Locating offshore seismic events with precision is very difficult due to a lack of nearby seismic stations, limited azimuthal coverage, and uncertain velocity structures. Thirty-four ocean bottom seismometers (OBSs) were deployed in August of 2010 in order to attenuate these issues and gain a better understanding of offshore seismicity patterns. I am scheduled to help recover these OBSs in September so that the data it has been collecting for almost a year can be examined and analyzed.
In preparation for the research cruise, I will map data from the Southern California Seismic Network (SCSN) earthquake catalog and other scientific sources using the Generic Mapping Tools (GMT). I will also calculate polarity-based focal mechanisms from waveform data using a Fortran computer program. The results from these two analysis techniques will hopefully show a correlation between offshore seismicity and known/suspected offshore faults. They will also help us gain a better understanding of offshore seismicity patterns, fault structure, and rupture dynamics.
I'm now back at school and caught up on work from missing two weeks of class for the research cruise. But I think it was worth it because I really enjoyed the cruise despite being sea sick. We managed to recover 32 of the 34 OBSs and I was truly fascinated by the recovery process. The OBS Techs on board the ship used a device to electronically send signals down to the particular instrument we were attempting to recover. If communication was established, they would then send a different signal that would allow the instrument to release its anchor and begin rising to the surface. We would calculate what time we believed it would arrive on the surface based on an average rise rate and have people in the bridge and on the deck looking for it. Luckily, the instruments are big and yellow with a bright orange flag so they're pretty easy to spot even from a distance. They also emit a radio signal when they get to the surface that shows up on the bridge's radar and they have a strobe light for recoveries in the dark. Once they are on the surface, the ship must be navigated to their position and people with lines and a crane are used to pull it out of the water and onto the deck. I helped out with this portion of the process a lot because I enjoyed the physical part of the recovery.
Once the instruments were recovered, the data chips were removed and formatted, filtered, and viewed. I luckily got to look at some of the data as well and got P-wave polarities that will hopefully be able to improve my focal mechanism calculations. I plan on using this new data to re-run HASH for the local earthquakes that I calculated focal mechanisms for to see how it changes the results - if our theory is correct, this additional coverage should improve the results. I'm just worried that the new data won't improve my calculations because the polarities were very difficult to pick for the small, local earthquakes so the data may be incorrect. I guess I'll just have to wait until I find time to actually run the program and see what happens. Either way, I will be discussing it on my poster for the AGU conference.
I saw lots of dolphins while I was on the boat - especially since the signals used to communicate with the OBSs was at a frequency similar to the one that dolphins used and attracted them to the ship. I also saw some seals, sea birds, and squids, but unfortunately no sharks or whales. It was nice to sit on the deck and look for sea life when the weather was nice and the seas weren't too rough, especially since there wasn't too much else to do on the ship (besides tons of schoolwork!).
The research cruise helped me decide that I do actually want to study physical oceanography in graduate school and am willing to go on another cruise. Since I definitely know what I want to do know now, I just have to get through this semester, study and take the GRE (which I'm finally signed up to take in November), and apply to grad schools. I'm very glad that I did this internship - it was a priceless experience and I learned a lot about myself as well as seismology.
Hello all! Its been a while since the last time I blogged and so much has happened. While I was home, I survived an earthquake and a hurricane - neither of which are common occurrences in New Jersey. I also survived my first week of classes, which ended up only being two days long because of the hurricane. I'm really excited about some of my classes this semester; they look really interesting and I like all of my professors. I'm also continuing some independent research that I've been doing with my advisor at TCNJ and will be presenting that research at the AGU conference as well.
But now I'm back in sunny California and aboard the R/V New Horizon in preparation for the research cruise that will conclude my IRIS internship. The whole crew will spend the night on board the ship at the port and then we push off for open water at 8 AM tomorrow morning. I've been assigned the 4-8 shift, which means that I will be at the main lab with two other student volunteers taking logs and helping with the OBS recovery from 4 AM to 8 AM and 4 PM to 8 PM. These would be ungodly hours under normal circumstances, but since I am adjusted to East Coast time they actually work out pretty well for me. I am, however, worried about sea sickness especially since this is a small ship (only 170').
