Western U.S. Topography Origins: Part 2
Awhile back we contrasted two papers that addressed the presence or absence of dynamic topography, here defined as topography generated by stuff below the lithosphere. The two are Levandowski et al. in JGR in early 2014 and Becker et al. in EPSL in late 2013 (well, the publication date is 2014; it came online back in October). Now a third paper (well, kind of a network of papers) has emerged that bears on the same kind of balance but has yet another approach to getting there. We will focus here on the Kaban et al. G3 paper. [Note: this will get technical]
Quick reminder of how Levandowski et al and Becker et al approached this. Levandowski worked from S-wave structure of the top 150 km from combined ambient noise derived Rayleigh waves and ballistic Rayleigh waves with control from Ps receiver functions and more or less converted these to density. Becker et al. had several components to their approach but the most relevant here relied on Moho depths from receiver functions combined with estimates of crustal velocity based on CRUST2.0; the mantle lithosphere was controlled from a body wave inversion from Schmandt and Humphreys. And as we noted before, it seemed like the big difference was in the calculation of the contribution from the mantle lithosphere: for Levandowski et al., that left little residual topography, but Becker et al. found quite a bit which would then require a dynamic component to WUS topography. Both studies basically agree that variations in the crust are insufficient to explain the topography of the western U.S. [not an exceptionally new conclusion, but both come to fairly similar values from the crust]. Both papers are largely working from data collected at the uniformly-spaced Transportable Array.
Kaban et al. start from a set of refraction interpretations to create (in an associated paper, Tesauro et al., Tectonophysics 2014) a new crustal model for North America. In brief, North America is divided into provinces and the seismic wave speeds of upper, middle, and lower crustal layers are determined from data within the USGS database within that province. These values are then used as the basis for their crustal model, though there clearly are some other steps somewhere as their velocity maps do not mirror their province maps (for instance, there are substantial differences in the inferred crustal velocities between eastern Kansas and Illinois despite both areas being well within their Great Plains province, which seems entirely reliant on points in Missouri).
A concern is the distribution of the refraction profiles used:
Some areas are clobbered at this scale, others not so much. And again, at least to this reader, why things vary within a province is a mystery. Perhaps the “adaptive interpolation” algorithm used in Mooney and Kaban’s 2010 predecessor provides the explanation: more continuously sampled variables like topography are used to guide interpolation. There are several other issues one can have with the crustal model, but let us move on for now.
Kaban et al. then use the crustal model to derive a density model (using Christensen and Mooney values, apparently–hopefully with the corrected regression in deep crust!) and from this calculate a gravity model and a residual topography model. It is the latter we are interested in:
Although it isn’t well advertised, this also includes “the dynamic contribution from the anomalous masses below 325 km” (p. 14, apparently from Petrunin et al., 2013–though that paper really is describing a technique and lacks any estimated dynamic topography GG can find; it does calculates geoid variations). This gives us our best window for comparing with the earlier two papers:
This takes a minute to digest as the color scale in Kaban et al. is inverted from the other two (and has removed some dynamic component) and the Levandowski et al map on right is not simply observed topography – crust-supported topography but seismically inferred topography assuming a purely thermal upper mantle.
At a broad scale, things are pretty similar. Basin and Range needs about 2-3 km of support from the mantle relative to areas far to the east. In detail there are lots of differences. The Colorado Plateau, for instance, needs about 1 km less support from the mantle than Northern Rockies in Kaban et al., about 0.5 km in Becker et al., and needs a bit more support from mantle in Levandowski et al. SE Wyoming needs about 2 km of mantle support in Kaban et al., something under 1 km in Levandowski et al. NE Montana is about the same as the Colorado Plateau in Becker et al. but needs about 1 km more support from mantle in the other two models.
[If the dynamic topography included in Kaban et al. looks at all like that from Fig. 7c of Becker et al (which uses the same seismic model but perhaps to a shallower depth), then there is a deep mystery as that model predicts about a 3 km gradient across the SW US–if that is really the case, it is hard to see how the Kaban et al. map looks anything like either of the other two maps above.]
Because of the assumptions used, Kaban et al. assign the about 1-1.5 km of topographic variation in the western U.S. seen above to the uppermost mantle (above 325 km depth) (because they use gravity and the residual topography shown above to separate thermal and compositional components in the mantle, presumably they fit the topography perfectly). They find that nearly all of the variations seen above are thermal (consider their Figure 11, which shows the results from two different mantle seismic models which have strong thermal variations and small and quite different compositional variations in the area shown above); this is nearly identical to the results of Levandowski et al., who did not need compositional variations in the mantle to reproduce topography in this region. In contrast, Becker et al. found that most of the topography seen in the figure above requires a dynamic origin as the predicted contribution from the seismic model of Schmandt and Humphreys does not account for that residual topography.
How to best look at all this? Everybody seems to agree there is something below the Moho responsible for holding up the western U.S., especially the Basin and Range; the differences in estimating the crustal densities are not enormously different at the large scale. Kaban et al. and Levandowski et al. seem to put most of this in thermal variations in the upper mantle at lithospheric depths; Becker et al. argue that most of this is coming from dynamic effects. It would seem that the challenge is to clear up the story in the uppermost mantle.