UC IrvineUC Irvine Previously Published Works

TitleThe effects of North Atlantic SST and sea ice anomalies on the winter circulation in CCM3. Part II: Direct and indirect components of the response

Permalinkhttps://escholarship.org/uc/item/47q6528j

JournalJournal of Climate, 17(5)

ISSN0894-8755

AuthorsDeser, CMagnusdottir, GSaravanan, Ret al.

Publication Date2004-03-01

DOI10.1175/1520-0442(2004)017<0877:TEONAS>2.0.CO;2

Copyright InformationThis work is made available under the terms of a Creative Commons Attribution License, availalbe at https://creativecommons.org/licenses/by/4.0/ Peer reviewed

eScholarship.org Powered by the California Digital LibraryUniversity of California

1 MARCH 2004 877D E S E R E T A L .

q 2004 American Meteorological Society

The Effects of North Atlantic SST and Sea Ice Anomalies on the Winter Circulation inCCM3. Part II: Direct and Indirect Components of the Response

CLARA DESER

National Center for Atmospheric Research,* Boulder, Colorado

GUDRUN MAGNUSDOTTIR

Department of Earth System Science, University of California, Irvine, Irvine, California

R. SARAVANAN AND ADAM PHILLIPS

National Center for Atmospheric Research,* Boulder, Colorado

(Manuscript received 16 January 2003, in final form 6 October 2003)

ABSTRACT

The wintertime atmospheric circulation responses to observed patterns of North Atlantic sea surface temper-ature and sea ice cover trends in recent decades are studied by means of experiments with an atmospheric generalcirculation model. Here the relationship between the forced responses and the dominant pattern of internallygenerated atmospheric variability is focused on. The total response is partioned into a portion that projects ontothe leading mode of internal variability (the indirect response) and a portion that is the residual from thatprojection (the direct response). This empirical decomposition yields physically meaningful patterns whosedistinctive horizontal and vertical structures imply different governing mechanisms. The indirect response, whichdominates the total geopotential height response, is hemispheric in scale with resemblance to the North AtlanticOscillation or Northern Hemisphere annular mode, and equivalent barotropic in the vertical from the surface tothe tropopause. In contrast, the direct response is localized to the vicinity of the surface thermal anomaly (SSTor sea ice) and exhibits a baroclinic structure in the vertical, with a surface trough and upper-level ridge in thecase of a positive heating anomaly, consistent with theoretical models of the linear baroclinic response toextratropical thermal forcing. Both components of the response scale linearly with respect to the amplitude ofthe forcing but nonlinearly with respect to the polarity of the forcing. The deeper vertical penetration of anomalousheating compared to cooling is suggested to play a role in the nonlinearity of the response to SST forcing.

1. Introduction

Sea surface temperature (SST) and sea ice cover inthe North Atlantic have undergone pronounced trendsin recent decades. Although driven in large part bychanges in atmospheric flow (cf. Deser et al. 2000; Seag-er et al. 2000), the trends in SST and sea ice may exerta significant feedback upon the atmospheric circulation.The nature of this feedback is the subject of the presentstudy and a companion study (Magnusdottir et al. 2004,hereafter referred to as Part I). In this work, we examinethe wintertime atmospheric circulation responses to ob-served patterns of North Atlantic SST and sea ice cover

*The National Center for Atmospheric Research is sponsored bythe National Science Foundation.

Corresponding author address: Dr. Clara Deser, Climate and Glob-al Dynamics Division, NCAR, P.O. Box 3000, Boulder, CO 80307-3000.E-mail: cdeser@ucar.edu

trends by means of experiments with an atmosphericgeneral circulation model, version 3 of the CommunityClimate Model (CCM3). As shown in Part I, a notableresult of these experiments is the similarity of the re-sponse to trends in ice cover and to trends (with reversedsign) in SST, despite large differences in the spatialpattern and amplitude of the two forcing parameters. Inaddition, these responses resemble the model’s leadingmode of internal variability, the simulated equivalent ofthe observed North Atlantic Oscillation or NorthernHemisphere annular mode patterns (e.g., Wallace 2000).Previous modeling studies have also found a corre-spondence between the patterns of internal variabilityand the response to an imposed midlatitude heating per-turbation (Peng and Robinson 2001; Hall et al. 2001)or to increasing greenhouse gas concentrations (Kushneret al. 2001).