I'll try to post a couple of blogs about our progress throughout the cruise, but the internet is not entirely reliable so I'll post when I can. I also have some school work to get done while I'm on the ship since I will be missing two weeks of classes. Between work and all of the television shows and movies on my hard drive I should have plenty of things to occupy me in my down-time. I think this is going to be a really great experience to have before going to graduate school and I think its going to be a lot of fun as well!
I've been home for a week now, so I thought it would be appropriate to write a blog post reflecting on my summer goals and the research experience as a whole. I can't believe its already been eleven weeks since I arrived in New Mexico for orientation and met the other IRIS interns in person for the first time. I feel that that week in NM was important for learning and bonding for all of us since we were scattered all over the country (with a couple exceptions) for the remainder of the summer, yet having very similar research experiences. Getting to know the other interns created a built-in support system for when we were separated.
I really enjoyed my time at CalTech - I learned a lot while I was there and accomplished nearly all of the goals I set for myself at the beginning of the summer. Going in I said that I wanted to become comfortable with various computer programming languages. Once I actually started, I realized that I was already really comfortable using Linux and SAC because I had used them for previous seismology research. That allowed me to focus on learning GMT, which became very important for plotting my results for viewing and analysis. While I will never know every function that GMT is capable of, simply because they are innumerable, I do feel confident using it to create detailed maps. Being able to problem solve and trouble shoot issues without any help gave me confidence in myself as well.
By the end of my time in Pasadena, I had more results than we had originally hoped to get. In my last post, I mentioned that I had calculated focal mechanisms through 2009. But during my last week there, I managed to calculate mechanisms for 2008 as well, for a total of twenty-nine by the time I submitted my abstract to AGU. Those 29 events produced significant results that showed patterns in terms of with vs without offshore seismic stations and geographic correlation with known fault locations. I'm very proud of myself for accomplishing so much during my short time at CalTech and I'm very excited that I got some interesting results to present at AGU. The webinar we all attended on communicating research will especially come in handy for preparing my poster and actually presenting at the conference to scientists with a wide array of knowledge.
Although I have not begun working on my poster or studying for the GRE (yikes!), I will hopefully have plenty of time to do so onboard the ship for my research cruise. My research experience is not complete, so I will post again when I return to California in several weeks, but I wanted to give a review of my summer while it was fresh in my mind and have time. I start my senior year of college in only two weeks so I intend to relax before having to get back to work. I am already very excited about going to the AGU conference in December and seeing all of my fellow interns again, but I have plenty of things to get done and keep me busy before then.
Last week, I finished calculating focal mechanisms with and without the island stations for all twenty-two seismic events that I identified as being relevant to my project. (I realize that I said 23 in my last post, but apparently I am incapable of counting correctly.) There were several events in 2009 that were greater than M4.0, which were exciting to look at. I find it truly remarkable that a small difference in magnitude makes a rather significant change in the P-wave amplitude. This made picking the polarities for these events a lot easier, especially since the noise from stations in the LA Basin usually overpowers the signal at lower magnitudes. After comparing the focal mechanism solutions for the same event with and without the island stations, several things became quite apparent. While including the island stations increased the accuracy of the solutions (by decreasing errors in fault plane uncertainty and focal mechanism probability), it did not always make the solutions look visibly different. Those that did look significantly different also showed no pattern in becoming either more strike-slip or more thrust-like.
Since my calculations and analysis are done for the most part, I've been able to focus on writing my abstract so that it is ready for submission to AGU on Thursday. I have found that writing an abstract for research is a lot different than writing an abstract for something like a lab report or term paper, where there is a defined answer to the problem you are trying to solve. I'm trying to represent my research and findings in a concise, positive light without making it sound like I have made the greatest discovery of the century - not exactly the easiest task I've had all summer.
This week I also need to add bathymetry data to all of the maps that I have created so that some of the offshore faults, scarps, and other tectonic features are more prominent. This is especially important since some of the faults are not included in the fault model that I am using, but are obvious in the bathymetry data. I'm going to work on going further back in time with my focal mechanism calculations so that the seismicity patterns are even more apparent, although this is not a priority.
