There was a clear sequence of seismic events that led up to last night's M6.9 earthquake.
 
The first event of this seismic sequence occurred on 22 April with a magnitude 4.9 quake. The following day, there were many more small and moderate sized events, including a M5.9 quake. We call this sequence a seismic swarm because it contained many similar sized events occurring in the same location and within a short time of each other. There is some indication a number of these events may have occurred on splay faults that branch from the main subduction fault (see cross-section below).

I came across a video from 8th April 2017 of a live news report being filmed whilst an earthquake strikes in the Philippines. Videos of news broadcasts whilst earthquakes strike are not rare. During the video, two periods of shaking are experienced, one minute and 30 seconds apart in time. Originally, I thought that the initial shaking was caused by the arrival of the weaker P-waves, whilst the second arrival was caused by the slower-travelling but stronger S- or surface-waves from a single event. Such a long time difference would imply that the earthquake occurred several hundred kilometres away. Due to the dissipation of seismic energy over long distances, it is unlikely that P-waves could be felt so strongly by people.

However, on further inspection, what makes this video so unique is that ​it turns out that the location of the filmed news report was pretty much right at the earthquake's epicentre (up to 5 km [3 miles] away). Even more intriguing is that not only one earthquake occurred, but actually, two separate earthquakes struck.

A quick search of the USGS catalogue shows that two earthquakes occurred at the same location, very close to the Port of Batangas. The events struck 1 minute 26 seconds apart. The first quake had a magnitude of 5.5; the second, stronger event had magnitude 5.9. According to the USGS, the first event caused a maximum shaking intensity of Level 5 (moderate shaking), whereas the second stronger quake resulted in shaking intensity of Level 7 (very strong). You can see this difference in shaking intensities in the video (for a description of the different levels of shaking, see this page). So far, there are no reports of significant casualties caused by these earthquakes.

With a re-watch of the video, you can hear screams from the first round of shaking at 0:53; screams from the second are heard at 2:22. This gives a time difference of 1 minute 29 seconds; this corresponds almost exactly to the time difference between the two earthquake origin times given by the USGS locations.

For the purpose of this post, I am setting forth that magnitude 5.5 and 5.9 can be considered to be 'of a similar size'. With their similar magnitude and because the two events occurred in the same location and within a short time of each other,  we call them an earthquake doublet (alternatively, some seismologists may consider the M5.5 event to be a foreshock of the M5.9 quake).

Doublet earthquakes are incredibly important from a seismic hazard point of view because they lengthen the overall time of ground shaking. Already weakened buildings could become damaged further, possibly leading to structural collapses. Furthermore, doublet earthquakes (particularly where the second earthquake is a similar size or slightly smaller than the first event) present a real challenge to real-time seismic warning systems. Where the second event is the same size or smaller, and when the two events are very closely spaced in space (e.g. less than 10 km apart) and in time (e.g. less than 30 seconds apart), the second event may be hidden inside the series of waves from the first event. 

Past studies have ​shown that stress changes causing doublet earthquakes may trigger large tsunamis. In fact, my own research [#ShamelessSelfPromotion] shows that doublets can occur within only 10 seconds of each other and within very short distances (under 30 km). The doublet that I studied comprised rupture on an underwater fault at shallow depth, which could have potentially caused a locally damaging tsunami. In this case, the doublet may have been completely concealed from global earthquake monitoring systems.

From a scientific viewpoint, doublet earthquakes are interesting because they demonstrate that stresses caused by the first quake were probably enough to have helped push another part of the fault, or a completely separate fault, closer to failure, triggering a second rupture. Triggering may be caused by the transfer of stress to surrounding rocks due to physical movement along the fault (called static stress or coulomb stress triggering). Alternatively, triggering may be due to energetic S-waves radiating from the first event shaking up adjacent faults. Static stress triggering occurs over longer timescales (from hours and days to years and decades). On the other hand, dynamic triggering relates to the speed of high amplitude S-waves - typically a few seconds to tens of seconds. In the case of the Philippines doublet, it is possible that both static and dynamic processes were at play. More research and higher resolution studies will be needed to find the exact processes responsible for this doublet. What we do know though is that the second fault must have been already close to failure and was simply 'pushed over the edge'.

Fortunately, the 8 April 2017 Philippines doublet occurred at a relatively deep depth of 42 km (26 miles). Had these events been much shallower, then the shaking intensities could have been much higher, possibly leading to collapsing masonry and structures. Also, because the second earthquake fortunately was the larger event, people were already alert and in the right place. 

