Hannah is a student at Case Western Reserve University currently completing her research at Los Alamos National Laboratory under Dr. Ellen Syracuse.
My project is a part of a larger one that is using Rayleigh wave ellipticity (H/V ratios), geodetic, and gravity measurements to identify temporal changes in groundwater in southern California. I am working on a subset of the H/V ratio analysis. The Rayleigh wave ellipticity measurements of the region show anomalous behavior in the longer period data (15 seconds) at several different points in time. At these times, many stations measure a larger H/V ratio due to an unknown cause, unrelated to the subsurface structure. My task is to identify when, within the week-long average of measurements, the anomaly causing phenomena occurred. This will be done by performing 9 component cross correlation between all station pairs, stacking these results, converting from the frequency to the time domain, rotating to the radial-transverse-vertical frame, and calculating H/V measurements using the rms amplitudes of the waveforms near the maximums of the envelopes of each component, for increasingly smaller time frames (week, day, and, finally, hour). Once the time of interest is identified, I will find possible causes of the anomaly and decide which is most likely.
Wow! I can’t believe this internship experience is already ending. I finished everything I needed to, including writing a 500+ line bash script, which is a feat I would have never thought I could accomplish a few weeks ago. I have never felt surer that I want a PhD. in geophysics. I loved every aspect of the research. I loved learning the subject, figuring out what to look for in data, running and editing code, making posters and abstracts, having meetings, and even debugging!
On my last day of work, I finished debugging the code that pulls my project together, then Ellen took me out to lunch. We sat there for around two hours talking about all sorts of things and enjoying ourselves. I really look up to Ellen, and am so grateful I ended up with her being one of my mentors. I came in very nervous and insecure, and Ellen and Andy were always so kind and knowledgeable. I really flourished with her guiding me. Ellen and I filled out my final skill rubric during this lunch. The last time we filled out the rubric, she emphasized that there is always room for growth and that I was doing a good job. This meeting, she told me how she saw my growth and labeled me as “proficient” in everything applicable to my project. She saw my potential, and with her help, I did succeed the way she thought I could.
School is often really hard for me because I get too anxious during tests to perform well, and I compare myself too much to others. This summer, I proved to myself that I am smart and can really prosper outside of the classroom environment. I feel so much better going to into next semester knowing that I do have talent, and that grades aren’t everything.
I will miss my project here at the lab, and I will also miss the interaction with my mentors that taught meso much. Ellen and I are hoping to write a short paper on this summer project, so I will still be in collaboration with her and Andy.
Goodbye Los Alamos! Thank you for an amazing summer!
I am pretty excited - I won a recognition award in my division at the symposium! There were hundreds of students and roughly 35 were recognized. The symposium itself was extremely long and tiring, but I enjoyed every second of it. I have attached a picture of the poster below. The deputy of the earth and space sciences division visited my poster and we discussed it for more than 15 minutes. He was so engaging, and he even pulled over an Earth sciences group leader (from a group different than my own) to talk with me too. I adore communicating my science; I was given such helpful feedback, and many of the people visiting my poster helped me better understand it. After the symposium, the deputy sent me an email (and cc-ed my mentors and group leader) to tell me I did good work. I have struggled with such crippling insecurity in college related to my intelligence and performance in classes, but now, I believe the success I have met here will really help me perform better academically.
The beginning of this week was spent finalizing figures and tweaking my poster for printing, all of Wednesday was spent at the symposium, and Thursday and Friday were spent writing my code for removing anomalous measurements. I really enjoyed writing this code from scratch. It is a bash script that looks through tens of thousands of earthquakes, compares them to removal criteria, calls to a python code that calculates how long anomalous measurements would be taken following a given earthquake, and copies every hour of data that should be excluded from our calculations to a txt file. After I leave, Ellen can use this txt file to easily move any supposedly anomalous data out of the folder that we run our calculations in. Writing this code was nothing but fun for me, and I was really surprised by how easy it felt. I love coding. I could code for hours on end and not even notice how much time has passed. I mentioned this to Ellen, and she had some great insight. She talked about how she thinks that coding is an exercise in logic. You have a toolbox given to you in the form of a coding language, and you create a series of logical steps that eventually bring you to your end goal. She is really right. If you are a logician, you can code. It doesn’t matter what you know of the language or how much experience you have. You just need to make a blueprint of steps that need to be taken, figure out your toolbox and be patient. I would love to one day have a job like this where I code a lot.
