Seamless Poleward Atmospheric Energy Transports … · 3706 JOURNAL OF CLIMATE VOLUME 16 q 2003 American Meteorological Society Seamless Poleward Atmospheric Energy Transports and

3706 VOLUME 16J O U R N A L O F C L I M A T E

q 2003 American Meteorological Society

Seamless Poleward Atmospheric Energy Transports and Implications for theHadley Circulation

KEVIN E. TRENBERTH AND DAVID P. STEPANIAK

National Center for Atmospheric Research,* Boulder, Colorado

(Manuscript received 7 January 2003, in final form 20 May 2003)

ABSTRACT

A detailed vertically integrated atmospheric heat and energy budget is presented along with estimated heatbudgets at the surface and top-of-atmosphere for the subtropics. It is shown that the total energy transports areremarkably seamless in spite of greatly varying mechanisms. From the Tropics to about 318 latitude, the primarytransport mechanisms are the Hadley and Walker overturning circulations. In the extratropics the energy transportsare carried out by baroclinic eddies broadly organized into storm tracks and quasi-stationary waves that covaryin a symbiotic way as the location and activity in storm tracks are determined by, and in turn help maintainthrough eddy transports, the quasi-stationary flow. In the upward branch of the Hadley cell, the predominantdiabatic process is latent heating that results from convergence of moisture by the circulation itself. Hence largepoleward transports of dry static energy are compensated by equatorward transports of latent energy, resultingin a modest poleward transport of moist static energy. The subsidence warming in the downward branch iscompensated by cooling in the subtropics that mainly arises from energy transport to higher latitudes by transientbaroclinic eddies that are stronger in the winter hemisphere. Effectively, the outgoing longwave radiation tospace is distributed over middle and high latitudes and is not limited to the clear dry regions in the subtropics.Further, some of the radiative cooling in the subtropics is a consequence of the circulation. Hence the coolingby transient eddies in the subtropics is a fundamental driver of the observed Hadley circulation and realizes theseamless transport from Tropics to extratropics, while tropical sea surface temperatures over the oceans determinewhere the upward branch is located. The relatively clear skies in the subtropics further provide for ampleabsorption of solar radiation at the surface where it feeds strong evaporation, which exceeds precipitation, andsupplies the equatorward flow of latent energy into the upward branch of the Hadley circulation as well as thepoleward transports into midlatitude storm tracks. The evaporation is sufficiently strong that it is also compensatedby a subsurface ocean heat transport that in turn is driven by the Hadley circulation surface winds.

1. Introduction

A primary driver of the atmospheric circulation is therequirement to balance the global heat and energy bud-gets. The sun–earth orbital geometry greatly determinesthe distribution of incoming solar radiation, while theoutgoing longwave radiation (OLR) is more or less uni-formly distributed, and these fundamental factors man-date a poleward heat transport by the fluid componentsof the climate system: the atmosphere and ocean. Otherinfluences, such as the surface albedo and cloud effects,are of secondary importance, and so the global heatbudget provides a strong constraint on the climate sys-tem (Stone 1978; Trenberth and Stepaniak 2003). Ob-servations show that not only is the total poleward heat

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

Corresponding author address: Dr. Kevin E. Trenberth, ClimateAnalysis Section, NCAR, P.O. Box 3000, Boulder, CO 80307-3000.E-mail: [email protected]

transport continuous with latitude, so too is the atmo-spheric transport (Trenberth and Stepaniak 2003). Yetthe mechanisms for carrying out the transport varygreatly. The large-scale overturning Hadley circulationis dominant in low latitudes while the baroclinic tran-sient eddies, assisted by the quasi-stationary planetarywaves in the Northern Hemisphere winter, are dominantin midlatitudes (Trenberth and Solomon 1994). Yet mostof the theories (discussed further below) of the over-turning circulations have an abrupt cutoff at about 308latitude at the poleward edge of the Hadley circulationfor the heat transport. So how is it that the polewardheat transports are so seamless from the tropical to theextratropical regime?

The total poleward heat transport peaks near 358 lat-itude in each hemisphere, while the poleward atmo-spheric transport peaks near 408 latitude (Trenberth andCaron 2001). Heating occurs in the Tropics, where it isalready hot, and cooling at higher latitudes where it iscold, and heat can become manifested in various formsof energy. Strictly speaking, there has to be a transportof energy by the atmosphere and ocean from source to

15 NOVEMBER 2003 3707T R E N B E R T H A N D S T E P A N I A K

sink. In the atmosphere, most of the transport occurs asmoist static energy (MSE). Hence, overall, air parcelsmoving poleward must have a higher MSE than air par-cels moving equatorward. Trenberth and Stepaniak(2003) document the global vertically integrated at-mospheric heat budget as zonal means for the annualmean, mean annual cycle, and the interannual variabilityfor 1979–2001. They further examine the transports andtheir divergences partitioned into contributions fromwithin-month transients and quasi-stationary compo-nents; in the Tropics the latter correspond to the large-scale overturning global monsoon and the embeddedHadley and Walker circulations. The results highlightthe strong cancellations between the transports of latentand dry static energy in the Tropics as moisture is con-verted into latent heat, and also between quasi-station-ary waves and transient components in the extratropics.Hence the total atmospheric energy transports and di-vergences are remarkably seamless with latitude and thetotal interannual variability is substantially less than thatof the components.

From a diabatic heating diagnostic standpoint, theHadley cell overturning is driven by heating in the deepTropics and cooling in the subtropics. It is widely as-sumed that the primary heat balance in the subsidingbranch of the Hadley circulation is between the adiabaticwarming from subsidence and the diabatic effects ofinfrared radiative cooling to space. Yet how can this be,given the seamless poleward heat transports? Insteadthe continuity of the heat transports across latitude im-plies that other dynamical mechanisms are also playingkey roles. In particular, as we show here, the coolingin the subtropics must also arise from heat transports tohigher latitudes by the transient baroclinic eddies andquasi-stationary waves. Effectively, energy is trans-ported from the subtropics to higher latitudes, and thusthe radiation to space is distributed over middle andhigh latitudes and is not limited to the clear dry regionsin the subtropics. This discussion therefore raises keyquestions about the role of the transients in setting upthe heating gradients that drive the Hadley and mon-soonal circulations.

The relatively clear skies in the subsiding air in thesubtropics mean that solar radiation is abundant at thesurface, and it heats the ocean. The subtropical highsare also the regions of highest evaporation on the globe(Trenberth and Guillemot 1995, 1998) not coinciden-tally, and some moisture is transported poleward bytransient eddies and some equatorward by the lowerbranch of the Hadley and Walker circulations to be re-alized as latent heat of condensation in the intertropicalconvergence zone (ITCZ) and South Pacific conver-gence zone (SPCZ). In turn this fuels the overturningand controls the size and magnitude of the upwardbranch as a powerful feedback. While the low-levelwinds transport latent energy equatorward, thereby di-minishing the total poleward energy transport, they pro-vide the dominant fuel to drive the upward branch of

the overturning. Accordingly, the overturning transportsare seen more clearly in the dry static energy (DSE)than in the MSE because of large compensation thatoccurs in the latter as moisture fluxes are converted intosensible heat through release of latent heat (Trenberthand Solomon 1994; Trenberth and Stepaniak 2003).However, a second powerful positive feedback also op-erates in the subtropics because the establishment of thesubsiding air and its relative dryness enhances radiativecooling to space, which helps compensate for the adi-abatic warming in the downward branch of the circu-lation (Hoskins 1996).

