Large igneous province locations and their connections with the core–mantle boundary - Volcanism and Global Environmental Change - page 38

on the tomography model used, and in most of our papers the PGZs have been
approximated by the 1% slow SMEAN tomographic contour near the CMB
(Torsvik et al ., 2006 ). The SMEAN model is an average of three shear-wave
velocity anomaly models but we have also examined numerous other models (e.g.
Torsvik et al ., 2008b ). They all provide broadly similar characteristics near the
CMB; thus, the choice of tomographic model is not critical, although different
models lead to slightly different statistical correlations. We have also compared
whole-mantle shear-wave velocity models with S-SKS models (Burke et al ., 2008 ,
Figs. 3 and 4), globally at the CMB, and have found that the 1% slow (negative)
contour of the SMEAN model compares with the 0.96% slow contour in the Castle
et al .( 2000 ) model ( Figure 3.2a ), and the 0.77% slow contour in the Kuo et al .
( 2000 ) model.
The PGZs were similarly located using shear-wave velocity cluster analysis by
Lekic et al . (2010) who produced a ' voting ' map, which described footprints on the

CMB for five global tomographic models. They determined whether a geograph-

ical data point was located in a seismically slower-than-average region. The data
were contoured in such a way that a contour score of 5 stated that all the models
agreed about the de rms

that the Siberian Traps originate from the PGZ of an LSVP (Burke et al ., 2008 ),
dubbed Perm by Lekic et al . (2010). This smaller anomaly is also discernible in the
SMEAN model (

nition of a data point. The voting map ( Figure 3.2b ) con

0.5% slow in Figure 3.2a ).
Below, we reconstruct 30 LIPs ( Table 3.1 ) in different ways and compare our
results with SMEAN, S-SKS models and the seismic voting-map contours.

-

3.4.1 Mantle reference frame
Using a hybrid reference frame, i.e. a moving hotspot frame back to 125 Ma and a
TPW-corrected palaeomagnetic frame before that time, we

nd that the recon-
structed LIPs plot on average 7.9 from the 1% slow SMEAN contour ( Table 3.2 ,
Figure 3.3a ); however, there are notably large deviations ( Figure 3.3b ) for the
Columbia River Basalt ( c .39 ), North Atlantic Igneous Province (NAIP, c .19 ),
the Alpha Ridge (or High Arctic LIP, c .34 ) and the Siberian Traps ( c .36 ). The
voting-map contours (C1 C5 in Figure 3.3 ) reduce this discrepancy considerably

because they extend the African LLSVP further northwards as well as identifying
the Perm anomaly. Seismic contour 5 provides a very good

-

fit. LIPs plot on
average 5.4 from that contour ( Table 3.2 , Figure 3.3a ). The best fit is yielded

by the Castle contour (4.4 ) because it also includes a small anomaly (LSVP) at the
CMB below the Columbia River Basalts ( Figure 3.2a ) that has not yet been
identi ed in global shear-wave tomographic models. The overall similarity of the

several seismologically de ned PGZs tested is remarkable as is made clear in

Figure 3.3a and Table 3.2 .