The massive International Seismological Centre data set of the past 20 years and the two-station method are used to determine Pn velocities in the mantle lid beneath the Colorado plateau. In this method the event is located at distances where Pn is the first arrival (2¿--16¿) and the path is in or very neat the azimuth of the two-station pair and crosses the plateau. This technique to a large extent minimizes the hypocenter mislocation effect and possible errors due to variations in the crustal structure near the source, since only the difference in travel times at the two stations is used. However, this technique has a few underlying assumptions and possible sources of errors (such as the quality of the Pn data base and station delays caused by varying crustal structure) that require an extremely careful application of the method. A detailed study of the source of errors and a methodology of selection of the data are presented. Application of this method to the Colorado plateau using all possible two-station pairs from 53 stations located within or along the margin of the plateau yields an average high Pn velocity of 8.12¿0.09 km/s.

This value is considerably larger than the average value of 7.83 km/s based on available but very limited seismic refraction profiles but is remarkably similar to the average value of 8.1 km/s for the relatively stable midcontinent region. Our new Pn velocity for the Colorado plateau eliminates the paradox in the literature that emphasizes the rather close similarity between average Pn velocities beneath the Colorado plateau and the Basin and Range Province while their tectonic and magmatic Cenozoic history is dramatically different. Previous models for the structure and evolution of the plateau have used the low Pn velocity as an important constraint on density and thermal state of the lithosphere. Hence such models should be reexamined on the basis of this new uppermost mantly P2in velocity determination. There are two main models that have been proposed to explain the 2-km uplift of the Colorado plateau. One is based on a combination of thermal thinning of the lithosphere and crustal thickening, and the other involves a combination of the delamination of the subducted, subhorizontal Farallon oceanic plate from the overriding North American plate and crustal thickening. We show that the delamination model is more readily consistent not only with our velocity determination and the elevation of the plateau but also with varied geological observations reported in the literature that concern the Cenozoic evolution of western North America. ¿ American Geophysical Union 1989