As the weeks begin to wrap up I would like to say Ellen has been such an amazing mentor, and I am so lucky to have been paired with her. I loved working with someone who is so good at what she does and also so good at explaining things to someone still trying to learn. Andy, my co-mentor is also an amazing person to work with. My project is less in his domain than Ellen’s, but he has taught me a lot about geophysics as a whole, and is always ready to help me when I come to him with questions. I am fortunate to have met them and been placed here at Los Alamos.
Figure 1: My poster! It is small and hard to read this way, but I just wanted to show what it generally looks like. It's a winner!
This was a week of many figures and great anticipation for the poster symposium next week. This week I was still waiting on a bit of data to finish running, so I can’t say I completely finished my analysis, but it is practically done excluding a few additional data points. Next week is the student symposium, and my poster is really coming together. I have spent a great deal of time on this poster, and I hope it wins recognition in my division. It has been a great learning experience to create a poster so detailed. I have made posters in the past, but I have never grasped the full “story” as well as I do in this case.
Back to my research! One interesting thing about these Rayleigh wave H/V ratio anomalies is that they last very long, but raw waveforms do not show signal lasting nearly as long. Most earthquake signals do not persist beyond a few minutes on raw waveforms, but we know earthquake signal is present long after the quake. For example, Rayleigh waves from large earthquakes will often circumnavigate the earth several time, taking roughly 3 hours per complete trip. While we may not see much remaining energy in the raw waveform, the spectrograms still show an increased amount of low frequency signal for as many hours as we see our anomalies persisting. This is shown in Fig. 1 in greater detail. This depiction of long duration change in frequency composition helps confirm that R1, R2, R3, etc. phases are what are causing the long duration of anomalous measurements following large quakes. Neat stuff! I love my project. I am so excited to present next week, and to have finalized figures. I additionally will be starting the code that creates a list of hours to exclude from our calculations due to their likelihood of being contaminated by earthquake Rayleigh waves. Time is flying!
Figure 1: Colorful figures on the top are spectrograms and pictures below are seismograms. The three times shown are: July 29th, 2016 17:00 UTC, 4 hours before a large earthquake, July 29th, 2016 22:00 UTC, 30 minutes after the large earthquake, and July 30th, 2016 10:00 UTC, 12 hours after the large earthquake. Raw waveforms show no significant difference in signal even 30 minutes later, while spectrograms show increase in low frequency measurements beyond 12 hours after the earthquake.
I was hoping I would be done with a lot of the re-running of everything I had to do by the end of the 6th week, but it carried over a lot into this week. This is ok though, because now I have a wealth of data. One new thing I did was create a spreadsheet of 40 + anomalies, their earthquakes, and features of their earthquakes. With this new catalog of data, the next step is clear: look for relationships between anomaly length, earthquake magnitude, earthquake depth, and distance from the earthquake epicenter to the seismic network. By the end of this week the catalog was nearly finished, and I began to look into these relationships. One new thing I noticed is what seems to be a relationship between station elevation and sensitivity to anomalies. This really interests me, so next week I will be writing a script to explore it. Additionally, this week I edited my poster more and started working on my AGU abstract. I already wrote an abstract for the student symposium that I am doing in two weeks, and I thought that having an abstract already written would help me, but it actually hindered me! It was hard to stray from the language I already used even though I was trying to make this abstract different. This week was probably my slowest week here at the lab. I felt like I was never doing enough, since I was really just waiting on my data and trying to occupy myself in the meantime. Weeks like this happen though, and getting through all of the slowness will let me have a really fast-paced and productive week next week!
On a personal-life-note this Friday had a bit of a hitch in that I stepped on a rusty nail on my way to work! It turns out I hadn’t gotten a tetanus shot since 2005, so even though it hardly punctured the skin, I still needed to get one. Insurance was a big problem on Friday, and the whole ordeal made me miss home, where my insurance is accepted, stepping on a nail costs less than $100, and my family pampers me when I get hurt. The weekend lifted my mood, however, because my friends organized a “Bob Ross Painting Party”, where we all watched Bob Ross’ show and painted on brown paper bags as best we could. Attached is my STUNNING artwork. I like Los Alamos and the friends I have made a lot. As these final weeks draw nearer, I am sad that I will be leaving this great work, my wonderful mentors, and my new friends, but I am getting really excited to see what 4th year will bring at Case and where I will end up for graduate school. Really whirlwind times are ahead; I kind of feel like this week will be the last “slow” week until I retire – and this is good!
