REPORTS Dynamic Topography Change of the tiallyimportant ...geosci.uchicago.edu/~kite/doc/Rowley_et_al_2013.pdfgravity, plate velocities, and core-mantle...

Dynamic Topography Change of theEastern United States Since3 Million Years AgoDavid B. Rowley,1* Alessandro M. Forte,2 Robert Moucha,3 Jerry X. Mitrovica,4

Nathan A. Simmons,5 Stephen P. Grand6

Sedimentary rocks from Virginia through Florida record marine flooding during the mid-Pliocene.Several wave-cut scarps that at the time of deposition would have been horizontal are nowdraped over a warped surface with a maximum variation of 60 meters. We modeled dynamictopography by using mantle convection simulations that predict the amplitude and broad spatialdistribution of this distortion. The results imply that dynamic topography and, to a lesser extent,glacial isostatic adjustment account for the current architecture of the coastal plain andproximal shelf. This confounds attempts to use regional stratigraphic relations as references forlonger-term sea-level determinations. Inferences of Pliocene global sea-level heights or stabilityof Antarctic ice sheets therefore cannot be deciphered in the absence of an appropriate mantledynamic reference frame.

Thecontinental margin of the East Coast ofthe United States is the archetypal Atlantic-type or passive-type continental margin(1). Such margins have been thought to overlay amantle that is entirely passive (2). As a conse-quence, passive-type margins are generally inter-

preted as having simple stratigraphic historiescontrolled by the interplay between thermallydriven subsidence, sediment loading, compaction,and sea-level variations (3, 4). Flexural responsesof the lithosphere resulting fromoff-shore sedimentloading (5, 6) and, less frequently, onshore ero-

sional unloading (7) are also recognized as poten-tially important (4–6). These assumptions underpinthe rationale for the use of the U.S. East Coastmargin in determining global long-term [≥ 0.1 mil-lion years (My)] sea-level variations (4–6, 8, 9).

The mantle is not a passive player. Mantleflow influences surface topography, through per-turbations of the dynamic topography, in a man-ner that varies both spatially and temporally. As aresult, it is difficult to invert for the global long-term sea-level signal and, in turn, the size of theAntarctic Ice Sheet by using East Coast shorelinedata (10). Factors that need to be considered in-clude flow associatedwith the negative buoyancyof the subducted Farallon slab (10–14) and thecoupled shallower westward flow of hotter man-

1Department of the Geophysical Sciences, 5734 South EllisAvenue, The University of Chicago, Chicago, IL 60637, USA.2GEOTOP—Université du Québec à Montréal CP 8888, suc-cursale Centre-Ville Montréal, Québec H3C 3P8, Canada. 3De-partment of Earth Sciences, 204 Heroy Geology Laboratory,Syracuse University, Syracuse, NY 13244, USA. 4Department ofEarth and Planetary Sciences, Harvard University, 20 OxfordStreet, Cambridge, MA 02138, USA. 5Atmospheric, Earth, andEnergy Division, Lawrence Livermore National Laboratory,Livermore, CA 94551, USA. 6Jackson School of GeologicalSciences, University of Texas at Austin, Austin, TX 78712, USA.

*Corresponding author. E-mail: [email protected]

Fig. 1. Post–mid-Pliocene warping and incision of theCoastal Plain. (A) Present topography based on ETOPO1emphasizing the incised, low-relief, mid-Pliocene floodingsurface of the East Coast Coastal Plain, highlighting (blacksolid and dashed line) the locations of the Orangeburg,Chippenham (CS), Thornburg (TS) wave-cut scarps and TrailRidge (TR) and (blue dots) sites with preserved mid-Pliocene(Yorktown, Duplin, Chorlton, and Cypresshead Formations)strata. Superimposed are contours showing an estimate ofthe amplitude of the post–3-Ma change in the dynamictopography based on TX2007 V2 model. The dynamic to-pography change is shown by the contours with a 10-mcontour interval, white where they are superimposed on thetopography. (B) Inset graph of the height of the OrangeburgScarp as a function of latitude (solid line) based on (26) andthe highest preserved mid-Pliocene marine sedimentaryrocks as a function of latitude (dots). (C) Contours of GIA-induced relative sea level change based on the V2 viscosityprofile (supplementary text) with a 5-m contour interval. TheOrangeburg, Chippenham, and Thornburg Scarps areindicated by the thick black line; the Fall Zone is the thickdashed gray line.

