion.
You can also see that the P wave at the beginning is not just one single pulse. Instead, there is a clear start, but then shaking continues for tens of seconds. That is because the rupture continues to produce P waves as it travels along the fault. The same is true of S waves and surface waves. The other thing to note for this seismograph is that the strongest shaking is not well recorded at this station — it clips. There are different kinds of seismometers that can be used to target different kinds of shaking (weak, strong, low-frequency, high-frequency). A seismometer intended to capture the details of weak shaking will not do a good job of recording strong shaking, and vice-versa. Figuring out what kinds of seismometers to install where is its own branch of seismology. Ultimately, for a person who experienced this earthquake, it probably felt like one intense event, not two distinct earthquakes, although the specific pattern of shaking would depend on the location. The question of whether the two pulses were in fact two separate earthquakes is a bit of a technicality, in our opinion. Back-projectionAlthough we ourselves are not seismologists, we can benefit from the rapid work of others using advanced seismology to understand earthquake patterns. Back-projection is one such technique. This method uses giant, far-away seismic arrays to identify the locations of a series of bursts of high-frequency energy produced during the rupture. Because the seismic networks are far away, the absolute locations of the energy sources are hard to estimate, but their relative locations can be determined pretty reliably. The results from different networks covering different parts of the globe can also be compared to assess the robustness of the observation. Here is a back-projection calculation for the M7.5 earthquake. This was produced by Prof. Dun Wang’s group at the China University of Geosciences, Wuhan, and shared on LinkedIn by Dr. Baoning Wu at UC San Diego. The map below shows two back-projection calculations, from two different seismic networks — one in North America, the other in Europe.
The map on the right shows the locations where high-frequency seismic energy was produced. The symbol shapes represent the networks, and the colors represent the time since the earthquake started. The symbol sizes on the right presumably represent the amount of energy released. Overall, it is clear from this data that the rupture started in the west and moved east, approaching Caracas, and it seems that the total energy release represented here took about 60-65 seconds (although we are not sure if that include the M7.2 or not). Satellite imagingWe now have access to early radar-based satellite imaging of the rupture region. This type of data — called InSAR — is used to create maps of topographic change, and is able to detect vertical ground deformation on the order of centimeters. We wrote about how and why these data are collected, and how to interpret them, after the M6.8 earthquake in Morocco in 2023. As we describe in that post, InSAR first comes “wrapped” (essentially a map of displacement contours), and must then be “unwrapped” (to convert it into a map of absolute displacement towards or away from the satellite; since the satellite is high up in the sky, this is usually treated as up or down for casual purposes). Below, we show only “wrapped” images. Researchers at the USGS have a project in development to systematically invert InSAR data to interpret slip patterns in earthquakes (gCent; Shea and Barnhart, 2022). At present, these are unreviewed response products, not yet included on USGS event pages but sometimes used to inform other products. However, they are very useful for people like us just want the quick scoop on likely rupture scenarios. Below are the rapid InSAR maps from gCent. The top image covers the epicentral area. This is an “ascending” image: captured by the satellite flying northward, looking towards its right. The bottom image covers the part of the rupture that was offshore, with the satellite flying southward, again looking towards its right.
It can be tricky to read InSAR images like this. The general idea is that the rainbow bands (fringes) are areas where there was flexing or bending of the ground surface. The more closely spaced the fringes are, the more intense the bending. The areas where the colors are blotchy are noise, like in the western part of the top map — these color changes do not reflect any sort of actual change. In areas where the ground was actually offset, or where there were landslides, or where the changes are too intense to be continuously tracked using this wavelength, we get decorrelation: grainy colors all mixed together.
So, what can we see in these images? The upper map shows the area where the earthquakes started. The two stars represent the epicenters of the M7.2 and M7.5. They are located right where the fringes curve around and pinch out. This is compelling evidence that the earthquakes started here and then ruptured unilaterally eastward — there are no obvious offsets or decorrelation west of the mapped stars (epicenters). The lower map shows fringes that parallel the coast. These fringes are more widely spaced to the south, and tighten towards the north, then disappear into a wide zone of decorrelation. It is not possible at this time to know what happened in that decorrelation zone. One possibility is that the deformation in this region was too large to map with this satellite. In the maps above, each fringe represents 2.8 centimeters of movement towards or away from the satellite. For this method to work, the change from pixel to pixel has to be significantly less than that wavelength. There are other InSAR satellites with longer wavelengths (up to ~12 centimeters) that are intended to track that larger deformation, but we have not seen images from them yet. Alternatively, this decorrelation might be related to landsliding. The USGS estimates that there was a lot of landsliding across this region. (We’re not sure why the background map isn’t plotting; seems to be a glitch — sorry.)
