Advances in exploration and extraction technologies have greatly improved our abilities to locate and extract more oil and natural gas from reserves. Currently, hydraulic fracturing (also called "hydrofracking" or simply just "fracking") is the main method used to recover these resources. Fracking works by injecting a pressurized fluid into various rock layers containing oil and/or natural gas to fracture the rocks and provide pathways for the oil and gas to migrate toward reservoir rocks where we can then extract them. The injected fluid is water, typically with sand and other chemicals added to it to help keep the fractures open and allow for easier migration of the oil and gas. Once it has become too contaminated to further serve its purpose, the injected water becomes wastewater. Since the mid-late 1950’s, this wastewater has been injected several kilometers beneath the surface for storing. However, this wastewater has weakened the strength of the local rocks containing it, and many earthquakes have since been recorded near wastewater injection wells and are thought to be induced by excessive fluid pressures. The fractures and faults on which these earthquakes occur create an anisotropic medium through which seismic waves can propagate. Polarized waves, such as shear waves, in which oscillation can occur in more than one orientation, tend to split into two polarized waves upon incidence with an anisotropic medium. This phenomenon is called shear wave splitting, or seismic birefringence. One of the new waves will orient itself parallel to the cracks and travel faster than the wave that attains some other orientation. These two polarized waves are called the fast wave and slow wave, respectively. The observed time delay between the two contains information about the density of the cracks and their orientations. Temporal variations in the degree of anisotropy will manifest in shear wave splitting measurements, from which one can can infer stress variations along fractures and faults induced by fluid injection. The scope of my project is to utilize shear wave splitting measurements to learn more about the fault structure and source mechanics of the Oklahoma earthquake sequence that occurred in November 2011.
Hello world, and no, I have never been to anger management -- yet. But I love sarcasm and the amazing ability it has lighten your mood when frustration gets you down. I would like to take this opportunity to share with you a recent piece of work from the right side of my brain. I believe it reflects something that I have truly come to acknowledge this past week.
Perhaps all scientists should be required to go through computer science boot camp. This would help prepare them for battle in the war against the machines. (By the way, I hope you like my artwork. I'm thinking of making this a weekly thing.)
In all seriousness, a computer is one of the best friends a scientist can have…if you speak its language. Having problems with your real friends? Talk to a computer! A computer will do whatever you tell it to do. Literally, no strings attached. You really couldn’t ask for a better friend. In fact, once you get to know you’re computer, you’ll learn that it can be quite a character. And, if you’re relationship with the computer doesn’t enter the fail state, it will continue to return zero complaints while running in loops to process your data. Else…………….
Ctrl + C
Well I’m glad I ended that mess -- that input stream wasn’t going anywhere. I may have already stated this in a previous post, but much of my work this summer will be processing and analyzing data. Therefore, getting to know my computer and all of its problems is of utmost importance in order for me get the most out of this research project. Some of the earlier processing and picking of seismic wave arrivals will be done with SAC, which (thankfully) has a much smaller learning curve than ANTELOPE. Additionally, some data conversion will be done using MATLAB scripts (I have to say I like MATLAB the best out of all the programs I will be using this summer; it understands humans the best). The traveltime and ray path calculations will be done with TauP, and SPLIT cluster analysis code will be implemented by MFAST to do the actual shear wave splitting measurements. Lastly, I’m sure any pretty maps or figures I make this summer will be done with GMT. I’m really going to be counting on fellow interns to help me get a GMT script that works, as I was very unsuccessful during orientation in writing my own. Otherwise, I may have to substitute my very sophisticated artwork (as seen above) in place of GMT, and be the only presenter at AGU this December with stick figure maps and diagrams -- that would be AWKward.
At the conclusion of my second week of the internship, I find myself staring at the computer screen watching the shell script spit out line after line of output -- it’s been running for three days now. This is a good thing. It means that no errors have been encountered yet in converting the data to a format that will eventually be analyzed using the MFAST code. Given that this batch of data takes about a week or so to process, let’s hope that everything goes correctly the first time! You may be wondering what on Earth could be taking this long to process? Let me explain:
From my experience, whenever I have heard a seismologist talk about processing the data for their picked events, they are usually only referring to a couple to several tens of earthquake events, maybe even on the order of a hundred if they’re ambitious -- that’s cute. Without a doubt, my mentor must be going for some kind of record in the seismology hall of fame (which doesn’t exist to my knowledge). We are processing -- brace yourselves -- sixty-one thousand three hundred and sixty-three events (61,363)! That is two orders of magnitude greater than the number of events even the most ambitious seismologists usually tackle in a research project! Then again, I’ve never heard a scientist say “You have too much data.”
When an earthquake occurs, sometimes it is preceded by a couple of foreshocks, and it is always followed by many, many aftershocks -- many, many...many aftershocks. Most people's attention, understandably, would be drawn to main event itself; the big climax of the earthquake story, the epic event that causes all the buildings to sway and bridges to collapse, with power plant explosions and fires and other exciting things (assuming the earthquake is large enough, of course). As implied, these "big events" usually only occur once in the story of an earthquake, and once it has passed most normal people usually lose interest and resume with their day. Seismologists, on the other hand, are not normal people. Oh no, we haven’t seen the end of the story yet! The plot goes on!