28 JUNE 2013 VOL 340 SCIENCE www.sciencemag.org1560

REPORTS

on

Janu

ary

24, 2

017

http

://sc

ienc

e.sc

ienc

emag

.org

/D

ownl

oade

d fr

om

http://science.sciencemag.org/

tle (10, 12, 15). The latter produces, at least lo-cally, changes in buoyancy and associated shorterwavelength changes in dynamic topography. Bothfactors confound local estimates of long-term sea-level variations (15).

The Coastal Plain is characterized by a se-quence of marine and nonmarine sedimentaryunits that range from at least Early Cretaceous topresent in age. These units generally thicken east-ward to more than 12 km (16–19). This packageunconformably overlies pre-Mesozoic crystallinerocks, as well as Triassic/Jurassic rift-basin strata,and pinches out to the west along the Fall Zone(Fig. 1A).Models of the depositional architectureof this margin have been developed on the basisof combinations of seismic stratigraphy and drill-ing (9, 18, 19) in order to better understand itsevolution (4–6, 20). Thesemodels have also beenused to infer global sea-level history by solvingfor the contributions of thermal subsidence, sedi-ment loading and compaction, flexural loading,and sediment delivery while assuming that theonly remaining unknown is the contribution fromchanges in sea level. Most attempts have fo-

cused on the New Jersey segment of this margin(4–6, 9, 21). A local and temporally limited sea-level estimate has been made by using the mid-Pliocene Orangeburg Scarp as a marker (8). Inthis particular example, after correction for ~50 T18 m of post–mid-Pliocene uplift derived froma local estimate of stream incision rate (22), theOrangeburg Scarp has been inferred to have hadan elevation of 35 T 18 m, which has been takento indicate the height of themid-Pliocene sea level(8). This height would imply collapse of theGreenland and West Antarctic Ice Sheets andpotentially considerable melting of the East Ant-arctic Ice Sheet during the mid-Pliocene cli-mate optimum (23–25). However, the Pliocenestrandline and immediately adjacent shallowma-rine sediments are not preserved at constant ele-vation along the Coastal Plain (Fig. 1B) (26).Thus, the Orangeburg Scarp is not a good refer-ence for sea-level determinations for the Plio-cene. Instead, we used the scarp as a marker forcharacterizing the processes that have warped thecontinental margin subsequent to 4 to 3 millionyears ago (27).

To assess the processes responsible for thepost–mid-Pliocene warping of this margin, wedeveloped a model of the Coastal Plain that ac-counts for mantle dynamics (10) and glacial iso-static adjustment (GIA) (28). Formation of karstscan also induce uplift (29). However, becausecarbonates are scarce north of Florida, we ignoredthis effect. Potential contributions from flexuralwarping because of offshore sediment loadingand erosional unloading (7, 30) were assessed(supplementary text) but were deemed too un-certain to yield reliable estimates for the currentanalysis. In addition, it is shown below that themajority of the warping can be accounted for bydynamic topography and GIA alone.

Our analysis focused on the variably incised,mid-Pliocene, low-relief flooding surface that char-acterizes the geomorphology of the eastern sea-board coastward of theOrangeburg and equivalentwave-cut scarps that define the landward edge ofthis surface (Fig. 1A). The Orangeburg and cor-relative scarps would have been horizontal at thetime of formation; the adjacent mid-Pliocene shal-lowmarine rocks and associated flooding surfacewould have been largely undissected and wouldhave sloped gently eastward in a manner com-parable to the modern shallow shelf. Therefore,the warping (Fig. 1B) and incision of this low-relief flooding surface primarily reflects post–mid-Pliocene relative uplift together with erosionaldown-cutting by rivers and streams that traversethe eastern Coastal Plain. A reasonable test of ourmodeling will be the retrodiction of this floodingsurface to a configuration comparable to the mod-ern shelf.