Another way of comparing the DSE and MSE is tonote that the overturning is driven by the atmosphericdiabatic heating, Q1, but that this in turn is dependenton the condensational latent heating arising from mois-ture convergence. It thus depends on moisture transportas well as the moisture sources. Column-integrated la-tent heating is often referred to as Q2. Under a steadystate or where the tendency terms are small, the diver-gence of the atmospheric energy transports is = ·FA øQ1 2 Q2 (Trenberth and Stepaniak 2003).

The discussion has focused on the zonal mean pictureof the atmospheric general circulation and the polewardenergy transports across latitudinal zones and, althoughthese are dominant, zonal asymmetries are also impor-tant (Cook 2003). The latter arise from the distributionof land and surface topography, in particular, and giverise to regional monsoons and subtropical anticyclones(Trenberth et al. 2000; Rodwell and Hoskins 2001; Chenet al. 2001). Heat transports by the quasi-stationaryplanetary waves are substantial in the Northern Hemi-sphere in winter. In the Tropics, the zonal mean Hadleycirculation is but part of the deep tropospheric over-turning mode, called the ‘‘global monsoon’’ by Tren-berth et al. (2000), which also encompasses the Walkercirculation. The dominant energy outflow center in theatmosphere coincides closely with the region of highestsea surface temperatures (SSTs) in the oceans and mi-grates back and forth across the equator following thesun (Trenberth et al. 2000). But what determines thelocation of the downward branch of the Hadley circu-lation and its change with the seasons?

A number of studies have explored the extent towhich the Hadley circulation and more general ther-mally driven overturning circulations can be explainedwith fairly simple theories that exclude the role of syn-optic transient eddies and other high-frequency vari-ability; see Lorenz (1967) and Newell et al. (1974) forhistorical perspectives. Schneider (1977) developed amodel for a zonally symmetric steady state that wassubsequently expanded by Held and Hou (1980), Lind-zen and Hou (1988), and Hou and Lindzen (1992) toinclude effects of the distribution of heating, related tolatent heat feedback, and displacement of the upwardbranch away from the equator to be more representativeof the winter/summer extremes. Schneider (1987) andEmanuel (1995) generalized the model to include non-


zonal monsoonal-type circulations. These theories aresuccessful in accounting for certain features of the trop-ical tropospheric circulation, notably the width of theHadley circulation and the position of the subtropicaljet, which are controlled by geostrophy and conserva-tion of heat and momentum. Fang and Tung (1999) in-troduced time-varying heating and found the mean Had-ley circulation was about a factor of 2 stronger than thesteady-state case. Many of these studies recognized thatmidlatitude eddies play an important role in enhancingthe Hadley circulation (e.g., Schneider 1984). Kim andLee (2001) investigated the effects of baroclinic eddieson the zonally symmetric Hadley cell using a simplifiedmodel symmetric about the equator and found that theeddies contribute to a greatly strengthened Hadley cellthrough direct effects as well as indirect effects (feed-backs) on the diabatic heating and surface friction. How-ever, all of these theories and models use a radiative–convective equilibrium base state of some sort, perhapsgeneralized into some form of Newtonian heating/cool-ing. It is evident that latent heating arising from con-vergence of moisture into the upward branch of theoverturning is crucial in defining the detailed structureand intensity of the circulation, but such effects are notfully accounted for in these models.

The importance of the downward branch of the Had-ley circulation has also been greatly emphasized re-cently because of its role in drying the subtropical at-mosphere above the boundary layer, and the possiblerole this may play in water vapor feedback (Pierrehum-bert 1995, 1999; Spencer and Braswell 1997; Lindzenet al. 2001; Salathe and Hartman 1997). The latter studynotes the swirling paths of motions that carry moisturefrom source regions into the upper-troposphere sub-tropics and that such tracks are not adequately reflectedby a zonally averaged Hadley circulation. Yet the pre-dominant arguments rest heavily on the idea that steady-state large-scale subsidence, driven by infrared radiativecooling to space, is the primary governing process. Zon-al asymmetries and the origin of the subtropical anti-cyclones have been explored by Rodwell and Hoskins(1996, 2001) and Chen et al. (2001) using idealizedmodels with specified heating and implied cooling (suchas through relaxation to a specified basic state, or be-cause the zonal mean has to be zero). Although descentoccurs into the subtropical anticyclones, the associatedadiabatic warming has to be balanced either by diabaticprocesses or horizontal atmospheric transports, and inthese models it is a by-product of the way the modelsare forced.

In addition, the Hadley and Walker circulations arefar from steady and the level of the variability in mon-soon areas is quite high. Trenberth et al. (2000) showthat the mean vertical motion in the peak monsoon sea-sons is only about 60% of the standard deviation of thedaily mean values in the monsoon areas—although thisis much greater than in other regions. Many transientsplay a role of mixing, both in the vertical and horizontal

and, as we show here, extratropical storms commonlyhave an influence in winter into the Tropics. Rodwelland Hoskins (1996) point out, theoretically and in agree-ment with observations, that the subtropics are whereboth vertical and horizontal advection are of comparableimportance in balancing diabatic heating. Therefore keyquestions of interest are to determine what the dominantterms and processes are in the heat budget of the sub-tropics.

There are a number of interesting issues related tothe vertical structure of the atmosphere and the energytransports and how the transports come about, such asthrough the Hadley and Walker overturning circulations,but those issues will be taken up elsewhere as resultsdepend on the datasets used. Instead, in this work, weexplore what the latest datasets can tell us about thepoleward heat transports and their divergences in a ver-tically integrated framework. Trenberth and Stepaniak(2003) provide the background to this paper. They out-line the datasets used and the theoretical framework,and document the global vertically integrated atmo-spheric heat budget as zonal means for the annual mean,mean annual cycle, and the interannual variability. Here,we examine in detail the full geographical structure ofthe heating, the atmospheric transports, and their di-vergences, with a special focus on the implications forthe Hadley circulation dynamics.

Accordingly, the heat budget is analyzed in detail inthe Tropics and subtropics and the relative roles of theoverturning and transients in the heat transport fromsource to sink are determined. Section 2 briefly dis-cusses the datasets used. The physical background andmathematical expressions for the transports and theircomponents are given in Trenberth and Stepaniak (2003)and briefly outlined in section 3. The vertically inte-grated heat budget and its breakdown into the mean andtransients, the components of dry static energy, latentenergy, and kinetic energy, and the rotational and di-vergent circulation components are presented in section4. Detailed estimates of heat budgets are made in thesubtropics of both hemisphere, not only for the atmo-sphere, but also at the top-of-atmosphere (TOA) and thesurface. Section 5 discusses the results.

2. Methods and data

The overall energy transports are derived from theNational Centers for Environmental Prediction–Nation-al Center for Atmospheric Research (NCEP–NCAR) re-analyses (Kalnay et al. 1996) as given in Trenberth etal. (2001) because they are most consistent with theoverall heat budget based upon TOA and ocean mea-surements (Trenberth and Caron 2001). Hence we usethose reanalyses from 1979 to 2001 to examine the com-ponents (see Trenberth and Stepaniak 2003).

