Friday, March 26, 2010

Accretionary Wedge #23: The Thing That Eats My Time

Hello, world! I realize I've failed at the day-to-day writeup of AGU that I promised back in December (eep), but what better way to come back from an unannounced work-induced hiatus than to describe the work I've been doing, hm?

I don't think I've blogged much about the research I do. I know I post about about interesting places I visit, visible and tangible fault features, and outreach, but my actual research doesn't have a field component yet at all. I do work in earthquake dynamics - the physics of how faults rupture. My method so far has been numerical modeling, specifically with both 2d and 3d finite element codes. This method divides both sides of the fault into a grid of elements of a designated size, then applies equations of motion, stress transfer, and wave propagation to each element over each timestep, then sums up the result. Each model represents a single earthquake on a fault; aside from a forced nucleation point, the magnitude and intensity of the quake are determined by the model parameters. There are many existing codes that do this, and I've been working with two of the newer ones. I've also very recently started doing some multi-cycle quasi-dynamic models. These take a fault system and put it through multiple earthquake cycles, including interseismic application of tectonic stresses. What makes them quasi-dynamic is the ruptures work by way of stress state, and don't include dynamic wave propagation.

There are a lot of fault parameters you can tweak in dynamic models, and my work focuses on the effects of fault geometry on rupture behavior and ground motion. Models investigating other parameters might use a fault that's a straight line, but straight lines almost never happen in nature. Faults have branches, bends, discontinuities, and stepovers, and they've proven to be important factors in determining where earthquake rupture stops. There's one field study of mapped surface ruptures (Wesnousky, 2008) that shows that a significant number of surface ruptures die out near a geometrical discontinuity in the mapped fault trace. On the other hand, events like the 1992 Landers earthquake or the 2002 Denali earthquake show that rupture can traverse some considerable discontinuities between faults. Understanding how rupture behaves when it encounters a geometrical discontinuity is therefore very important in determining the hazard associated with individual faults or fault systems.

I already have one fault geometry paper (which also happens to be the first chapter of my Master's thesis) in review. It's a parameter study of whether or not rupture will propagate through a bend in a fault of a given length, with a given connecting angle. Unsurprisingly, steeper angles and longer bends are more likely to halt a rupture, but exactly how steep or long varies depending on whether the step is extensional or compressional (that is, whether the bent segment is pulled apart by the fault's direction of slip, or whether it's crunched together), on the overall size of the fault system, and on the orientation of the stress field acting on the fault. That third criterion is particularly important - in some orientation, dynamic effects control the rupture far more than static ones, but in others, static effects can overcome dynamic ones. I still haven't heard from any reviewers on this paper, but I'm sure I'll post here excitedly when it goes into press.

I'm working on a second paper now, determining how fault geometry affects the intensity and distribution of peak ground motion in an earthquake. In this case, I'm using fault systems that consist of two segments with no linking segment, meaning that rupture has to re-nucleate on the second segment as opposed to just traversing an unfavorable part. I'm looking at both compressional and extensional systems here, with a varied distance between the fault segments, a variety of stress drops, a variety of rupture velocities, and a variety of different rock types. So far, I'm finding that every single one of those parameters affects the pattern and intensity of motion. The talk I gave at AGU (which I will write about, really!) was about preliminary results of this work, and I'll be giving a talk at the Seismological Society of America meeting in Portland in April about the specifics of the material contrast cases.

So that's my research - as I transition into working on a PhD, I'll start investigating more complex geometries and looking at real fault systems, as opposed to hypothetical ones. All that while still taking a bunch of classes to catch up on my background!

2 comments:

Andrew Alden, Oakland Geology blog said...

This may be an odd question or a fruitful one: How much of your thinking gets done by writing a paper, as opposed to doing the research for it?

Julian said...

I definitely get most of the thinking done before I write a single word. I generally don't even start writing until I've fully hashed out the analysis in discussions with people. As a result, once I actually sit down to write, the words come out very quickly. There's just always the issue of putting the first words on the page; it's sometimes hard to make myself start, even if I've thought the whole thing through already.