Download as pdf or txt
Download as pdf or txt
You are on page 1of 28

The ultimate goal of seismic hazard assessment is to inform policy

makers on how to design policies that reduce the number of fatalities


during an earthquake and to increase the resilience of a state in the
aftermath of the event. We cannot prevent earthquakes from
happening, nor can we predict when an event will occur. We can
forecast the probability of an event above a certain size happening in a
certain area in a certain time period and model th elikely effects of that
event. This still has large uncertainties, and in many areas even the raw
data needed is sparse.
First: How do earthquakes happen? They occur on FAULTS, planes of
weakness in the rocks and release elastic energy stored gradually as
the rocks on either side of the fault are bent by the movement of the
plates relative to each other. The movement between two plates can
be taken up almost entirely on as single fault at the plate boundary (as
often happened in subduction zones as shown above) or it can be
spread over lots of faults over a broad area. Some slowly deforming
faults can be far from any plate boundary.
Earthquakes occur because to begin with frictional forces acting on
the fault are too great to let the two sides move relative to each other:
The fault is LOCKED. The plates are still moving relative to each other
so…
The hanging wall is dragged down by the footwall (for a reverse fault
like this) bending it and storing elastic energy. This part of the
earthquake cycle is the INTERSEISMIC (between earthquake) phase
This means the shear stress on the fault is getting larger and larger
until…
The shear stress exceeds the critical shear stress, the fault breaks, and
the earthquake occurs releasing the stored elastic stresses. This
means that energy stored gradually over hundreds or thousands of
years is released in a few seconds.
This is the COSEISMIC phase (with
the earthquake) Note that the
hangingwall has sprung up as the
stored elastic stresses are released.
This is what Simeulue (and island of Sumatra) looked like before the
2005 earthquake (the one that caused a large tsunami that inundated
many places around the Indian Ocean). Note that the trees
(deliberately planted in rows by farmers) have died because they are
below the high-tide mark. This is because the island is being pulled
down by the interseismic movement of the plates.
After the earthquake the Island has popped up out of the ocean,
moving the coastline far from the dead trees.
To assess the hazard of an area: The first thing we would look at is the
number and magnitude distribution of earthquakes in that area. To
grossly oversimplify in rapidly deforming areas like plate boundaries
some areas are more active than others and these can be identified by
looking at where there have been numerous large earthquakes in the
past. We have good data on the location of moderate-large
earthquakes globally since the 1960s. This is largely because a global
network of seismometers was set up to police the 1963 Nuclear Test
Ban Treaty. Explosions can be detected by these instruments just as
easily as earthquakes….
We measure the size of earthquakes using the MOMENT MAGNITUDE
SCALE (not the Richter Scale, which hasn’t been used for decades
despite what you will here in some news articles). This scale has a
direct physical meaning which you can think of as being equivalent to
the amount of energy released by the earthquake (or rather the amount
of work done). You calculate it using the top equation which is the area
of the patch of the fault that slipped multiplied by the average amount
it slipped by multiplied by the elastic shear modulus of the rock. Once
you have calculated this you will likely have a very large number with
many zeros which is not very intuitive (Mo). To give a more intuitive
number (usually between 1-9) you can convert this using a log scale to
give Mw. This gives you something roughly comparable to the older
local magnitude scale created by Richter. (FYI the problem with the
Richter scale was that rather than being directly related to a physical
property of the earthquake it was instead linked to the displacement
generated on a particular instrument a particular distance away from
the event. It also would give different values for the same energy
release in different places due to differences in the efficiency in how
the rocks in each area transmitted the seismic waves. That is why it’s formal
name s local magnitude – it is only strictly comparable within a single area.)
Because of the 2/3 factor each unit of Mw corresponds to the 32x
increase in energy release. The number of earthquakes in a particular
area (N) is related to their magnitude (M) by the Gutenberg-Richter
relationship ( log10 N = a-bM) . A and b are constants that vary by area,
but b is usually ~1 meaning that there are usually a tenth of the number
of M6 earthquakes than M5 and a tenth again of M7. This relationship
means that measuring the size and frequency of the small earthquakes
that occur in an area does tell you something about the likely
frequency of larger events, even if you haven’t observed any yet.
Because the time period we have good observations for the location
and magnitude of earthquakes is short (~50 years) compared to the
earthquake cycle (100s-1000s of years) it is important to find other
ways of constraining the hazard. One way is to identify the signature of
faulting in the landscape. Repeated ruptures on a fault generate
cumulative offsets of many metres or even kilometers. These can be
horizontal offsets as shown by these rivers crossing a strike-slip fault
or have a vertical component building topography (as in normal and
reverse faults). Recall the size of the earthquake is related to the area
(and so the length) of the fault that failed. By maaping faults we can
get a good idea of how large an earthquake they could generate.
This is a local example of a fault scarp found near Sesriem in Namibia.
It is up to 7m high, 35 km long and likely failed in a M7 earthquake at
some point in the last 10-100,000 years. It is a normal fault and the
hanging wall on the right has moved down relative to the footwall on
the left.
This is a Digital Elevation Model of the same fault, You can see the