What that means for our interpretation is that although we know where the earthquake started, and we know where it didn’t go (west), we cannot actually see where the rupture went in its eastward section. The gCent model, which aims to match the part of the data that we can see, suggests a fault that lies just offshore, with ~1.1 meters of slip. That seems a little low for a magnitude 7.5, but again, the data is incomplete and this is a very preliminary result. GeologyFortunately, we aren’t flying blind. We can compare these results to the underlying geology, which has been painstakingly mapped over the course of decades. When we do that, a clear image starts to emerge. The map here is from a USGS Open-File Report from 2023 by Audemard and others. To it we have added the focal mechanism at the epicentral location, and a first guess about the fault rupture based primarily on the InSAR. The earthquake appears to have started on the northern part of the Boconó fault (VE-06a on the map) and then moved onto the connected San Sebastián fault, which is largely offshore but crosses onto land right near Caracas. That gives us a total length of ~160-170 kilometers — a fairly normal value for a M7.5.
We noticed that in the FUNVISIS catalog from Venezuela, the M7.2 and M7.5 earthquake epicenters are located about 10 kilometers apart from each other, with the second earthquake to the north. The fault map actually does show several parallel faults in this area. It is possible that the two earthquakes occurred on separate, parallel faults. Alternatively, this distance could reflect uncertainties in epicentral locations (certainly there have been a lot of uncertainties). If that happened, it would support the argument that these should be treated as two separate events (although there is no strict rule that one earthquake = one fault, and multi-fault or multi-segment ruptures are regularly observed).
What to expectMany aftershocks have already occurred, and more can be expected. Although most of these aftershocks will be small, some could be larger and damaging. For people located close to a large aftershock, the shaking could be quite intense; further building damage (especially of already weakened structures) is possible. There is a small possibility that an earthquake could be triggered that is larger than the mainshock. For a large event like this, we often look at the ends of the fault rupture as areas of heightened stress, and consider whether new ruptures could be triggered at these locations. In this case, that would be the parts of the fault near Caracas and San Felipe. A new earthquake that started near San Felipe could rupture southwest along the Boconó fault; a new earthquake that started near Caracas could rupture east along the San Sebastián fault — although the chance of this is quite small. The USGS aftershock forecast currently estimates a 3% chance of a M7+ earthquake within the next year. We note that the largest instrumentally recorded earthquake in Venezuela, the 1900 ~M7.7 San Narciso earthquake, is thought to have caused by rupture of the San Sebastián fault east of Caracas. We can hypothesize that the earthquake in 1900 relieved a significant amount of stress on that part of the fault, and that the termination of the recent M7.5 may be related to the residual stress shadow of the 1900 earthquake. However, 126 years is a long time; long enough for new stresses to have accumulated. It is possible for the section of fault that ruptured in 1900 to rupture again. (Again, unlikely!)
We can also expect new scientific data and modeling in the days, months, and years to come.
Early results will be informal, but many peer-reviewed research papers will dig into this event, trying to extract as much information as possible. This work will happen in parallel with rescue and recovery efforts, which are the main focus at this point. We note that this post would not have been possible without the work of many scientists who already have contributed to the result shared here, both within the last few days and over the decades preceding, to map and understand the faults in this region. Please do let us know if you are aware of any other data about this earthquake. Earthquake Insights is a reader-supported publication. To receive new posts and support our work, consider becoming a free or paid subscriber.