The purpose of Part II of this study is to further ex-amine the relationship between the forced response totrends in SST and sea ice cover, and the dominant pat-

878 VOLUME 17J O U R N A L O F C L I M A T E

FIG. 1. Jan (left) SST and (right) sea ice forcing used in the SST25 and ICE2 perturbationexperiments, respectively. In the SST panel, dark (light) shading denotes positive (negative)anomalies relative to the control run; the contour interval is 1.5 K and the zero contour has beenomitted. In the sea ice panel, dark (light) shading denotes grid boxes where sea ice has beenremoved (added) relative to the control run.

tern of internally generated atmospheric variability. Spe-cifically, we partition the total geopotential height re-sponse into a portion that projects onto the leading modeof internal variability (the indirect response) and theresidual from that projection (the direct response). Thisempirical decomposition is shown to yield physicallymeaningful patterns whose distinctive horizontal andvertical structures imply different governing mecha-nisms. Kushner et al. (2001) also found such an ap-proach insightful for understanding the zonally averagedatmospheric circulation response in the Southern Hemi-sphere to increasing greenhouse gas concentrations ina coupled climate model.

It was noted in Part I that the total geopotential heightresponse to trends in SST and sea ice is nonlinear withrespect to the polarity of the forcing. Here we expandupon this issue by examining the sensitivity of both thedirect and indirect components of the response to thesign of the imposed SST and sea ice anomaly trends.We also investigate the vertical distribution of anoma-lous heating compared to cooling in these experiments,a factor which may play a role in the nonlinearity ofthe response.

The outline of the paper is as follows. Section 2 pro-vides a brief overview of the experiments analyzed. Sec-tion 3 presents the decomposition of the total geopo-tential height response into direct and indirect compo-nents, and documents the vertical distribution of anom-alous heating associated with the imposed SST and seaice anomaly trends. Section 4 discusses and summarizesthe results.

2. Experimental design

A complete description of the atmospheric generalcirculation model (CCM3) and experimental design is

given in Part I. The two primary experiments analyzedhere are termed SST25 and ICE2. In SST25, the forc-ing is defined as the observed monthly SST anomalytrend during 1954–94 over the North Atlantic, multi-plied by a factor of 25 (i.e., the polarity of the observedtrend is reversed and its amplitude is magnified five-fold). In ICE2, the forcing is the observed monthly trendin sea ice extent over the North Atlantic during 1958–97, magnified by approximately a factor of 2. In eachcase, the forcing is applied as an anomaly upon the meanseasonal cycle. Each experiment, as well as a controlsimulation forced with the climatological seasonal cycleof SST and sea ice cover, is integrated for a minimumof 61 yr; we analyze years 2–61 from each run.

Figure 1 shows the January SST and sea ice anomaliesin SST25 and ICE2, respectively; anomalies in the oth-er winter months are similar (not shown). In SST25,the primary forcing is a positive SST anomaly centerlocated in the subpolar gyre (maximum amplitude ;7K), with a weaker negative anomaly off the east coastof North America (maximum amplitude ;3 K). In ICE2,the forcing consists of a reduction in sea ice cover inthe Greenland Sea and an extension of the ice edge inthe Labrador Sea. Note that a grid cell is either ice freeor 100% ice covered; there is no fractional ice concen-tration.

We shall also make use of additional experiments thathave been conducted to test the sensitivity of the re-sponse to the magnitude and polarity of the imposedthermal forcing. For SST forcing, these are termedSST15, SST12.5, and SST22.5; for sea ice forcingthese are ICE1, ICELAB, and ICEGRN. SST15 is iden-tical to SST25 except for a sign reversal (e.g., the ob-served SST trend magnified by a factor of 5); SST12.5(SST22.5) is identical to SST15 (SST25) except thatthe observed trend is multiplied by a factor of 2.5

1 MARCH 2004 879D E S E R E T A L .

FIG. 2. Decomposition of the (left) total Dec–Apr Z500 responses in (top) SST25 and (bottom)ICE2 into a component that projects onto the leading EOF of (middle) the control run and (right)the residual from that projection. The contour interval is 10 m in all panels; positive (negative)contours are solid (dashed) and the zero contour is the thin solid line. The locations of the SSTand sea ice anomalies are indicated by shading in the residual panels.

(22.5); ICE1 is identical to ICE2 except that the arealextent of the observed sea ice trend has not been mag-nified; ICELAB (ICEGRN) is identical to ICE2 exceptthat only the Labrador (Greenland) Sea portion of thesea ice cover is altered.

