Full Page Version | Step-by-Step version

Introduction to Earth Sciences I

Earthquakes, Seismic Waves, and the Earth's Interior

Objectives:

Earthquakes are one of the most profound expressions of the Earth's dynamics. They provide the most direct evidence that the rigid outer shell of the Earth is in motion. They are also the cause of huge tragedy as seen in earthquake losses in the last two years (see Case Studies below). The objective of this part of the course is to provide some basic information on earthquakes as physical phenomena, what makes them happen and why they happen where they do, and how the propagation of seismic energy generated by earthquakes propagates through the Earth and can be used to infer information about the interior properties of the planet. So I hope you will come away from this course having a basic understanding of -

1. What an earthquake is, why they occur, and why they occur where they do.

2. How the seismic waves generated by earthquakes are used to image the Earth's interior and some things we have learned from these images.

Your text is reasonably good on the basics of this subject (Chapters 14 and 15) but neither of the book's author's are seismologists and you can find better treatments of most of the material at Web sites created by researchers at earthquake research centers. The four that I think will be most useful are:

http://vearthquake.calstatela.edu/ which is from Cal State LA and goes through a number of simple exercises in finding earthquake locations etc that we will do in class or in the lab.

http://scecdc.scec.org/eqcountry.html and http://www.scecdc.scec.org/Module/TableofC.html from the Southern California Earthquake Center, a consortium of universities in Southern California (but including Columbia) who combine to do research on earthquakes and their associated hazards. I have had the most difficulty getting to this site. We will use quite a lot of materials from this site.

We will discuss two new discoveries about the deep earth that seismic analysis has revealed. One is the "super-rotation" of the inner core. A description of that can be found at
http://www.ldeo.columbia.edu/press_releases/song/pr.html which includes an animation. The scientific paper is published in the journal Nature and would be too technical for our purposes.

We will also study the nature of the so-called D'' Layer which is at the boundary between the outer core and the lower mantle. It exhibits a surprising degree of heterogeneity and structure suggesting that the core/mantle boundary is a very dynamic interface. This was published in Scientific American in May 1993 by Jeanloz and Lay. It's a very accessible article that you should be able to read without much difficulty.

4.1 Some Essentials

4.1.1 Earthquakes Occur On Faults

Three pages at SCEC provide some good definitions.

Earthquakes occur when the Earth breaks. The breakage is called a fault. They are the instantaneous failure of a place on the Earth that is under stress caused by the far field action of the major lithospheric plates that are in constant motion driven by the Earth's heat engine (see Topic 2). This is an important point that the popular press often gets wrong -- earthquakes result from the instantaneous localized release of stresses that are broadly distributed and essentially uniform. They do not result from an instantaneous increase in stresses. In this way they are analogous to the landslide on Bak's sand pile in that the avalanches occur as instantaneous isolated events driven by the steady drip of sand from the sand faucet.

Earthquakes usually occur on faults that were known to exist before the earthquake happened, and have had a prior history of earthquake activity. Many earthquakes occur on the San Andreas Fault, for instance. In a simple way this is a type of prediction. We don't really expect earthquakes to occur in places where there are no faults. Put another way, earthquakes usually occur in places where they have occurred before -- on "active" faults. This is where the sandpile analogy breaks down. To be accurate the sandpile would have to be pre-figured with the scars of previous landslides leading to a tendency for new landslides to occur at the locations of old landslides. The Earth has a memory of where it broke before.

Activity: From SCEDC Section #2 Activity #1 using Seismicity for 1932-1996 chart

Although in detail faults really come in all configurations and accommodate stress relief in many different ways, there is a basic classification of faults into three types -- normal faults, reverse faults and strike-slip faults. Your book has some reasonable diagrams that show the motions associated with each type of fault.

Normal faults relieve tensional stresses and result in two parts of the crust moving apart.

Reverse (or thrust) faults relieve compressional stresses and result in two parts of the Earth moving together. They look much like a normal fault on which the motion has been reversed. Hence the name. SCEDC Section 1, Page 14 and Page 15

Strike-slip faults relieve lateral stresses and result in two parts of the crust sliding along by one another. Strike-slip faults can be right lateral or left lateral.. The difference is easy to recognize -- imagine yourself standing on one side of the fault, the other side moves to the right, the fault is right lateral, left lateral if it moves left. The best known fault in the United States, the San Andreas Fault, is a strike-slip fault. SCEDC Section 1, Page 16

It is rare for one type of fault to exist alone. Strike-slip faults often have compressional stresses associated with the crust on either side of the fault, especially if it isn't perfectly straight (and they never are) so normal and reverse faults are commonly found as secondary features associated with the main strike-slip structure. SCEDC Section 1, Activity 13 and map showing sense of slip density

4.1.2 Faults don't slip all at once

The animations of fault motion shown above are intentional simplifications designed to show the basics of how the different types of fault motions take place. In the real Earth faults do not move all in one simple motion. Typically one segment of fault moves leaving the remainder without any motion. The San Andreas is a huge fault several thousand kilometers in length. When an earthquake occurs on that fault only a very small portion of that fault actually moves.

Generally slip on a fault nucleates at one place and ruptures a segment of the fault surface. The fault block model shows this idea. The earthquake is said to occur at the nucleation point and the area affected by fault motion is called the rupture zone. Large earthquakes have very large rupture zones, perhaps several square kilometers, small ones may effect only a few square meters of the fault. The hypercenter of an earthquake is the place where the earthquake nucleates and the epicenter is the location on the earthís surface directly above the hypercenter.

