We employ the recently published Crust 5.1 model of global crustal structur
e (Mooney ef al. 1998) to estimate the topographic contributions of isostat
ically compensated crustal loads and, in turn, infer a long-wavelength fiel
d for Earth's dynamic surface topography. Our inference of dynamic surface
topography is characterized by peak-to-peak variations of the order-of 4 km
. We argue that the square root of age variation of the oceanic bathymetry
has a dynamic origin. Remarkably, continental regions are the site of deep
(approximately 1.5 km) dynamic surface topography depressions. The power sp
ectrum of the field is dominated by spherical harmonic degrees l=1, 4 and 5
, suggesting the importance of continent-ocean differences, rather than dee
p lower mantle heterogeneity, to its origin. We model the Crust 5.1-based i
nference of dynamic surface topography within the context of seismic tomogr
aphy-based internal loading theories. We compute the depth dependences of d
ynamic surface topography response functions for various viscosity profiles
that we consider in our analyses. Our best descriptions of the Crust 5.1-b
ased inference achieve total variance reductions of the order of 70 per cen
t in the spherical harmonic degree range l=1-8. These are obtained for whol
e-mantle circulation models that preclude lower mantle heterogeneity from m
aintaining significant vertical stresses on the outer surface. These models
are characterized by a significant increase in viscosity across the depth
of the 660 km seismic discontinuity, by a factor of at least 50 relative to
the average viscosity of the upper mantle. We also consider results of a l
ayered circulation model but, in this case, we are unable to reconcile cons
traints provided by large-scale superswell topography. We propose that the
antipodal Pacific and African superswells are dynamically maintained by pos
itively buoyant lower mantle superplumes imparting on the outer surface a l
arge-scale, low-amplitude, predominantly degree 2 pattern of vertical stres
ses. Finally, we note that the Crust 5.1-based inference of dynamic surface
topography provides useful geodynamic constraints on the nature of deep su
bcontinental structure: optimal descriptions of the data require deep subco
ntinental keels composed of anomalously dense material.