3. Results

a. Geopotential height response

The left-hand panels of Fig. 2 show the total geo-potential height response at 500 hPa (Z500) for the winterseason December–April for SST25 and ICE2, obtainedby subtracting the 60-yr mean of the control simulationfrom the 60-yr mean of each perturbation experiment(April is included in the definition of winter due to thestrong similarity between the March and April responsesin both experiments; not shown). Values exceeding ap-proximately 10 m in absolute value are significant atthe 95% confidence level according to a local t test (notshown). Although the SST and sea ice forcings differin location and magnitude, the responses are remarkablysimilar in both amplitude and spatial pattern. Both re-sponse patterns are hemispheric in nature (although theforcing is confined to the Atlantic sector), with positiveheight anomalies at high latitudes and negative heightanomalies at midlatitudes of the Atlantic and Pacific.Maximum height anomalies are ;80 m in the polarregion and ;2(20–40 m) in midlatitudes.

As noted in the introduction, there is an emergingconsensus that the internal variability in an atmosphericgeneral circulation model may play a strong role in shap-

ing the pattern of the forced response. The middle panelsof Fig. 2 show the leading EOF of winter Z500 from thecontrol run, which accounts for 33% of the variance(and is well separated from the second EOF) over theNorthern Hemisphere north of 308N. Cosine weightinghas been used in the covariance matrix to compute theEOF. (The scaling of the EOF patterns, which differsslightly in the two middle panels, is described shortly.)The similarity between the leading EOF of the controlrun and the response patterns in both experiments isevident: the pattern correlation is 0.70 for SST25 and0.84 for ICE2. The principal component time series forthe control run EOF is approximately normally distrib-uted about a mean of zero (not shown but see Part I).

Although there is a high degree of similarity betweenthe leading Z500 EOF in the control run and the Z500

responses in the two experiments, there are some subtledistinctions in the relative amplitude and location of thecenters of action. To examine these differences, we pro-jected the response onto the EOF using linear leastsquares spatial regression, and then subtracted the pro-jection (the EOF scaled by the regression coefficient:middle panels of Fig. 2) from the response to obtain theresidual. We excluded the forcing region (908W–458E)from the calculation of the spatial regression coefficientso as to optimally compare the features of the far-fieldresponse that are common to the internal variability;however, similar results are obtained when the full hemi-spheric domain is used (not shown). In both experi-ments, the total response is dominated by its projectiononto the internal variability; however, a well-defined

880 VOLUME 17J O U R N A L O F C L I M A T E

FIG. 3. As in Fig. 2 but for different levels in the troposphere: (a) ICE2, (b) SST25, and (c)SST15.

residual pattern that is closely tied to the location of theforcing is also apparent (right-hand panels of Fig. 2).In SST25, the residual response consists of an anom-alous ridge over and slightly downstream of the warmSST anomaly, with maximum amplitude ;60 m (10 mK21). In ICE2, the main feature of the residual responseis an anomalous ridge centered over the Greenland Seain the region of sea ice removal and extending over thepolar cap, with maximum amplitude ;50 m. The anom-alous ridge response in the residual fields in both ex-

periments is consistent with simple physical reasoning(e.g., a raising of geopotential height surfaces withinand above an anomalously warm column of air), lendingcredence to our diagnostic technique and the notion thatthe total response may be partitioned into direct (e.g.,residual) and indirect (e.g., projection onto the leadingmode of internal variability) components. We note thatneither residual pattern resembles any of the higher-order EOFs in the control run, suggesting that the re-sidual is indeed a direct, forced response.

1 MARCH 2004 881D E S E R E T A L .

FIG. 3. (Continued )

The vertical structures of the total, indirect, and directresponses are shown in Fig. 3. The indirect responseswere computed by regressing the geopotential heightanomalies (relative to the 60-yr mean) in the controlrun at each level upon the leading 500-hPa principalcomponent time series in the control run, and then scal-ing those regression patterns (denoted Zreg) by the spatialregression coefficient between the total height response(experiment-minus-control) and Zreg at each level sep-arately. In this way, we ensure that the indirect responses

shown in Fig. 3 represent the projections upon a singlemode of internal variability as represented by the lead-ing PC time series at 500 hPa. However, using the lead-ing EOF computed separately for each level in the con-trol run in place of Zreg yields nearly identical resultsdue to the high temporal correlation (.0.96) betweenthe leading PC time series at each level with that at 500hPa (not shown).

