My Research

Brief Introduction to Seismology, Earthquakes and Volcanoes

Kinematic Inversions for the Earthquake Source

Seismology, Earthquakes and Volcanoes

So what is seismology? One can define seismology as the study of elastic waves that propagate through the solid Earth. These waves can be generated by various sources: earthquakes, volcanoes, explosions, atmospheric and oceanic phenomena (e.g., storms), etc. Basically, any shear or pressure disturbance within the Earth generates waves that travel through the interior and surface of the globe, causing the ground to shake in more or less perceptible ways. With seismometers we record the ground motion at the surface of the Earth (and sometimes also in boreholes). These records of ground motion are the main data we use to infer what the interior of the Earth looks like, to investigate the source processes, to understand how the waves are distorted near the surface and how ground shaking affects the many structures built on the surface of the Earth.

Seismic waves propagating though the Earth.
(image from SEDI's Image and Movie Library)

Different types of faults.
(image from The American Heritage Dictionary of the English Language)

What are earthquakes? If you grab a piece of rock and try to pull opposite ends in opposite directions, at some point - and if you're strong enough

- the rock will break and the two resulting pieces will slide past each other. In a larger scale, this is what happens in an earthquake: two blocks of "rigid" material slip one past the other. We refer to the slipping interfaces as faults. Most commonly earthquakes are a consequence of the motion of tectonic plates (but other forces can also cause earthquakes). Tectonic plates grossly behave as rigid blocks that drift in their own directions: they collide, they separate, and they move laterally past each other. Earthquakes are one of the ways that the uppermost layer of the Earth has to keep up with the plate motion; they occur because the two blocks on each side of the fault are trying to move in different directions. When the earthquake finally happens, a burst of energy is released. Some of this energy travels away as seismic waves, some is used to break the rocks around the fault (faults are usually surrounded by damage zones), and some energy is just dissipated (we don't know exactly how...).


And what about volcanoes? Volcanoes bring material from the interior of the Earth up to the surface. Some volcanoes occur near convergent plate boundaries. Under certain conditions, when two tectonic plates collide, one of them subducts under the other (subduction = "diving" into the interior of the Earth). When the subducted surface material reaches a certain depth, pressure and temperature become so high that the water contained in the subducted plate is released from the rock matrix. This rock dehydration reaction generates melt, which then ascends to the surface (the melt is warmer and therefore more buoyant than the surrounding material). Once it reaches the surface, the melt is expelled forming volcanoes. Other volcanoes are located far away from plate boundaries. They are called "hot spot" volcanoes, and we don't really understand the mechanism that generates them and continues supplying them with magma (we know there's a pocket of magma down there...). But there are many theories...	Schematic draw of a subduction zone. (image from whylifes.org)

Check out these links if you want to learn more about seismology:

Moment Tensor Inversions

Coming Up Soon

Historic Seismology

Coming Up Soon

Earthquake Source Dynamics

Dynamic rupture model for the 2004 M6 Parkfield earthquake (Ma et al., in preparation). Both figures show the fault plane. The top panel displays the stress drop that occurred during the earthquake, and the bottom panel displays the total slip.

If we look up in a dictionary (take the OED), "dynamics" is defined as "the branch of physics dealing with movement and force". In other words, dynamics concerns the relation between forces and movement. For example, if a body is subject to a given force, how will it move? If a body is moving in a given way, which forces are acting upon it? "Earthquake source" is a very broad and poorly defined term used to describe all the processes involved in the rock failure phenomenon that is an earthquake (e.g.: mechanical fracture, friction, rock failure and healing, rock behavior under stress, etc.). Thus, in earthquake source dynamics we investigate how earthquakes are generated, how they develop and eventually stop, under given stress (force) conditions and taking in account the properties and behavior of the rocks.

Work in this field may be divided into three main groups: theoretical, experimental and numerical modeling. My research in earthquake source dynamics has been mainly numerical modeling. Some of the issues that I have investigated are:

How does heterogeneity in the material surrounding the fault affect the dynamic rupture process?
Can we use b-values (an empirical"measurement" based on the number and size of earthquakes that occurs in a region) as a proxy for stress in the crust (and particularly, in faults)?

Advisors and Collaborators:

Dr. Ralph Archuleta -
Dr. Shuo Ma -

Papers: in preparation.

Interesting links:

Kinematic Inversions for the Earthquake Source

If we keep using the dictionary (OED), we find that "kinetic" is defined as "produced by movement". Thus, in kinematic studies we don't care about forces, we just want to know which movement leads to a given final result. In particular, in kinematic inversions for the earthquake source, our goal is to find which pattern of slip on the fault (space-time slip distribution = rupture model) gave rise to the ground motion that we observe at the surface of the Earth. We call it an "inverse" problem because we're trying to infer the cause (the slip on the fault) from the result (the ground-motion). In opposition, in a "forward" problem we would try to compute the consequence (the ground-motion) from the cause (slip on the fault).

