Hunkering down for Isaac at Tulane University, I am finally realizing that my amazing summer of research is over. I have been able to learn so much from this summer. With the help of my wonderfully patient IRIS hosts, I have been able to grasp concepts that I thought would be impossible to learn after the first week in New Mexico. Becoming comfortable with both looking to papers for answers and learning something I had never heard of before is something I will be able to take with me into grad school or any job I choose to have after college.
I was especially appreciative of the internship because there is no geophysicist at Tulane and so I had never seriously been introduced to seismology. Because I enjoyed the internship so much, I plan to continue to do research in the same area through an independent study here at Tulane. I will continue my project by calculating synthetic receiver functions to further investigate a change in the amplitude ratio of the LAB to the Moho over different frequency ranges and then analyzing receiver functions for different geologic settings (Craton, orogenic belt, rift, ocean island). I am very lucky to have such supportive hosts that will help me throughout the semester.
Along with continuing the project this next semester, I will be applying to grad school for seismology. I would have never found that I enjoyed seismology so much had it not been for the IRIS internship and I am excited to see everyone in December at AGU!
Last week I was able to finally finish common conversion point stacking and get some really need figures. I was hoping to be able to only use the TA stations when creating these, but I found including the permanent stations from the New Madrid Seismic Zone as well as from USArray and IU stations has really helped the CCP stacking shown below.
Also last week I had the opportunity to visit a TA site installation. The amount of time and equipment that go into each TA station is amazing! It made me really appreciate of the stations I had to use ( about 50) and how much time went into getting them up and running.
As this is the final week of my internship, it is time to reflect on some of the goals that I had set back in the first week. My Goals were
1. learn how to use software to create graphs and visual representations of my research. - I have not needed to use GMT, which I thought I might need at the beginning of the summer. I have been able to create some figures with the matlab codes I was given. I think that I will continue to learn more about this as I begin to create my AGU poster.
2. produce a well written abstract to submit to AGU- with much editing help from my advisors, I was able to submit an abstract just in time.
3. read at least one paper every week that has to do with my research or any interesting research being done in seismology- I have read at least one paper a week and sometimes more! I think reading papers was the best way to find information on similar research that has been done. I was happy to discover that the more papers I read, the easier it got.
4. become comfortable with both water level and iterative deconvolution and be able to discuss it in an academic environment. - I believe that I have become relatively comfortable with many types of deconvolution, I hope to continue to learn how to talk about my research in an academic environment. A big test of if this has been accomplished will be at AGU.
5. try and visit all of the Smithsonian museums and become comfortable using the metro- Considering the number of museums in DC, I think I have done pretty well. I found that I kept going back to the natural history museum to look at the rocks and minerals and of course the seismology section but I was able to visit most of the museums. I have also become really comfortable with public transportation in DC and I take the metro everyday!
Overall the internship has been an amazing opportunity and has made me really consider continuing seismology in grad school.
This past week began a little disappointing as I ran the CCP stacking code and which produced an image where it was hard to pick out the Moho and LAB. The first time I ran the code I only used TA stations which have not been in the area for a long time. Even with the "ideal parameters" I tried to find in the first part of the project, some of the TA stations on unconsolidated sediment had only 10-15 Ps and Sp paths that could be used in calculating the receiver function. In order to try and improve the stacks, I am now working on adding some permanent New Madrid, US, and IU network stations that are nearby in the hopes of filling out the stack and hopefully making them easier to interpret.
Other than that this week I worked on writing my abstract which I found difficult especially since I do not have great CCP stacks to interpret. I also found it difficult to summarize all the work I've been doing. At first I wanted to write every detail about everything I've learned, but after a few hundred words realized that I needed to summarize and pull out the most important points or no one would want to read the whole thing.
To close out the work week I went to a delicious restaurant called Tonic about two blocks from my apartment for the summer and ....
WOAH! I unexpectedly ran into Mya!! We were both a little flustered and surprised to see each other but I was really happy to get to see a fellow IRIS intern way before AGU in December. Apparently DC is the place to be!
This coming week I hope to actually submit my abstract and really get some good CCP stacks that I feel would be worthy of posting to the blog. This coming week I hope to observe a TA station being put in with some people from IRIS HQ. It would be exciting to have a trip to the field, even if its just for a day!
