Katherine Anarde

August 31st, 2010

A Joint Rayleigh and Love Wave Analysis for the Hawaiian PLUME Project

During the two-phase Hawaiian PLUME (Plume-Lithosphere Undersea Mantle Experiment) seismic deployment from January 2005 through June 2007, we collected continuous seismic data at thirteen (ten temporary and three observatory) land stations and nearly 70 ocean bottom sites. Most of these sites were occupied with broad-band 3-component seismometers (Laske et al., 2009). This provides the ideal basis to analyze both Rayleigh and Love waves across a broad frequency band, between 10 and 50 mHz. Our analysis explores the radially anisotropic shear velocity structure of the region through comparison of long-period teleseismic Rayleigh and Love waves recorded from phase 2 of the PLUME project which spans the time interval of May 2006 through June 2007.  The collection of earthquakes suitable for a fundamental mode surface wave analysis consists of 190 shallow events.

Using the two-station method, we determine the path-averaged phase velocity for over 60 paths from 1322 Rayleigh wave and 749 Love wave phase measurements. Most of the high quality Love wave data collected exists between 2 land station paths: BIG2-KIP and KIP-POHA. Here, we present a detailed comparison of the vertical shear velocity, $V_{SV}$, and the horizontal shear velocity, $V_{SH}$, between these 2 two-station paths. From the path-averaged dispersion curves, we use the weighted average, where the weight is determined by individual error bars, to perform 1D inversions. We determine the percentage radial anisotropy as $((V_{SH}-V_{SV})/V_{SV})\times100$, as a function of depth.

In order to distinguish between effects caused by mantle anomalies and those caused by variations in bathymetry and crustal thickness, we explore the changes in phase velocity through forward modeling. Compared to predictions for 52-100 Ma old lithosphere, our dispersion data reveals high phase velocities for Love waves and relatively low phase velocities for Rayleigh waves, along both station paths. This suggests that $V_{SH}$ in the lithosphere and possibly the asthenosphere must be higher than $V_{SV}$. Our inversions for shear velocities as functions of depth confirm this. We observe radial anisotropy mainly in the lithosphere to be up to 10\%. We infer from this rather high value that radial anisotropy beneath the island chain reflects mostly horizontal fabric and flow in the shallow mantle.

References:
Laske, G., Collins, J.A., Wolfe, C.J., Solomon, S.C., Detrick, R.S., Orcutt, J.A., Bercovici, D. and Hauri, E.H., 2009. Probing The Hawaiian Hot Spot With New Ocean Bottom Instruments, EOS Trans. AGU, 90, 362-363.

URL:
[url=http://igppweb.ucsd.edu/~gabi/plume.html]http://igppweb.ucsd.edu/~gabi/plume.html[/url]
 

August 9th, 2010

As part of my summer internship I was able to take a tour of the IGPP OBS lab. Like the construction of land seismometers at PASSCAL, the OBS lab technicians start out with an intact sensor ball, a Trillium 240. Within the sensor ball lies a spring system with X, Y and Z components that sense perturbations. Ironically this massive ball does not store any of the data it collects nor does it contain its’ own power source. These are located in the lager and battery bottles. The lager stores both the memory (through a simple flash drive) and the clock on memory boards as well as serves as a battery source. Active source seismometers can suffice from the power stored in the lagers (short deployment) where as passive experiments require additional battery bottles. The clock used in the SIO OBSs, called a c-scan, has the precision of 50 parts per billion, meaning it loses only ~1.5 seconds/year.The technicians use a Gimbel system for quality control to level the sensor ball (precautionary measure as the sensor ball contains an additional level). The Gimbel system also serves as a skeleton for the sensor ball to hang from a frame. An SIO OBS frame houses several features. The yellow container holds four glass balls used for flotation in device retrieval. The lager, battery bottle and a pressure sensor sit below the glass balls. The frame is connected on the bottom to an anchor the weighs the system to the ocean floor. When an OBS is ready for retrieval, an acoustics system responds to an individualized sequence of pins at a particular frequency and burns off a link that holds the system together. Malfunctions in the acoustic and flotation system can cause device loss (a particular problem in the PLUME deployment).

This yellow lager can weigh
up to 90 pounds when full!

