After graduating from Kansas State University (May 2005) I have been actively involved in four different projects with two of them having a strong field component. 

The Cascades have been extensively studied in the past in order to answer questions pertaining to the evolution of the High Lava Plains and how it could fit into the timeframe of the Yellowstone hotspot track. This study provides new constraints on variations in crustal thickness and composition across the southern Cascadia Subduction zone.
                This study utilizes radial receiver functions determined from CASC array waveform data to evaluate in tandem both crustal thickness and Vp/Vs ratios. I use a grid search method where in I vary the crustal thickness and  Vp/Vs over a broad range to look for values producing the maximum stacking amplitude. This method is more reliable as it does not depend as heavily on crustal  velocity models.
               
The above figure shows the tectonic setting of the Pacific West and the location of the Cascadia Array (CASC). The map also shows the plate motion direction of the Juan de Fuca and Gorda plates as they subduct beneath the North American plate. Figure to the right shows the entire CASC array with boxed area representing the stations from the array that are used in this particular study. The stations to the right could not be used as we could not detect clear later arrivals such as negative PSmS phases.


Methods:

I compute ~600 robust radial receiver functions which are selected based on clear P arrivals. I use the Moho converted phase ( Figure below) PmS as well as crustal reverberations PPmS and PSmS.  I use the water level deconvolution method to compute individual radial receiver functions. An example of a good receiver function recorded at stations A13 is shown below.

             


Teleseismic events are extensively used for lithospheric studies because the wavefront is essentially planar at great distances. A teleseismic source is defined as that between the range of 30 to 90 degrees ( 3300 km to 10500 km ) away from the receiver [Burdick and Langston, 1977]. The seismograms recorded include the source-time function  of the source, near source reverberations and the effects of propagation and attenuation. In order to decipher the near receiver velocity structure of the crust we need to eliminate the common source function and instrument response, referred to as Source Equalization. This process enables clear observation of later arrivals which are conversions and reverberations from the underlying Moho.   
        At velocity interfaces beneath each station a phase conversion from P to S takes place, thus producing unique response on the horizontal and vertical components. Langston [1977] assumed that the vertical impulse response of the earth's structure is a delta funtion and thus the earth's structural response can be isolated by deconvolving the vertical component from the radial and transverse component. The resultant of this process is a seismogram having only the receiver effects and thus called as receiver functions (Figure above). The Figure to the below (left) shows the epicentral location of the 100 events used in this study which yielded clear receiver sunctions. The bulk of the dataset is derived from events from the NW and SE, but SW backazimuths are also well-sampled.




The derived receiver functions were observed visually and those exhibiting a substantial amount of P wave energy were selected for further evaluation. A reliable determination of Vp/Vs depends on the utilization of  PmS, PPmS and PSmS arrivals, as a near complete tradeoff is observed when only one is used (colored figure above to the right). I therefore used all 3 arrivals in this study, weighting arrivals as shown in the panel of the figure. Example receiver function from one station is shown in the figure above (center). The single red trace on top is the result of simple time domain summation ( without moveout correction) of the individual traces. Triangles are theoretical arrival times for PPmS and PSmS calculated using Zhu and Kanamori, 2000. Squares are arrival times calculated using higher Vp/Vs ratios (mafic values). Comparison of the plotted traingles and squares realtive to observed peaks suggests its proximity to the appropriate value.

                                               


I stack these high quality individual receiver functions at each station assuming moveouts based on candidate Moho depths and Vp/Vs values and perform a grid search over  crustal values from 30 km to 75km and Vp/Vs ratios from 1.65 to 1.85, representing an appropriate range of crustal compositions. I searched for combinations of Moho depth and Vp/Vs which resulted on the maximum stacking amplitude of the PmS phase.

Results and Discussion:

           



Preliminary results from this study indicate crustal thickness values ranging from ~35km to ~65km . Three stations within the Cascade Range exhibit crustal thickness calues ranging from 65 km to 70 km while stations within the Willamette Valley exhibit crustal thicknesses on the order of ~45 km. Preliminary results from this study also indicate Vp/Vs ratios ranging from 1.70 to 1.85. These Vp/Vs values are consistent with intermediate to felsic crustal compositions. The crustal thickness values recorded across the array are suggestive of a somewhat uneven Moho across the region. The inconsistency in crustal thickness values across the Cascade Range suggests that further work is required to evaluate the potential for very low velocity in the upper crust related to the volcanic plumbing system of this region, which would also give rise to overestimates of crustal thickness values.