Last week was spent finishing up more single receiver function parameter changes and started writing about each specific case. It was helpful to have some time to actually sit back and think about what all of the work and calculations I've done really mean. It was also helpful to think about the specific geology of each of the areas. In addition to the Mississippi Embayment/Realfoot rift area, I had been hoping to have time to look at CCP stacks of the area under the Mid-Continent rift. To at least start this I have created a list of TA and permanent stations that I will use this week to request data and hopefully have time to get to before the AGU abstracts are due. I was able to also start preparing data for CCP stacking which I have been really excited to get to. A new code was just given to me today that would make the CCP stacks better than using the iterative time domain calculations that I had been preparing to use. The new method is called the Extended-Time Multi taper Receiver Function Estimation Technique and was created by George Helffrich (2006) and improved by Takuo Shibutani, Tomotake Ueno, and Kazuro Hirahara (2008).
This week I hope to look into what this code actually is doing by reading these papers and talking to my host about the code. While I figure out what this code is doing I hope I will be able to prepare all of the data and get CCP stacks that are interpretable. I am a little nervous that the sediment layer on the Mississippi embayment be a huge problem in interpreting the data despite the large number of stations. All I can do for now is hope that it will all work out well.
This week is also the last full week before AGU abstracts are due! This means that in addition to finishing everything I can I also will have to start writing about what I can present at AGU. This will be the first abstract I am writing so I am a little nervous but also excited!
Until next week!
Scarlet fever unfortunately made the last week a little shorter than expected and I was not able to get as much done as I was hoping. After a ride to the doctor and some antibiotics, I am ready to start playing a little bit of catch-up this week as the AGU abstract deadline is getting uncomfortably close! Before getting sick, I was able to go to a DC summer classic "screen on the green" event showing Butch Cassidy and the Sundance Kid next to the Smithsonian!
Related to my project, I was able to use illustrator to create some graphs of the parameters I have been talking about, and this week I hope to really get cranking on CCP stacking and analyzing it for the New Madrid area. I hope to have a bit more to share at the end of this week !
I spent this week running the matlab code for receiver function parameters. It has taken me much longer than predicted. Unfortunately, I have been having problems creating illustrator plots, but resolving this problem will be part of my goals for next week. While the codes have been running I have taken the time to write up the general goal of this part of my project as well as what changing each parameter does. The basic idea is that to get good receiver functions that can be most easily interpreted requires knowledge of the parameters that can be changed and how changing each parameter will affect the data. When changing parameters the goal is to minimize the noise without changing the integrity of the deconvolution.
For the parameter changes, only the parameters being tested will change and all other variables will remain as constants. There have been three stations chosen to undergo parameter changes. The first is CBKS, which is a permanent US station located in Cedar Bluff, Kansas. This station has been active for 17 years and therefore has a lot of data to offer. Underneath the station lies basement geology of consolidated sediments. The second station is MPH from the New Madrid (NM) seismic network and is located in Memphis, Tennessee. This station has been collecting data since 1998, and unlike the US station CBKS it lays on unconsolidated sediments along the Realfoot rift. The final station is V42A of the Transportable Array (TA) network and has only been around since 2011, so it has not collected much data. Like the NM station MPH, it also lies on unconsolidated sediments around the Realfoot rift.
The goal is to compare each station with itself as different parameters are changed as well as to compare the receiver functions of each station against one another. Things to consider about the stations when they are being compared to each other include the length of time the station has been active and the geology underneath each station. After all of this has been analyzed, the next task is to determine the minimum magnitude event that can still provide a significant number of viable seismograms for each station.
Like Greg mentioned in his blog, I’m really glad that there are so many good papers on receiver functions. Although I was getting bored of reading papers at the beginning, it has really helped to gain a background in receiver functions and it’s exciting to know exactly where to look when you aren’t sure about something. On that note, I’m going to take the rest of this wonderful Sunday afternoon to go read about the crustal structure along the North Anatolian Fault zone with Albert down the street!
Al's such a great listener!
Until Next Week !
I hope everybody enjoyed a short week due to the fourth of July and was able to see some awesome fireworks!