We calculated global mantle convective flowby following the approach of (10, 15, 31, 32)with the tomography models TX2007 (33) andTX2008 (34), in which global seismic data to-gether with a range of convection-related observ-ables (present-day surface topography, free airgravity, plate velocities, and core-mantle bound-ary excess ellipticity) were jointly inverted to yieldthe three-dimensional (3D) distribution of densityin the mantle (34) that is consistent with seismic,geodynamic, andmineral physics data. The under-lying physical basis of this model is described indetail by Forte (35). We considered two differentmodels of the radial distribution of viscosity, V1and V2 (10, 31, 32), and these, together with thetwo different inversions (TX2007 and TX2008)for mantle density, provide four alternative mod-els for predictions of time varying dynamic to-pography (see supplementary materials).

The 3D distribution of mantle buoyancy, whenintegrated with estimates of the radial distribu-tion of viscosity, allows the computation of theinstantaneous global flow field (31). With this inhand, we iteratively computed a global backwardadvection solution brought forward in timewith afull convection calculation to estimate the verticalstresses acting on the base of the crust arisingfrom flow in the mantle (10, 32). These time-dependent vertical stresses generate a globally dis-tributed dynamic topography that warps Earth’s

SE Georgia Embayment

SWGeorgia

Embayment

Charleston Embayment

AlbemarleEmbayment

SalisburyEmbayment

RaritanEmbayment

South NewJersey Arch

SouthFlorida

Embayment

40°

36°

32°

28°

84° 80° 76° 72°

0 20 40 (m)-40 -20Dynamic Topography Change Since 3 Ma

Fig. 2. Calculated dynamic topography change since 3 Ma. Locations of features associated withEast Coast Coastal Plain geology after (36). The Fall Zone marks the approximate landward erosional edgeof the Early Cretaceous to Cenozoic Coastal Plain strata. Color image is the distribution of retrodicteddynamic topography change based on TX 2007 V2 results. Dashed gray rectangular boxes outline theunderlying resolution of the Simmons et al. (33, 34) joint seismic-geodynamic tomography inversion. TheOrangeburg and correlative scarps pass over the center of the maximum of retrodicted dynamictopography change since 3 Ma.

www.sciencemag.org SCIENCE VOL 340 28 JUNE 2013 1561

REPORTS

on

Janu

ary

24, 2

017

http

://sc

ienc

e.sc

ienc

emag

.org

/D

ownl

oade

d fr

om

http://science.sciencemag.org/

surface. The difference between the present-daydynamic topography and estimates of past dy-namic topography yields the change in dynam-ic topography as a function of time (Figs. 1 and2 and fig. S1). The variations in height of theOrangeburg Scarp and related sediments (Fig. 1B)is well correlated, in latitude, with our estimatesof the dynamic topography change since 3 mil-lion years ago (Ma). Both are high in Florida, de-crease toward the north in the Southeast GeorgiaEmbayment (~31°N), and then rise again farthernorth in the vicinity of the Cape Fear, Neuse, andNorfolk arches (Fig. 2).

The height of the Orangeburg Scarp risesmore quickly starting north of 32°N than the es-timates of dynamic topography change since 3Ma(fig. S1). We attribute this misfit primarily to dif-ferences in the spatial scales of these data setsand to uncertainties in the tomography-based flowcalculation. The joint seismic-geodynamic con-strained tomography model has a minimum hori-zontal spatial resolution of 270 km by 270 km(34) (Fig. 2). It thus resolves mantle heteroge-neity on a length scale substantially greater thanthe geological data being considered, whose var-iations are known to less than a kilometer resolu-tion. In this regard, it is important to emphasizethat our estimates of dynamic topography changeare derived from full global mantle convectionsolutions and have not been adjusted or in anyway tuned to yield better fits to the observedwarping of the Orangeburg and correlative scarps.