SCEDC Section 1, Page 6 (Epicenter/Hypercenter diagram)

Slip on a fault is typically progressive. The fault ruptures in one place and the area slipping moves along the fault from one place to another. The two examples below illustrate this. They show the evolution of slip on two recent earthquakes in southern California, the Northridge and Landers events. Particularly in the Landers case we see the slip progressively move from south to north along the fault triggering activity on three separate fault segments.

SCEDC Section 1, Activity 3 (Rupture Model Animations including Landers rupture and energy models and Northridge rupture model)

We are starting to learn some surprising things about the evolution of fault slip. It now seems that motion on one fault can trigger activity on other faults that are quite some distance away and that new seismic activity may occur quite some time in the future.

What happens is, in principal quite easy to understand. A place like southern California is a complex deformation zone with many different faults of different type and in different states. Some have recently had earthquakes and some have not had earthquakes for many years. Because an earthquake represents a release of stress in the earth this complex array of faults are in different states of stress, some ready to rupture. When an earthquake releases stress on one fault or segment the earth motion may transfer the stress to another fault region. The release of stress in one place may increase stress in another place. This newly stressed region may then become prone to new earthquake activity.

4.1.3 Damage Can Occur At Considerable Distance From An Earthquake.

The great San Francisco earthquake of 1907 did not occur on a fault that ran directly through San Francisco. The largest ground motions associated with earthquakes occur on the faults themselves but a great deal of damage is often reported a considerable from the trace of the fault. Sometimes the fault itself can be quite difficult to locate and not an especially dramatic feature.

This distant damage results because the energy that is released at the fault by the earthquake cannot stay localized to the region of the fault itself. The energy propagates away in seismic waves which travel through the Earth and disturb the crust as they pass. These waves are something like the waves one sees in a body of water if a stone is dropped in. The ripples move away from the site of the dropped stone in circles and as they do they reduce in size. At some great distance one would never know that the stone had been dropped into the water. The same is true of earthquakes in that as distance increases from the earthquake the disturbance reduces in size. A large earthquake will cause disturbances at distance from the quake that are capable of significant damage, just as a large object thrown into water will cause large waves that travel considerable distances.

The motion associated with an earthquake is very complex and the waves that travel away from the earthquake are similarly much more complex than a simple ripple in a pond. Several different types of waves are generated simultaneously and are associated with different types of ground motion. For instance, to effect a city located north of an earthquake the seismic waves obviously must travel north to reach the city. But when they arrive the motion of the ground will be in both a north-south direction (the direction the waves traveled) and east-west, exactly orthogonal to the direction of travel. And the two types of disturbance don't arrive at the same time. Those making motion in the direction of travel of the wave arrive ahead of those making motion in the orthogonal direction. The first type are called Primary or P-waves, the second are called Secondary or Shear waves designated S-waves. Both are elastic waves and both can travel from one side of the Earth to another from sufficiently large earthquakes. Even small earthquakes send waves to great distance and can be detected by sensitive instruments called seismometers that can pick up very tiny ground motions much too small to be detected by people. These two types of seismic waves will prove to be very useful when we come to use seismic energy to learn about the interior of the Earth.

SCEDC Section 3, Page 9 (Seismic Waves Identified)

SCEDC Section 3, Activity 2 (Measuring Earthquakes -- Seismic Waves)

SCEDC Section 3, Page 10 (P.S. I Rayleigh Love U)

SCEDC Section 3, Page 11 (P is for Primary Waves)

SCEDC Section 3, Page 12 (Secondary Shear Waves)

4.1.4 Earthquakes Have A Distinct Global Distribution

Even a cursory look at a map of the world showing the locations of earthquakes makes it clear that they are anything but uniformly distributed. They occur in distinct bands. In fact, most of the Earth is relatively free of earthquakes. A map showing the locations of the deepest earthquakes we see that their locations are even more restricted. Deep earthquakes are essentially absent in the US, for instance, but common along the west coast of South America. If instead of looking at a plan view, we were to make a section through the Earth at one of the locations where deep earthquakes occur we see that they form a very distinct band penetrating deep into the Earth.

Go to Pre-Assembled Event Information page.

These zones of intense earthquake activity mark the major lithospheric plate boundaries -- those places where the plates are being created, destroyed and are sliding past one another. The latter two types of boundary create the most intense activity. The most destructive earthquakes to have taken place on Earth occurred at subducting plate boundaries where the lithosphere is being thrust deep down into the mantle. Where one plate is diving beneath the other tremendous forces are at play. One plate literally rubs against the other, sometimes sticking and building stresses released in earthquakes, sometimes sliding. In a subduction zone dipping at a relatively shallow angle a large area of the two plates is in contact and when such a large area slips instantaneously there is the potential for a very large earthquake. The larger the rupture area the larger the earthquake will be. So if a very large section of a strike-slip fault gives way, a very large earthquake is also possible.

The type of faulting associated with the three different plate boundaries is consistent with the overall motion of the plates -- normal faulting dominates at divergent boundaries where plates are being created, reverse faulting at subduction systems and strike-slip faulting at transform boundaries. It is possible to analyze the seismic waves from an earthquake recorded at a number of locations around the world to determine the location, depth, magnitude and focal mechanism (the type of motion that gave rise to the earthquake). http://vearthquake.calstatela.edu/ is a good source for how to do earthquake location and magnitude determination. That is, we do not need to be on top of an earthquake to know where it occurred and why. Today, there is a world wide network of identical seismometers operated to monitor earthquakes and quickly determine their size and location. An earthquake larger than about magnitude 4 anywhere in the world will be detected by this network and quickly located. Smaller earthquakes are recorded on local parts of the network. The accuracy of earthquake location depends on the number of seismometers that recorded the event and how well they are distributed with respect to the quake. Generally, the more seismometers that record the event the better it will be located but the best locations are obtained if the seismometers actually surround the earthquake.