The atmospheric transports vector FA can be split intorotational and divergent parts and a potential functionx can be defined such that = ·FA 5 ¹2x. The rotational


component is large but, as it is not linked to the diabaticforcings, we focus on the divergent component of thetransports. Plots of x depict a highly smoothed versionof the divergence, which is not especially revealing.Instead we have experimented with different smoothingsin order to emphasize the large-scale features of thecirculation and we settled on a T31 truncation with ataper to avoid ringing effects for the figures. Experiencesuggests that the features of interest are well picked upalthough their intensity and some gradients are under-estimated. While higher resolution adds greater reality,it also adds complexity and makes the figures less read-able. However, all computations are performed at T63resolution.

The transports and components are split into thoseassociated with the time mean circulation, given by theoverbar, and the transients, given by the prime, so thath 5 1 h9, for example. The transients are defined tohbe the within-month variability and the quasi-stationarycomponent includes the interannual variability.

Complementary estimates of the heat budget are ob-tained for the surface from the Southampton Oceano-graphic Centre (SOC) heat budget atlas based on surfacemarine data (Josey et al. 1998, 1999). Sampling uncer-tainties and systematic biases arising from bulk fluxparameterizations are substantial in the results (Tren-berth et al. 2001), but they are nevertheless the bestavailable and can be useful if physical constraints areutilized, as is done here. At the TOA we use the EarthRadiation Budget Experiment (ERBE) measurements ofradiation (Trenberth 1997). We also use the ClimatePrediction Center (CPC) Merged Analysis of Precipi-tation (CMAP) precipitation estimates from Xie and Ar-kin (1997). For temporal consistency, we use monthlyvalues from the SOC and CMAP data for the ERBEperiod February 1985–April 1989.

3. Physical framework

The mathematical expressions for the vertically in-tegrated energetics of the atmosphere are given in Tren-berth and Stepaniak (2003). The total atmospheric en-ergy AE can be broken up into components of kineticenergy KE, internal energy IE, potential energy PE, andlatent energy LE; the total energy AE 5 PE 1 IE 1 KE

1 LE is conserved in the absence of forcings. However,when the transports of energy are considered, there isa difference between the total energy and the quantitytransported that arises from the pressure–work term inthe thermodynamic equation and the conversion of ki-netic to internal energy. We thus need to consider theMSE transports, which in turn can be broken up intocomponents from the DSE, and latent energy (LE):

]AE 5 2= · F 1 Q 2 Q 2 Q , (1)A 1 f 2]t

where the total atmospheric transport

F 5 F 1 FA M KE

and F is the MSE flux. The total energy flux includesME

the kinetic energy component (second term) but, asshown in Trenberth et al. (2002), this is small. Conse-quently the atmospheric energy transports FA predom-inantly correspond to the MSE transports:

F 5 F 1 F ,M D LE E E

where the terms on the rhs are the DSE and LE fluxes.The atmospheric diabatic heating is˜Q 2 Q 5 R 2 R 1 H 1 LP 5 R 2 F 1 Q ,1 f T s s T s 2

where Q2 5 L(P 2 E) is the column latent heating; Qf

the (small) frictional heating; RT and Rs the net down-ward radiation at the TOA and the surface respectively,Hs the sensible heat flux at the surface, and Fs 5 LEs

1 Hs 2 Rs the net upward flux at the surface. The tilderefers to vertical integrals (see Trenberth and Stepaniak2003).

4. Results for the energy budget

To set the stage for subsequent analysis, we first pre-sent some aspects of the divergent circulation to definethe realms of interest. Trenberth et al. (2000) provideglobal seasonal figures of the Hadley circulation andglobal monsoon up to 10 mb, and here we focus on themean divergent circulation from about 408N to 408S forthe zonal average (left) and a Pacific sector 1408E–1408W (right) for December–February (DJF) and theannual mean below 100 mb (Fig. 1). As the zonal meancirculation is closed, streamfunctions of mass flow con-tours are added to show the Hadley circulations. For thePacific sector, only the average flow can be indicated,weighted appropriately by mass (Trenberth et al. 2000).These figures reveal the extent of the Hadley circulationand where the subsidence in the subtropics transitionsfrom that identified with the Hadley circulation to thatassociated with the Ferrel cell at midlatitudes. In DJFthe Hadley circulation extends from 138S to 318N bothregionally over the Pacific and for the zonal mean. Forthe annual mean, the subsidence extends to 318N and318S, with the rising motion centered north of the equa-tor between 108S and 158N and a somewhat strongersouthern Hadley cell. Accordingly, we will later focuson zones on both sides of about 308 latitude.

a. Annual means

The spatial patterns of annual mean heating and en-ergy transports and their divergences are given in thissection. The annual means of the TOA radiation, basedon ERBE measurements from February 1985 to April1989, are given in Fig. 2. Although the absorbed solarradiation (ASR) and the OLR both depict distinctivesignatures of clouds, the net radiative heating exhibitswell-known but nevertheless remarkable cancellation


FIG

.1.

The

(top

)D

JFan

d(b

otto

m)

annu

alm

ean

dive

rgen

tci

rcul

atio

nfr

omab

out

408N

to40

8Sfo

rth

e(l

eft)

zona

lav

erag

ean

d(r

ight

)14

08E

–140

8W.

Str

eam

line

sof

the

mas

sfl

owar

esh

own

inth

ele

ft-h

and

pane

lsw

ith

cont

our

inte

rval

303

109

kgs2

1an

dva

lues

grea

ter

than

60un

its

shad

ed.

The

mas

ssc

ale

vect

oris

inup

per-

righ

tco

rner

ofea

chpa

nel.


FIG. 2. Annualized mean TOA ERBE measurements for the period Feb 1985–Apr 1989 for theASR, OLR, and net in W m22. The color key is under each plot and the contour interval is 20W m22. Zonal mean profile panels are given at right.

for most clouds; the main exception being the low stra-tocumulus cloud decks such as in the eastern Pacific.

Figure 3 provides the inferred annual mean diabaticheating Q1 2 Qf , latent heating Q2, and net heating Q1

2 Qf 2 Q2. The latter is equal to the net downwardradiation at TOA (Fig. 2) plus the net surface flux ofheat into the atmosphere. Hence the difference with thenet radiation in Fig. 2 over the oceans is due to ocean

heat transports. Over land, for the annual mean, differ-ences should be small, as there is only a very smallchange in energy storage. These aspects are discussedin detail in Trenberth et al. (2001) and Trenberth andCaron (2001), who note problems especially in moun-tain areas. However, such problems are greatly reducedover the oceans. For instance, the equatorial minimumin the zonal mean in the lowest panel of Fig. 3 is seen


FIG. 3. Annual mean heating components for 1979–2001 in W m22. Shown are the (top)atmospheric diabatic heating Q1 2 Qf (middle) column latent heating Q2, and (bottom) totalcolumn atmospheric heating Q1 2 Qf 2 Q2. Zonal mean profile panels are given at right.

to arise mainly from cooling over the cold tongue ofSSTs in the equatorial Pacific where radiation enters theocean in relatively clear skies. Perhaps the most strikingfeatures in the lowest panel are the large positive regionsoff the east coasts of the northern continents, signalingthe surface sensible and latent heat fluxes from the oceaninto the atmosphere in these regions in the winter halfyear. Again, the difference with the lowest panel in Fig.2 highlights the fact that these are compensated by oceantransports of heat.