Despite the longer-wavelength character ofthe seismic tomography constrained mantle flowcalculations, there is a good spatial correlationbetween the maxima of the estimated changes indynamic topography since 3 Ma and relative in-cision of the mid-Pliocene flooding surface (Fig.1). Regions in Georgia with limited retrodictedchanges in dynamic topography are character-ized by limited fluvial incision into this surface.In contrast, farther north, where the retrodictedamplitude of dynamic topography change in-creases, the intensity of dissection increases con-comitantly, and both reach a maximum in thevicinity of Chesapeake Bay (Figs. 1 and 2). Theamount of incision of the low-relief floodingsurface is about 50 T 10 m in this region, in ac-cord with the retrodicted amplitude of dynamictopography change since 3 Ma. This impliesthat a large fraction of the Coastal Plain geo-morphology, at least shoreward of the Orange-burg Scarp, is a result of the interaction betweenflooding-related planation and subsequent dy-namic topography induced uplift and fluvialincision within the last 3 My.

The principal outstanding feature of the dy-namic topography retrodictions is the pattern ofvariable uplift along the East Coast of the UnitedStates (fig. S1). The origin of this uplift can bedirectly traced to the existence of hot, buoyantmaterial in the shallow (50 My) start-ing from the present, and thus anymisfit withmore

Fig. 3. Retrodicted paleogeography of the Coastal Plain at 3 Ma. Paleogeographic reconstructionof the eastern United States at 3 Ma. Retrodicted topography from which differential dynamic to-pography based on TX 2007 V2 results and a GIA signal have been subtracted. Scale bar is in meters.No attempt has been made to remove effects of subsequent river and stream incision. Thick dottedline is the shoreline inferred geologically (23, 29) that essentially follows the Orangeburg andcorrelative scarps. The thin blue line is the +25 m contour on the retrodicted topography. Blue dotsare locations for which there are independent outcrop or borehole constraints on the presence ofPliocene marine sediments. Blue stars in southern Delaware and New Jersey are locations of Plioceneestuarine sediments (38, 42).

28 JUNE 2013 VOL 340 SCIENCE www.sciencemag.org1562

REPORTS

on

Janu

ary

24, 2

017

http

://sc

ienc

e.sc

ienc

emag

.org

/D

ownl

oade

d fr

om

http://science.sciencemag.org/

recent times (

originally published online May 16, 2013 (6140), 1560-1563. [doi: 10.1126/science.1229180]340Science

2013) Mitrovica, Nathan A. Simmons and Stephen P. Grand (May 16, David B. Rowley, Alessandro M. Forte, Robert Moucha, Jerry X.Since 3 Million Years AgoDynamic Topography Change of the Eastern United States

Editor's Summary

years.mid-Atlantic and Southern United States coast varied by 60 meters or more during the past 5 millionpublished online 16 May) used a model of flow in the mantle to show that the topography of the

(p. 1560,et al.Rowley that the coastal plain has deformed in response to flow in Earth's mantle. been used to infer changes in global sea level through the Cenozoic. However, recent work has shown

hasmostly to the weight of deposited sediments. As a result, the fine-scale stratigraphy of the sediments The Atlantic coastal plain of North America has been thought of as a passive margin, responding

By the Sea Side

This copy is for your personal, non-commercial use only.

Article Tools

http://science.sciencemag.org/content/340/6140/1560article tools: Visit the online version of this article to access the personalization and

Permissionshttp://www.sciencemag.org/about/permissions.dtlObtain information about reproducing this article:

is a registered trademark of AAAS. ScienceAdvancement of Science; all rights reserved. The title Avenue NW, Washington, DC 20005. Copyright 2016 by the American Association for thein December, by the American Association for the Advancement of Science, 1200 New York

(print ISSN 0036-8075; online ISSN 1095-9203) is published weekly, except the last weekScience

on

Janu

ary

24, 2

017

http

://sc

ienc

e.sc

ienc

emag

.org

/D

ownl

oade

d fr

om