This week also happens to be my last week at CalTech. This campus and Pasadena, in general, are both gorgeous and I'm going to miss it. But, having been away for over two months now, I'm more than ready to go home. I do get to come back to California fairly soon for the research cruise in early September, which means that my research will not be completed by the time I fly back home to New Jersey on Saturday.
I have made much progress since my last post, so I think its time to get you all up to date. After outputting the table of distances, depths, and angles that I discussed last time, I made some very helpful discoveries about how HASH actually calculates the take-off angles. When you first run the program, it assumes that the earthquake location is unknown so it creates a table with a large range of source depths and source-station distances. It then does a ray trace for every combination of depth and distance to calculate a take-off angle and populate the rest of the table. HASH then reads the input files containing the earthquake's actual depth and distance from each station. The take-off angle for each of the stations is then calculated from the table using double interpolation. I checked this particular calculation for several stations by hand to make sure that this step was correct and it was.
Since the take-off angle calculation was clearly not the issue, I then had the program output some different information - the range and depth at each layer in the velocity model for a single seismic station so that I could plot the actual path of the ray trace. While I discovered that it always has the ray turn at the mid-point for rays that are initially down-going (which is a fair assumption), I did not find any problems with this calculation either. After pouring over all of this extensive output for several days by myself, I then looked it over with Monica to see if she saw anything different than I did. She agreed with my conclusions that the program was doing everything just as it should be. Therefore, the odd circle effect on the focal mechanisms was simply a manifestation of the large distance and wasn't actually wrong, just ugly. Because of this, we decided that it would be best not to plot the polarities on the focal mechanisms any more.
Sorting out the apparent non-issue gave me the go-ahead to run HASH for all of the data that I had already prepared and to download some more. Since then I have re-calculated focal mechanisms for the eight events that I had done before trying to debug the program, as well as for an additional six events. I've downloaded waveforms for even nine more, which will bring my total up to 23 (more than our original goal of 10-15), although I have not gotten to pick the first-motion polarities for those quite yet. I'm now working my way back in time (already in 2009) so that I will have more seismicity to plot on maps to help determine which offshore faults are active and potentially dangerous.
I was also given the task to begin preparing some data and maps for the research cruise in September. I was finally given the exact locations and depths of the 34 OBSs to work those into the ALBACORE HASH subroutine for when we actually have the data available. (Which is a whole other story - apparently the Navy needs to look at all of the data the stations have collected before we do to make sure we haven't detected anything we shouldn't have. This kind of makes me feel like I'm in a James Bond movie...) I also created a map, below, showing the azimuthal distribution of the M6.0+ earthquakes that have occurred since the OBS deployment; there were a lot because of the Japan earthquake. The hydrophones on the OBSs should be strong enough to detect those events, so it will be interesting to look at that data someday.
I haven't done much of anything exciting outside of work since the midnight showing of HP7P2. This is partially because I've re-injured my knee working out, but also because its sometimes nice just to relax in my room on the weekends. However, I am greatly looking forward to this upcoming Monday/Tuesday because I get to do some field work! I've been invited by the Geological and Planetary Sciences (GPS) department at CalTech to participate in their field days. We're going out to the San Jacinto fault region to download data, remove stations, and prepare them for shipment (probably to PASSCAL). I'm sure that it will be just about as hot as it was during orientation in New Mexico but it will be a nice chance to get out of the office for a couple days and get some more field experience.
In my last post, I mentioned that I thought I had the "mysterious circle issue" figured out. The plot I made of source-station distance vs. take-off angle suggested that limiting the range would provide enough data with a variety of take-off angles to successfully compute focal mechanisms. So after limiting the range and running the program for all of the events that I have first motion polarities for, its giving me the same problem as before. Even worse, its barely using data from enough stations to compute a focal mechanism at all and when it does, its giving me awful errors. Here's an example of what I'm talking about:
This particular focal mechanism was created using my average velocity model for a M3.69 event that occurred on 2011/05/21. I have plotted the polarities from various stations on the "beach ball" - where a circle is a negative polarity and a plus is a positive polarity - according to their azimuth and take-off angle. Since the program calculated nearly identical TOAs for most of the stations, they form a circle (especially on the upper right-hand side). This solution gave me a fault plane uncertainty of 49 degrees and a mechanism probability of only 43%. You can tell that the fit is terrible just by looking at it, since the negative polarities are supposed to fall in the white quadrants and the positive polarities in the blue quadrants.
