Mantle plumes: heat-flow near Iceland

1 Foulger s (2002) paper in the last issue ofAstronomy & Geophysics(A& ) illustrates the debate over whetherhotspots regions of long-lived excess volcan-ism such as Iceland , Hawaii or Yellowstone result from plumes of hot material upwellingfrom great depth in the Mantle (Morgan 1971).In the plume model, plate motion over fixed orslow-moving plumes causes age-progressive lin-ear volcanic chains and topographic swells thatidentify plumes and yield inferences about theirproperties. This model has been widely acceptedbecause it gives an elegant explanation of howdiverse volcanic regions have similar origins,and an absolute reference frame describing platemotions relative to the deep , many hotspots deviate from theexpected behaviour.

2 Some hotspots move sig-nificantly relative to each other and the spinaxis (Tarduno and Cottrell 1997), changes insome volcanic chain orientations do not corre-spond to the expected plate motion changes(Norton 1995), and some chains show no clearage progression (Schlanger et ). A viewis emerging that at least some hotspots, notablyYellowstone, are not due to deep Mantle plumes(Humphreys et , Christiansen et ), and the entire plume model is being chal-lenged (Anderson 2000, Hamilton 2002). Iceland is a focus of these discussions, as thetype example of a hotspot on a mid-ocean the plume model, the elevation and thickcrust relative to typical mid-ocean ridges resultfrom melting by a hot plume (White 1999),whereas in Foulger s (2002) non-plume model,temperatures are not unusually high but excessmelting of more fertile material occurs, consis-tent with petrologic arguments (Korenaga andKelemen 2000).

3 Seismological results for themaximum depth of the low velocity anomaly,the strongest discriminant between a deepmantle plume and an upper Mantle meltinganomaly, are discordant (Foulger et ,Shen et ) because seismometers onIceland have limited resolution for structure atdepth owing to the island s small heatflowGiven this interest, we examined seafloor heat-flow data from the Iceland region. The small orabsent heat-flow anomalies at other hotspotsplay a role similar to that of the dog whose fail-ure to bark helped Sherlock Holmes locate themissing racehorse Silver Blaze. Originally, theuplift at Hawaii and similar midplate hotspotswas thought to reflect a hot plume causing heat-ing to about 50 km of the surface (Crough1983, McNutt and Judge 1990).

4 Such heatingpredicts heat-flow significantly higher thanfrom the usual cooling of oceanic lithosphere asit spreads away from the mid-ocean ridgeswhere it formed. Although anomalously highheat-flow was initially reported, subsequentanalysis showed that most, if not all, of theapparent anomalies resulted from comparingdata to thermal models that underestimatedheat-flow elsewhere (Von Herzen et ,Stein and Abbott 1991, Stein and Stein 1993).Hence subsequent models generally assume thatthe uplift results from the dynamic effects of ris-ing plumes (Liu and Chase 1989, Sleep 1994)and the associated compositional buoyancy,whose thermal effects are concentrated at thebase of the lithosphere and raise surface heat-flow at most slightly, because conduction to thesurface takes tens of millions of years.

5 heat-flow has played little role in the debateabout hotspots like Iceland , which are on ornear mid-ocean ridges, for two reasons. Thefirst is that predictions for heat-flow have notbeen offered, because such hotspots are thoughtto reflect an interaction between upwellingplumes and nearby spreading centers (Ito et ) more complex than at mid-plate hotspotswhich are generally attributed to a simpler(albeit not yet understood) interaction of aplume with a plate interior. Second, seafloornear on-ridge hotspots is young, less than40 Myr old. In young seafloor, measured heat-flow is significantly lower than expected purelyfrom conductive cooling of the lithosphere,because some heat is transported by hydro-thermal circulation of sea water through thecrust ( Stein and Stein 1994).

6 Hence it wasunclear how to characterize normal heat flowand assess possible models imply that heat-flow should beabove the normal in several ways. The mostimportant is likely to be an indirect effect ofplume material migrating along the Mid-Atlantic Ridge (White 1999). This should raisetemperatures along the ridge by up to severalDiscussion: Mantle 2003 Vol 44 Mantle plumes: heat-flow near IcelandIn the first of four pieces arising from Gill Foulger s challenge to the Mantle plume hypothesis (last issue), Carol Stein andSeth Stein join the debate with some data and comment on heat-flow around heat-flow near Iceland on theNorth American side of the Mid-AtlanticRidge is comparable to that for oceaniclithosphere elsewhere, and thus shows noevidence for significantly highertemperatures associated with a mantleplume.

7 heat-flow is higher on the Eurasianplate than on the North American plate,an intriguing asymmetry opposite to thatexpected from models in which Icelandformed over a Mantle :Bathymetry andheat-flow for theIceland shown asheat-flow fraction,observed valuesnormalized by globalaverage values forthat lithospheric age(figure 2).Lithosphere youngerthan about 35 Myrindicated by the positions of magnetic anomaly 13 (solid line) after Muelleret al.(1997) orapproximated by dashed flow fraction< > (m) 0 1100 2000 3000 4000 5000hundred degrees, depending on distance fromthe plume, so that lithosphere formed on eitherplate would have higher plume should also have direct effects onheat flow. First, outward-flowing plume mat-erial should heat the base of already-formedlithosphere.

8 This effect would be similar to thatat Hawaii, but larger because heat is added atthe base of the lithosphere, which is thinnernear Iceland because of its relative youth. Henceincreased heat-flow should occur on both sidesof the Mid-Atlantic second direct effect could result from thehistory of relative motion between the plume,Mid-Atlantic Ridge, and the two this history is more complex thanalong the Hawaiian Emperor seamount chain,where the history of volcanism is used to inferthe history of the plume. In contrast, the Icelandplume s history cannot be inferred directly fromthe elevated Iceland Greenland and Iceland Faroe plateaus extending westward and east-ward from Iceland (figure 1), because modelsassuming various hotspot sizes and motions offer non-unique solutions that could be usedto explain a plateau of any location, origin, andage progression (Vink 1984).

9 To address this ambiguity, Vink (1984) usedplate reconstructions assuming that plumes arefixed to predict that the plume presently underIceland was under Greenland 45 Myr ago. Sincethen, westward motion of the Mid-AtlanticRidge relative to the plume has brought Icelandover the plume. During this time, plume mat-erial flowed laterally beneath the NorthAmerican plate to the Mid-Atlantic that plume material flowed to theclosest point on the Mid-Atlantic Ridge, whereplateaus formed by excess volcanism and weretransported away in opposite directions as thetwo plates spread, this matches the observedtrends of the plateaus. Alternatively, White andMcKenzie (1989) argued that such lateral flowwas not possible.

10 Instead, they proposed that anewly formed plume initiated the rifting of theGreenland margin and the opening of the NorthAtlantic, such that the paired plateaus formeddirectly above, via ridge jumps that kept theMid-Atlantic Ridge above the plume s core(White 1999). Although these plume historymodels differ, and only the first reflects detailedkinematic modelling, we expect that both pre-dict heat-flow near Iceland higher on the NorthAmerican (west) plate than for lithosphere ofthe same age on the Eurasian (east) thus examined heat-flow data for siteswithin 500 km of Iceland to see if they showedeither expected effect abnormally high heat-flow on either side of the ridge, and higherheat-flow to the west.