Recall that the diabatic heating includes the latentheating from precipitation but not from evaporation atthe surface. The figure reveals the strong zonal asym-metries associated with the land distribution that makethe zonal means (Trenberth and Stepaniak 2003) onlya part of the picture. The ITCZ and SPCZ are evidentin the top two panels over the Pacific, along with thecooling and moistening in the South Pacific high. To-gether, the South Pacific high and the SPCZ form partof the Walker circulation. The evaporative sources of


FIG. 4. Annual mean divergent energy transport for 1979–2001 asvectors and the divergence in W m22. Shown are the atmosphericenergy transports for the (top) total, (middle) quasi-stationary, and(bottom) transient components.

FIG. 5. Annual mean divergent northward DSE transport for 1979–2001 as vectors and the divergence in W m22. DSE transports forthe (top) total, (middle) quasi-stationary, and (bottom) transient com-ponents.

moisture in the subtropical highs are strongly evidentin the second panel (Fig. 3). Typical values of E through-out the tropical oceans range from 3 to 5 mm day21

(Trenberth and Guillemot 1998) and lack much spatialstructure. They correspond to about 290 to 2145 Wm22 of cooling, as is seen in the subtropical anticy-clones. Hence the structure in Q2 arises primarily fromthe P field. Monsoon-related heating over central andnorthern South America and central Africa are present,and the similarity between patterns in the top two panelsindicates the primary role of precipitation and thus latentheating.

The bottom panel of Fig. 3 looks almost exactly thesame as the divergence of the atmospheric energy trans-ports, given in Fig. 4 in the top panel, as it should, as

the difference is the change in storage (tendency term),which is small for a long-term average. However, thisfigure also allows presentation of the actual divergentvector transports and their partitioning into components.Note the flow of atmospheric energy from the sourceregions to the sinks, many of which are at higher lati-tudes. For instance these include the source in the west-ern tropical Pacific and the sink over the tropical easternPacific, where the energy flow is via the Walker cir-culation.

Figure 4 also provides the quasi-stationary and tran-sient components. The latter highlights the storm tracksover the North Pacific and North Atlantic and aroundthe Southern Hemisphere as dipole structures with di-vergence on the equatorward side and convergence at


FIG. 6. Annual mean divergent northward LE transport for 1979–2001 as vectors and the divergence in W m22. LE transports for the(top) total, (middle) quasi-stationary, and (bottom) transient com-ponents.

higher latitudes, as the baroclinic storms transport heatand energy poleward. The quasi-stationary componentdominates throughout the Tropics but provides a strongcomplement to the transient component in the extra-tropics. Hence the pronounced negative correlation, not-ed by van Loon (1979) for the geostrophic sensible heattransports and by Trenberth and Stepaniak (2003) forthe zonal means, also carries over to local spatial struc-tures.

The breakdown of these three panels into the DSEand LE components is given in Figs. 5 and 6. Onceagain, both of these feature the ITCZ and SPCZ struc-tures in the total and quasi-stationary parts that are notapparent in Fig. 4 for the MSE. Of particular note isthe strong region of convergence of DSE by the quasi-

stationary flow in the subtropics of the Northern Hemi-sphere over the Pacific and Atlantic (Fig. 5). There isa direct link from this back to the tropical source in theITCZ and other monsoonal rainfall regions. Meanwhile,the moisture flow (Fig. 6) is from the subtropics to therain regions.

b. Regional and seasonal relationships

In the Southern Hemisphere, the stronger zonal sym-metry allows the storm track transient statistics toemerge from the zonal mean statistics, along with theirrelationship to the quasi-stationary component. Never-theless, in the subtropics about 22% of the zonal meanof the Southern Hemisphere is land (Australia, SouthAmerica, and southern Africa). However, in the North-ern Hemisphere, the storm tracks are distinctive onlyover the Pacific and Atlantic. The Atlantic storm trackis oriented somewhat northeast–southwest whereas thePacific storm track is more zonal (lower panel Fig. 5).Accordingly it is relatively easy to isolate and relatefeatures in the Pacific by simply extracting zonal meansover a section such as 1408E to 1408W, which is overthe ocean away from the influences of Asia and Japanalthough it misses the storm track entry region. Asshown in Fig. 1 it is also a region where the Hadleycirculation is stronger than the zonal mean by a factorof 2 or so but with the same dimensions. Note, however,that the interpretation of regional sector results is com-plicated by the presence of the rotational transports andthe apparent overturning mass flows do not depict thoseof particles.

The zonal kinetic energy transport is significant inDJF (convergence ;30 W m22 near 308N) and for theannual mean but has been omitted from the plots in theinterests of simplicity. Figure 7 shows the annual meantotal energy (called TE in figures) across this sectionfor the total, quasi-stationary, and transient componentsalong with their breakdown into DSE and LE parts. Thetotal, quasi-stationary and transient components are thenshown together on the same plots for the four seasonsDJF, etc., in Fig. 8.

The pattern for the transients in all seasons is quitesimilar to that of the annual mean (Fig. 7). Divergenceof DSE and LE out of the subtropics and convergenceat higher latitudes demonstrates the poleward transportsacross midlatitudes and with the LE component nearlyin phase with, but shifted somewhat equatorward of, theDSE contribution. However, the transient role is muchweaker in summer (JJA) and shifted farther poleward.Typically between 208 and 608N the quasi-stationarycomponent of DSE divergence has opposite sign to thetransient component but does not completely cancel it.In particular, in DJF strong divergence of DSE from 308to 458N is complemented by LE divergence out of thesubtropics occurring in the stationary waves, with con-vergence at higher latitudes.

In the Tropics, the total divergence (Fig. 8) closely


FIG. 7. The divergence of the (top) annual mean total, (middle)quasi-stationary, and (bottom) transient energy for the total energy(TE) and the components from the DSE and LE zonally averagedacross the Pacific 1408E–1408W. Units are W m22.

matches that of the quasi-stationary component and fea-tures the DSE and LE with opposite signs, but with anet divergence, all of which is a characteristic signatureof the Hadley circulation. Of note is the double humprelated to the ITCZ in the Northern Hemisphere and theSPCZ in the Southern Hemisphere, with the latter stron-gest in DJF near 118S for this zone, but with the northernmaximum strong in both JJA and SON and farthest northat 108N in SON. In this sector of the Tropics, ratherthan migration of the region of upward motion typicalin monsoon areas, modulation of the intensity of quasi-permanent features occurs.

For DJF the picture is governed by the substantialextratropical stationary wave transports associated withthe Aleutian low and Siberian high pressure systems.These transports are almost entirely of the DSE, andthey set up the mean state that the baroclinic stormswork on. Hence the stationary waves are responsiblefor the location of the storm track (Branstator 1995).The maximum divergence of total energy at 358N (Fig.8) arises from offset peaks in both the transient and thequasi-stationary components. The divergence out of thesubtropics by the transients is partially compensated byconvergence in the downward branch of the local Had-ley circulation, but energy is also gained from the sur-face (see also section 4c), especially in the form of latentenergy evaporated from the oceans.

In SON (Fig. 8) the pattern is somewhat similar ex-

cept the quasi-stationary wave DSE divergence is re-duced from 190 to 120 W m22 and shifted polewardfrom about 408 to 508N. The transients remain verystrong and, because SSTs are near their maximum inSeptember, feed on the warm surface ocean waters forenergy. In MAM (Fig. 8), the pattern is closest to theannual mean and the magnitude of the transient com-ponent is reduced from winter and autumn as the oceanis depleted of heat. In summer (JJA) the transients areweakest but still distinctive with the storm track near488N. Monsoonal overturning helps feed the stormtrack, and LE emanating from the large subtropical highis significant.