4.1.5 Locating Earthquakes and determining their magnitude.

The Web site http://vcourseware5.calstatela.edu/VirtualEarthquake/VQuakeExecute.html provides easy to follow exercises on how to locate earthquakes and determine their magnitude. You can follow the instructions to do your own location and magnitude determination. It uses real seismograms from actual seismograph recording stations. The basic principal of location is fairly simple. We know that earthquakes generate both p-waves and s-waves and that the p-wave energy travels faster than the s-waves. Like two cars starting from the same place with one going faster than the other, the faster car will reach its destination sooner, and the longer the cars are driving the further ahead the fast car will be. With the cars both heading for the same very distant destination, the amount of time the slower car is behind the faster will increase as the destination distance increases. For seismic energy the s-to-p separation time is a measure of how far away the earthquake must be - the greater the separation the greater the distance to the source. Of course, a single recording cannot tell us the direction the energy came from and, as the Web exercise demonstrates, a minimum of three recordings is needed to determine the location. In fact, accurate locations require about 10 independent recordings and there is always some uncertainty associated with the location because, as you will see in the exercise, the s-to-p separation time cannot generally be determined with complete precision.

Determining the magnitude is a little more difficult. Any recording of an earthquake at some distance from the source will produce a wiggle on the seismograph. But for a given wiggle, how can we know that it resulted from a small earthquake nearby, or a large earthquake at a large distance? The estimate of magnitude has to include some knowledge of both the distance and the amplitude of the wiggle. That's the key. The Web site shows how it's done.

Given that earthquakes generate a lot of different types of energy (p-waves, s-waves, surface waves) its not surprising that seismologists have come up with several different magnitude measures. The well known Richter scale measures the s-wave amplitude.

4.1.6 It Is Not Hard To Tell An Earthquake From A Nuclear Explosion

The worldwide network of seismometers was set up in large part in response to the need to establish a mechanism to verify compliance with the Comprehensive Nuclear Test Ban Treaty. But how can we distinguish a nuclear explosion from an earthquake? In fact it is not terribly difficult.

Figure 4.1.1

Think about the mechanism of any earthquake. Regardless of whether it is generated by normal, reverse or strike-slip faulting what is always taking place is the sliding of one part of the crust against another. This produces very distinct seismic energy radiation patterns. Imagine yourself standing in front of a normal fault. The ground motion you felt (as would a seismometer at that location) would be down and toward you as the crustal block slides down the inclined fault. On the other side of the fault the opposite motion would occur and to the side some shearing motion would take place. So the motion you feel depends on where you are with respect to the fault and the seismic energy radiated from the fault is similarly different in different directions away from the fault. It is this information that allows us to determine what sort of fault caused an earthquake.

Now think about an explosion. Unless the explosive agent is distributed in a line or other pattern we would expect a fairly uniform distribution of energy in all directions from the explosion. There is no reason why any direction away from the explosion would receive more energy than any other. No fault-generated earthquake can produce that sort of radiation pattern. Also. because there is no real sliding motion associated with an explosion very little shearing occurs and hence very little S-wave energy is present. The ratio of S-wave to P-wave energy from an explosion is therefore unusually low and quite unlike a fault-generated earthquake.

Figure 4.1.2

Figure 4.1.2 illustrates the differences in how seismic signals are generated for both earthquakes and explosions. The P-waves associated with an earthquake are a four-lobed pattern of compressions and rarefactions due to the shearing motion across a fault, while those associated with an explosion radiate out uniformly in all directions due to the relatively uniform pressure applied to the walls of the cavity created by the explosion.

Another way to determine whether seismic signals are due to an earthquake or a nuclear explosion is to look at the ratio of surface wave magnitudes, M_s, to body-wave magnitudes, m_b. Body waves are seismic waves that travel through Earth's sphere, while surface waves travel around Earth's surface. Underground nuclear explosions produce seismic signals that have different surface to body wave ratios than those produced by earthquakes. Figure 4.1.3 below is a M_s: m_b diagram. The diagram separates earthquakes and explosions into two distinct populations which makes identification reliable.

Figure 4.1.3

 

4.2 An Earthquake Or A Clandestine Nuclear Test

On August, 16 1997, there was a small seismic event in the Kara Sea, near a nuclear test site on the far northern island of Novaya Zemlya, used by Russia for nuclear weapons research. Within days, scientists determined that the event was in fact an earthquake, having occurred 130 km from the test site and beneath the bottom of the Arctic Ocean. However, US agencies took longer in reaching that conclusion, claiming that the event was an underground nuclear explosion. It took the CIA 10 weeks to finally concede that the event was not a nuclear explosion, as it had previously insisted. More about this: Nature Article

 

4.3 Earthquakes Disasters and Prediction: Three Case Studies

Living in New York there seems to be little need to worry about earthquakes. We hear about earthquakes but they always occur somewhere else, either in a distant part of the world, or on the opposite of the US. Few people living in Manhattan would rate earthquake risk very highly among the hazards to life of living in the city. One is much more likely to be run down by a taxi than to be injured in an earthquake. So why worry?