References:Audemard, F.A., Machette, M., Cox, J.W., Dart, R. L., Haller, K., 2000. Map and database of Quaternary faults in Venezuela and its offshore regions, U.S. Geological Survey Open-File Report 2000-18, https://doi.org/10.3133/ofr0018 Bradley, K., Hubbard, J., 2025. Remarkable video captures fault slip in the Myanmar earthquake. Earthquake Insights, https://doi.org/10.62481/01cd039c Colón, S., Leal, A., Audemard, F.R.A.N.C.K., Vásquez, R. and Rodríguez, J.A., 2019. Análisis macrosísmico del sismo Mw~ 7, 6 de San Narciso del 29 de octubre de 1900, aplicando la escala medio-ambiental de intensidad sísmica ESI 2007. Universidad Central de Venezuela. Revista de la Facultad de Ingeniería, 34(1), p.17. Colón, S., Audemard, F.A., Beck, C., Avila, J., Padrón, C., De Batist, M., Paolini, M., Leal, A.F. and Van Welden, A., 2015. The 1900 Mw 7.6 earthquake offshore north–central Venezuela: Is La Tortuga or San Sebastián the source fault?. Marine and Petroleum Geology, 67, pp.498-511. Hubbard, J. and Bradley, K., 2023. Did the High Atlas really grow 20 cm in the Moroccan earthquake?. Earthquake Insights, https://doi.org/10.62481/9cb23fd9 Shea, H.N. and Barnhart, W.D., 2022. The Geodetic Centroid (gCent) Catalog: Global earthquake monitoring with satellite imaging geodesy. Bulletin of the Seismological Society of America, 112(6), pp.2946-2957. https://doi.org/10.1785/0120220072 Wells, D.L. and Coppersmith, K.J., 1994. New empirical relationships among magnitude, rupture length, rupture width, rupture area, and surface displacement. Bulletin of the seismological Society of America, 84(4), pp.974-1002.https://doi.org/10.1785/BSSA0840040974 |
To view this discussion visit https://groups.google.com/d/msgid/psnlist/CALjL13rk8CdbFjuUqhaa8NP5vmQDRFSzqQHf0EHKCmML0d6B-A%40mail.gmail.com.
Stephen Hammond
Bob Hammond
|
Can you help? We are working to make earthquake science accessible to the public, but this effort is entirely supported by readers like you. If you can, please consider supporting our work by 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! |
This is our third post about the June 24 earthquake in Venezuela. With the passing hours and days since the devastating earthquake in Venezuela, the official report of fatalities has crept up — from 200, to 500, to 900, to 1400. Those numbers on their own are tragedy enough. But the larger question looms: what about the missing, currently numbered at 68,900? We hope that many of this number are just lost or unable to contact their loved ones. The possibilities are daunting. Meanwhile, the “golden window” for rescue — typically 2-3 days — is closing, although news reports do continue to highlight stories of people pulled from the rubble. Meanwhile, aftershocks continue to shake the region. Footage and satellite images begin to describe the scale of devastation, with many large, concrete buildings violently flattened. Some of these buildings collapsed floor-by-floor, a process known as “pancaking”. While the strong shaking and soft soils contributed to this destruction, there are also questions about whether the buildings themselves were substandard. While some of the structures that collapsed predated modern building codes, some of them were newer and should reportedly have been more resilient. While the legacy of old, vulnerable buildings is a persistent problem globally, there are options for remediation and retrofitting that can make a real difference. However, enforcing these kinds of changes requires significant political willpower. Many people, both inside and outside the country, are watching the unfolding situation with horror, uncertainty, and grief. This is truly an unspeakable tragedy; one that will echo through the lives of those affected and through the history of Venezuela — and through earthquake science. Most of what we know about the hazard of large earthquakes comes from studying events like this one. Here, we continue to share the ongoing scientific study of this event. Our goal is to provide clarity about the physical process of the earthquake, and to explain how we know what happened. Satellite imagingAbout six hours after the earthquake struck, the Directorate-General for European Civil Protection and Humanitarian Aid Operations (DG ECHO) triggered on-demand mapping by the Copernicus satellite system. As of yesterday, 1,303 impacted buildings had been identified using this data. The map below shows a sample of what these data look like — collating the damage to specific buildings visible from the satellite. This approach is very useful as a first-pass, but it is limited to what the satellite can actually see. A building that is internally weakened but remains standing might be unsafe, but not marked here. This kind of information can be used to direct rescue efforts and supplies.