Now, I'm going back to look at the code some more to make sure that it is properly computing the take-off angles. I'm having the program output a table of distances, depths, and angles from the P-wave ray trace to my screen so that I can plot these numbers. Hopefully this will allow me to see a graph of the path the rays are taking through the earth and determine if it looks realistic. If the waves are not behaving as expected, I may have to alter either the velocity models or the code for the calculation. I guess I'll have to wait and see where this takes me.
I got a lot of work done while my advisor was on vacation, but things have unfortunately slowed down since then. I downloaded the waveform data from the SCEDC website for ten events with a local magnitude greater than three that have occurred since the OBSs were deployed last summer. I was able to successfully pick the first-motion polarities for seven of those ten events. The other three events had a magnitude very close to 3.0 and their signal-to-noise ratio was absolutely atrocious, so it was nearly impossible to distinguish the signal from the noise. I also picked polarities for a couple of the larger earthquakes from the previous year.
With all of that data formatted, organized, and ready for use, I had to finally make a decision about which velocity model(s) to use with HASH. I ran the program with all of the standard inputs except for the maximum source-station distance, which I increased to accommodate for the fact that my offshore sources are quite far from most of the onshore stations. I then recorded and compared the error outputs for all of the velocity models. I had to re-read the user manual and the academic paper about the computer program to actually understand what the various error sources were telling me, but finally managed to properly interpret it. It became quite clear that the RMS fault plane uncertainty and mechanism probability were the best outputs to look at as they give a quantitative measurement of the fit of the data and the constraint of the focal mechanism solution. I then created my own velocity model based on the other velocity models that yielded the best results and probable Moho depth for the Borderlands region.
I called this file "vz.average" (creative, I know) and ran the program for the Santa Barbara event using this as an input. I was quite surprised to find that it yielded great results - nine degrees of fault plane uncertainty and 100% mechanism probability. Although these results were not replicated for the other events that I had data for, it still gave better output than most of the other velocity models. But after plotting the polarities on the focal mechanisms produced, I realized that something was not right. The polarities were forming a perfect circle on the focal mechanism. This indicated that the program was calculating the exact same take-off angle for many of the stations, which would not normally happen.
I worked on this problem all day today and I think I might have gotten it all sorted out. Monica suggested that I plot source-station distance versus take-off angle to see if there is a correlation. After doing this for several velocity models for one event, I discovered that there was. The program was calculating nearly identical take-off angles for stations with a greater range because the P-wave velocity varies very little at greater depths. So in increasing the search radius earlier on to use as much data as possible, I was actually skewing my results. After limiting the radius based on the plot that I made, the mysterious circle issue seems to be resolved. Hopefully, this means that I have finally sorted out all of the kinks with HASH and I can just start cranking out focal mechanisms.
During the extended Fourth of July weekend, I thankfully got to take a much needed break from my computer screen. My parents and I went up to Santa Barbara for some shopping, beach time, and surfing. Every new place I visit in California is more beautiful than the last and SB was certainly not an exception. I really enjoyed getting to relax at the beach and finally go surfing in the Pacific Ocean! I also loved getting to see my parents for the first time in over a month, although it made me miss home more than I thought it would. This past weekend I went to the beach in Malibu with a couple of friends. It was quite cold out, but there was a swell so it was cool getting to watch the surfers that are much braver than I am out on the big waves. Now I'm looking forward to the midnight premier of the final Harry Potter movie on Thursday night, although it will be sad saying goodbye to the characters that were such a huge part of my childhood and teenage years.
I have been very busy working with the HASH program this past week. Obtaining the first-motion polarities for all of the seismic stations associated with the Santa Barbara event took much longer than it probably should have. I drastically over-filtered the waveforms my first time through so I got all of the polarities incorrect and the focal mechanism that the program calculated looked inside out and backwards from what was expected. I had much more success my second time through - I decided not to filter the waveforms at all unless absolutely necessary. Most of them were surprisingly "clean" (not noisy) anyway and I could easily pick off the first-motion.
