Auxiliary Material for paper 2006JB004608 Global mantle flow and the development of seismic anisotropy: Differences between the oceanic and continental upper mantle Clinton P. Conrad Department of Earth and Planetary Sciences, Johns Hopkins University, Baltimore, Maryland, USA Mark D. Behn Department of Geology and Geophysics, Woods Hole Oceanographic Institution, Woods Hole, Massachusetts, USA Paul G. Silver Department of Terrestrial Magnetism, Carnegie Institution of Washington, Washington, D. C., USA Conrad, C. P., M. D. Behn, and P. G. Silver (2007), Global mantle flow and the development of seismic anisotropy: Differences between the oceanic and continental upper mantle, J. Geophys. Res., 112, B07317, doi:10.1029/2006JB004608. Introduction Figures S1 and S2 compare predictions of ISA and PI (e.g., Fig. 5 in the main text), which were made using a model resolution of 157 km in the horizontal direction and 25 km in the vertical direction above 350 km, to models that use both higher and lower resolutions. Fig. S1A shows results for a coarser resolution of 210 km by 33 km while Fig. S1B shows results for a finer resolution of 105 km by 17 km. Basic patterns of predicted ISA and PI are similar regardless of resolution. This is confirmed by the detailed comparisons of PI and ISA between models of varying resolutions (Fig. S2). 2006jb004608-fs01.eps Figure S1. Comparison of the Infinite Strain Axis (ISA, bars) and the Grain-Orientation Lag (PI, colors) for two different model resolutions throughout the lithosphere and asthenosphere: shown are a "low" resolution calculation (A and C, with 210 km horizontal and 33 km vertical resolution) and a “fine” resolution calculation that halves the spacing of nodes (B and D, with 105 km horizontal and 17 km vertical resolution). For comparison, this paper employs a model with 157 km horizontal and 25 km vertical resolution, which is intermediate between the two resolutions shown here. Both models use laterally-varying viscosity and a combination of plate-driven and density-driven flows (beta=0.5), and thus can be compared to the calculations shown in Figs. 5A and 5B, except that the planforms shown here (A and B) are plotted for 200 km depth (instead of 225 km in Fig. 5A) because that depth is common to both resolutions. Both planforms (A and B) plot the ISA approximately every 840 km (every 4th node for A and 8th node for B). Differences in resolution are apparent in the vertical dimension of the cross sections (C and D, and Fig. 5B), but in the horizontal dimension we interpolate the output of both models to 300 km spacing because the nodes do not generally fall along the cross section trace. The similarity between the output for both resolutions, and to the output for Fig. 5, shows that the basic spatial patterns of both PI and ISA are robust and independent of the resolution used. 2006jb004608-fs02.eps Figure S2. The difference in azimuth of the ISA direction (A and C) and the base 10 log of the ratio of the PI values (B and D) for comparisons between the 200 km model results at different resolutions. A and B show a comparison between the "intermediate" resolution calculation used in this paper (157 km horizontal and 25 km vertical, e.g., Fig. 5A) and the "low" resolution calculation of Fig. S1A (210 km horizontal and 33 km vertical). C and D compare the intermediate to the “fine” resolution calculation of S1B (105 km horizontal and 17 km vertical). When comparing the ISA directions (A and C), regions for which PI>0.5 are masked because the ISA does not accurately represent LPO for these points. The median directional difference for the remaining (unmasked) points is 2.62 degrees for the comparison between low and intermediate resolutions (A) and 4.25 degrees for the comparison between intermediate and fine resolutions (C). For the comparisons of PI (B and D), the higher resolution estimate of PI is shown relative to the lower resolution estimate. The median of the absolute value of this ratio is 1.8 for the comparison between intermediate and coarse resolutions (B) and 1.92 for the comparison between fine and intermediate resolutions (D). Similar resolution comparisons of plate-driven or density-driven flows by themselves show even closer agreement. These comparisons quantify the broad similarities in the spatial patterns of PI and ISA at different resolutions (Fig. S1). In particular, we can infer that our calculations include a model uncertainty of about 5 degrees for the ISA direction and about a factor of 2 for the estimate of PI.