In addition to mapping normal buildings, it is critical to evaluate major infrastructure, especially infrastructure that could itself pose a danger (like petrochemical plants) and transportation hubs, which are necessary to facilitate the rescue and recovery efforts (airport, seaport, highways, etc.). The NASA NISAR Urgent Response processing system was also activated to monitor this earthquake. In our last post, we showed a “wrapped” InSAR image, but it suffered from decorrelation due to the short wavelength of the radar. The NISAR image below has a longer wavelength, so it is able to resolve larger amounts of deformation. This is an “unwrapped” image: instead of seeing rainbow bands, there are red and blue colors that represent the total displacement (towards or away from the satellite). On the map we have drawn in the mapped trace of the Boconó and San Sebastián faults, which were responsible for this earthquake. This map was produced by Dr. Yuankun Xu at Caltech and shared by Dr. Eric Fielding at NASA JPL. There is a clear transition from red to blue very close to the trace of the fault as it approaches the coast, as expected. The satellite has detected ~80 centimeters of relative offset — note that the satellite can only measure movement in the direction of its line-of-sight, so this is not a complete number. In the offshore area, the map shows tens of centimeter of movement of the coast to the west, but there are no observations on the other side of the fault, since that area is under water.
Two earthquakes or one, againQuestions continue to circulate about whether this should be considered one earthquake (~M7.6) or two earthquakes (M7.2, M7.5, as reported by the USGS, separated by 39 seconds). We reached out the USGS to find out what their thought process looked like. Dr. William Yeck, who has been closely involved in these decisions, wrote back. He told us that the event was first identified as a magnitude 7.1, based on the initial seismic signals. Indeed that is what was first published on the USGS website. But it soon became clear that something else was going on. The analysts at the USGS (yes, they work 24 hours a day!) quickly noticed that the waveforms of the Venezuela earthquake(s) were different from a typical M7 earthquake. The images below show how the Venezuela waveforms started with a long period of low amplitude shaking, interrupted by an abrupt increase. In contrast, the M7.4 Japan earthquake below has a quick increase in amplitude soon after the waves arrived. The time between the start time and the peak shaking was much longer for Venezuela than Japan.
This prompted a lot of discussion internally. The earthquake could be treated as one event, or it could be considered as two distinct ruptures. Indeed, they were able to model is as two point sources (with some difficulty); these events are currently included in the USGS catalog. “Our guess,” writes Dr. Yeck, “was that the smaller rupture was on a splay that triggered the mainshock.” This approach of modeling the earthquake as two ruptures is actually a more advanced form of seismology than what the USGS usually does — it typically takes more time and more work to determine results about earthquake pulses, or sub-events. Dr. Yeck continues: “In terms of keeping it one our two events, there are varying opinions internally. We decided to stick with two at the NEIC because we thought the source-time function of the 7.2 would be short enough that it was distinct from the 7.5 give the 40 second separation.” In other words: a magnitude 7.2 earthquake typically has a rupture that lasts less than 40 seconds. If the rupture only took 20 seconds, say, then there would have been a pause of 20 seconds between the end of the first rupture and the beginning of the second one. Of course, calculating the source-time function of an earthquake isn’t easy. The image below (shared by Dr. Yeck) shows the high frequency envelope calculated at a bunch of stations, adjusted in time to account for the travel time from the hypocenter to the station. By adding up all of these curves (“stacking” them), you can get a general sense of how much earthquake moment (~energy) was released at every second.
There is definitely a step up in moment release at ~40 seconds in the stack. That is the second pulse, or the M7.5. There isn’t a clear “pause” before that, but it would be hard to see in any case. So, does it really matter? Practically speaking, not really. For people on the ground, the shaking was probably continuous — shaking does not stop when the rupture stops. For seismologists, one-vs-two is a matter of semantics. A seismologist’s view of an earthquake goes way beyond magnitude: they look at the earthquake as a rupture over time, investigating the action every second along the way. Advanced modeling of this sequence will necessarily capture the whole thing — starting with the smaller event and continuing through the larger one. The impact of this discussion has mainly been in the media, which has highlighted both the two-event interpretation and the apparent disagreement. Two earthquakes sound worse than one. Ironically, sometimes having two earthquakes close together can actually reduce impacts compared to having them far apart in time. In the 2023 earthquakes in the Türkiye, the 9 hours of separation between the first and second earthquakes meant that many people had evacuated before the second earthquake hit. This kind of effect could not occur in Venezuela, with its continuous shaking. However, as Dr. Yeck notes, having this discussion in the media has provided an opportunity for education about how earthquakes work. “I think the good thing about this it has helped highlight the discussion in the media on how earthquakes are complex.” We agree — this has created much more room for talking about fault rupture, and the timeline of an event. Journalists have been much more engaged on this topic than usual! Slip modelSpeaking of work at the USGS, their efforts have continued. Their latest product is a slip model — one of those more advanced ways of looking at the earthquake. Like all USGS event page slip models, this is a preliminary result. We can expect to see many more slip models published in research papers in months and years to come. At the top of the USGS website it says: ‘This is a combined inversion for both the M7.2 and M7.5 events. Teleseismic body and surface waves and InSAR observations were used. Fault geometries were constrained by InSAR using the gCENT approach.” In other words, just like seismologists more generally, the modelers at the USGS think it makes sense to look at this as a cohesive event, for a research product like this one.