If we were to analyze the story of a typical earthquake, we would see a relatively short period of small rumblings, a big jump in activity, and then a much longer period of continued smaller rumblings. However, all earthquakes are unique in some manner or another. For example, some earthquakes exhibit detectable foreshocks, whereas others do not. Or perhaps the earthquake has two or three large events instead of just one. Using this story timeline as an analogy, my dataset focuses on the small rumblings after the big event...except there’s a twist.
On November 5, 2011 a magnitude 5.0 (M5.0) earthquake ruptured the Wilzetta fault in Oklahoma. Less than a day later, a M5.7 earthquake occurred less than 2 km away from the first. On November 8, another M5.0 earthquake occurred. Each earthquake generated thousands of its own aftershocks. In this sequence of events, the first earthquake is classified as a large foreshock, the second the main event, and the third a large aftershock. The story of the Oklahoma earthquake sequence therefore has a large event in each section of the timeline, and each earthquake is thought to have activated a separate portion of the Wilzetta fault system. My dataset consists of the thousands of aftershocks recorded by 47 seismometer stations in the days and months following the first M5.0 earthquake.
The first week of the internship is almost over, and I have already accumulated a dense pile of research papers thicker than my copy of Atlas Shrugged -- thank you Danielle for the light reading! Several of the papers served the purpose of providing a historical background on similar case studies of potentially induced earthquakes near wastewater injection sties, including a 200-some-page white paper on the topic released by the National Academy of Sciences; other papers served the purpose of expanding on the methods of measuring shear wave splitting (SWS) and structural anisotropy. Finally, I am working my way through the manual on how to implement the code -- called MFAST -- that I'll be using this summer to analyze shear wave splitting measurements. MFAST stands for Multiple Frequency Automatic Splitting Technique, which is essentially a hodge-podge of code using different shell scripts, Seismic Analysis Code (SAC), our dear old friend Generic Mapping Toolkit (GMT), SPLIT cluster analysis code, TauP toolkit, Fortran, and various UNIX tools, all of which come together and make processing these measurements really fast and mostly free of user bias. Sounds good to me!
Below is an example of some recorded wiggles from a practice data set that I have been playing around with using SAC.
The first and second rows show recorded horizontal ground motions in the E-W and N-S directions, respectively. The third row shows recorded vertical ground motions. As you may have already guessed, the P-wave arrival (marked by the IPUO line) appears first in the vertical component and the S-wave arrivals (marked by the T0 line) appear later and more clearly in the horizontal components. A careful eye might notice that the arrival times of the E-W and N-S S-waves do not exactly match up. In fact, the E-W shear wave arrives about a quarter of a second earlier than the N-S shear wave. Identifying and measuring this time lag is a significant component of my research this summer, for this time lag tells me something about the degree of anisotropy in the Earth. Additionally, by noting which horizontal component of the S-waves arrives first (in this example the E-W component), we can infer that the E-W shear wave must have been the fast wave, and therefore the fast (or preferred) direction of the anisotropic medium must also be E-W. Wherever the crust is fractured and faulted, the orientations of the cracks determine the fast direction. For a given preferred orientation, if the time lag between the two horizontal component shear waves changes over time, one can infer that the stresses acting along the faults must be changing; the cracks will dilate and contract as stresses increase and decrease, and therefore cause the speeds of the fast and slow waves to change. Fluid injection has the potential to induce such changes in stress along faults, and if enough fluid pressure is exerted, an earthquake can occur.
This image from Wikipedia demonstrates the basic idea of shear wave splitting.
Hello readers of my first blog post! Orientation week has come and gone, and now my internship in Pasadena is starting!
I really enjoyed staying in Socorro this past week. I met a lot of great people coming from a myriad of different backgrounds, and now we've already split up to different parts of the country! In addition to making new friends, I also picked up some useful skills that will most certainly be necessary for a successful internship. A lot of my research will involve computing, so it was really great to get an introduction to MATLAB and expand my UNIX background. I also learned the basics of Inverse theory and how to make geophysical models from data. I really enjoyed installing my first seismometer as well as conducting a field survey utilizing an accelerated weight drop. Analyzing the data we collected was also very useful practice for what is to come this summer. As always, I loved the many field excursions we took to explore the local geology around Socorro, in particular the visit to the Magdalena Ridge Observatory and the subsequent three-hour hike down the mountain! This past week was truly eye-opening to the geologic complexity of rift systems, in particular the structural and volcanic histories associated with crustal extension.
Without a doubt, hands-on seismology is orders of magnitude more interesting than learning about it in a classroom. However, I can't help but feel I that I would have been overwhelmed this past week had I not been given such a great introduction to seismology in my previous classes (kudos to Dr. Zhou and Dr. Hole at Virginia Tech!). Lastly, the many fascinating aspects of geology have always been my motivation for the use of geophysics to learn more about the Earth and other terrestrial bodies in the Solar System. Thank you Dr. Law for the strong background in structural geology that helps me see more than "just rocks".