scale interactions on diurnal to seasonal timescales and

ANNALS OF GEOPHYSICS, VOL. 46, N. 1, February 2003

139

Scale interactions on diurnal toseasonal timescales and their relevance

to model systematic errors

Julia Slingo, Peter Inness, Richard Neale, Steven Woolnough and Gui-Ying YangNCAS Centre for Global Atmospheric Modelling, Department of Meteorology, University of Reading, U.K.

AbstractExamples of current research into systematic errors in climate models are used to demonstrate the importance ofscale interactions on diurnal, intraseasonal and seasonal timescales for the mean and variability of the tropicalclimate system. It has enabled some conclusions to be drawn about possible processes that may need to berepresented, and some recommendations to be made regarding model improvements. It has been shown that theMaritime Continent heat source is a major driver of the global circulation but yet is poorly represented in GCMs.A new climatology of the diurnal cycle has been used to provide compelling evidence of important land-seabreeze and gravity wave effects, which may play a crucial role in the heat and moisture budget of this key regionfor the tropical and global circulation. The role of the diurnal cycle has also been emphasized for intraseasonalvariability associated with the Madden Julian Oscillation (MJO). It is suggested that the diurnal cycle in SeaSurface Temperature (SST) during the suppressed phase of the MJO leads to a triggering of cumulus congestusclouds, which serve to moisten the free troposphere and hence precondition the atmosphere for the next activephase. It has been further shown that coupling between the ocean and atmosphere on intraseasonal timescalesleads to a more realistic simulation of the MJO. These results stress the need for models to be able to simulatefirstly, the observed tri-modal distribution of convection, and secondly, the coupling between the ocean andatmosphere on diurnal to intraseasonal timescales. It is argued, however, that the current representation of theocean mixed layer in coupled models is not adequate to represent the complex structure of the observed mixedlayer, in particular the formation of salinity barrier layers which can potentially provide much stronger localcoupling between the atmosphere and ocean on diurnal to intraseasonal timescales.

1. Introduction

Scale interactions describe the influence ofsmall spatial scales and high frequency temporalscales on the large scale, low frequency varia-bility of the climate system and vice versa.

Weather in the tropics is dominated by the effectsof cumulus convection, be it the daily cycle ofrain over the continents, tropical cyclones or theseasonal monsoon rains. In turn, the heatingassociated with cumulus convection is thedominant driving mechanism for the tropicalcirculation. In the last two decades, observationsof the tropics by satellites have revealed a richtapestry of space and time scales of convectiveactivity in which the various scales interact.These scales range from individual clouds, tocloud clusters associated with synoptic scaledisturbances (e.g., easterly waves), through tosuper cloud clusters, which display intraseasonalbehaviour associated with the Madden-Julian

Mailing address: Prof. Julia Slingo, NCAS Centre forGlobal Atmospheric Modelling, Department of Meteorology,University of Reading, Earley Gate, Reading RG6 6BB, U.K.;e-mail: [email protected]

Key words diurnal cycle − MJO − convection −ocean-atmosphere interaction

140

Julia Slingo, Peter Inness, Richard Neale, Steven Woolnough and Gui-Ying Yang

Oscillation (MJO). On the planetary scale,tropical convection shows seasonal and inter-annual variability associated with, for example,monsoons and the effects of the El Niño/SouthernOscillation (ENSO).

This paper will discuss the importance ofscale interactions between diurnal, intraseasonaland seasonal timescales for addressing systematicerrors in current atmospheric climate models.Two particular errors will be the focus of atten-tion and will be used as the basis for presentingideas on how information from observations oftropical variability on diurnal to intraseasonaltimescales can be used to suggest new approachesto improving model simulations in the tropics.The two errors in question are the dry bias overthe Maritime Continent and the inability to simu-late the Madden Julian Oscillation. The ideasbeing presented in this paper derive from a syn-thesis of the results from a variety of researchprojects undertaken by the authors, including thatsupported by the EU SINTEX Project, which aredescribed in more detail in the referenced papers.

The main tools used in the research describedin this paper are the Met Office Unified Model(UM) and a variety of global observational datasets. Coupled (HadCM3) and atmospheric(HadAM3) versions of the UM (see Pope et al.,1999) are used for the various modeling exper-iments. Precipitation data from the ClimatePrediction Center’s (CPC) Merged Analysisof Precipitation (CMAP; Xie and Arkin, 1996)and the European Centre for Medium-rangeWeather Forecasting (ECMWF) 15-year Re-analysis (ERA-15) are used to evaluate the modelperformance. A new multi-year global data setof window (11 µm) brightness temperature frommultiple satellites has been developed by theEuropean Union Cloud Archive User Service(CLAUS) project. The utility of such data forinvestigating the space-time variability oftropical convection has been demonstrated by,for example, Salby et al. (1991). In our case, theavailability of 3-hourly data has enabled theconstruction of a climatology of the diurnal cyclein convection, cloudiness and surface tem-perature for all regions of the tropics (Yang andSlingo, 2001). This has proved critical foridentifying potential processes in the tropics thatare not well represented in climate models.