For ICE2 (Fig. 3a), the vertical structure of the totalresponse pattern is approximately equivalent barotropic,

882 VOLUME 17J O U R N A L O F C L I M A T E

FIG. 3. (Continued )

with embedded small-scale structures evident at 1000hPa coincident with the location of the sea ice anom-alies. The indirect response is also equivalent barotropic,as expected and consistent with previous studies (e.g.,Ting and Lau 1993). Note the absence of localized ice-related structures in the indirect response at 1000 hPa.The direct response is baroclinic between 1000 and 850hPa over the regions of anomalous sea ice cover; forexample, where sea ice has been removed in the Green-land Sea, the height anomalies are negative at 1000 hPa

and positive at 850 hPa, indicative of a local warmingof the boundary layer. Above the boundary layer, thedirect response consists of a primary center of abovenormal heights over the Greenland Sea and spreadingto much of the Arctic. This anomalous ridge grows withheight from 850 to 500 hPa, with no further amplifi-cation between 500 and 300 hPa. A weaker anomaloustrough is also present over and downstream of the en-hanced ice cover in the Labrador Sea.

For SST25 (Fig. 3b), the vertical structures of the

1 MARCH 2004 883D E S E R E T A L .

FIG. 4. As in Fig. 2 but for (top) SST22.5 and (bottom) ICE1.

total and indirect responses are approximately equiva-lent barotropic, similar to those for ICE2. The residualresponse is baroclinic within the boundary layer directlyover the positive SST anomaly (i.e., an anomaloustrough at 1000 hPa and near-zero height anomalies at850 hPa), also similar to the results for ICE2. The re-sidual response in the free atmosphere consists of anequivalent barotropic anomalous ridge over and down-stream of the warm SST anomaly. Note that this anom-alous ridge amplifies considerably between 500 and 300hPa, unlike the corresponding feature in ICE2, whichattains its maximum amplitude at 500 hPa.

We have also carried out the decomposition forICEGRN, ICELAB, and SST15. The total, indirect, anddirect responses for ICEGRN (not shown) are very sim-ilar to those for ICE2 (except for the lack of a localboundary layer response over the Labrador Sea due tothe absence of sea ice forcing there), while the ICELABresponses (not shown) are weak and not statisticallysignificant (except for the boundary layer response overthe Labrador Sea). These results indicate that it is theremoval of sea ice in the Greenland Sea rather than theaddition of sea ice in the Labrador Sea that is primarilyresponsible for the circulation response in ICE2 (seealso Part I).

The total, indirect, and direct responses for SST15are shown in Fig. 3c. While similar in spatial and ver-tical structure (albeit with opposite sign) to their SST25counterparts, their magnitudes are considerably weaker.For example, the amplitudes of the total and indirectresponses throughout the troposphere are approximatelya factor of 3 smaller for SST15 than SST25. In ad-dition, the anomalous trough in the direct free-atmo-sphere response to SST15 exhibits less amplification

with height than the analogous ridge in SST25: themagnitude of the maximum height anomaly at 300 hPais ;30 m in SST15 compared to ;90 m in SST25.The fact that our empirical technique for decomposingthe total response yields meaningful patterns even forthe small amplitude response in SST15 lends furthersupport to its credibility.

In summary, the total response to positive surfacetemperature anomalies, either in the form of SST chang-es or reductions in sea ice cover, is dominated by thespatial (hemispheric) and vertical (approximately equiv-alent barotropic) structure of the leading mode of in-ternal variability of the model’s wintertime atmosphericcirculation. The structure of this ‘‘indirect’’ componentof the response contrasts with that of the residual com-ponent, which is more locally tied to the region of forc-ing and is baroclinic within the boundary layer whileamplifying with height in the free atmosphere. The am-plitude of the model response (total, direct, and indirect)is much greater for a positive surface temperature per-turbation than a negative one, both for SST and sea iceforcing.

In contrast to the asymmetry in the strength of theresponse with respect to the polarity of the forcing, themodel responds in a nearly linear fashion with respectto the amplitude of the forcing. Figure 4 shows the total,indirect, and direct Z500 responses for SST22.5 andICE1 that may be compared with their doubled-forcingcounterparts SST25 and ICE2, respectively (recall Fig.2). The spatial patterns of the responses are similar be-tween the two SST experiments and between the twosea ice experiments, and the amplitudes scale approx-imately linearly with the magnitude of the forcing. Thespatial regression coefficients between the Z500 respons-

884 VOLUME 17J O U R N A L O F C L I M A T E

FIG. 5. Vertically integrated heating rate anomalies for (top) the entire depth of the modelatmosphere (1000–10 hPa), (bottom) the boundary layer (1000–850 hPa), and (middle) the freetroposphere (750–250 hPa) for (left) SST15 and (right) SST25. The contour interval is 17.5 Wm22 in all panels and the zero contour has been omitted.

