Imaging the Icelandic plume - what are the velocity perturbations?

Richard M Allen, Michael M Deal, Guust Nolet

Dept of Geological and Geophysical Sciences, Princeton University, Guyot Hall, Princeton, NJ 08544

Abstract submitted to AGU Spring Meeting, May 1996.

___________________________________________________________________________

A preliminary teleseismic tomography study of the Icelandic Plume has been conducted using data from the permanent SIL network. The inferred P-wave velocity model indicates the plume is not vertical but instead dipping to the northwest.

A total of 344 high quality arrivals were used from 31 events recorded on 19 stations. The residuals are obtained using the cross-correlation technique proposed by VanDecar and Crosson. These were used to constrain velocity perturbations in the model box which has a surface area of 1000 x 1000 km and extends down to 500 km depth. Station corrections, event corrections and event re-locations are also considered as model parameters. Use of the cross-correlation technique allows an accurate estimate of the variance of each residual which are used to create the data covariance matrix. The model covariance matrix is constructed a priori taking into consideration the variable variance of the different model parameter types. The linear system is inverted using the LSQR algorithm allowing variable damping for each model parameter type.

Our analysis shows that the arrivals are delayed relative to the background model by an average of 3.1 sec. This is due to the fact that the stations are all on top of the plume head and all sample some of the low velocity plume. The inversion techniques currently in use assume the arrivals are distributed around zero delay and thus cannot account for this bias. If we are to obtain correct P-wave velocity perturbations this must be accounted for and we demonstrate how this can be done.

The model shows many similarities with that of Tryggvason (1983); a low velocity zone (lvz) extending from 100 km to 450 km depth below which we loose resolution. The lvz is 100-150 km in diameter at 100 km and broadens out to a region ~150 km by ~300 km at depth. The column is not vertical but instead dipping to the northwest.

_______________________________________

© Richard M Allen