Aside from the science of this earthquake, it is interesting to see the reaction of the people to the shaking in this video (apart from those who hid under the transit structure). First and foremost, the reporter maintains excellent composure. More importantly though, in my opinion, the people in the video respond well to the shaking. There isn't too much panic, people steadily move away from structures that could fall, and many people crouch down to ensure that they keep their feet steady and do not fall over. The reaction of people in this video highlights how people in the Philippines are accustomed to feeling strong earthquakes, which occur all too often in the country. In fact, the same area in the video was hit by a magnitude 5.1 earthquake four days earlier. Perhaps this event could be classed as the true foreshock ...

On Monday 4 April 2017 at 17:40 UTC, a magnitude 6.5 earthquake struck in central Botswana, Africa. The rupture occurred at an approximate depth of 30 km (19 miles). Very strong shaking (Intensity 7) was felt close to the epicentre and weak shaking (Intensity 3) was reported 500 km away in Johannesburg, South Africa. This earthquake is the largest to have struck mainland South Africa for over eleven years.

The central and western areas of Southern Africa are assumed to be tectonically very stable. The geology of the region is typically referred to as the 'Kaapvaal Craton'. Craton means a strong region of a continental tectonic plate that are typically stable for over a billion years. The Kaapvaal Craton contains rocks that are 3.6 to 2.5 billion years old. Large earthquakes in such stable tectonic environments occur very rarely.

Comparing the location and mechanism of the Botswana earthquake with a regional geological map reveals that the earthquake's epicentre may have occurred close to the boundary between Kaapvaal rocks and those of the Limpopo Belt - a distinct group of metamorphosed rocks with a slightly younger age of ~1.8 billion years old. This may simply just be a coincidence. Furthermore, given the lack of nearby seismic stations, this exact location may be uncertain. Nevertheless, it is worth considering what implications such a relationship may have.

Dr Eddie Dempsey is a geologist from the University of Durham; he specialises in analysing the structure of rocks that have been deformed deep within the crust. He agrees that there may be an influence on the stresses and resulting strains caused by rifting in East Africa. Stress may eventually be released along old but relatively weak tectonic suture zones to form occasional quakes like the Botswana event. Such old cratonic rocks, however, are typically very layered (high seismic anisotropy), which may complicate our analyses of this quake using seismic waves.

One research paper from the 1980s studied the nature of the geological contact between Kaapvaal and Limpopo further to the east in South Africa. Based on small changes in gravity across the two groups of rocks, the study interpreted that this geological contact demarcates an ancient (pre-Cambrian) plate boundary between two colliding plates.

Is it therefore possible that the inherent weakness of this ancient, billion-year-old plate boundary ruptured during the 2017 Botswana earthquake? Whilst the craton as a whole stands strong, discrete weak points could still be exploited by shifting tectonic stresses.  Yet our first estimates of the earthquake's location are somewhat uncertain; therefore, much more research including on-the-ground field studies, as well as analysis of satellite and geophysical data will be needed in the next few months to better answer this question.

What we do know though is that earthquakes caused by intraplate deformation will always be surprising to us; they have the potential to produce very large and damaging earthquakes. Dr Susan Hough, a seismologist at the United States Geological Survey says that stable continental crust may occasionally produce earthquakes as large as magnitude 7.5. 

Everyone around the world, even those living in the most stable tectonic areas, should try to familiarise themselves with some simple procedures to follow if they experience shaking caused by an earthquake should the unexpected ever occur.

I have made a quick map showing the distribution of ridges, rifts and micro-plates in Southern Africa and how these relate to earthquake activity in the region. Overall, the picture is very complicated but shows the main network of structures that may have contributed to the Botswana earthquake.

I have found a research paper from 2002 written by Ranganai et al. that maps small changes in gravity over south-eastern Botswana to interpret the location of ancient geological provinces and faults.

The April 2017 earthquake occurred in a highly deformed region separating the Kaapvaal Craton from the Zimbabwe Craton. The paper calls this region the Limpopo–Shashe Belt. This belt is composed of many northwest-southeast trending shear zones and reverse (thrust) faults that were formed as the Kaapvaal and Zimbabwe cratons slammed into each other billions of years ago (during the Archean eon). The authors speculate that this collision might have been similar to the same mountain building process that has resulted in the present-day Himalayan chain.

As the map shows below, the Mahalapye Shear Zone may be related to the fault that ruptured in April 2017. The orientation of the fault is parallel to the orientation of the focal mechanism. However, the rupture was extensional - not compressional like the mapped fault. Therefore, it is possible that the regional extensional rift tectonics in Southern Africa have reverse the sense of motion of these faults in a so-called stress inversion.

​With the possible complexity of this earthquake that was suggested by the early data, it is possible that more than one of these closely-spaced faults ruptured during the magnitude 6.5 earthquake. More detailed analysis will be needed to look at this possible scenario.

Further information

• Earthquake Report from Dr. Jay Patton
• USGS earthquake page
• EMSC earthquake page
• UC Berkeley Seismo Blog