es in SST22.5 and SST25, a measure of the relativeamplitudes of the response patterns are 0.50, 0.46, and0.49 for the total, indirect, and direct components, re-spectively, and those between ICE1 and ICE2 are 0.50,0.54, and 0.38, respectively.

b. Nonlinearity of the response: The role of thevertical distribution of anomalous heating

One factor that may play a role in the stronger modelresponse to positive versus negative SST anomalies isthe nonlinear dependence of evaporation upon SST ac-cording to the Clausius–Clapeyron relation. Indeed, themagnitude of the latent heat flux anomaly averagedwithin the 2 K (absolute value) contour of the main SSTanomaly center is larger by a factor of 1.4 in SST25than SST15 (see maps in Part I figures). However, thisratio is not large enough to directly account for theapproximately three-fold increase in the amplitude ofthe total response in SST25 compared to SST15.

Another factor that may be important is the verticaldistribution of the anomalous heating. Figure 5 showsthe anomalous heating rate fields integrated over theentire depth of the model atmosphere (1000–10 hPa),

the boundary layer (1000–850 hPa), and the free tro-posphere (750–250 hPa) for SST15 and 2SST15. Inboth experiments, the anomalous heating integrated overthe depth of the atmosphere is localized to the regionsof anomalous SST. There is also anomalous heating overthe far eastern North Atlantic that is of opposite signto that farther west; we speculate this is due to thethermodynamic adjustment of the low-level southwest-erly flow to the upstream SST perturbation, as in theexperiments of Hall et al. (2001). The magnitude of theanomalous heating in the free troposphere over the cen-tral North Atlantic is approximately 2.5 times larger inSST25 than in SST15. The anomalous heating patternswithin the boundary layer are relatively linear by com-parison.

Further detail on the vertical heating distribution isgiven in Fig. 6, which shows cross sections of the anom-alous heating rates in SST25 and SST15 averaged overthe latitude band (438–658N) as a function of longitudeand height. The top panel shows the total heating rateanomalies and the lower panels show the individualcomponents that make up the total: the sum of the ra-diative and vertical diffusion terms, condensationalheating in shallow convection, and condensational heat-

1 MARCH 2004 885D E S E R E T A L .

FIG. 6. Vertical cross sections for the lat band 438–658N of anomalous heating rates for(left) SST15 and (right) SST25 due to (middle top) deep convection, (middle bottom)shallow convection, and (bottom) radiation plus vertical diffusion. (top) The sum of thelower three panels. The contour interval is 0.3 K day21 in all panels; positive (negative)values are indicated by solid (dashed) contours, and the zero contour has been omitted.

886 VOLUME 17J O U R N A L O F C L I M A T E

FIG. 7. Vertical profiles of the total anomalous heating rate averagedover the main SST anomaly center (438–658N, 508–308W) in SST15(dashed) and SST25 (solid). The SST15 profile has been multipliedby 21.

ing in total precipitation minus that in shallow convec-tion (termed ‘‘deep convection’’ in the figure). It is clearthat the total heating rate differences in the free atmo-sphere (e.g., deeper penetration of anomalous heatingcompared to cooling) are almost entirely due to differ-ences in the deep convective component that peaks inthe midtroposphere near 650 hPa. At this level, the mag-nitude of the anomalous heating due to deep convectionis approximately a factor of 3 larger for SST25 thanSST15. The anomalous heating rates due to shallowconvection and radiation plus diffusion are nearly linearby comparison. Figure 6 also shows that the eastwardtilt with height apparent in the total heating rate is dueto the eastward displacement of deep convection relativeto shallow convection.

A summary of the vertical structures of anomalousheating compared to cooling over the main SST anomalycenter in SST25 and SST15 is given in Fig. 7, whichshows area averages for the region (438–658N, 508–308W; note that the sign has been inverted for SST15for easier comparison with SST25). Although theanomalous heating within the boundary layer is similarin magnitude for the two cases, it is considerably stron-ger and deeper in the free atmosphere for the positiveSST anomaly compared to the negative one. For ex-ample, the anomalous heating at 600 hPa is ;0.7 Kday21 in SST25 compared to ;20.2 K day21 inSST15, and the level to which the anomalous heatingextends is ;300 hPa in SST25 compared to ;500 hPain SST15.

