Anne's blog

Rayleigh-wave Group Velocity Tomography in the Vicinity of the Hawaiian Hotspot

Anne E Strader¹, Gabi Laske², John A Orcutt², Cecily J Wolfe⁴, John A Collins³, Sean C Solomon⁵, Robert S Detrick³, David Bercovici⁶, Erik H Hauri⁵

1. Bucknell University, Lewisburg, PA, USA.
2. Scripps Institution of Oceanography, La Jolla, CA, USA.
3. Woods Hole Oceanographic Institution, Woods Hole, MA, USA.
4. University of Hawaii at Manoa, Honolulu, HI, USA.
5. Carnegie Institute of Washington, Washington, DC, USA.
6. Yale University, New Haven, CT, USA. 

We present maps of long-period Rayleigh wave group velocity maps for the area spanned by the Hawaiian PLUME (Plume-Lithosphere Undersea Mantle Experiment) project. Specifically, we used observations from the second deployment of ocean-bottom and land broadband instruments that operated from April 2006 through May 2007. The recording network consisted of13 land stations with ten temporary and three observatory instruments and 38 ocean bottom sites that were equipped with 4-component broad-band instruments. With an average station spacing of approximately 200 km, this network had an aperture of nearly 1300 km.

Since the last blog, I've nearly finished my AGU poster, final report and the abstract, which will be submitted tomorrow. The last major step in the PLUME project (at least for the past 10 weeks) is error analysis. Because using group velocity to detect low-velocity anomalies within the PLUME network is a recent addition to the ongoing research, it is important to assess the effectiveness of the programs and parameters used throughout the process.

Minimum Error in Group Velocity Maps

When the path-averaged dispersion curves were calculated, there were several individual dispersion curves containing suspiciously inaccurate values with low error bars. This presents a problem in that velocities with lower error bars are given more weight when calculating the weighted group velocities displayed in these maps. Therefore, in order to observe the impact of faulty data not effectively accounted for by the program, it was necessary to generate another set of group velocity maps using a variety of minimum error bars. I tested 11 mHz, 26 mHz and 50 mHz to observe the effect of error bars on the full frequency spectrum. In order to see a significant difference in the group velocity maps for 26 mHz, I had to decrease the minimum error from 2.25 to 1.80, indicating that 2.25 was an appropriate minimum error bar to use. However, I only had to slightly alter the minimum error bars for 11 and 50 mHz to alter the corresponding group velocity maps. Increasing the minimum error bar resulted in defining the low-velocity anomaly northwest of Hawaii more clearly. The good news is that this supports that the anomaly is an actual structure. The bad news is that the anomaly was "smeared", or appeared over a larger portion of the network than was accurate.

Resolution Testing

For the past week, I've been determining how several factors contribute to the low-velocity anomalies I've noticed in the first-inversion maps. Although the presence of a low-velocity area can signify a magma plume or other geologic structure, velocities can also be heavily influenced by bathymetry and crust thickness. Additionally, the type of velocity model chosen within the model roughness/data misfit tradeoff curve has significant impact on the smoothness of the map, and may inaccurately amplify or subdue velocity anomalies.

Sensitivity Kernels

In order to understand how different frequencies are affected by these factors, it's important to keep in mind that higher frequency waves are more sensitive to shallow depths, while lower frequency waves are sensitive to deeper depths. Therefore, the frequencies that are most affected by a factor such as ocean depth or crust thickness depend on the depth of the layer that is altered in the velocity model. Here is a diagram of some sensitivity kernels (for phase velocity) which display p- and s-waves at various frequencies:

Bathymetric Mapping

I was recently notified by my advisor that, until some of the figures are actually in a publication, I can't post figures directly pertaining to my results online for everyone to see. That includes most of my figures from the last blog.

Basically, the PLUME project is controversial in that several scientists "don't believe in magma plumes" and will go so far as to reject papers that, with extensive research and recent technology, support their existence (aka the report I've been working on). Broadcasting results before they're reliably in a publication for this case would therefore be a bad idea. I'm not even allowed to discuss my results in too much detail (or at least prove them).

Figures I drew on AI, dispersion curves, velocity model maps (next blog), and stick maps are all ok because they're "wishy-washy" enough that one can't draw definite conclusions from looking at them. However, all inversion maps (the good stuff) had to be taken offline.

Sorry, readers. This stinks and the figures would help me explain some of the concepts and progress, but at least everything will be on the AGU poster in December. Also, hopefully there will be a successful publication as well for those who can't make it to the AGU presentation.

Also, I'm about to write a blog about the velocity modeling I've been doing for the past week which will be posted today.