During this past week I have moved back to IRIS headquarters for awhile as my host at Maryland travels around Europe (not all for fun and games, apparently some seismology is involved as well). I spent the beginning of the week updating the receiver function codes on the computer at IRIS so that they matched changes I made in the codes at the University of Maryland. Much of the rest of this week has been spent working and starting a number of small projects. A few that I started working on are creating a KML file of the TA stations I will be using and deciphering CCP stacking code I will be using on the TA stations in the future.
I felt as I was starting too many new things before I was finishing anything so at the very end of this week I have decided to finish up single receiver functions for the permanent stations. Throughout the process of calculating receiver functions, I have had to make decisions on what the parameters should be and have spent time figuring out which parameters will work best for my data. Some of these parameters include what the signal to noise ratio is, what frequency range is included, what the minimum allowed frequency (or water level) is , what the Gaussian is that is used in the deconvolution, and whether to use the time or frequency domain in the deconvolution. I wanted to wrap up this part of my project by producing some final figures to outline the effects of changing each parameter and write a small paragraph explaining what changing each parameter does. I hope to add this as a small part of my end project. My goal for next week will be finishing all of this up and having some great (hopefully poster ready) figures.
As for the weekend, I plan to attend a free jazz in the garden concert at the sculpture garden just a few blocks from IRIS headquarters and to also just stay out of the heat as much as possible ( its supposed to be 105 F tomorrow ! )
Happy forth of July from the nation's capital! The title of this weeks blog is not only for the fourth for overcoming the first problems I had encountered with the matlab code during the second week of the project where the azimuth was being calculated incorrectly. Luckily, since then I have had no major problems figuring out the code and running it for other stations.This week as been challenging as I have run the same calculations many times, but I am glad to finally be done (for now) with calculating the receiver functions for the permanent stations. I am working on writing a KML file for Google earth that will show where all the stations are located and have some basic information about each one that I hope to be able to add to one of my next few blogs!
After calculating all of the permanent receiver functions I was happy to to have some time to look through them . I was able to see that the New Madrid seismic zone stations produced receiver functions that were much harder to interpret. To get better results I plan to request data of earthquakes down to 5.6 instead of 6.4 For the other stations I ran an updated phase weighted stacking code to replace H-k stacking. The results on the good stations were about the same, but for some of the less clear stations, the phase weighted stacking made a great improvement and the moho depth and Vp/Vs were much improved. Also, this week I picked out TA stations from around the Mississippi Embayment/Reelfoot rift as well as around the mid continental rift which I will hopefully use later to create a CCP stack of the data.
Outside of lab I was able to head to a beach in Maryland this weekend and enjoy some fireworks and friends. I was able to have mussels and crab for the first time and was supervised to find that they were delicious and fun to eat! Although the beach was great I'm ready to turn in my flip flops and towel and get back to working on receiver functions.
It is officially summer weather in Washington DC! What started off as a rainy week finished up today with a temperature of about 94 degrees and 50% humidity. I never thought that DC would be hotter than New Orleans (which only reached 89 degrees today). Although it made walking to work miserable, at least the heat has encouraged me to stay indoors and work on my project.
Last week I had just calculated receiver functions for 5 IU stations after some difficulty with the matlab code. I was hoping to calculate some of the US stations receiver functions in both the time and frequency domain as well as some NM seismic stations receiver functions and have time to really look over and compare them all. Unfortunately, I have only made it up to the NM stations and so I have not been able to look at how the differing basement geology effects the quality of the receiver functions. However, I have made some progress on something that I wasn't expecting to work on this week.
I had previously read a paper by Lupei Zhu and Hiroo Kanamori entitled, " Moho depth variation in southern California from teleseismic receiver functions" where they were able to laterally map out the Moho under southern California using a method called H-k stacking. H-k stacking involves performing a modified frequency deconvoluion to produce a radial receiver function. It is possible to determine the arrival times of the Ps, PpPs, and the PpSs+PsPs waves from that reviewer function. Taking those arrival times, you weight them in an H-k stacking equation given in the paper to create a graph which will show you where the maximum stacking occurs. This point determines what the values of H (crustal thickness) and k (Vp/Vs) are. The resultant graph looks something like :
From the graphs we can see that the most weighted part occurs at around 45 for H (Moho Depth) and 1.8 for k (Vp/Vs).
