David Watkins is a student at Indiana University of Pennsylvania currently completing his research at Miami University of Ohio under Dr(s). Harmony Colella and Michael Brudzinski.
Slow slip events have now been observed in subduction zones around the world. They clearly represent a transition zone between the shallower seismogenic zone and the steady creep that takes place at greater depths and temperatures. However, the controls on slow slip events are still poorly understood. Duration of slip events and recurrence intervals vary substantially around the world, and along the strike of individual subduction zones. Subduction zones generate the largest and most destructive earthquakes, and slow slip plays an integral part in the stress accumulation and release that leads to these quakes. Better understanding slow slip is critical to helping predict when these events will strike. My project will be focused on simulating these events, using the long-term earthquake simulator RSQSim. The current record of slow slip events is quite short; simulations allow us to generate records orders of magnitude longer. RSQSim allows us to vary parameters such as fault geometry, frictional properties, normal stresses, and slip speed to see their affect on slow slip events. A model that reproduces existing observations will help us see how these events change over much longer timescales. Some questions we hope to address are: -Are slow slip recurrence intervals stable, or do they vary over time? -What variables cause different recurrence intervals within and between subduction zones? -What geologic factors change these variables?
Last week the refurbished seismometers that will be heading to Mexico (hopefully along with me!) finally arrived from Guralp in the UK. They're a bit different than what we installed at orientation - there is no other 'box' that they connect to - everything is controlled from a laptop via Guralp's interface. That made things a bit more complicated, but after a couple email exchanges Mike got it figured out, and several of us got them all hooked up to test to make sure they were all working. Right now they are all set up in Mike's lab for a huddle test, all powered by a single battery via an impressive array of spliced power cables:
We ran the GPS units out the window to the grass outside so they could get some sky-view. They've gotten many weird looks from passers-by already.
I'm now officially halfway done with my internship. I feel like I have a good grasp of how to run our different fault models, and am starting to explore the different parameters that we can vary more in depth. Although it doesn't always feel that way, we have generated some useful results already. I started with zones of different normal stresses across the fault, thinking that higher normal stresses would cause longer recurrence intervals of slow slip events in that section. However, it turns out that this effect only shows up between different faults; variations in normal stress on continuous faults don't have significant affects on the each section, since events nucleate in one section and often migrate to another. I don't think we have a way of sorting events by where the most slip occurs. This probably suggests that normal stress either isn't one of the main factors, or that there are tears or other features in the subducting slab which prevent sections of the fault from interacting with each other. We will be exploring this later in the summer with some different fault geometries, as well as changing the frictional properties of the fault.
Each of these model runs takes about 2.5 days to complete, and we can only run one or two at a time on Harmony's computer, so there is a fair amount of time I don't have a whole lot to do. It's not a bad problem to have I suppose, but I do sometimes feel like I should be working more than I am. That could change here soon, since I will need to start my AGU abstract soon, and also I may be starting to write the introduction for the paper we are planning from this. I'm not real worried about writing an abstract, since I have done that already, but the paper will be new. I wrote two different proposals for classes this past school year, but an actual paper will be different, although I feel like it could be easier in some respects since I've actually done what I'm writing about. The writing will run through the fall semester most likely. I'm planning on continuing my project through the school year as my senior project at IUP. It should lend itself well to that, since I work remotely a lot of the time already.
Last week was spent largely dealing with computer problems. We set up my own account on Harmony's computer, which involved re-installing all the random Unix tools involved with running RSQSim. Suffice it to say that things didn't always go smoothly. Lots of things had to be figured out which should have been written down, a valuable lesson for the future. Coincidentally, I also heard from my mentor last summer asking about some issues I had with the program was running last summer. Should have done the same thing there!
By the end of the week we did get down to running the actual fault model in RSQSim (and dealing with some issues there as well, of course). Basically, I will be running simulations of a shallow thrust fault, representing the plate interface of a subduction zone, with a seismogenic zone, slow slip zone, and a creeping zone. RSQSim generates a record of relatively large events (>Mw 4 or so) over hundreds to thousands of years. This will be the primary 'data' I will be working with this summer, although of course model results are not 'real', so I'm not sure you can really call it that. We will of course be comparing it to real data, to make sure the results are realistic. The main advantage of working from a modeling perspective for this issue is the longer time-scales you can work with in a model, orders of magnitude longer than we have from actual seismic data. I will be varying different things with different runs of the model, so in that sense it will be entirely unique work. RSQSim was originally designed for strike-slip faults, but Harmony has adapted it to model subduction zones. At this point I have a working (I hope) knowledge of how to run it and mess with different parameters, and should be getting to into the meat of my project in the next few days.
Well, I've now spent my first full week at my internship, and things are starting to fall into place for the summer. I've found myself in yet another place that will be confusing to explain to people; I go to school in the state-within-a-state town of Indiana, PA, and now I'm at the school that's supposed to be in Florida (Miami), but is in fact in the slightly less exotic locale of Oxford, OH. I admittedly wasn't excited about still being in Ohio (my home state) for the summer, especially after spending last summer in Minneapolis, but I'm very happy with my project here. My advisors are great, and I've met a lot of the grad students here already. The town of Oxford is amusing similar to Indiana in some ways, although perhaps even more of a backwater geographically, complete with the train running through one side of town. The gritty Appalachian feel is gone, replaced by a decidedly more upscale finish, although the college-town culture (or maybe lack thereof?) is still here. On the plus side, I was able to drive since Oxford was (relatively) close to home, I was able to bring my mountain bike with me, and will be trying out the local singletrack. Some other interesting weekend trips should be in the works as well. Also, on the work side of things, there is the prospect of a trip to Oaxaca sometime this summer for some broadband deployments. It seems the equipment is in limbo somewhere in Europe at the moment though, so it's still up in the air. The only real thing to complain about so far is my apartment still being off the grid, since Time-Warner is taking their time shipping my internet equipment.
As far as my actual project is concerned, I have largely spent the week plowing through literature and working on Mike's Unix tutorial. There will definitely be a learning curve with Unix, as I never used before orientation, but it seems relatively straightforward so far. My semester of Java last fall wasn't completely irrelevant at least. Getting comfortable with Unix and RSQSim, the main program I will be using, will be probably the biggest thing for me the next few weeks. RSQSim is a long-term earthquake and slow-slip simulator that can produce hundreds of thousands to millions of events on a fault segement, allowing us to generate much larger datasets of recurrence intervals than what is available from actual seismic records. I will be dealing with recurrence intervals of slow slip events in Cascadia, via RSQSim, and hopefully later getting into the geological controls on this. I haven't actually used it yet, so I don't have anything original to post yet data-wise. Besides moving forward with my project this summer, I will be getting my AGU abstract ready sometime later next month, and also at least outlining a paper. We are hoping to publish in Geophysical Research Letters with our results from this summer.
That's all for now. Tune in again next week, when I use RSQSim for the first time!
This week has been intense, but I've learned a lot. The most interesting to me was the modeling and data manipulation lectures. I had already had most of the seismology material in my geophysics class this spring, but I had never written Matlab code before or done anything whatsoever with Unix.