Given the stronger and deeper anomalous heating in

SST25 compared to the anomalous cooling in SST15,it is now evident why the direct component of the geo-potential height response in SST25 amplifies withheight at a considerably greater rate than that in SST15(recall Fig. 3) because the thickness between two geo-potential height surfaces must be proportional to thetemperature anomalies in the layer. The degree to whichthe vertical profile of the anomalous heating or coolingaffects the strength of the indirect response cannot bedefinitively answered without a more complete under-standing of the dynamical mechanisms responsible forthe leading mode of internal variability in the model.However, qualitatively we expect that a deeper heatingprofile will project more strongly onto the equivalentbarotropic vertical structure of the leading internal modeof variability, thus exciting it more efficiently. In thiscontext, Hall et al. (2001) have examined the sensitivityof the response of a simplified AGCM to different ide-alized vertical profiles of midlatitude heating and findthat ‘‘the pattern of the global response is not very sen-sitive to the changes in the shallow heating profile, butas the heating becomes deeper the response is strongerboth locally and in the equivalent barotropic telecon-nections.’’ Indeed, we find that the indirect response isapproximately 3 times larger in magnitude for SST25than SST15, a factor that is similar to the differencein strength of the anomalous heating above the boundarylayer in the two experiments (recall Figs. 5–7).

Of course, the anomalous heating pattern itself maybe affected by the circulation response, in particular theindirect component. To examine this possibility, we de-composed the vertically integrated heating rate responsein SST25 into indirect and direct components by re-gression upon the leading Z500 PC time series in thecontrol run [this time series is highly correlated (r 50.86) with the leading PC of the vertically integratedheating field in the control run]. The projection onto theleading internal mode (Fig. 8) exhibits a north–southdipole in the Atlantic, with enhanced heating along 408–508N and diminished heating along 608–708N associatedwith a southward shift of the mean westerly jet (see alsoFig. 10a in Part I). However, this indirect component isapproximately a factor of 5 weaker than the total re-sponse, so that the total heating rate response is dom-inated by the direct component, unlike the case for geo-potential height (recall Fig. 2). Thus, the indirect cir-culation response does not substantially impact the totalvertically integrated heating response. This result in-dicates that other processes besides heating, such aseddy momentum fluxes, are important for the mainte-nance of the internally generated variability, consistentwith previous studies (cf. Ting and Lau 1993 and Fig.14a in Part I). Analogous results are obtained for SST15and ICE2 (not shown).

Why does the anomalous heating extend so muchhigher in the warm SST experiment compared to thecold one? The answer may be simply that cooling frombelow is an inherently stabilizing process while heating

1 MARCH 2004 887D E S E R E T A L .

FIG. 8. As in Fig. 2 but for the vertically integrated total anomalous heating rate in SST25.The contour interval is 15 W m22 in all panels (the values in the middle panel have been multipliedby 5) and the zero contour has been omitted.

FIG. 9. Long-term mean winter temperature profiles. The two cold-est profiles are for the Greenland Sea region (748–858N, 208W–1008E)from the control run (solid) and ICE2 (dashed). The three warmestprofiles are for the north-central Atlantic (438–658N, 508–308W) fromthe control run (solid), SST15 (thin dashed), and SST25 (thickdashed).

from below is a destabilizing one, conducive to con-vective overturning and deeper vertical penetration. Insupport of this notion, we show mean vertical temper-ature profiles averaged over the main SST anomaly cen-ter from the control, SST25, and SST15 experiments(Fig. 9). It is clear that the positive temperature pertur-bation is sufficiently buoyant relative to the control runlapse rate that it can extend into the upper tropospherewhile the negative one is mainly confined to the lowertroposphere.

