Larry J. Ruff
Some of the toughest questions from students are about earthquake depth. A typical one goes something like this: "You told us that earthquakes at midocean ridges occur no deeper than 15 km, but I just saw one on the NEIC Web page that was at 33 km--is this a major discovery?" To help teachers respond to questions such as these, this EduQuakes column is focused on earthquake depth, and the EduPhase section treats pP. Although not a comprehensive survey of hypocentral depth, at least it provides a starting point for students to learn about the nagging problems with hypocentral depths in global seismicity catalogs.
Earthquakes occur within the Earth, and the depth of rupture initiation is the hypocentral depth. Most earthquakes are "shallow", except in subduction zones, where earthquakes can be as deep as 700 km. There are a few funny spots around the world with deep earthquakes and no subduction zone, but most seismologists think that geologically recent subduction explains these events.
How shallow is "shallow" for global seismicity? There are two answers to this question. The scientific answer is that it depends on the tectonic environment. In regions of the world that can be characterized as spreading, rifting, volcanic, and transform faulting, detailed seismological studies show that most earthquakes have a hypocentral depth of 15 km or less. In subduction zones, most seismicity occurs along the plate boundary down to a depth of about 40 km. Thus from a scientific perspective, "shallow" is 40 km or less in subduction zones, and 15 km or less nearly everywhere else.
The technical answer to the above question is that "shallow" means somewhere above about 70 km or so. Given the technical difficulty in determining hypocentral depth more precisely than this, the depth is usually assigned the default shallow depth--33 km is the most common depth assigned by the USGS NEIC. If you search through a global seismicity catalog, you might think there is something magical happening at a depth of 33 km, but alas, it is just the numerical equivalent of "somewhere between 0 and 70 km."
There are a few places around the world where seismologists operate a dense network of seismographs on top of shallow seismicity. In these places, it is technically possible to determine a more precise depth from observations of P and S arrival times; in detail, seismologists hope to have a seismograph located within ten kilometers or so of the epicenter to get a precise depth. (Educational Web page activities related to this style of earthquake location can be found at the USGS site at http://quake.usgs.gov/info/eqlocation/ and the Virtual Earthquake pages at http://18.104.22.168/VirtualEarthquake/). However, most global seismicity occurs beneath the ocean in remote parts of the world; this is why most shallow earthquakes in the global seismicity catalog are assigned the default depth. Since there is no comparable default depth for deeper (i.e., depth more than 70 km) earthquakes in the catalog, does this mean that deeper depths are better determined? If so, how is that possible? Is there any hope to improve hypocentral depths for shallow global seismicity?
pP is the best answer to the above questions. The best way to determine depth is to look in the seismogram just after the P wave arrival to find its reflective cousin: the pP arrival. As it name implies, pP initially goes up from the earthquake source, reflects off the Earth's surface, and then follows closely behind the P wave to arrive at the seismograph. How soon after the P wave does pP arrive? Well, that depends directly on the hypocentral depth--hence the great value of pP! Take the simple ray geometry of a P wave that leaves the source nearly straight down to travel to a faraway station. Then the additional travel time for pP is simply twice the vertical travel time from hypocenter to the surface. We can write the extra travel time as (pP - P)=2d/v, where (pP - P) is the travel time difference, d is hypocentral depth, and v is the average P wave velocity above the source. In the EduPhase Web pages (start at http://www.seismosoc.org/education/, or go directly to http://www.geo.lsa.umich.edu/%7EMichSeis/eduPhase/eduPhase.html), the expression for the (pP - P) time is shown for the more general case of the P wave leaving at some angle to the vertical. Even for the more general case, there is a direct linear dependence between (pP - P) time and hypocentral depth. Thus, just use a standard value for the P wave velocity--and our velocity models are usually correct to within 10%--and we can make a fairly precise estimate of hypocentral depth from observations of (pP - P) time. Indeed, the best way to determine hypocentral depth for global earthquakes, shallow and deep, is to use pP. Another indication of the importance of pP is that seismologists commonly refer to it as a "depth phase." Yes, there is another "depth phase": sP goes up from the earthquake as an S wave and converts to a P wave upon reflection at the Earth's surface, and then follows closely behind the pP phase.
If pP is so proficient at yielding hypocentral depth, why are most shallow earthquakes still assigned the default depth by the USGS NEIC? The EduPhase Web pages will illustrate why with example seismograms, but here I will just say that it can be quite tricky to pick the pP arrival correctly. There are many bumps and wiggles on the seismogram after the P wave. Which one is pP? When you look at just one seismogram, it can be very difficult to identify pP properly. The USGS NEIC and ISC earthquake location services do collect and report the pP picks, but they have learned to be careful in using these picks for their standard earthquake locations. Inclusion of an incorrect depth phase pick can cause huge errors in the depth that are not apparent in the formal uncertainties of the hypocentral parameters. Dr. R. Engdahl at the USGS and colleagues have gone back through the seismicity catalog with careful consideration of depth phases to relocate many earthquakes. This work is producing detailed pictures of the deep Wadati-Benioff zone seismicity.
If you have a global collection of seismograms for an earthquake, you can exploit the systematic behavior of depth phases to ensure an accurate depth determination. Unfortunately, for hypocentral depths less than 35 km, pP arrives within ten seconds or so after the P wave, and the seismogram can appear to be just a tangled mess from the P arrival past the depth phases. For these shallowest events, seismologists need to use fancy techniques like "waveform inversion" to determine a good depth. These problems with shallow earthquakes are quite evident in the examples given on the pP Web pages. Thus, we end with the surprising result that the deepest earthquakes have the best-determined hypocentral depths, while the exact depths of most "shallow" earthquakes are still unknown!
SRL encourages guest columnists to contribute to "EduQuakes." Please contact Larry Ruff with your ideas. His e-mail address is email@example.com.
Posted: 22 July 1999