For the most part there is probably little need for Manhattanites to be too concerned about earthquake hazard. Earthquake have occurred in the northeast and some have been very destructive. Destructive earthquakes are sure to occur again in the area and when they do the potential for property damage is very real and the possibility of loss of life is also real. It is extremely unlikely, however, that loss of life in a Manhattan earthquake will rival that which occurred in two very recent earthquakes that we will use in this class as examples of how different types of earthquakes in different setting can cause loss of property and life. The first is the earthquake and tsunami that took over 2000 lives in a small village on the north coast of New Guinea in July 1998. The second occurred in August 1999 in a densely populated part of Turkey and took more than 12,000 lives, and the count is still increasing. The causes of these earthquakes are very different.

 

4.3.1 Sissano Lagoon, July 17, 1998

The USGS Web site provides an excellent description of this earthquake event. The lives that were lost in this earthquake were not caused by the earthquake itself, but by a submarine landslide that the earthquake caused. This is a very dramatic example of a very common phenomenon - earthquakes don't usually kill people. It is extremely rare that people are swallowed up in gaping chasms caused by earthquakes (Superman movies notwithstanding) or crushed by the movement of the Earth itself. Usually people die in the ruins of buildings that collapse on them or by being struck by falling objects. There is an old adage that says .... earthquakes don't kill people; buildings do. In Sissano Lagoon the earthquake of July 17 didn't harm anyone. It was felt by many people, but no damage was done and no-one was injured. What killed more than 2000 people was a tsunami (Tidal wave) that resulted from the earthquake.

Off Sissano Lagoon the sea floor slopes very steeply and rapidly attains depths of 4000 meters (http://walrus.wr.usgs.gov/tsunami/itst.html). Sediments that collect on the steep slope are unstable and can slide down the slope in a phenomenon that is the submarine equivalent of a landslide or avalanche (a term restricted to the description of snow slides). Witnesses have described how air is pushed in from landslides creating strong winds. When a submarine slide, called a slump, moves down-slope a large amount of water is displaced. Because the surface of the ocean is free to deform the displaced water tends to cause the sea surface to bulge upward and a wave of risen water moves out in all directions from the area of the slump. Tsunami waves created in this manner can travel at frightening speeds of as much as 400 miles per hour or more. The figure below shows the progress across the Pacific Ocean of a hypothetical tsunami wave origination off Hawaii along with the wave heights.

Figure 4.3.1

Traveling away from the coast tsunamis cause no discernible effect but traveling toward the coast they are natural hazards of immense destructive force. These waves break as they enter shallow water just as normal ocean waves do at a beach. When they move to shallower regions and begin to break they become much larger, in the sense that the water piles up into a mountainous steep wave. The only good news is that on entering shallow water the waves actually slow down some, but generally not enough for people to escape unless they knew it was coming a long time before it made landfall. An excellent Web site describing many aspects of tsunami waves is at the University of Washington site www.geophys.washington.edu/tsumani. The site includes a description of the tsunami warning system that is used to detect these great waves and alert populations that might be in harms way.

The villagers of Sissano Lagoon, gathered in the evening for their meal, would have seen a wall of breaking water as much as 50 feet high coming directly at them so fast that escape was impossible. Four villages were completely wiped out and others badly damaged. While the loss of life of 2000 seems gruesome enough, the coastline here is very sparsely populated. A similar event off Miami, Seattle or New York would kill tens of thousands. Is this impossible? The answer is "no".

The offshore slope of the sea floor adjacent to many coastlines is often quite steep. Sediments are thick and unstable. Almost all continental slopes exhibit some real potential for unstable behavior.

4.3.2 Turkey, August 17, 1999

At perhaps the worst possible time of day, 03:02 a.m., on August 17^th, a magnitude 7.4 earthquake struck near the town of Izmit in Turkey. At that time of day almost everyone is in bed asleep and unable to take any actions to evade the catastrophe. Loss of life in earthquakes is strongly dependent on the time of day at which the earthquake strikes. People die in houses and other buildings. The more people who are out of their houses during an earthquake, or at least awake enough to get out of their houses or find a safer place within, the lower the death toll. The fact that structures, not the Earth, kill people was bought home graphically and sadly in Izmit.

The New York Times ran headlines on August 18^th through the 21^st on the quake together with an article in the Week in Review Section on Sunday the 22^nd, and an article in Science Times on Tuesday, the 24^th. Watching the changing emphasis of these pieces it becomes increasingly apparent that the social cost of this event was due to many factors under the control of humans. In a 7.4 magnitude earthquake people will die and property loss will be high. The state of our science and the understanding of earthquakes does not presently allow any useful prediction skill - we don't know when an earthquake will occur or how large it might be, but we do have a refined knowledge of where earthquakes will occur. The Izmit earthquake took place on the Anatolian Fault, a structure of known historic earthquake activity. Seismologists knew that large earthquakes would occur in this region - not when or how large exactly - but that they certainly would occur. Knowing that, we can take some preventive actions. Primarily we can build earthquake resistant houses and buildings. The second day of reporting in the Times, the headline featured the key issue, "Recrimination and Rescue in Turkey; Toll hits 4,000. Shoddy Housing Vulnerable". What is always apparent in an earthquake is that the damage seems capricious. Some structures stand almost unscathed, others completely collapse. Why ?