In DJF and MAM a strong symbiotic relationship isapparent between the transients and the subsidence inthe subtropical highs (convergence of DSE 208–308Nin Fig. 8) associated with the downward branch of thelocal Hadley circulation in which the transients feed onthe warmth, while helping to drive the subsidence. Re-markable cancellation also exists near 508N where con-vergence by the transients feeds the quasi-stationarywave component and, at the same time, the divergenceof energy by the quasi-stationary waves is balanced bythe transport of energy by the transients into the regionand determines the location of the storm tracks.

c. Subtropical heat budgets

Given the new estimates of the atmospheric trans-ports, we have also estimated the TOA, surface, andatmospheric heat budgets for the ERBE period in thesubtropics to provide a more complete view of the pro-cesses involved. Based on Fig. 1, we have chosen tofocus on the oceans in two zones, from 258 to 308 and308 to 358 latitude in both hemispheres. The former isclearly part of the Hadley cell subsidence.

We have devised heat budgets of the quantities illus-trated in Fig. 9. The net radiation RT is the sum of theASR and OLR from ERBE with appropriate sign; ASRis directed downward and OLR is upward. The sensibleheat (Hs) flux and surface shortwave radiation areswRs

computed from the SOC atlas for the same period, andthe amount absorbed in the atmosphere Ra is also given.Estimates of the latent energy from evaporation are tak-en from the SOC atlas and the NCEP reanalyses, andfor precipitation from the SOC atlas and CMAP, butthese are used only as guidelines, and adjustments areimposed to fit the total Q2 based upon E 2 P from thereanalyses. The surface longwave radiation is thenlwRs

estimated as a residual from Q1 and the total atmosphericheat budget. The net radiation at the surface Rs com-bined with the evaporative losses and Hs gives the netsurface flux Fs 5 LEs 1 Hs 2 Rs. The large arrowsindicate divergences of heat transports by the atmo-sphere for the DSE (left) and the LE (right), with thecomponent from the transient eddies (top) and quasista-tionary component (bottom). Not shown are the kinetic


FIG. 8. The divergence of the (left) DSE and (right) LE mean total (solid), quasi-stationary (dashed), and transient (dotted) energy zonallyaveraged across the Pacific 1408E–1408W for the four seasons, DJF, MAM, JJA, and SON. Units are W m22.

energy divergences. Monthly means are averaged to getannualized values.

Because these values are based upon different data-sets, they are not compatible, and so compromises havebeen made to achieve a balance. In some cases theseare nontrivial although we believe the result is usefuland illustrates the uncertainties. For instance, for theannual mean from 258–308N, 1408E–1408W, the esti-mates of LEs are 111 W m22 from SOC and 127 W m22

from NCEP; estimates of LP are 83 W m22 from SOCand 75 W m22 from CMAP, yet our estimate of L(E 2P) of 62 W m22 greatly exceeds the SOC value of 28W m22. Our compromise values place more weight onthe LP estimates and hence LE and LP are chosen as134 and 72 W m22. In DJF our estimate of L(E 2 P)of 102 W m22 also greatly exceeds the SOC value of44 W m22, requiring adjustments as large as 18% inLEs. Adjustments are much smaller at other latitudes.

Figure 10 presents our estimates for the Northern

Hemisphere Pacific region for annualized means (top)and DJF (bottom) for the two zones 258–308N (left) and308–358N (right). Similarly, in the Southern Hemispherewe present the zonal means over the oceans for the sametwo zones but for the annual mean and JJA (winter)(Fig. 11). In all regions and times, except for DJF 308–358N in the Pacific, the divergence of DSE by the quasi-stationary component is negative, suggesting domina-tion by subsidence and hence convergence of energy bythe Hadley circulation. At 308–358N in DJF, this sub-sidence contribution is overcome by the divergence inthe quasi-stationary waves (Fig. 10). Both hemispheresat all times have substantial transports of DSE out ofthe zones by the transient waves that average about 100W m22 in the northern winter in both northern zonesand, for the annual means, 30–40 W m22 in the SouthernHemisphere and up to 73 W m22 in the Northern Hemi-sphere. The annual cycle of the atmospheric transportsin the Southern Hemisphere is small for 308–358S and


FIG. 9. Schematic layout of the energy flows in zones through theTOA (dotted line, subscript ‘‘T’’), surface (scalloped line, subscript‘‘s’’), and atmosphere (stippled, subscript ‘‘a’’). Arrows or bracketsindicate direction of flows. Here R indicates radiation flows withamounts in square boxes and amount deposited in ovals for shortwave(superscript ‘‘sw’’) or longwave (superscript ‘‘lw’’). The large hor-izontal arrows indicate divergence of energy by the atmosphere forthe (left) DSE and (right) LE broken down into transient and quasista-tionary components. All units are in terms of energy, hence latentheating from evaporation is given here by LEs (to distinguish it fromlatent energy LE) and precipitation by LP, while the sensible heat isHs and the net surface flux is Fs.

the main changes are in ASR and ocean heat uptake andrelease. At 258–308S in winter, however, there is over50% increase in DSE export by transient eddies and asubstantial increase (to 83 W m22) in convergence ofenergy through subsidence in the Hadley circulation.

Evaporation exceeds precipitation in all zones andtimes (Figs. 10, 11), although more so from 258 to 308latitude. Transient eddies carry latent energy poleward(Figs. 7 and 8) while the quasi-stationary componentcarries similar amounts out of the zones equatorwardin the lower branch of the Hadley cell (Fig. 8). Theexception is the Pacific from 308 to 358N, where quasi-stationary waves converge moisture in the region farthernorth to feed the Mei Yu and Baiu convergence zonesin summer (see Fig. 8).

Large seasonal variations in Fs (Figs. 10, 11) arisefrom the uptake of heat by the ocean in summer and itsrelease in winter; however, the annual mean values in-dicate that the net transports of heat into these zones bythe ocean range from 11 W m22 in the Southern Hemi-sphere to 61 W m22 for 308–358N in the Pacific.

Figures 10 and 11 leave out kinetic energy transports,which are a factor in the North Pacific mainly becauseof zonal transports, and so there is an imbalance of asmuch as 33 W m22 in DJF at 308–358N. For the annualmeans the imbalance is 13 W m22 in the same zone,while there is a balance in the overall budgets to within;1 W m22 in the Southern Hemisphere (and note someroundoff errors).