c. Sea ice versus SST forcing

In the previous section, we hypothesized that thestronger model response to positive versus negative SSTanomalies is mainly due to the deeper penetration ofanomalous heating compared to cooling. Does this ar-gument apply to the sea ice experiments, and in partic-ular, does it explain why the response to reduced icecover in the Greenland Sea (e.g., surface warming) isso much larger than that to enhanced ice cover in theLabrador Sea (e.g., surface cooling; not shown but seePart I)? Figure 10 shows the anomalous heating ratefields integrated over the entire depth of the model at-mosphere, the boundary layer, and the free tropospherefor the ICE2 experiment. The anomalous heating rateintegrated over the depth of the atmosphere exhibitslarge negative (positive) values over the enhanced (re-duced) ice cover in the Labrador (Greenland) Sea, withweaker values of opposite sign directly downstream dueto the modification of the air mass as it encounters openwater (see Part I). Peak values associated with the icecover changes are ;2300 W m22 over the LabradorSea and ;150 W m22 over the Greenland Sea (the high-er values in the Labrador Sea may be due to the strongercross-ice component of the near-surface winds that en-hances the surface turbulent energy flux anomaly; notshown). These peak values are considerably larger thanthose in the SST anomaly experiments, although theyare restricted to a smaller area. In contrast to the resultsfor SST25, the anomalous heating over the GreenlandSea is confined almost entirely to the lower troposphere.We note that the shallowness of the anomalous heatingprofile over the Greenland Sea is consistent with thevertical structure of the direct response (recall Fig. 3a)in that the anomalous ridge directly over the reducedice cover does not amplify above ;700 hPa.

The results shown in Fig. 10 raise two questions: 1)why does the anomalous heating associated with re-duced ice extent in the Greenland Sea not penetrate asdeeply as that associated with a warm SST anomaly in

888 VOLUME 17J O U R N A L O F C L I M A T E

FIG. 10. As in Fig. 5 but for ICE2.

the North Atlantic, and 2) if the anomalous heating overthe Greenland Sea is so shallow, why is the indirectcomponent of the geopotential height response so large(as large as the response in SST25)? The mean verticaltemperature profiles over the Greenland Sea (748–858N,208W–1008E) in the control and ICE2 experiments,shown in Fig. 9, provides some insight into the firstquestion. When sea ice is present (in the control run),the lower troposphere is nearly isothermal and conse-quently, very large positive surface temperature pertur-bations are needed to overcome the strong static sta-bility. Indeed, a surface temperature anomaly of 17 K(ICE2-minus-control) is only large enough to penetrateto ;700 hPa. In contrast, a surface temperature pertur-bation of only 4 K in SST25 is sufficient to penetrateto ;350 hPa due to the weaker stratification over theice-free North Atlantic (Fig. 9). Thus, we speculate that

the mean static stability in the region of the surfacetemperature perturbation plays a decisive role in howdeep the anomalous heating can penetrate.

Given that the anomalous heating over the GreenlandSea in the ICE2 experiment does not penetrate as deeplyas that over the North Atlantic in SST25, why are theindirect responses in free atmosphere comparable in thetwo experiments? Put another way, why is the leadingmode of internal variability excited so strongly in theICE2 (and ICEGRN) experiment? Although we do nothave a definitive answer to this question, we speculatethat the ice anomaly in the Greenland Sea may be lo-cated in a region to which the model’s internal vari-ability is particularly sensitive. In this context, we notethat the Greenland Sea ice perturbation is broad in zonalextent and located near the Arctic center of action ofthe leading Z500 EOF in the control simulation, perhapsresulting in an efficient projection upon the spatial pat-tern of the dominant mode of internal variability. An-other possible factor is the interaction of the flow in theforcing region with the upstream topographic barrier ofGreenland, which may result in a strong projection uponthe internal mode. Further work is planned to investigatethese issues.

4. Summary and further discussion

We have decomposed the simulated total geopotentialheight responses to North Atlantic SST and sea ice coveranomalies into a portion that projects onto the leadingmode of internal variability (the indirect response) andthe residual from that projection (the direct response).We suggest that this empirical decomposition yieldsphysically meaningful patterns with distinctive hori-zontal and vertical structures, as discussed below.

a. The direct response

The direct component of the response to a positiveNorth Atlantic SST anomaly (SST25) exhibits a bar-oclinic structure in the vertical, with a surface troughdirectly over the SST anomaly and an equivalent bar-otropic ridge in the free atmosphere over and slightlydownstream of the SST anomaly. This structure is qual-itatively consistent with simple theoretical models ofthe linear baroclinic response to extratropical thermalforcing (see, e.g., the review by Kushnir et al. 2002).A similar pattern of opposite sign is found for the directresponse to a negative SST anomaly (SST15), but theamplitude is weaker by a factor of 2–3. The weakerresponse in SST15 than SST25 may be a result of thereduced magnitude and vertical extent of the anomalouscooling associated with a negative SST anomaly com-pared to anomalous heating associated with a positiveone. We postulate that the shallower vertical extent ofa negative heating anomaly compared to a positive oneis due to the stabilizing effect of surface cooling versusthe destabilizing effect of surface heating. Unlike the

1 MARCH 2004 889D E S E R E T A L .

asymmetry in the strength of the direct response to pos-itive versus negative SST forcing, the direct responseis approximately linearly proportional to the magnitudeof the imposed SST anomaly.