In last week’s blog, I showed a figure that I called “confusing” for two reasons. The first reason I found it confusing was because the H and V amplitudes were being measured at separate points in time. My bringing up this issue with Ellen, and showing her how wide spread of a problem it was, led her to talk to Jack Muir, a postdoc at Caltech we collaborate with, and fix the problem. She had already noticed the problem and wondered about it, but the fact I also noticed it and had so many examples of it being an issue, particularly in anomalous time periods, pushed it to be fixed. Previously, the amplitudes chosen for H and V were picked as follows: the arrival time of the fundamental wave Rayleigh waves were calculated, so a time wouldn’t be chosen before their arrivals, then the time, after the arrival, with the largest amplitude would be highlighted for the H and V measurements separately. In more well-behaved wave packets of shorter period, these large amplitude time windows for H and V usually occurred simultaneously, but during anomalies, specifically with longer period measurements, H and V do not always have simultaneous time windows of largest amplitude. We now do this entire procedure the same way, but additionally, the signal to noise ratios (SNRs) are compared between the two time windows chosen in the H and V measurements, and the time window containing an amplitude measurement with a better SNR is the time the amplitude that is measured in both H and V. Before, it was unclear what these untimed measurements were physically showing, but now we have valuable H/V measurements.
The second reason the figure last week was confusing was because I thought it was showing an anomaly before the anomaly began. The ellipse that was created from data hours before the anomaly was flat and long, like a pancake, which would be the shape of an ellipse with a high H/V ratio. We had always said anomalous measurements are those with higher H/V ratios. Some further analysis then showed a change in the ellipse on the onset of an anomaly, but the opposite change I would expect. The ellipse during an anomaly was very round, and this would indicate a H/V ratio closer to 1. I showed this to Ellen, and we discovered that some place in our code we had switched H and V, so all previous results were showing V/H ratios! This was a simple fix, but had I not thought critically about it as soon as I saw the figures, we may have not caught it until much later in the project, when many more figures were made. After these little problems were found and solved, I had a long week of rerunning many codes and recreating all my figures. Overall, it was a good week of science. I feel very happy that I have gained the critical thinking skills to detect some issues!
Additionally, this week all of the IRIS interns shared a short, two-minute presentation of what they have done so far. It was really great to hear how knowledgeable everyone has become on their projects. I delivered my presentation and felt like I did a pretty good job. I went back and listened to my presentation, and I do think it was good, but I also noticed that I say “um” too much, so now I am working on that before I present my poster at the student symposium.
On a personal life note, my host family was away for 3 weeks, so I took care of the house, plants, and animals (two cats and one dog who I love very much). It was an added responsibility that at first worried me (I have never been the caregiver to an animal on my own, let alone 3 animals!), but I ended up loving all of the cleaning, gardening, and tummy rubbing.
This week Ellen and I completed my mid-project self-assessment together, and I really enjoy her viewpoint of “there is always room for growth”. She assured me that she is happy with my work, which was a good thing to hear. We talked about the things I haven’t gotten to show my skill in yet, and she encouraged me to keep voicing ideas. Additionally, I made my abstract and poster for the student symposium here at LANL. I think it looks pretty good, and I’m pretty proud of my abstract soon. Over the next few weeks I need to do some scripting of what to say and practice a lot! The symposium is judged, so there is the potential to win an award for my poster and presentation, which would be a dream come true.
This week we were prompted to discuss a paper that has been the cornerstone of our research, and I have quite a few. One paper titled “Rayleigh-Wave H/V via Noise Cross Correlation in Southern California”, by Jack B. Muir and Victor C. Tsai, is the paper written by one of our collaborators, Muir, that set the precedent for this project. Within this paper, Muir’s methodology is identical to ours when looking at ellipticity measurements. His use of these methods was to create a velocity model while ours will be to compare trends to things like precipitation. An additional paper I have referred to a lot is titled, “Processing seismic ambient noise data to obtain reliable broad-band surface wave dispersion measurements”, by G.D. Bensen et al. This paper is a review of standard procedure for analyzing ambient noise. This paper contains alternative methods of preprocessing that I would like to one day test on the raw waveforms. Possibly, one of these methods could better remove high magnitude distant earthquakes. I really enjoy mining for papers. The process spawns in me many new ideas that have really been driving my curiosity.