In general there are two reasons:

1) Poor construction - although people live in areas that have experienced repeated earthquakes over centuries, and their description appears in the historic records of those people, construction practices for structures that house people put those individuals in danger. In general, the knowledge of engineering practices that allow buildings to resist earthquake shaking is quite new. So, many older structures in earthquake prone areas are highly vulnerable to damage by shaking and loss of life occurs from building collapse. Today it is possible to make most structures earthquake resistant, even large sky scrapers such as the Petronas Towers in Malaysia. Part of the secret is to allow the building to experience a controlled yielding to the shaking motion. Buildings have internal members that absorb the motion. Think about how a car would be shaken if there were no shock absorbers at the wheels. In a car the shock absorbers take up a large amount of the motion that would be transmitted to the car due to bumps in the road. Building design now includes shock absorbing elements in their frame structure which act the same way. Smaller, domestic structures can be improved in many ways also. One simple measure is to ensure that the main frame of the house is firmly attached to the foundation. There have been many instances where houses have literally shaken off their foundations. In the Kobe earthquake one of the series of issues was that many traditional Japanese style houses have fairly sparse, light frame construction with heavy tiled roofs. This works if the structure is static, but if shaken the roof often collapses the light frame beneath, completely squashing the underlying structure.

In the final analysis, however, there are few structures that can withstand a direct hit from a huge earthquake. The job of the earthquake engineer is to design structures that minimize damage - not eliminate it. In the Izmit earthquake it appears that human greed was the major cause of loss of life. Although earthquake building codes existed many new buildings were built in ways that did not conform to code using poor construction materials. Corrupt local officials have been blamed, but the outcome of inquiries is not yet known.

2) Poor underlying ground conditions - No matter how strongly a building is constructed it will suffer major damage if it is not built on solid ground. In many graphic news photos of earthquake damage we see many in which a building appears to have toppled over almost intact. In most instances this occurs because the ground beneath the structure has failed. Most often the ground is not solid rock but a weak soil and the foundations are set into this weak material. The shaking associated with an earthquake causes the weak material to become even weaker - in fact, it can lead to a liquification of the soil in places. Where the foundation sits on the liquefied soil it will sink causing the building to topple over.

As an aside, the leaning tower of Pisa is leaning because it is built on very weak material and it rains a lot in the Pisa area. The water table is quite high so the weak soil is often saturated. It is in a near-liquid state. The soggy soil is somewhat weaker on the Southern side so the tower, which has sunk as a whole several meters into the ground, has simply sunk more on one side than the other, causing the lean.

The condition of the material on which a building resides is an essential indicator of how the building will perform in an earthquake. The USGS publishes soil maps such as the ones for the San Francisco Bay area that can be seen at http://ncweb-east.er.usgs.gov/prepare/hazards.html. Particularly poor materials include landfill - the land produced when soil is dumped in to bays and riverbanks to create "new" land. The material is poorly compacted and water logged and very weak. Buildings often topple essentially intact. The Notre Dame site http://www.nd.edu/~quake/ shows some excellent examples of buildings and other large structures that have fallen over as the ground beneath them fails to support their weight when shaken by the earthquake. Go to the part of the site that shows the Java Applets of shake table experiments. One of the movies shows a model building being shaken and toppling over followed by the surfacing of a buried object like a fuel storage tank that erupts out of the ground. These surfacing events are well known from earthquake damage. Under the Education part of the site there is a section called the "Virtual Laboratory for Earthquake Engineering". It shows several different ways in which buildings can be isolated from the motion of the ground using huge shock absorbing. You can change various parameters and see how well they work. In all cases the motion of the building is initially isolated from the ground motion then catches up with continued shaking. The objective is to isolate the building long enough that the shaking from the earthquake has stopped as the seismic wave passes. Note that the isolation is from the back and forth motion associated with p-waves.

The Izmit event provides an example of how, although earthquakes cannot be predicted with precision some aspects of the way they behave can give clues to where they might occur next. ( http://www.sciencenews.org/sn_arc99/8_28_99/fob2.htm). The Izmit event occurred on the North Anatolian Fault; a feature that has experienced significant historic activity. In fact, there is a distinct pattern to the seismicity. Major earthquakes have been marching east to west over the last century. Each earthquake ruptures a certain length of the fault and in doing so releases stress in that region. What appears to be happening is that each stress release event also increases the stress on another part of the fault. On the North Anatolian fault this progressive release/loading phenomenon is moving west implying that the area that has been loaded is now to the west of Izmit. Istanbul, an ancient city with many old structures, lies to the west of Izmit. It is generally thought that the Izmit earthquake has made Istanbul more vulnerable to earthquakes in the future. Just when, where and how strong the quake will be is not known.

Figure 4.3.2

Figure 4.3.3

Earthquakes marched west along the North Anatolian fault this century (top).

These shocks created stress, in red, in the region of the August 17th, 1999 quake (bottom).

4.3.3 The Parkfield Prediction Experiment

The Parkfield experiment is a long-term earthquake research project on the San Andreas fault in California. The ultimate goal of the project is to better understand the physics of earthquakes and provide a scientific basis for earthquake prediction. An elaborate and dense network of instruments has been set up in Parkfield in order to allow scientists to study what actually happens on the fault and in the surrounding region before, during, and after an earthquake.

Figure 4.3.4
The San Andreas fault in central California. A "creeping" section (green) separates locked stretches north of San Juan Bautista and South of Cholame. The Parkfield section (red) is a transition zone between the creeping and southern locked section. Spotted areas mark the surface rupture in the 1857 Fort Tejon earthquake.