The direct response to reduced sea ice cover in theGreenland Sea is analogous to the direct response inSST25 except that the anomalous ridge in the free at-mosphere does not penetrate as deeply (e.g., does notamplify with height above ;700 hPa). We suggest thatthis is due to the more limited vertical extent of theanomalous heating resulting from the higher mean staticstability in the lower troposphere associated with thepresence of sea ice in the control run. As in the SSTforcing experiments, the direct response to anomaloussea ice cover is nonlinear with respect to the polarityof the forcing and approximately linear with respect tothe amplitude of the forcing.

b. The indirect response

The indirect response, or the component of the re-sponse that projects onto the leading mode of internalvariability, dominates the total geopotential height re-sponse in both SST25 and ICE2. Like the direct re-sponse, the magnitude of the indirect response scaleslinearly with the amplitude of the forcing but nonli-nearly with the sign of the forcing. Unlike the directresponse, the horizontal structure of the indirect re-sponse is hemispheric in scale, resembling the NorthAtlantic Oscillation or Northern Hemisphere annularmode, and the vertical structure is approximately equiv-alent barotropic from the surface to the tropopause.

Many outstanding issues regarding the indirect re-sponse remain to be investigated, including an under-standing of the dynamical processes responsible for theleading internal mode of variability including essentialinteractions between the transient and stationary eddiesand the mean flow, the factors that determine the polarityof the indirect response and the relative strengths of theindirect and direct responses, and the mechanisms re-sponsible for the nonlinear behavior of the indirect re-sponse with respect to the sign of the forcing. Recently,Peng et al. (2003) have proposed that nonlinear inter-actions between the anomalous heating and eddy-in-

duced components of the flow can lead to asymmetriesin the response to positive versus negative SST anom-alies. Idealized experiments are planned to examinethese issues, including the sensitivity of the indirect re-sponse to the location and polarity of the forcing andto the vertical profile of the associated anomalous heat-ing, as well as possible nonlinear interactions betweenthe anomalous heating and eddy forcing.

Acknowledgments. We wish to thank Dr. Michael Al-exander for useful discussions during the course of thiswork and the three anonymous reviewers for helpfulcomments. We are also grateful to Dr. Christophe Cas-sou for his technical assistance. GM was supported byNOAA OGP under Grant NA96GP0420.

REFERENCES

Deser, C., J. E. Walsh, and M. S. Timlin, 2000: Arctic sea ice var-iability in the context of recent atmospheric circulation trends.J. Climate, 13, 617–633.

Hall, N. M. J., J. Derome, and H. Lin, 2001: The extratropical signalgenerated by a midlatitude SST anomaly. Part I: Sensitivity atequilibrium. J. Climate, 14, 2035–2053.

Kushner, P., I. Held, and T. L. Delworth, 2001: Southern Hemisphereatmospheric circulation response to global warming. J. Climate,14, 2238–2249.

Kushnir, Y., W. A. Robinson, I. Blade, N. M. J. Hall, S. Peng, andR. Sutton, 2002: Atmospheric GCM response to extratropicalSST anomalies: Synthesis and evaluation. J. Climate, 15, 2233–2256.

Magnusdottir, G., C. Deser, and R. Saravanan, 2004: The effects ofNorth Atlantic SST and sea-ice anomalies on the winter circu-lation in CCM3. Part I: Main features and storm track charac-teristics of the response. J. Climate, 17, 857–876.

Peng, S., and W. A. Robinson, 2001: Relationships between atmo-spheric internal variability and the responses to an extratropicalSST anomaly. J. Climate, 14, 2943–2959.

——, ——, and S. Li, 2003: Mechanisms for the linear and nonlinearNAO responses to the North Atlantic SST tripole. J. Climate,16, 1987–2004.

Seager, R., Y. Kushnir, M. Visbeck, N. Naik, J. Miller, G. Krahmann,and H. Cullen, 2000: Causes of Atlantic Ocean climate vari-ability between 1958 and 1998. J. Climate, 13, 2845–2862.

Ting, M., and N.-C. Lau, 1993: A diagnostic and modeling study ofthe monthly mean wintertime anomalies appearing in a 100-yearGCM experiment. J. Atmos. Sci., 50, 2845–2867.

Wallace, J. M., 2000: North Atlantic Oscillation/annular mode: Twoparadigms—One phenomenon. Quart. J. Roy. Meteor. Soc., 126,791–805.