Running this data through the program with the standard inputs was extremely simple once it was properly formatted. But that would make for a rather uninteresting project, so my challenge of the week was altering the HASH program itself as well as some of the input parameters to better suit my offshore data. I had to write a subroutine to the main program that runs parallel and nearly identically to a currently existing subroutine. The primary difference is that the one that I had to create will identify the ALBACORE (the chosen name of the OBS network deployed last summer) station names rather the standard SCSN station names so that the main program will be able to perform calculations using that data once it is available. I have data from one of the OBSs as well as a hydrophone for the Santa Barbara event to test the program with. So far, it is not running quite as I would like it to, but is functioning enough to allow me to move on and come back to it when I have a bit more time. This is especially important as part of my project is to compare focal mechanisms calculated with and without the island stations, which includes the OBSs.
I also went back through a few scientific papers about the seismic history and geology of the region of interest offshore California, known as the Continental Borderlands. These papers allowed me to come up with eight additional velocity models to use with HASH, which comes with a standard five. I have run the program for the Santa Barbara event with each of these individual velocity models to determine which best describes the borderlands and will, therefore, give me the most accurate focal mechanisms. Based on error outputs, the program seems to be favoring certain velocity models that are not compatible with the offshore region. I need to re-read the less-than-helpful user manual and go through the code again to learn how it is calculating the error because the results would make far more sense if I were just interpreting them incorrectly. Once I have a better understanding of the error, I will be able to select or create the best velocity model to use with all of my data.
My advisor, Monica, is on vacation this week and part of next so she has left me with plenty to work on while she is away. She would like me to start calculating focal mechanisms with and without island stations for all of the magnitude 3.0+ earthquakes that have occurred offshore since the OBSs were deployed last August. This means that I have to download the waveforms for all of these events from the painfully slow SCEDC STP website. I had hoped to download and interpret the waveforms for at least two events today, even if it meant working later than usual. I only got through one. Unfortunately, the website went from sluggish to a complete halt once the other students on campus got back from work and started using up all the bandwidth. At least now I know that I should download all the waveforms I need for the day in the morning when the internet is fastest.
This past weekend I got to see one of my friends from my internship last summer who I haven't seen in a year! It was really nice getting to see him and will hopefully get to hang out with him at least a few more times this summer before I go back to New Jersey. I've also been making good use of the pool out here and have been trying to go swimming at least twice a week (injured shoulder permitting). One of my new friends from the hall even goes with me sometimes. I'm very excited for this weekend as my parents are flying into California this evening. We're going to Santa Barbara for the weekend and I have been promised surfing!! I haven't even bothered to check surf reports because the waves couldn't possibly be any worse than they are at the Jersey shore, although the water is probably colder. Don't be surprised if the first pictures I post on my blog are of some gnarly waves
My second week of research has felt less productive than the first because I was extremely frustrated for most of the week. My last blog entry explained that I was a bit apprehensive about having to use a Fortran program called HASH to calculate focal mechanisms and, unfortunately, my fears were not completely unfounded. While I had no trouble compiling and running the program, I was having difficulties plotting the seemingly overly complicated output. However, this led to me reading through all several thousand lines of code multiple times in order to determine what the program was doing and what it was outputting, so I do feel that I have a much better understanding of both the program and the focal mechanism calculation technique now. I did eventually figure out how to plot the output using GMT so that I can actually view the focal mechanism that the program has calculated. In the upcoming week, I hope to write a short script that will allow me to compile, run, view output, and plot all in one short step to save time and typing - oh, the powers of Linux!
Now that I know how to use the program and understand its many different components - including input files, subroutines, the main program, and the output files - I get to start using my own data and not just play around with the provided examples. Monica showed me how to download the waveforms for a given seismic event from the SCEDC website. Using the graphical interface on the website is a bit cumbersome, so I think I'm going to look into downloading the Seismogram Transfer Program (STP) that can be run directly from the command line for future usage. After downloading the waveforms for the August 24, 2010 Santa Barbara event that we are currently interested in, I viewed the files in SAC to determine which stations might be tossed out and which need to be filtered.