Only 30%–40% of the OLR might be considered to

come from the surface (in reality, of course, this maybe absorbed and transported away and replaced by otherenergy), the rest arises from emission from greenhousegases and cloud tops and originates from the absorbedsolar radiation and latent heating in precipitation, plussmall contributions from sensible heating. In most cases,the realized latent heating in the same zone also exceedsthe net surface longwave radiation, hence the hydro-logical cycle provides the primary means of transferringenergy away from the surface.

d. Interannual variability

Trenberth and Stepaniak (2003) explored interannualvariability of the zonal means and showed that El Nino–Southern Oscillation (ENSO) provides the main signal.Moreover, the quasi-stationary wave transports arefound to be weaker than normal about 208 to the northand south of where they are strong, and at exactly thesame locations that the transient eddy transport is stron-ger than normal. Hence the location and strength of thetransients are largely determined by the quasi-stationarywaves, which in turn are altered by the resulting eddytransports. This result therefore applies not only to themean fields but also the variability. These conclusionsare also valid for the Pacific sector except that the mag-nitudes of the changes are enhanced by at least a factorof 2. Examination of the NCEP reanalysis monthly meanv fields at 500 mb reveals that, for the quasi-stationarycomponent, they are very strongly positively correlatedwith divergence of LE and negatively correlated withthe DSE component throughout the Tropics (not shown),confirming the association of the variability in the Had-ley circulation with the energy transports.

The variability on monthly and longer timescales isquite small compared with the annual mean divergences(see Fig. 6 of Trenberth and Stepaniak 2003), which areabout 55 W m22 at 258–308S and 45 W m22 at 258–308N. In southern winter (JJA) the standard deviationof monthly zonal mean anomalies of divergence of DSEby transient eddies is 7 W m22 at 258–308S and 3 Wm22 at 258–308N. The southern divergence is correlated0.6 with zonal mean 500-mb v (i.e., cooling accom-panies downward motion) and corresponds to a linearregression of 4 3 1024 Pa s21 per W m22. It furthersignificantly correlates with upward motion at 08–108Sand subsidence centered near 188N, confirming the linkbetween subtropical eddy cooling and the Hadley cir-culation.

In the northern winter (DJF), transient eddies in thesubtropics of both hemispheres play a role (standarddeviation of monthly zonal mean anomalies of diver-gence of DSE by transient eddies is 5 W m22 for 258–308S versus 8 W m22 for 258–308N). Near 308N, di-vergence (cooling) is again accompanied by subsidencelocally, upward motion near the equator and further sub-sidence near 158S. However, cooling near 308S is sig-


FIG. 10. Energy flow diagrams for the North Pacific sector 1408E–1408W for (left) 258–308N and (right) 308–358N, (top) annual and(bottom) DJF for the ERBE period Feb 1985–Apr 1989 in W m22.

nificantly correlated with subsidence near 308S and108S, and upward motion near the equator.

5. Discussion

The development of theories of the atmospheric gen-eral circulation are very nicely summarized by Lorenz(1967) up to that time. He reviews the attempts to ex-plain the general circulation by zonal mean overturningcells, the realization that eddies are important, and con-firmation through utilization of upper-air measurements.The need for and observations of the three cells of me-ridional overturning in each hemisphere were estab-lished and simulated quantitatively in the famous nu-merical experiment of Phillips (1956). In terms of theheat budget, baroclinically unstable transient eddies pro-vide the primary transports poleward in midlatitudes,while the Hadley cell prevails in the Tropics. It has onlybeen since the late 1970s that several attempts, reviewedin the introduction, have been made to develop a more

complete theory of the Hadley circulation. In manyways these have been very successful, but they are alsoincomplete because of assumptions made about the heat-ing.

It is well established that the Ferrel cell in middlelatitudes is driven by the transient eddy transports ofheat and momentum, and can be thought of as a meansof quickly adjusting the atmosphere so that it remainshydrostatic and geostrophic. The transformed Eulerianframework (e.g., Andrews et al. 1987) is an alternativeway to build in some of this automatic compensation.Accordingly, the poleward transient eddy heat transportsare to some extent compensated by subsidence in thesubtropics poleward of the Hadley circulation. However,the picture presented here is one where the average sub-sidence is relatively seamless, and extends equatorwardof 318 into the domain of the Hadley cell. The heatdivergence out of these regions by advection by thetransient eddies causes dynamical cooling that compen-sates subsidence warming, which is linked instead to


FIG. 11. Energy flow diagrams for the Southern Hemisphere zonal mean over the oceans for (left) 258–308S and (right) 308–358S, (top)annual and (bottom) JJA for the ERBE period Feb 1985–Apr 1989 in W m22.

the Hadley circulation. In turn, this also changes thecloudiness and water vapor distribution and makes fora relatively cloud-free and dry subtropics that is a con-sequence of the circulation. Despite substantial absorp-tion of solar radiation at the surface in the relativelyclear skies, large evaporative surface cooling is com-pensated in part by heat from ocean transports. Theatmospheric moisture then helps feed the upward branchof the Hadley cell. Therefore, additional aspects of thedriving of the Hadley circulation are the latent heatingin the equatorial regions and the radiative cooling in thesubtropics, which are feedbacks from the circulation andnot fundamental drivers of it. A schematic view of theprocesses involved is given in Fig. 12.

The picture put forward here applies mostly to theoceans and is observationally based. Cook (2003) ex-plores the role of continents and surface friction on theHadley circulation. The global reanalyses have enabledus to provide new quantitative insights; however, theyare diagnostic relationships, and the heat balance has to

occur by design, so it is not from these diagnostics alonethat we can infer cause and effect.

For instance, one perspective is that the subsidencein the subtropics associated with the downward branchof the local Hadley circulation provides heat (energy)that the transients feed on and transport toward higherlatitudes. Another, just as legitimate, viewpoint is thatthe transients drive the subsidence through advectivecooling. The latter can be argued as the main processthat is operating because both are dynamical conse-quences of the baroclinic instability that occur on thesame timescale, whereas radiative processes have asomewhat longer timescale. What is less obvious is thelink to the upward branch of the Hadley circulation untilwe consider the heat budget and where the heat in thesubtropics comes from.

Many phenomenological studies support the secondview. In particular, there is now considerable evidencethat cold surges from East Asia over the South ChinaSea lead to an intensification of the local Hadley circu-


FIG. 12. Schematic of the Hadley circulation and the heat budget in the subtropics along with key processes.

lation (Ramage 1971; Chang and Lau 1980, 1982; Lauet al. 2000; Compo et al. 1999; Yang et al. 2002) andthese are well simulated in models (Slingo 1998). Toquote from the latter ‘‘. . . tropical east Pacific convectionis initiated by the ascent and decreased static stabilityahead of an upper tropospheric extratropical trough,which has penetrated the deep Tropics in the region ofupper-level equatorial westerlies. Over the maritime con-tinent, convection is enhanced by cold-surge events, inwhich the increased near-surface northerlies substantiallyenhance the air–sea interaction over the South China Sea.These cold surges are triggered by the movement of theSiberian anticyclone in association with the passage ofmidlatitude weather systems.’’ In fact such cold surgesare not unique to East Asia but are also found east ofthe Rockies and Mexican Sierras, and east of the Andesin the Southern Hemisphere, and they move from themidlatitudes well into the Tropics to about 208N in about4 days (Garreaud 2001). The cold surges have a majorcooling and drying effect and represent a major sink ofenergy for the Tropics (Garreaud 2001).

Evidence for disturbances going in the other directionare not as abundant. The main evidence is from theintraseasonal oscillation with a period of about 2 monthswhere cloud perturbations are traceable from 108 to358N in the East Asian sector, mainly in the northernsummer (e.g., Kang et al. 1999).