Kinematic inversions of earthquake ruptures are highly under-determined problems because we only have a few records of ground-motion and the slip on the fault can be very heterogeneous both in space and time. My work in this area has been fairly specialized. I have focused on the robustness of kinematic inversions and the main factors affecting it:

How does the distribution of seismic stations around the fault (i.e., the locations where ground-motion records are obtained) influence the quality of the inferred rupture models?
How do site effects (near-surface distortions of the seismic waves) affect kinematic inversions?
A complete inversion for an earthquake space-time slip distribution is a non-linear problem that yields a non-unique solution. What is the variability between equally good rupture models?
How can we combine GPS displacement data (which records the static offset caused by an earthquake) with the seismic acceleration data (which records the dynamic field due to the passage of seismic waves) in the best possible way, i.e., according to their resolution abilities?

I have also used kinematic inversions to study the two most recent (1966 and 2004) earthquakes that occurred in the Parkfield section of the San Andreas Fault, California. Parkfield is known for its "characteristic" earthquakes. I.e., it had been suggested that earthquakes in Parkfield occurred with very regular recurrence times (~22 years), and they always started in the same hypocenter and grew into similar size earthquakes (M6), by rupturing the fault in the same direction (to the southeast). The 2004 Parkfield earthquake did not behave as a characteristic earthquake, in that it did not nucleate at the expected hypocenter and it ruptured the San Andreas Fault to the northwest. With kinematic inversions, we investigated if the patches on the fault that slipped during the 1966 and 2004 Parkfield earthquakes were identical. We concluded that these two earthquakes ruptured different parts of the fault, thus not being characteristic in their rupture pattern.

Map of California showing the location of Parkfield.
(image from the USGS)

Rupture models for the two most recent (1966 and 2004) M6 Parkfield earthquakes, obtained by inversion of seismic records. Both figures show the cumulative slip that occurred during the earthquakes projected into the fault plane. Slip in the two events occurred in different areas of the fault (Custodio and Archuleta, in press).

Advisors and Collaborators:

Dr. Ralph Archuleta -
Dr. Pengcheng Liu -
Morgan Page -

Papers:

Custódio and Archuleta (in press). Parkfield earthquakes: characteristic or complementary?, JGR.
Liu et al. (2006). Kinematic inversion of the 2004 Mw6.0 Parkfield earthquake including an approximation to site effects. BSSA.
Custódio et al. (2005). The 2004 Mw6.0 Parkfield, California, earthquake: Inversion of near-source ground motion using multiple data sets. GRL.

Links:

Volcanic Tremor

Google Earth image of the Fogo island, Cape Verde.

This spectrogram of ambient seismic noise (showing a time period of 5 days) recorded in the Fogo island displays harmonic tremor modulated by tides (Custodio et al., 2003).

In volcanoes we observe two very different types of seismic signals: volcano-tectonic signals and low-frequency signals. Volcano-tectonic signals are normally associated with earthquake-like events, that is, with the actual fracturing (brittle failure) of rock as the volcanic building deforms to accommodate magma moving in its inside. Low-frequency signals are normally not associated with the fracturing of rocks, but rather with the movement of fluids (e.g., magma, steam) inside the volcanic conduits and cracks. Of course, we also observe some hybrid seismic events, which present both a volcano-tectonic and low-frequency signature. Harmonic tremor is a particular kind of low-frequency signal, in which the ground moves at very specific frequencies. The most common analogy is that with a pipe organ, where the air passing through tubes of different sizes produces sound waves of specific frequencies (tones). In the same way, fluids moving in the interior of volcanic conduits and cracks of different sizes and shapes produce seismic waves (ground-shaking) with very well-defined frequencies.

My work in volcano seismology focused on harmonic tremor and tides. We observed that harmonic tremor in the Fogo island, Cape Verde, was sometimes modulated by tides. Such clear effect of the tides on seismo-volcanic activity had never been observed before, and reveals a great sensitivity of the seismicity to pressure variations (tides cause only tiny pressure variations on the Earth).

Advisors and Collaborators:

Dr. João Fonseca -
Dr. Nicolas d'Oreye -
Dr. Bruno Faria -
Dr. Sandra Heleno -

Papers:

Custódio et al. (2003). Tidal modulation of seismic noise and volcanic tremor. GRL.

Links:

VIGIL - Fogo Volcano Geophysical Monitoring