scarp as a white line of increased slope. It splits into two strands in the
right image. This DEM was made using stereo-photogrammetry based
on two high resolution satellite images. In addition to estimating the
geometry of the fault and the size of the earthquake it is possible to go
to the fault and date material deposited before and after the last
rupture to determine when it happened. Sometimes it is possible to
date a whole sequence of ruptures on the same fault so you can get
an idea of the repeat time. This is called paleo-seismology. It is very
useful because it allows us to investigate what is happening on similar
timescales to the seismic cycle rather than just the last few decades.
Another way of estimating how often a fault will rupture is to look at the
rate that the two sides are moving relative to each other. Each of these
arrows represents a GPS station which has been monitored for many
years. This whole region contains many faults, and by looking at the
relative motion between the two sides of each fault we can estimate
how quickly the elastic strain is building up (and so approximately how
long it will be be between earthquakes of a certain size. This motion –
the slip rate – is usually of the order of a few cm or mm per year for a
major fault.
Historically we describe the intensity of shaking experienced in an
earthquake by ranking the reported effects on an Intensity scale. This
is the widely used Mercalli Intensity Scale (worth a read, just to
imagine what being close to the epicenter might be like), but these
scales are quite culturally specific depending on the architecture of the
area of interest. These observations can be used to create historical
catalogues covering hundreds or even thousands of years. There are
large uncertainties using these techniques. The strength of the shaking
does not just depend on the magnitude of the earthquake, but also its
depth (deeper earthquakes will result in less strong shaking at the
surface) and the attenuation of the region (how well the rock transmits
the seismic waves – some energy is lost (or attenuated) with each
oscillation and in some areas the shaking dies away distance more
quickly due to higher attenuation).
Often (especially in countries with appropriate building codes) relatively
few people die directly from the ground shaking of the earthquake.
Fires start from broken gas mains, and even natural fires triggered by
falling rocks. When the fault is under the ocean then any vertical
movement displaces the water column potentially leading to a tsunami
(as in the 2005 Sumatra earthquake). The tsunami from the 2011
Tohoku earthquake flooded the emergency generators of the
Fukushima nuclear reactor (which were in the basement) leading to the
loss of the coolant circulation, nuclear meltdown and the largest
nuclear disaster since Chernobyl. As has already been said landslides
are extremely common during earthquakes. Landslides can cause
debris dams which eventually fail leading to rapid flooding
downstream. Finally Liquifaction can be a major problem in certain
areas.
Liquefaction occurs when water-bearing sand in the subsurface is
subjected to cyclical loading by seismic waves. Shearing results in a
volume decrease (a reduction in pore space) leading to saturation and
high pore-fluid pressures. The high pore-fluid pressures reduce the
effective normal stresses between the grains to near-zero leading the
to be unable to support shear stresses and causing them to liquify. The
liquified sand often erupts to the surface along fissures leading to
significant and irregular subsidence.
This is what liquefaction after a large earthquake can look like…
The factors above mean that the amount of shaking experiences can
vary significantly over short distances. With a good understanding of
the properties of the subsurface we can produce maps showing which
parts of a country or a city may be at higher risk and tailor building
codes appropriately.
Putting together factors like attenuation, the location of past
earthquakes, mapped faults, and strain accumulation rates measured
by GPS you can produce maps like the one above. The hot-spot in the
East is there because of the New Madrid earthquakes – three very
large earthquakes which occurred between 1811 and 1812. GPS show
the strain rate in the region is very low <1mm/yr and there have not
been any large earthquakes in the region since. This shows the
importance of historical records in earthquake hazard…
Most earthquakes occur in plate boundary zones, but some occur in
the center of the plates as well. South Africa is not on a plate boundary
(the nearest is the East African Rift) but that does not mean that there
are no earthquakes here.
There have been a number of damaging (and even fatal) earthquakes
in South Africa over the last hundred years. Underground miners are
particularly vulnerable to even small events. These are the three
largest. Another event (possibly M~6.5) took place in the Milnerton
area in 1809. The suspected fault line that caused by the Milnerton
event passes close to Koeberg, which is constructed to withstand an
earthquake up to M7. The reactor is built on a neoprene raft to isolate
it from ground shaking.
The earthquakes described here are significant, and need to be
incorporated into design, but they are not so large that they would
cause widespread devastation and very high casualty rates.
An even larger earthquake occurred in the Eastern Cape around
10,000 years ago on the Kango fault.
This shows the location of recent earthquakes and faults that have
been identified as potentially active in South Africa. The cluster of
earthquakes in Gauteng is likely to be, at least in part, due to the
effects of the deep mining there. The cluster in the Cape Mountains
are aftershocks from the Ceres-Tulbagh event. There are also other
‘hotspots’ such as Koffeefontein in the free state.
This is a recent seismic hazard map. It is primarily based on the
instrumental records of earthquakes over the last 50 years. It is with
noting that the region of the large rupture on the Kango fault in the
Eastern Cape (which was recognized from the landscape) is shown as
very low risk – in slowly deforming areas with long repeat times it is
very difficult to know where the next large event might be…

You might also like