A technician working on a Trillium 240

An SIO OBS frame (stored at IGPP)...must turn your head to the right 😊

July 23rd, 2010

It's true.  My advisor is leaving me to the domestic beasts of the barnyard (my lab).  Joking aside, these next three weeks I'm on my own to finish up the final 2D tomographic inversions, GMT plots and my abstract and poster.  So much to do in so little time.  This past week has been chaotic, filled with crash courses on GMT, inversions and such.   Here's a glimpse of my journal (w/o figures because I can't figure out how to upload eps files):

Through comparing the averaged dispersion curves of both Rayleigh and Love waves I attempted to distinguish areas of radial anisotropy. As evident through Figure 20, the region between stations 84-85 (BIG2 and KIP) exhibits high phase velocities for Love waves and relatively low phase velocities for Rayleigh waves (in comparison to the predictions of age-dependent lithosphere).  This relationship is indicative of horizontal mantle flow: VSH>VSV. Station path PL85-PL86 (KIP-POHA) indicates the same relationship with high Love wave phase velocities and relatively low Rayleigh wave phase velocities. Other explanations for these velocity perturbations include a change in water column depth and/or crustal thickness. Both of these possibilities were explored through forward modeling in section 3A and 3B, respectively. Forward modeling indicates that changing the water column depth does not affect Love waves and only affects Rayleigh waves at short periods. Variations in crustal thickness cause minimal perturbations for both Love and Rayleigh waves and again only affect the Rayleigh dispersion curves at short periods. As such, the observed dispersion curves in Figures 20 and 21 cannot be explained solely through variations in water column depth or crustal thickness and must be interpreted as the result of shear wave velocity perturbations. In addition, the magnitude of the offset for each dispersion curve at ALL periods suggests that the velocity anomaly must be in the lithosphere and perhaps extends into the asthenosphere. This is particularly evident for both Rayleigh dispersion curves (when compared to forward models in section 3C).

This would make a lot more sense with pictures.  I'd really like to upload my GMT stations plot as well.  Nothing seems to be working...



 

July 19th, 2010

Since my last blog I have flown through a TON of data processing. As I stated earlier I've been keeping a research journal. Here's a selection from this past week (not all figures not included):

The first step in data analysis involved measuring the relative phase of both Rayleigh and Love waves against a reference station for the land, SIO and WHOI data. The program Intdisp enabled me to align the phase of the transfer functions from each event per station and isolate it for further processing. After completing all the measurements I found that less than ~5% of the OBS data (SIO and WHOI) could be recovered for the next step due to noise. As such I decided to focus on the measurements from the land stations which were less noisy and provided more accurate measurements. Figures 18a-c juxtapose Love wave measurements from a land station, SIO and WHOI data from an event on day 219 in 2006.

From the phase measurements I then used the program unwrapn to convert to phase velocity perturbation. Through unwrapn I corrected each event for 2π problems that could affect the accuracy of the dispersion curves.

In the final stage of data preparation I used several programs to create dispersion curves for both Love and Rayleigh waves between all possible station path combinations. The program run.newlleg first calls an fcell with all the events and sorts which events are within 10º of the specified station path (ex. 75-76). Then the program run.events calculates all of the dispersion curves and provides the azimuth, distance from the great circle and the correction factor for each event per station path. The correction factor is particularly important because it shows the correction for off-great circle propogation, meaning that all earthquakes do not actually approach the station along the two station great circle. The final program, run.avepha, averages all of the dispersion curves for each path and plots 2 new curves: a simple and weighted average dispersion curve (Figure 19).

Figure 18a. A Love wave phase measurement using Intdisp from the land station POHA. The wave in the middle window is from the reference station BIG2.

I am currently working on the Love wave processing which has proven far more difficult than the Rayleigh waves. This will likely take me another day to complete and then I will start mapping. I also will be taking a tour of the Scripps OBS lab later today so more to come tomorrow...

July 7th, 2010

I JUST WITNESSED MY FIRST EARTHQUAKE!!!!  Unbelievable.

At first it was an erie feeling.  I didn't realize until about 5 seconds into it that it was an earthquake.  I was sitting in my computer lab and I looked up out the window and the window beams were shaking.  Truly bizarre for this inlander from Colorado.  For the past half hour the lab has been patroling the USGS site.  So far we know it was a magnitude 5.4 from the San Jacinto fault which is somewhere NE of here.  My dad felt it in LA so I'm guessing some of you guys father north also felt it.  What a fantastic experience!