Parkfield was chosen as an ideal location because of its unique earthquake history. Moderate-size earthquakes of about magnitude 6 have occurred on the Parkfield section of the San Andreas Fault at fairly regular intervals - in 1857, 1881, 1901, 1922, 1934, and 1966 (a repetition rate of around 22 years). In addition, the seismographs recorded from the 1922, 1934, and 1966 earthquakes are strikingly similar, suggesting that these earthquakes all ruptured the same area of the fault. Adding to this pattern of repetition, similar-size foreshocks occurred 17 minutes before both the 1934 and 1966 Parkfield earthquakes. These observations suggest that there may be some predictability in the occurrence of earthquakes, at least in Parkfield. In the figure below, the similarity of the seismographs recorded in the 1922, 1934, and 1966 earthquakes are shown. This similarity is only possible if the ruptured area of the fault is the same for all three events.


Figure 4.3.5

The investigations at Parkfield have advanced scientists' understanding of the earthquake process. For example, the link below allows you to view earthquakes at Parkfield in 3-D. Rotate and zoom with your mouse. These earthquake locations were obtained with the "double difference" method, which provides extremely high resolution and reveals structure not seen in earthquake catalogs produced with conventional methodology. http://quake.wr.usgs.gov/3D/parkfield.html?file=parkfield.3d&Submit=Start+Display.

Many hypotheses have been proposed to explain the regular behavior of the Parkfield earthquakes as well as why there has not been one of significant size since 1966. In addition, scientists have been reworking prediction models to take into account the information they have learned from the Parkfield research. The Parkfield Prediction Experiment predicted that a moderate-size earthquake would occur at Parkfield between 1985 and 1993. This prediction was unusual both in its precision (as to location, time and magnitude) and high degree of confidence (95% within the 9-year window). However, there has not been a moderate size earthquake in Parkfield since 1966.

While the Parkfield experiment can no longer serve as a confirmation of time-predictability, it is very much alive as a test of slip-predictability, in which the longer the wait, the larger the next earthquake. Some argue that an earthquake of magnitude 6.6-6.9 would balance the moment deficit that has accumulated since 1966, and the magnitude increases with each passing year. When the next Parkfield earthquake occurs, if its magnitude approaches this expectation, the Parkfield experiment would be transformed from a lesson in patience to a successful tracking the nucleation and propagation of a much larger earthquake with Parkfield's dense array of instruments. This would be a scientific prize. Perhaps in the end the delay will appear beneficial and Parkfield will turn out to be an inspired long-term investment for science.

4.4 Earthquakes And The Interior Of The Earth

The analysis of the seismic waves sent out by earthquakes provides us with some of the greatest insights into the interior structure of the Earth. They provide perhaps the most direct way to view the Earthís interior since they can be used to create an image of the interior.

This is because the interior of the Earth is composed of layers and structures that propagate seismic energy at different speeds. We learned above that an earthquake gives rise to many types of seismic waves, the P-wave and S-wave being the first two to arrive at any site from an earthquake. The interior layers of the Earth propagate these waves at different speeds, generally increasing downward in the earth. At the top of the mantle the p-wave velocity is about 8 km/sec and this increases downward to about ____ km/sec. The S-wave speeds vary from ___ to ___ km/sec.

Most important for deriving an understanding of interior properties, S-waves do not propagate in liquids. Remember that the S-wave is a shear wave that results from a motion in the earth that literally tries to shear the rock. Liquids cannot be sheared, of course, so S-waves cannot pass through them. So S-wave velocities increase in the mantle until the core is reached then they go abruptly to zero in the liquid outer core.

Figure 4.4.1

Summary of the average seismic-wave velocity and density profiles through the Earth according to the PREM model. The velocities of compressional (VP) and shear (VS) waves are given on the left, density on the right, and pressure as a function of depth on the top scale (see Table 2). From : Encyclopedia Britannica view this link for more context.

 

The seismic waves bend (refract) as they pass through different layers and bounce off the interfaces between layers of very different material just as sound echoes off the walls of a canyon. They refract around the liquid outer core giving rise to shadow zones on the opposite side of the earth from the location of the earthquake.

Figure 4.4.2: From Geophysical Properties of the Earth

By recording and analyzing the many seismic waves that are recorded at many locations throughout the Earth we can build up a picture of the structures beneath in much the same way as is done in medical imaging. The technique is called seismic tomography and is borrowed directly from the CAT scan techniques of medicine.
A detailed description of Seismic Tomography: http://www.geof.ruu.nl/~bijwaard/abstracts/vakidioot/vak_uk.html

Tomography using S-waves or P-waves or a combination of the two has in recent years been used to create remarkable images of the interior structure. What is being images are changes in seismic velocity which are associated with temperature variations in the Earth. Particularly for S-waves propagation speed is very sensitive to temperature with faster regions being associated with colder material, warmer with slower velocities. Thus, we are able to image the decent of a slab at a subduction zone because that material is relatively cold compared to its surroundings. Similarly regions of hot upwelling mantle appear as distinct regions of slower velocity material. One thing we have learned most recently is that some slabs seem to penetrate much deeper into the mantle than most scientists had thought possible. In fact, they appear, in places, to penetrate to the core itself!

Figure 4.4.3

Figure 4.4.4

Figure 4.4.5

Figure 4.4.6

As well as these general pictures of mantle structure emerging from tomographic analysis, special properties of seismic waves can be used to reveal features of the deep structure in surprising detail. Here are some examples.