The next step will be to go through the waveforms again once they have been made presentable and record the first-motion polarities for each station. This information will need to be formatted so that HASH can read it and calculate a focal mechanism for comparison to one that is pre-existing. I will need to repeat this process for waveforms from an OBS that was deployed last summer, as well as for a hydrophone. HASH does not know the locations of these monitors as they are not part of a recognized seismic network, so I will have to create another input file and a subroutine so that it can process the data accordingly. Eventually, I will also have to make my own velocity model (yet another input file) that better describes how waves travel and bend in the oceanic crust. Some of the current models that the program uses - such as those for the Los Angeles Basin and San Gabriel Mountains - would be very inaccurate for this type of calculation and would, therefore, create a focal mechanism with a great deal of error.
Earlier this week, I finally moved in to my permanent, on-campus housing for the summer. That turned out to be quite an ordeal as well because the lock on my door was broken and took three hours to fix. But I'm quite glad that I decided to stay on campus because it adds an important social aspect to my research experience. I've already gotten to know some of my floormates and I'm happy to have people to hang out with at night and on the weekends. I also finally got a chance to use the pool this week. I haven't gone swimming in so long and it felt absolutely wonderful to be back in the water. Hopefully I'll get a chance to say the same about the ocean and surfing someday soon!
I've only been at CalTech for one week and I already feel as though I've learned so much! Aside from gaining a better understanding of what my research project is actually about, I am also starting to grasp how to write scripts using GMT (Generic Mapping Tools) to create high quality images. Since I am already comfortable using Linux, my research advisor Dr. Monica Kohler had me jump right in to learning how to use GMT. After downloading some earthquake data from multiple sources, I was able to plot the offshore seismicity, focal mechanisms, and faults for Southern California during the time period covered by the OBS deployment. I also created comparable maps for data from the past ~20 years. These maps will hopefully give us an idea of the type of data that we will recover from the OBS stations when we go on the research cruise in September.
I have also had plenty of reading to do this week. I read several scientific papers on the historical offshore seismicity and faults. Coupled with the maps I created, these articles could help us identify the active faults off the coast, which could indicate a potential earthquake and tsunami risk to the Southern California community. I also read articles on the techniques used to relocate earthquake data and calculate focal mechanisms.
These last articles will be of more use in the upcoming weeks, as I attempt to use a Fortran program to calculate my own focal mechanisms for some of the observed offshore seismicity in the past year or so. It will be interesting to see how this goes as Fortran and I have not always been friendly in the past. However, I feel that my programming and problem solving skills have improved greatly since I last tried to use the program, so it should be a much greater success.
In other news, I am greatly enjoying living in Pasadena. I have been walking much more than I usually do, but the weather is a sunny 75F everyday and the area is beautiful so I certainly don't mind the fresh air and exercise. With so many awesome places to eat in the area (including a cheap, on-campus coffee shop), I may gain some weight despite the extra exercise. Jamba Juice and Chipotle and Stone Cold, oh my!
This past week in Socorro, New Mexico has certainly been an adventure that I will never forget. In the span of only one week, I have made amazing new friends that I've already had to say goodbye to, learned lots of new programming and geology concepts, gained invaluable field experience, and had plenty of car trouble (that's an understatement). I also got to visit a state that is unlike anywhere else I have ever traveled. I will truly miss this beautiful place and the wonderful people that I've met here. But, my next stop is sunny Pasadena, California to work with Dr. Monica Kohler at the California Institute of Technology. I feel so lucky to get to spend my second summer in a row doing seismology/geophysics research in Southern California.
While I am extremely excited about beginning something new and learning in the process, I am feeling a bit nervous. I feel that the best way to alleviate some of this anxiety is to set some summer goals for myself to stay on track and keep focused. The following is my academic and personal goals:
These are some important beginning goals to keep in mind as I begin my research on Monday. I'm sure that these goals will evolve and become more focused as I gain a better understanding of what I'm actually doing this summer. Now, I should go find my terminal so that I don't miss my plane to California - that would certainly put a damper on things!
(written in the Albuquerque airport)