A problem with this method is that sometimes the maximum is smudged or there are two maxima that occur. One way to get rid of this problem would be to use a phase weighted stack. The phase weighted stack will look at the coherency of all of the peaks when stacking them. It provides an assigned phase weight to each peak based on how it closely it aligns with other receiver functions. This prevents large peaks that are incoherent to have no effect on the resultant data. Hopefully, I will be able to use this system to get better H-k stacking results for some of the US and NM stations.
In the upcoming week I hope to finish calculating reviewer functions for the IU, US and NM stations and have some time to really sit back and compare them. Also, I hope the weather will cool off as DC begins to prepare for the nation's 236 th birthday!
This week I jumped onto the blue and orange line, changed to the red line, and took the green line to the second to last stop to arrive on the beautiful campus of the University of Maryland College Park. I spent this second week working on my project with Dr. Veda picture and a few helpful undergraduate researchers. Below is a picture of our lab group!
While at the University I was also able to attend a masters thesis defense by Sara Peek about the, "Geochemistry and geochronology of late Ediacaran and early Cambrian carbonate rocks in Siberia and South China". It was helpful to see the results of a masters degree and what can be accomplished in grad school.
Last week I was very excited to have produced some graphs using a receiver function matlab code that I was given. At the beginning of this week, we noticed some problems with the resulting receiver functions as well as some of the seismograms. After digging around the code, we were able to find the source of the problem. For all of the data we had requested, the longitude values were represented as positive numbers when they were actually negative. This meant that I had to re-request the data and start from scratch. The mistake actually turned into a positive thing as it made me learn how to look at my data in more detail and has helped me produce better looking receiver functions.
After finishing the receiver functions again, we found another error. The azimuth that was calculated in the code was not calculated correctly and so after the code was fixed, I will have to re-prep and re-calculate the data to get even better looking receiver functions. Below is the first example of a receiver function that is the most correct so far. It is located in Waverly, Tennessee which is on top of consolidated sediments .
I was satisfied to find that the moho right above 50km resembles that in an undergraduate senior thesis called, "Using the FLED array to determine crustal parameters" by Ellen Syracuse
Next week I hope to finish up the IU stations, and become more familiar with the matlab code I am working on so that I can be better prepared for any future errors and so I can manipulate it myself without the fear of making the code crash. I also hope to look at some stations from the US national seismic network and see how the receiver functions compare.
Until next week !
The IRIS internship started out with a wonderful orientation week in Socorro, New Mexico. It was great to meet all of the other interns, do a little bit of field work, and have lectures ranging from grad school to inverse theory. I never expected that I would be able to meet so many great people and learn so much about seismology in such a short period of time. I can't wait to see the results of all of my fellow interns work in December at AGU!
After orientation week, I arrived in Washington DC to begin working on my project. On Monday, I walked past my new neighbor at the the White House on my way to IRIS headquarters! The first few days I spent getting to know my environment and learning more about my project. I was able to meet both my hosts and get lunch from a food truck (which is apparently quite popular in Washington DC). I also visited the Natural History Museum which has a small section on seismology. Along with much help from my hosts and papers I have been reading I have been able to gain a better understanding of the methods I will be using to calculate receiver functions this summer.
Later in the week I requested data for some IU stations through the IRIS data management center. I have been using some UNIX programming to organize my data and using a SAC program to go through the data to do some 'quality control' , removing noisy data. I plan to use this data in a matlab program that will help me understand receiver function graphs in both the time and frequency domain.
A few goals I have for this summer would include the following:
I would like to learn how to use software to create graphs and visual representations of my research as well as produce a well written abstract to submit to AGU. I hope to also read at least one paper every week that has to do with my research or any interesting research being done in seismology to increase my appreciation for seismology. I hope to become comfortable with both water level and iterative deconvolution and be able to discuss it in an academic environment. I would also like to try and visit all of the Smithsonian museums and become comfortable using the metro.
Next week I will hop on the metro and head over to the University of Maryland to work with my other host and hopefully gain some more insight into receiver functions as I look at more and more data.