Additionally, this week, I created new types of figures to play with. One thing I created are plots of raw waveforms over the course of whole days, but these figures are too big and messy to include here. I created these figures to see if the raw waveform shows signal lasting as long as what the H/V gradient figures show. The raw waveforms had no obvious signal after a few hours, so the H/V measurements are picking up on something that isn’t obvious. Another figure I can make now is a plot of the ellipticity of the Rayleigh waves using the largest amplitude periods of the individual horizontal and vertical amplitudes (Fig. 1). These plots are already confusing because the amplitude measurements that are being used to plot the ellipticity are not occurring simultaneously. Additionally, the wave packets in the H and V directions are looking anomalous before the anomaly appears on the H/V gradient plot; this could be a plotting mistake or it could be pointing to something bigger. Ellen is going to make the ellipticity plot using a different method of choosing the sections of the H and V waveforms, so we will see how that changes things. Another good week of science!
It is crazy how quickly these weeks go by! I will be presenting at the student research symposium here at Los Alamos in less than a month! Now that I am nearly done stacking time periods with anomalies on smaller scales, I have begun looking at raw waveforms to see if they also show odd behavior for the same duration of time. So far, it looks like the raw waveforms do not show this behavior, making this issue even more interesting. Next week I will be making ZZ v. ZR component plots for certain station pairings over different time windows to get a different kind of visualization of the anomalies. I also will be working on my poster for the symposium to get it submitted for review by the 9th of July. I love when I have a lot to do!
This week we were prompted to discuss our frustrations and triumphs. It is hard to say that anything specific has really frustrated me, I love the lab and my project. I really enjoy the struggle of research, but overall, I used to be afraid I didn’t possess a mind marvel enough to produce research questions. When I saw professionals look at data and say “Hmm, that’s interesting!” I would feel blind to science and inadequate. I worried that I would never be able to make overarching connections. I have since begun to develop, and it is a great feeling. I now can often see things that are of interest and I am even naturally curious about certain things before they are brought up. This internship has really been showing me that I can’t force growth. If I go with the flow and let myself learn, growth happens naturally. Coming into this project I knew absolutely nothing about seismology, but before I knew it I was trying my best to propose new questions and try to create my own theories. Since coming here I have had several ideas that I’ve shared with my mentors, and none of my ideas were ever laughed at or instantly tossed aside. Many of the ideas I have had have led to great discussions and expansions of my project. My greatest frustration was a negative mentality I had adopted, and my triumph was letting my curiosity take the wheel for the first time in a long time and drive me to a place of growth.
I also wanted to share more on my research area. This is a project closely tied to drought, which is a major issue in southern California. As shown in Figure 1, we are using a subset of stations in the Southern California Seismic Network. Stations chosen had to have been running as broadband 3-component stations from 2000 through 2017.
This was the week that my mentor, Ellen, was out of town. I came in a little worried that I would get stuck and not be able to move forward, but the opposite happened! My first two days alone I successfully did everything I had said I was going to do this week. I narrowed an anomaly’s time to the resolution of 1 hour, and then I discovered a distant earthquake that happened the hour I found the anomalous measurements to begin in (Fig. 1). While I was happy to have moved forward and to have found a probable cause of these anomalous measurements, I became concerned that my project just collapsed on itself. These distant earthquakes most likely have focal mechanisms and locations that give for large Rayleigh waves hitting California. They simply slipped through preprocessing, and there is really no way to get them out of the signal. The best way to correct for these anomalies is to remove the days they occur in. The data is from 2000 until 2017, so the loss of about 40 days in those 17 years is negligible. A quick check of other weeks containing anomalies showed that they all contained major earthquakes as well. Would my project become simply cutting out data? I then began to think, this project doesn’t need these periods of time, but what if some other project needs the cut times back for their data? That’s when I got my new idea: apply machine learning algorithms. I pitched to my mentor, Andy, two options: 1) use machine learning to fill in the gaps of time we created in HV measurements or 2) Use machine learning to find a way to preprocess that could successfully remove these distant earthquakes. Andy liked option 2 and when I have free time, I have the go ahead from both him and Ellen to work on that. Meanwhile, for my original project, I will continue looking at more anomalies in detail. We hope to notice something unique about the earthquakes that make it
through preprocessing and automate a way to remove the time windows they are in. The H/V measurements seems to be influenced by these distant earthquakes for up to 12 hours. We aren’t sure why the signal lasts so long, so that could be interesting in itself.
Going into next week I am excited and nervous. I have never attempted to write a machine learning algorithm, and this doesn’t seem like a beginner type of opportunity, but I am so incredibly excited to have thought of something on my own, see that it is actually valuable, and more forward into it. If I continue down this path, I will have much less guidance than in previous phases of my project. This is scary, but also a great opportunity for growth.