On the other hand, when the interannual variabilityis considered, ENSO is the predominant mode. Changesin SSTs in the tropical Pacific determine the preferredlocation of the moisture convergence and organizeddeep convection, and alter the heating patterns of theatmosphere. In turn this drives divergence that helps toset up quasi-stationary Rossby waves and teleconnec-tions into midlatitudes, which are associated with chang-es in storm tracks (Trenberth et al. 1998). Nevertheless,the location of the downward branch is a key to theHadley circulation and this is driven by the divergenceof energy (cooling) by transient eddies in the subtropics,which is why the Hadley circulation is directed into thewinter hemisphere. It is also why there is a very weakconnection in JJA in the Northern Hemisphere but asomewhat stronger link in DJF into the Southern Hemi-sphere, making for a stronger southern Hadley circu-lation in the annual mean (Fig. 1).

6. Summary and conclusions

Results of new analyses of the atmospheric energy andheat budgets are presented. The approach is to analyzethe vertically integrated atmospheric energy transportsand their divergences, and the implied heating focusingespecially on the dry static energy, the latent energy, andtheir sum, the moist static energy. These components are


also partitioned into the within-month transient and quasi-stationary components, and the annual cycle and spatialpatterns are documented. The quasi-stationary compo-nent includes the long-term mean as well as the inter-annual variability and contributions beyond monthlytimescales. In the Tropics it corresponds primarily to thelarge-scale overturning global monsoon and the embed-ded Hadley and Walker circulations. In the extratropicsit includes the quasistionary planetary waves that areprimarily a factor in the Northern Hemisphere winter. Byexamining all these components the main processes thatcome into play are documented.

There remain considerable uncertainties in the heatbudgets in the subtropics, as seen in disparate estimatesof evaporation and precipitation. The atmospheric mois-ture budget was used to constrain E 2 P and hence toconstruct surface and TOA heat budgets in addition tothose of the atmosphere, for the subtropical regions,with foci on the North Pacific and the zonal mean overthe oceans in the Southern Hemisphere. The results arephysically consistent with all the available informationon the energy, mass, and moisture budgets, which is nottrue of the bulk fluxes computed in isolation.

There are several characteristics of the energy trans-ports that seem quite remarkable in the way they in-dividually feature strong signatures yet combine to givean overall seamless profile of energy transports and ofenergy divergences. In the Tropics, a characteristic ofthe Hadley circulation is the divergence of energy andpoleward heat transports of the dry static energy butstrong compensating equatorward moisture, or equiv-alently, latent energy transports. The convergence oflow-level moisture, on average just north of the equatorinto the mean ITCZ, provides the energy that predom-inantly fuels the upward branch of the Hadley circu-lation through latent heat release in precipitation.

In the extratropics, the transports of the latent anddry static energy are both poleward although the latentenergy is more important at low to midlatitudes. Thesetransports are carried out by baroclinic disturbances,often organized into storm tracks in some broad sense.However, there is a remarkable compensation betweenwhere the divergence of the quasi-stationary energy ispositive and where it is negative by the transient waves,showing the cooperative nature of these components sothat they contribute to the seamless character of theoverall transports. Hence the location and strength ofthe transients are largely determined by the stationarywaves that in turn are altered by the resulting transports.

Of particular interest is the interaction between thequasi-stationary component and the transients in thesubtropics. We have emphasized the importance of thereach of the energy transports by the transients well intothe Tropics so that cooling results (of order 60 to over100 W m22 in winter) and this is compensated by sub-sidence in the downward branch of the Hadley circu-lation. The resulting relatively clear skies produce ra-diative cooling to space, which does not change much

with season and should be regarded as a modest feed-back, not a fundamental forcing. It also provides forample absorption of solar radiation at the surface whereit feeds the strong evaporation that exceeds precipitationand feeds the equatorward flow of energy into the up-ward branch of the Hadley circulation as well as thepoleward transports by storm tracks (see Fig. 12). Nev-ertheless, the evaporation is sufficiently strong that italso has to be compensated by a subsurface ocean heattransport. Held (2001) argues theoretically that this isbecause the ocean overturning and associated heat trans-port is driven by the Hadley circulation winds. Thusboth the heating in the upward branch of the Hadleycirculation and cooling in the downward branch are de-pendent on the circulation and should properly be re-garded as feedbacks.

The seamless nature of the atmospheric poleward heattransports relates directly to how the Hadley circulationis driven and is linked with the midlatitude storms. Itis a picture consistent with views of the general cir-culation in Lorenz (1967) but nonetheless at odds withsome current perceptions about the Hadley circulation(discussed in the introduction). The latter emphasize theimportance of radiative cooling in the subtropics as adriver of the Hadley cell (albeit often deliberately toexplore the possible atmospheric states without eddies),whereas both the latent heating in the upward branchand the radiative cooling in the downward branch arereally processes that result in large part from the cir-culation. Instead, as suggested by Hoskins (1996) andRodwell and Hoskins (1996) and documented here, thecooling is more fundamentally caused by the transientbaroclinic eddies advecting cooler air into the region,and these eddies play a vital role in driving the Hadleycirculation. The theories of the Hadley circulation dis-cussed in the introduction are still pertinent, althoughthe cooling that drives the downward branch of the cir-culation should be properly interpreted and, like thelatent heating, a portion should be part of the solutionand not specified as the forcing.

This work has suggested intriguing relationshipsamong several quantities, which combine to produceoverall energy transports and divergences that are seam-less. These relationships also hold on a monthly meantimescale, and therefore it is necessary to resolve dailytimescales to better understand how they come aboutand the extent of lead and lag relationships, such asthose arising from cold surges. Research into these as-pects should provide a better basis for understandinghow the Hadley circulation responds to transient eddytransports, and vice versa.

Acknowledgments. This research was sponsored bygrants from NOAA Office of Global Programs and jointlyby NOAA/NASA under NOAA Grant NA17GP1376. Wethank Ed. Schneider and Ray Pierrehumbert for comments.


REFERENCES

Andrews, D. G., J. R. Holton, and C. Leovy, 1987: Middle AtmosphereDynamics. International Geophysical Series, Vol. 40, AcademicPress, 489 pp.

Branstator, G. W., 1995: Organization of stormtrack anomalies byrecurring low-frequency circulation anomalies. J. Atmos. Sci.,52, 207–226.

Chang, C.-P., and K.-M. Lau, 1980: Northeasterly cold surges andnear-equatorial disturbances over the winter MONEX area dur-ing December. Part II: Large-scale aspects. Mon. Wea. Rev., 108,298–312.

——, and ——, 1982: Short-term planetary-scale interactions overthe Tropics and midlatitudes during northern winter. Part I: Con-trasts between active and inactive periods. Mon. Wea. Rev., 110,933–946.

Chen, P., M. P. Hoerling, and R. M. Dole, 2001: The origin of sub-tropical anticyclones. J. Atmos. Sci., 58, 1827–1835.

Compo, G. P., G. N. Kiladis, and P. J. Webster, 1999: The horizontaland vertical structure of east Asian winter monsoon pressuresurges. Quart. J. Roy. Meteor. Soc., 125, 29–54.

Cook, K. H., 2003: Role of continents in driving the Hadley cells.J. Atmos. Sci., 60, 957–976.

Emanuel, K. A., 1995: On thermally direct circulations in moist at-mospheres. J. Atmos. Sci., 52, 1529–1534.

Fang, M., and K. K. Tung, 1999: Time-dependent nonlinear Hadleycirculation. J. Atmos. Sci., 56, 1797–1807.

Garreaud, R. D., 2001: Subtropical cold surges: Regional aspects andglobal distribution. Int. J. Climatol., 21, 1181–1197.