Upper Mantle Seismic Stucture of the Southern Basin and Range and Colorado Plateau:


        The american Southwest has been shaped by tectonic activities concurrent to the plate dynamics that have been dominant in the West.  Though understanding the evolution and stabilization of the continental American Southwest has been an area of research for quite some time, there are many unanswered questions pertaining to the structure and composition of the crust and mantle underlying these regions. The regions encompassing the Basin and Range and Colorado Plateau have been extensively studied and various models driven by factors such as crustal thickenning, lower crustal flow, lithispheric delamination and many more have been proposed to explain the intruiguing juxtapositioning of a highly deformed Basin and Range to a fairly undeformed Colorado Plateau.
 

                                                                   

                                                                                                                       IMAGE : USGS..GOV
        Although most of the models proposed individually answer questions pertaining to evolution and stabilization of the tectonic blocks, there are no consences on why these contrating blcoks have behaved so differntly under similiar deformational conditions. To begin addressing these questions we need to locate areas within these terranes that exhibit seimic velocity anomalies which in turn would provide constraints on factors that have contributed to the regional crust.
        My current research using the two station group and phase velocity inversion technique (by which i examine the shear wave velocity structure of the lithosphere beneath) will enable us to produce a realible seismic velocity model for the area that can be used to correlate surface geology and deeper mantle structures.


Preliminary Analysis

I use the fundamental mode Rayleigh waves recorded at the USArray staions installed in and around Arizona to determine the phase and group velocities. In preparation for using the USArray data, i produced an inventory of all possible station combinations based on stations locations. The two station method is grossly dependent on the theory that two stations along a common (approximate) great circle path have difference in waveform mostly due to path effects of the earths structure between the two stations. A threshold of three degrees has been fixed  for the backazimuth between these stations in order to minimize the difference in paths between the earthquake epicenter and the station. Also a minimum station pair separation of 200 kms is required to have one complete cycle of the waveform as observed in studies by Aki and Richards [2002]. The estimated error is propotional to the signal to noise ratio as well as the wavelength to interstation ratio. Although there are a large number of station pairs only a few satify the requirements for the analysis.

              

The figure to the left shows station pairs that satisfy requirements for the study and lines between station pairs represents the great circle path between the stations. Events having a backazimuth within 3 degrees of the GCP of a station pair is used for further analysis. The figure to the right shows an example waveform for a selected event and it has been filtered with different bandpasses varying from 0.005 to 0.1 Hz. The figure below shows the great circel path between the events and the selected station paths used in this particular study.

                                                       

Ideally a representative sampling of one or two stations is made and a frequency time analysis is perfored using methods by Dziewonski and Landisman [1969]. This helps identify a range of frequencies and velocities for the events to be used and is an important quality control step (filtered waveform figure above).

               

A time shift calculation using waveform correlation is carried out on selected wavelets to obtain group and phase velocites for a particular frquency as shown in figures to the left top and below. The figure to the right (top) is the output for a MFILT run with period and group velocity (U) on the x and y-axes respectively. The text to the right of the plot gives event information. The distance (delta) and backazimuth between the station and the epicenter is shown as well as the event depth. In the plot the x's are computer picked energy maxima for each period and the vertical lines span +- 1 dB. The trend of the maxima is what one wants for a surface wave, and the contours indicate the data should be usable from about 15s period to about 100 s period.

Image source: miaki
                                       

      
    In the figure to the top-right small circles are phase velocities calculated from the observed waveforms for the interstation path between stations Y13A and Q11A for event 20061119. The vertical error bars for each phase velocity is based on the coherence of the two waveforms after the near station waveform has been time shifted to the far station epicentral distance using the calculated phase velocities. I carry out the above procedure on different events for the same station pair. A composite phase velocity value for selected periods for selected station pairs can then be calculated.

Inversion using Neighborhood Algorithm:

     In order to compute the S-wave velocity struture from the onserved phase velocities I intend using the Neighborhood Algorithm (NA). Previous studies were based on the linearized least square inversion (LLSI) to determine the S-wave velocity structure. LLSI requires a starting/refernce velocity model to be used for inversion and the final model should be close to the starting one for assumptions of linerization to be valid [Nguuri, 2004]. As there are no thorough velocity models for the region under investigation  and using IASP91 (Model) and PREM would be a thorough deviation from the pursose of this study.
    NA does not require the reference model to be close to the final model as there is no linearization in this method to invert for the velocity structure thus, NA method for non-linear inversion will be used to extract information from the ensemble of models that fit the dispersion data, in addition to finding a single best but model.

Expected coverage for the area based on predicted location or completion of the first phase of USArray installtions is shown in the plot below.

Results from the analysis will be posted as progress is made (07/31/07)

                                                                                                                                                                                                                                               
                                                                      

For Phased movement of the array click on this link (Movie)

Shaji K Nair shaji {dot}nair {at} asu {dot} edu School of Earth and Space Exploration Arizona State University Tempe, AZ 85281

©Shaji Nair