This week we were asked to write elevator pitches describing our work. I wrote one for scientists and one for non-scientists. I thought both were easy to write because I personally need to be able to describe concepts in laymen’s terms to know I understand them, and I also have begun my abstract for AGU (if I can go; that week is finals week at Case).
Overall things seem to be going well!
This week the research began to take off!
If you read my project description you will see my overarching goal and a brief summary of the process. In order to figure out exactly when anomalous H/V ratios were measured and why occurred and weren’t removed in preprocessing, I need to stack my cross correlations over finer and finer increments of time. Since the data was at first stacked so every week was with weekly averages, I could only tell which weeks were of interest. After picking one of these weeks containing an anomaly, I was given the preprocessed data for the week of the anomaly and the “normal” week after. Next, I performed a 9-component cross correlation and stacked over the days of the weeks of interest, rather than the whole weeks. Ellipticity values were calculated from this data. Beginning next week, I will be plotting these daily H/V ratios and narrowing down my time of interest from one week to one day, then I will move forward into looking at hours of interest.
So far, I haven’t written any codes of my own and have just been editing existing code. In the future, I would like to write more from scratch, so hopefully once things really get going that might be an option. While I haven’t been doing intensive coding, I have been navigating the directories and running scripts in terminal quite a bit and have a pretty good working knowledge of bash scripting and Unix language in general. I once accidentally unpacked all of the data files from these huge subdirectories into their parent directory, making all of the data indistinguishable, but luckily, we have a snapshot folder that is kept just to clean up blunders like that. That was a very time consuming (and somewhat embarrassing) mistake, but it definitely made me more careful.
My mentor, Ellen, will be gone all of this coming week traveling for work, and I hope I have a lot done by the time she is back, and that I have enough to do. My other mentor, Andy, mentioned that another geophysicist in our department will be doing field work soon, and that I will probably be able to tag along for the experience. Details to come on that endeavor.
As for my living situation in Los Alamos: things have gotten better. I love where I live and the people hosting me, but it is hard not to feel isolated. To give myself some more freedom I ordered a dirt scooter that will help me get around town a little faster and more frequently. I order pizza every Friday and am reading at least one book a week and enjoying that a lot. Dory and Sydney came from Albuquerque this Saturday to pick me up and take me to Santa Fe with them. It was nice to interact with humans. We went to the Georgia O’Keefe Museum and Meow Wolf. Both places were really interesting and fun. Hopefully next weekend is busy too. Overall, I am happy with my research and think this experience will be tremendously valuable.
Orientation
To begin, I would like to say that being accepted into the IRIS program was so extremely exciting for me and an incredible honor. Preparing for my internship and the orientation beforehand, I had extremely high hopes and tremendous anxiety.I arrived at the Albuquerque airport and found the group of interns who landed around the same time as me. Everyone was so kind and fun. The first night of the orientation we looked at our week's plans, learned about the program, and did ice-breaking activities. The other IRIS interns stole my heart immediately, and the weeks agernda looked amazing. I don't want to give away all of the goings-ons at the IRIS orientation, but it was a week full of hiking, eating, laughing, chatting in the parking lot late at night, and, most of all, learning in a really fun way. I came in so nervous and very uncertain of my abilities to be successful in my internship, but my fellow IRIS interns and all of the preparation I received at orientation made me feel more excited and probably more confident in my abilities than I have been in years. Once it was time for us all to head separate ways it was bittersweet, but I know we will all be in contact and I also knew then that I was ready to do my best at Los Alamos.
My First Week at Los Alamos
I arrived in Los Alamos via shuttle and was in awe of the beauty of the terrain. The elevation is something to get used to (a bunch of blood vessels actually burst in my eyes due to my elevated blood pressure), but the amount of vegetation is something I wasn't expecting to see in New Mexico. My host family is amazingly kind to me and treat me like family already, so I feel very safe and at home. I can't talk much about my project yet because with my location being a national lab, there are certain avenues I need to go down before releasing any work details, however, my work is open access, so I will eventually be able to share every detail. What matters this week is that my mentors Ellen and Andy are very fit for the job of mentoring, and, already, I feel they are going to help me get the most out of this experience. This week was mostly paper reading along with a small amount of analysis, but next week the project will really kick into gear. Living in Los Alamos, the town, is a bit lonesome, but with time I will hopefully pick up a new hobby or find my friend group. I am remaining opptimistic because I know I have all of the ingredients that make for a great summer, and even if this summer is workcentric, I really enjoy the work and am most of all here to learn and leave my mark.