Held, I. M., 2001: The partitioning of the poleward energy transportbetween the tropical ocean and atmosphere. J. Atmos. Sci., 58,943–948.

——, and A. Y. Hou, 1980: Nonlinear axially symmetric circulationsin a nearly inviscid atmosphere. J. Atmos. Sci., 37, 515–533.

Hoskins, B., 1996: On the existence and strength of the summersubtropical anticyclones. Bull. Amer. Meteor. Soc., 77, 1287–1292.

Hou, A. Y., and R. S. Lindzen, 1992: The influence of concentratedheating on the Hadley circulation. J. Atmos. Sci., 49, 1233–1241.

Josey, S. A., E. C. Kent, and P. K. Taylor, 1998: The SouthamptonOceanography Centre (SOC) Ocean-Atmosphere Heat, Momen-tum and Freshwater Flux Atlas. Southampton OceanographyCentre Rep. 6, 30 pp.

——, ——, and ——, 1999: New insights into the ocean heat budgetclosure problem from analysis of the SOC air–sea flux clima-tology. J. Climate, 12, 2856–2880.

Kalnay, E., and Coauthors, 1996: The NCEP/NCAR 40-Year Re-analysis Project. Bull. Amer. Meteor. Soc., 77, 437–471.

Kang, I.-S., C.-H. Ho, Y.-K. Lim, and K.-M. Lau, 1999: Principalmodes of climatological seasonal and intraseasonal variations ofthe Asian summer monsoon. Mon. Wea. Rev., 127, 322–340.

Kim, H.-K., and S. Lee, 2001: Hadley cell dynamics in a primitiveequation model. Part II: Nonaxisymmetric flow. J. Atmos. Sci.,58, 2859–2871.

Lau, K.-M., K. M. Kim, and S. Yang, 2000: Dynamical and boundaryforcing characteristics of regional components of the Asian sum-mer monsoon. J. Climate, 13, 2461–2482.

Lindzen, R. S., and A. Y. Hou, 1988: Hadley circulations for zonallyaveraged heating centered off the equator. J. Atmos. Sci., 45,2416–2427.

——, M.-D. Chou, and A. Y. Hou, 2001: Does the earth have anadaptive infrared iris? Bull. Amer. Meteor. Soc., 82, 417–432.

Lorenz, E. N., 1967: The Nature and Theory of the General Circu-lation of the Atmosphere. World Meteorological Organization,161 pp.

Newell, R. E., J. W. Kidson, D. G. Vincent, and G. J. Boer, 1974:The General Circulation of the Tropical Atmosphere and Inter-actions with Extratropical Latitudes. Vol. 2. The MIT Press, 370pp.

Phillips, N. A., 1956: The general circulation of the atmosphere: Anumerical experiment. Quart. J. Roy. Meteor. Soc., 82, 123–164.

Pierrehumbert, R. T., 1995: Thermostats, radiator fins, and the localrunaway greenhouse. J. Atmos. Sci., 52, 1784–1806.

——, 1999: Subtropical water vapor as a mediator of rapid globalclimate change. Mechanisms of Global Change at MillennialTime Scales, P. U. Clark, R. S. Webb, and L. D. Keigwin, Eds.,Geophysical Monograph Series, Vol. 112, Amer. Geophys.Union, 394 pp.

Ramage, C. S., 1971: Monsoon Meteorology. Academic Press, 296pp.

Rodwell, M. J., and B. J. Hoskins, 1996: Monsoons and the dynamicsof deserts. Quart. J. Roy. Meteor. Soc., 122, 1385–1404.

——, and ——, 2001: Subtropical anticyclones and summer mon-soons. J. Climate, 14, 3192–3211.

Salathe, E. P., and D. L. Hartmann, 1997: A trajectory analysis oftropical upper-tropospheric moisture and convection. J. Climate,10, 2533–2547.

Schneider, E. K., 1977: Axially symmetric steady-state models of thebasic state for instability and climate studies. Part II: Nonlinearcalculations. J. Atmos. Sci., 34, 280–296.

——, 1984: Response of the annual and zonal mean winds and tem-peratures to variations in the heat and momentum sources. J.Atmos. Sci., 41, 1093–1115.

——, 1987: A simplified model of the modified Hadley circulation.J. Atmos. Sci., 44, 3311–3328.

Slingo, J. M., 1998: Extratropical forcing of tropical convection ina northern winter simulation with the UGAMP GCM. Quart. J.Roy. Meteor. Soc., 124, 27–51.

Spencer, R. W., and W. D. Braswell, 1997: How dry is the tropicalfree troposphere? Implications for global warming theory. Bull.Amer. Meteor. Soc., 78, 1097–1106.

Stone, P. H., 1978: Constraints on dynamical transports of energy ona spherical planet. Dyn. Atmos. Oceans, 2, 123–139.

Trenberth, K. E., 1997: Using atmospheric budgets as a constrainton surface fluxes. J. Climate, 10, 2796–2809.

——, and A. Solomon, 1994: The global heat balance: Heat transportsin the atmosphere and ocean. Climate Dyn., 10, 107–134.

——, and C. J. Guillemot, 1995: Evaluation of the global atmosphericmoisture budget as seen from analyses. J. Climate, 8, 2255–2272.

——, and C. J. Guillemot, 1998: Evaluation of the atmospheric mois-ture and hydrological cycle in the NCEP/NCAR reanalyses. Cli-mate Dyn., 14, 213–231.

——, and J. M. Caron, 2001: Estimates of meridional atmosphereand ocean heat transports. J. Climate, 14, 3433–3443.

——, and D. P. Stepaniak, 2003: Covariability of components ofpoleward atmospheric energy transports on seasonal and inter-annual timescales. J. Climate, 16, 3691–3705.

——, G. W. Branstator, D. Karoly, A. Kumar, N.-C. Lau, and C.Ropelewski, 1998: Progress during TOGA in understanding andmodeling global teleconnections associated with tropical sea sur-face temperatures. J. Geophys. Res., 103, 14 291–14 324.

——, D. P. Stepaniak, and J. M. Caron, 2000: The global monsoonas seen through the divergent atmospheric circulation. J. Cli-mate, 13, 3969–3993.

——, J. M. Caron, and D. P. Stepaniak, 2001: The atmospheric energybudget and implications for surface fluxes and ocean heat trans-ports. Climate Dyn., 17, 259–276.

——, D. P. Stepaniak, and J. M. Caron, 2002: Accuracy of atmo-spheric energy budgets. J. Climate, 15, 3343–3360.

van Loon, H., 1979: The association between latitudinal temperaturegradient and eddy transport. Part I: Transport of sensible heatin winter. Mon. Wea. Rev., 107, 525–534.

Xie, P., and P. A. Arkin, 1997: Global precipitation: A 17-year month-ly analysis based on gauge observations, satellite estimates andnumerical model outputs. Bull. Amer. Meteor. Soc., 78, 2539–2558.

Yang, S., K. M. Lau, and K. M. Kim, 2002: Variations of the EastAsian jet stream and Asian–Pacific–American winter climateanomalies. J. Climate, 15, 306–325.

Documents

Seamless Poleward Atmospheric Energy Transports … · 3706 JOURNAL OF CLIMATE VOLUME 16 q 2003 American Meteorological Society Seamless Poleward Atmospheric Energy Transports and