July 6th, 2010

I can't believe it's July.  I only have a little over a month left here and I'm feeling VERY behind!  I haven't even started working with actual data.  Hopefully I will get to that by tomorrow.  BUT, on a good note I am completely done with synthetic and forward modeling.  We found from the synthetic modeling that the kernels of the Love waves look too similar to the reference models so it's nearly impossible to figure out where an anomaly is located at depth.  As such, we now have a new focus for my project which seems more reasonable with the minimal time I have left here.  Below is an introduction from my journal:

The goal of this project is to quantitatively explore the radially anisotropic shear velocity structure of the region surrounding Hawaii from the analysis of Rayleigh and Love wave phase velocity data of the PLUME deployment.  From previous experimentation, we know that there is an anisotropic anomaly located in the central Pacific plate (Ekstrom and Dziewonski (1998)).  While these studies have used global isotropic S-wave velocity models to compare VSV and VSH at depth, we hope to more accurately distinguish the flow pattern of the upper mantle around Hawaii through two-step 3D inversions of Raleigh and Love wave dispersion measurements. In support of the plume model, we expect to find a change in anisotropy under Hawaii indicative of vertical flow (VSH<VSV).

More to come soon!

June 25th, 2010

This past week was far slower than the last.  I've been practicing phase velocity measurements on Rayleigh wave data to get acquainted with the programs/method.  Meanwhile, Gabi has been busy putting the final touches on the programs needed for inverse modeling of the PLUME love wave data.  Yesterday she showed me how to make synthetic models of Rayleigh wave data, similar to the forward modeling I did last week.  In this process I call a program to produce a synthetic model from a "tweaked" grid (i.e. I pick where and how large an anomaly should be).  From this model the program calculates dispersion curves which I further invert at the 24 iteration.  Another program allows me to play around with the tradeoff between the model roughness and X2, looking at iterations from 1-40.  I can also look at how frequency effects the "accuracy" of the velocity model.  Overall this process will give us a better idea of what quality of data to expect as proceed with the true data inversions.  Next week I should be able to continue the synthetic modeling on love waves and then move on into actual inversions the following week. 

The world cup has dominated my free time/lunch break.  I'm either at the pub, at the brewery or sleeping.  It's a great life 😊

June 18th, 2010

This week has been a busy week of forward modeling.  After lots of reading on surface waves, particularly love waves, I was finally able to get to work on my summer project.  Gabi and I decided that instead of looking at microseisms from the PLUME data, I would investigate love wave tomography with the same data set.  Tomography is more compatible with my interests and seems like a better and less complicated starting point for a beginning seismologist 😊  Analysis of the vertical components from the PLUME deployment has been completed so the Love wave analysis will HOPEFULLY provide more images to compliment or inhance the existing models.  After finishing my last forward model yesterday looking at variations in shear velocity at depth intervals, Gabi seemed a little apprehensive about starting the inverse modeling.  It's looking like the data might be difficult.  Either way, I'm excited to start some preliminary inverse modeling next week!  

As for my goals for the summer, it seems like we're taking one week at a time.  My personal goal for the summer is to complete all of my modeling by the end of July so I can use August to work on my AGU poster and abstract.  Throughout the summer I'm compiling a scientific journal of all of my figures and models along with introductions, analysis and commentary so by the end of July I should have a TON of material to use for my abstract.  Although the journal is tedious I really think it is going to pay off.  Plus I've found that writing everything down helps me understand it better!

AND as for the my free time, WHO ELSE THINKS THE REF IN THE SLOVENIA GAME NEEDS GLASSES?!?!?!?!?!?  Seriously, I'm fuming.  Just saying, should have been 3-2 USA. 

June 12th, 2010

So far San Diego has been pretty uneventful.  This past week I only went to campus twice because Gabi was out of town at the IRIS conference.  She left me with some light reading on the project.  Actually, we didn't decide until Friday what project I'm going to be pursuing this summer.  We were debating between a tomography or microseism project using her PLUME data from Hawaii. Ended up choosing the tomography project using love waves.  More info to come!