 

4.4.1 Inner Core Rotation

Lamont news site: http://www.ldeo.columbia.edu/press_releases/song/pr.html

This site describes an astonishing discovery published in 1996 about the Earth's inner core -- the Earth's inner core actually rotates inside the Earth. How can we possibly know such a thing? The first and essential piece of information is that it had been known for some time that the solid inner core of the Earth is anisotropic in the way it propagates seismic energy. What this means is that seismic body waves that pass through the inner core travel faster if they pass in one direction than in another. How do we know that in the first place? Imagine an earthquake that occurred in Chile (say) recorded by seismometers on exactly the other side of the Earth. The amount of time the seismic energy takes to make that long passage can be measured quite accurately. Imagine another earthquake that occurred further around the world in, maybe in Indonesia, and recorded also on exactly the opposite side of the Earth from Indonesia. Its time of passage through the Earth can also be measured. What is known is that the time of passage for seismic energy taking these trips from one side to the other varies quiet a bit depending on where the earthquake occurred. The surprising but clear cause of these time differences is that the energy travels faster through the inner core when traveling in certain directions. That's what is meant by anisotropy. Many Earth materials display this property and it is usually caused by internal layering - the energy will travel faster along a layer than it will across the layers. What was not known was that the inner core, made of solid iron, would display this property. Most scientists believe that it results from the preferred orientation of iron crystals in the core.

It is this property combined with other measurements that lead to the idea that the inner core is rotating. The web site gives the details but in simple terms we use measurements of seismic signals that travel all the way across the entire Earth as described above but in this case two waves from the earthquake are used; one that travels through the inner core and another that skirts the inner core and travels only through the outer core. There will be a difference in time of travel of these two signals because the one traveling through the inner core picks up speed because the inner core propagates energy faster than the outer core so this wave will arrive earlier than the other. Seismic signals of this type have been recorded for many years and a seismologist at Lamont chanced to compare seismograms recorded over the last several decades. To his surprise he saw that the time difference between the wave that traveled through the inner core and the one that traveled through the outer core appeared to systematically change. What could make the travel time of seismic energy propagating across the Earth change over the historic past? The answer is contained in the above Web site and is that the anisotropic inner core must be rotating, and thereby changing the orientation of "fast" direction for propagation.

Since the finding was first published in 1996, other seismic observations have been used to examine the same effect. The figure below shows three independent approaches; the middle one being the one discussed above. To the left we see a seismic wave reflected off the inner core. The core-mantle boundary is known to be rough (see discussion below) so the energy received from an earthquake (at the star location) to a receiver (at the triangle) will vary in strength over time as the inner core rotates. On the left we see a slightly different idea. The earthquake energy is scattered from many sources within the inner core, and will also vary in strength and character as the inner core rotates.

Figure 4.4.7

Since the original finding many approaches have been used, some have come up with negative results suggesting that there is no discernible rotation. Most have suggested that the rotation rate is slower than originally proposed. But the idea that the inner core might be rotating differentially with respect to the outer core and hence the Earth as a whole remains one of the most surprising findings of whole Earth seismology in the decade of the 1990's.

4.4.2 Structure of Mantle Discontinuities

So far in these lectures you have learned about some of the major divisions within the Earth - the crust, the lithosphere (involved in plate tectonics), the inner and outer core - but there are a number of other boundaries that are much more subtle. The major boundaries involve significant changes in the physical properties within the Earth and some are often associated with major chemical changes also. Several "discontinuities" occur in the mantle that are most likely to be associated with phase transitions. It is believed that the bulk composition of the mantle is of the material peridotite that is composed of an iron and magnesium silicate (Mg, Fe) SiO4. This is the material that lies immediately beneath the crust and is sometimes exposed in oceanic fracture zones and in other tectonically active parts of the Earth. Deep in the Earth the increasing pressure squeezes the peridotite and cause the atoms to form a more compact structure. No chemical reaction occurs, but the re-packing of the atomic structure causes the same atoms to become a new material known as spinel or Ringwoodite. It is still made of Mg, Fe, Si and O (magnesium, iron, silicon and oxygen) but it is a different material because of the atomic packing. This may be a hard idea to understand at first but remember that graphite (the material in your "lead" pencil), coal, and diamonds are all made of carbon. They are just in different atomic arrangements associated with pressure due to burial (diamonds come from very deep within the Earth). A second phase transition occurs from spinel to different material known as perovskite. The first phase transition causes a discontinuity in the mantle about 410 km beneath the surface and the second at about 660 km (they actually vary in depth depending on the tectonic setting.
The physical property changes across these boundaries that occur over quite short distances (on the scale of the mantle) are not large enough to be detectable by seismic tomography which gives a fairly low resolution, smooth picture of the structure of the mantle. To resolve the discontinuities we need to study the very weak direct reflections they cause or from refractions along their surface. Even trickier, there appears to be a very subtle boundary at about 520 km. It too is most likely caused by a phase transition like the 410 and 660 discontinuities. In fact it may be a pair of boundaries, one at 500 and the other at 560 km.

The difficulty with using waves to detect the physical world is that the quality of your image is limited by the wavelength you use. When using a particle as a probe, we need to use particles with short wavelengths to get detailed information about small things. To probe down to smaller scales, the probe's wavelength has to be made smaller.

Things with long wavelengths are analogous to the 'throwing basketballs in the cave to discover what is in growling in front of you' story because neither can provide too much detail about what they hit. Things with short wavelengths are like the 'throwing marbles in the cave to discover what is in growling in front of you' story in that they can provide you with fairly detailed information about what they hit. The shorter the probe's wavelength is, the more information you can get about the target.