The first step in building a slip model is building a model of the fault. The USGS has modeled the fault as two segments, to approximately match the curvature of the system. In the model, the earthquake started on the Boconó fault in the west. It behaved like a typical moderate-size earthquake: the rupture initiated at ~20 kilometers depth, and spread outward along the fault surface. The westward rupture front didn’t get very far — maybe 20 kilometers. The eastward rupture persisted, continuing east. Once that rupture front reached about 30-40 kilometers from the hypocenter, things really took off. The slip grew from half a meter to one, then two meters. Instead of just slipping at depth, the rupture occurred along the whole width of the fault — from near the surface down to ~30 kilometers depth. This larger rupture continued for about 170 kilometers before shrinking and eventually stopping. The average rupture speed seems to be ~2 kilometers per second (a fairly standard subshear rupture). Remember: preliminary. Here we are describing the features of the model, but these will change as more work is done. But the large-scale features — a small initial fault growing to a larger, more intense one; the total fault length — those will likely persist.
In parallel with the slip model, the USGS has released a source-time function. This represents how much earthquake moment (sort of but not really the same as energy) was released over time. This function shows that the whole rupture took about 95 seconds. The moment is broken down into two sources — the first segment/earthquake (blue), and the second one (orange); added together they make the black line. Curiously, in this case, the initial rupture was over in ~30 seconds — but because the second one started up at ~10 seconds, there is no gap between the two; the two sources overlap. That pattern is shown on the slip model above, too.
Note that this source-time function doesn’t quite look like the curve in Figure 4, which was supposed to also approximate the source time function. We believe (but remember that we are not seismologists) that this can be attributed to two main reasons. First, this is a more advanced research product that represents a broader view of the whole earthquake. Second, the image in Figure 4 only took into account a specific frequency of shaking, so it was actually looking at just one part of the moment. In this source-time function there is no obvious distinction in the moment release rate before and after 39 seconds. However, the slip model certainly shows a change in behavior, with a smaller rupture that triggered a larger one. The complexity of large earthquakesThe reality is that every large earthquake starts as a small earthquake — and small earthquakes happen all the time. There are typically more than one thousand M5+ earthquakes each year, somewhere in the world. As a result, much of the research surrounding earthquakes focuses not on “why do earthquakes occur?” but rather “when and why do small earthquakes grow into large ones?” In 2024, Ross Stein and Peter Bird wrote an article on this topic. They noticed that most of the very large continental strike-slip earthquakes nucleated on branch faults, and then triggered larger ruptures on the main faults. That isn’t a universal pattern — for instance, the 2025 M7.7 Myanmar earthquake doesn’t seem to display this behavior — but it does seem to mostly fit what happened in Venezuela. We still don’t know enough about the specific geometry of the connection between the Boconó fault and the San Sebastián fault to really evaluate whether the mechanism applies, but it will certainly be an avenue of research. The key take-away is: large earthquakes are complicated. Usually that complexity is hidden within a single number reported by the media: the magnitude. The fact that the magnitude here was divided into two parts has created space for this discussion. We will end with a comparison to the slip model of the 2023 M7.8 Kahramanmaraş earthquake in Türkiye. This earthquake has been on our minds in part because of the similarity of the footage (collapsed high-rise concrete buildings) and the potential scale of devastation (~60,000 deaths from this earthquake and its doublet, nine hours later). The figure below is a perspective view of the slip model in the first large earthquake in Türkiye, published by Barbot et al. just a couple months after the event.
That earthquake was very complex. It started on a smaller splay fault — the Narli fault — and then spread both northeast and southwest onto several other faults. Zones of higher slip can be seen on the model (orange/red), separated by areas of lower slip. The USGS source-time function for the Kahramanmaraş earthquake shows low values for the first ~20 seconds, with a rapid ramp-up after that. Those first ~20 seconds represent slip on the smaller splay fault, which could theoretically have been described as a separate earthquake.