The resolution of your measurement is the wavelength of light divided by the diameter of the aperture. This means that you would either need a small wavelength or a very large diameter to have good resolution.

Check out this website for a more detailed explanation. http://pdg.web.cern.ch/pdg/cpep/cave.html

In an innovative investigation shown below seismologists used shear waves that bounced underneath the boundary as shown below.

Figure 4.4.8

They used two energy paths, one that bounced at the Earth's surface and one that bounced off the "520" discontinuity. Because the latter travels a shorter path length it takes less time to reach back to the surface (about 20 seconds less). This second path makes a very weak wiggle in seismograms, but it is definitely present. In some places we see a single "520" wiggle; in other places there appears to be two wiggles suggestive of a splitting of the boundary into one at 500 and one at 560 km.
Detecting these very subtle boundaries requires the analysis of very high quality seismogram data from many different parts of the Earth. Every year it seems, new information about the structure of the mantle comes to light from these studies. One reason they are important is that although they are subtle they represent a partitioning of the mantle into different zones with different physical properties. In particular these properties are thought to have an influence on the pattern of mantle convection and the way in which subducting slabs penetrate into the deep mantle.

4.4.3 Structure of the Core-mantle Boundary

Open any High School or undergraduate science text book to the section on the Earth's deep interior and you will see the core represented as a fairly passive blob; more or less inert and not doing much. In fact we now know that a lot is going on in the core. It has been known for quite a while that vigorous convection in the liquid outer core is responsible for the Earth's magnetic field. It works something like an electric dynamo but the motions are very complicated and that statement simplifies things a lot (too much really). We also know that the solid inner core rotates with respect to the liquid outer core. In fact, it may be electromagnetic energy that is responsible for driving the rotation. The core is actually quite and active body.

One of the most dynamic parts of the Earth turns out to be the boundary between the core and lower mantle. This boundary (sometimes referred to as the Gutenburg discontinuity) places liquid iron in contact with the perovskite of the lower mantle. One thing that happens is that the core eats into and absorbs the lower mantle. As a consequence the iron outer core gets contaminated with the material of the mantle causing it to be perhaps 10% less dense than pure iron. Outer core and lower mantle also react chemically and produce new compounds as reaction products. The liquid iron is able to permeate upward into the lowermost mantle by a process called capillary action, the same process that trees use to draw water from the soil through their roots and up into the body of the tree. The liquid iron reacts with the lower mantle rocks and creates a thin reaction zone. The reaction products are new materials, even more exotic than those of the lower mantle. But the really intriguing part of the story involves the dynamic interaction with the mantle. Convection, as we learned in Topic #3, stirs the deep mantle and plumes rise from the surface of the core. These actions move the mantle around and stir up the reaction zone. These motions entrain reaction zone products and move them upward in regions of mantle upwelling, perhaps exposing mantle to the core and initiating further reactions. The reaction zone products are both liquid and solid (crystalline) and as they get entrained in the mantle flow and move upward the heavier solids will drop out and fall toward the core. The figure below from a Scientific American article by Ray Jeanloz and Thorne Lay in May 1993 gives a sketch of what might be going on.

Figure 4.4.9

The uneven layer of entrained reaction products is called D'' and is a highly irregular and dynamically variable feature of the core-mantle boundary. How can we possibly know this? One way is direct simulations of the conditions present at the pressures and temperatures of the core-mantle boundary. Jeanloz and Lay describe how a device called a diamond anvil can be used to bring tiny samples of materials typical of the mantle and core to the pressure and temperature of the core-mantle boundary. They react in the anvil and their reaction products can be studied.

Seismic evidence involves the study not of the time of arrival of seismic energy but the shape of the seismic waves. When seismic waves are emitted from their earthquake source they travel away in a fairly smoothly expanding sphere (actually a half sphere because they don't go into the air). So long as the structures in the earth through which they pass are also smooth the wavefront will stay smooth and the shape roughly spherical. But imagine the smooth wave encountering a rugged interface. The wavefront will become distorted. Seismic energy propagating such that it just glances the core-mantle boundary will pass along the highly variable D'' Layer described above and wavefronts will become distorted as they travel through. This is shown schematically in the figure below.


Figure 4.4.10

To detect these distorted waves we need many closely spaced seismic recording stations set out in a dense array. These arrays can track the passage of a seismic wave across the surface of the Earth and detect whether it is smooth or distorted. Studies of this type have established that the D'' layer is indeed highly variable in shape and thickness from place to place around the core-mantle boundary.

These studies tell us that D'' is quite variable but don't tell us much about its internal properties. For that we must look closely at individual seismic records from energy that has traveled for a while in D''. The figure below shows four different interpretations of what the region of the core-mantle boundary might be like. In the figure ULVZ means ultra low-velocity Zone.


Figure 4.4.11

The above sketches are the seismograms expected for the energy that travels in each of the possible D'' Layers. There are distinct but very small differences between the seismic signals even though the D'' structures are really quite different. The problem is one of resolution. At the depths of the core-mantle boundary seismic energy has very long wavelengths - many kilometers -- and structure can only be resolved to about a quarter of a wavelength.

Basically we are trying to examine structure in the deep Earth that is right at the limit of our ability to resolve.

Study Exercises

---

Topic 1 / Topic 2 / Topic 3 / Topic 5

---