In fact, many earthquakes can be described as multiple smaller events. As a rupture front propagates, it has to travel along a complex, rough fault system. Geometric barriers (like bends or fault steps), changes in fault properties (like friction or fluid pressure), and changes in underlying stress, can cause the rupture grow or shrink, or accelerate and slow down. By comparing the evolution of a rupture to known physical properties of the fault, scientists can try to unravel the cause-and-effect: what caused the earthquake to grow — and what caused it to stop? Those questions will drive much of the study of the Venezuela earthquake. It started on a branch fault (sort of) and then grew on the San Sebastián fault. It did not rupture to the southwest. Its eastern limit seems to match the westward end of the 1900 ~M7.7 San Narciso earthquake. What else can we learn? And, critically, what can we learn from the damage? The study of the rupture is important for earthquake science, but to translate that into societal relevance requires a rigorous analysis of how the rupture caused shaking; how the shaking caused buildings to collapse; how different types of rocks and soils responded to the shaking. The catastrophe we’re looking at in Venezuela is not a surprise. Geologists knew that the Boconó fault and the San Sebastián fault were both capable of generating M7.5 earthquakes. When the earthquake happened, the USGS PAGER told us right away: based on the magnitude of the earthquake, the vulnerability of the built infrastructure, and the population distribution, we should expect thousands to tens of thousands of fatalities. The news reports have been catching up to this reality this whole time. That delay — with news agencies reporting only the official count of confirmed dead — has consistently downplayed the severity of this event. Venezuela is not alone. There are many other hazardous faults around the world where large earthquakes are expected. In many of these regions, the infrastructure is not resilient, and a large earthquake can be expected to kill thousands to tens of thousands of people. In the quiet period between earthquakes, it is easy to lose the political will to prepare. Hopefully, people and governments will look at the images coming from Venezuela, and reflect on their own level of preparation. The time to take action is now. Once the earthquake strikes, it will be too late. |
Earthquake Insights is a reader-supported publication. To receive new posts and support our work, consider becoming a free or paid subscriber. References: |
Barbot, S., Luo, H., Wang, T., Hamiel, Y., Piatibratova, O., Javed, M.T., Braitenberg, C. and Gurbuz, G., 2023. Slip distribution of the February 6, 2023 Mw 7.8 and Mw 7.6, Kahramanmaraş, Turkey earthquake sequence in the East Anatolian fault zone. Seismica, 2(3). http://dx.doi.org/10.26443/seismica.v2i3.502 Hubbard, J., and Bradley, K., 2026. Catastrophic M7.5 earthquake strikes northern Venezuela. Earthquake Insights. https://earthquakeinsights.substack.com/p/catastrophic-m75-earthquake-strikes Hubbard, J. and Bradley, K., 2026. Early scientific picture of the deadly Venezuela earthquake emerges. Earthquake Insights. https://earthquakeinsights.substack.com/p/early-scientific-picture-of-the-deadly Pousse‐Beltran, L., Vassallo, R., Audemard, F., Jouanne, F., Carcaillet, J., Pathier, E. and Volat, M., 2017. Pleistocene slip rates on the Boconó fault along the North Andean Block plate boundary, Venezuela. Tectonics, 36(7), pp.1207-1231. https://doi.org/10.1002/ 2016TC004305 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 Thank you for supporting Earthquake Insights! Because of paid subscribers like you, we are able to devote time to these kinds of analyses. |
--
You received this message because you are subscribed to the Google Groups "Public Seismic Network Mailing List" group.
To unsubscribe from this group and stop receiving emails from it, send an email to psnlist+u...@googlegroups.com.
To view this discussion visit https://groups.google.com/d/msgid/psnlist/3a537f55-f464-487b-9987-19ced7c22736n%40googlegroups.com.
Stephen Hammond
Jeff Z
--
You received this message because you are subscribed to the Google Groups "Public Seismic Network Mailing List" group.
To unsubscribe from this group and stop receiving emails from it, send an email to psnlist+u...@googlegroups.com.
To view this discussion visit https://groups.google.com/d/msgid/psnlist/711ef7b6-bec8-40a7-909d-9a35f45db3f8n%40googlegroups.com.