2. The Maritime Continent and its role in the global circulation

The Maritime Continent, with its complexsystem of islands and shallow seas, presentsa major challenge to models, which tend tosystematically underestimate the precipitation inthis region (fig. 1a; Neale and Slingo, 2003).Neale and Slingo (2003) argued that the deficientrainfall over the Maritime Continent could be adriver for other systematic errors, such as theexcess precipitation over the Western IndianOcean. To demonstrate the sensitivity of globalsystematic model errors to the heating in thisregion, Neale and Slingo (2003) performed twoexperiments, one with the existing distributionof islands and a second where the island grid-points are replaced by sea grid-points with SSTinterpolated from existing adjacent grid-pointvalues. Both experiments were run for 17 yearsusing observed SSTs for 1979-1995, as in thesecond phase of the Atmospheric Model Inter-comparison Project (AMIP II). In the absence ofthe islands of the Maritime Continent, the localprecipitation increases, reducing the existingdry bias and bringing the model closer to obser-vations (fig. 1b). In response to this improvedheating distribution, precipitation decreases overthe West Indian Ocean and South Pacific Con-vergence Zone (SPCZ), reducing the systematicwet bias in these regions. This supports thehypothesis that tropical systematic errors are oftenrelated through vertical (Walker) circulations.

The extra-tropical response to changes in thetropical heat source is also well demonstrated bythese experiments. The enhanced heating andhence divergent outflow over the MaritimeContinent generates Rossby waves which havea significant impact on Northern Hemispherewinter surface temperatures across much ofNorth America and the North East Eurasianregion (fig. 2a,b). These changes in mid-latitudecirculation and surface temperature are such asto substantially reduce model systematic errorin these regions. These results reinforce thecritical role played by the Maritime Continent inthe global circulation, and emphasize theimportance of considering the global context ofmodel systematic error in which biases in thetropics may be a key factor.

141

Scale interactions on diurnal to seasonal timescales and their relevance to model systematic errors

It could be argued that this system of islandsis poorly resolved at the coarse climate resolutionof the model, but Neale and Slingo (2003) furthershowed that even with a three-fold increase inhorizontal resolution there is no improvement inthe dry bias of the model; if anything it seems toget slightly worse (fig. 3a-d). Additionally, theanomalously wet regions in the SPCZ and northand west of the Philippines are worse in theexperiments with the highest resolution. Theresults in fig. 3a-d demonstrate very clearly the

persistence of model errors and reinforce thenotion that increasing resolution is not a panaceafor model error, with the suggestion that de-ficiencies in the representation of the physicalsystem are primarily responsible.

The results summarized above clearly showthat the Maritime Continent heat source is akey component of the global climate, and thatimprovements in its simulation may have sig-nificant impacts on remote systematic errors.The fact that many models show similar prob-

Fig. 1a,b. a) Annual mean systematic error in precipitation (mm/day) from HadAM3 based on 17-year (1979-1995) integration with observed SSTs. b) Change in annual mean precipitation from an integration of HadAM3 inwhich the islands of the Maritime Continent have been replaced with ocean grid points. From Neale and Slingo(2003).

a

b

142


lems over the Maritime Continent suggests thatthey all lack some key ingredient. Neale andSlingo (2003) hypothesized that the diurnal cycleover the islands and the complex circulationpatterns generated by land-sea contrasts may be

crucial for the energy and hydrological cycles ofthe Maritime Continent and for determining themean climate. This suggestion was based on newresults pertaining to the diurnal cycle, which aredescribed in the following section.

Fig. 2a,b. Seasonal mean differences averaged for 1979-1995 between the sensitivity AMIP II experiment withthe Maritime Continent land grid-points removed and the control AMIP II experiment: a) DJF Northern Hemisphere500-hPa geopotential height (m), and b) DJF Northern Hemisphere surface air temperature (K). From Neale andSlingo (2003).

a

b

143


Fig

. 3

a-d.

Ann

ual

mea

n pr

ecip

itatio

n er

rors

fro

m f

our

Had

AM

3 A

MIP

II

inte

grat

ions

with

dif

fere

nt h

oriz

onta

l re

solu

tions

(m

m/d

ay);

a)

Clim

ate

reso

lutio

n (3

.75°

× 2.

5°);

b)

1.5

× cl

imat

e re

solu

tion

(2.5

° × 1.

67°)

; c)

2×

clim

ate

reso

lutio

n (1.875°×1.25°

); d

) 3

× cl

imat

e re

solu

tion

(1.2

5 °× 0

.83°

).Fr

om N

eale

and

Slin

go (

2003

).

ab d

c

144


3. The diurnal cycle and its role in the climate of the Maritime Continent

As noted earlier, the CLAUS data set providedthe basis for a detailed study of the diurnal cycleacross the global tropics. The diurnal cycle do-minates the sub-seasonal variability over tropicaland subtropical continents and systematical-ly modifies the precipitation over the tropical o-ceans (see Yang and Slingo, 2001 and referencestherein). In common with previous observationalstudies, the CLAUS climatology of the diurnalcycle has shown that oceanic deep convectiontends to reach its maximum in the early morning.Continental convection generally peaks in theevening, although there are interesting regionalvariations, indicative of the effects of complexland-sea and mountain-valley breezes, as wellas the life cycle of mesoscale convective systems.

A striking result from this analysis of thediurnal cycle in the CLAUS data set has beenthe extent to which the strong diurnal signal overland is spread out over the adjacent oceans (fig.4a,b). As noted by Yang and Slingo (2001), thisspreading out of the diurnal signal away fromtropical coastal regions has a coherent structuresuggestive of wave propagation away from thecoast. These coherent signals can be seen forseveral hundred kilometres and in some in-stances, such as over the Bay of Bengal, can leadto substantial diurnal variations in convection andprecipitation. Looking in detail at specific re-gions, fig. 5a-c shows the phase of the diurnalharmonic, expressed as the local time of max-imum brightness temperature (i.e. minimumconvection). In fig. 5a, lines of constant phasespread out south eastwards across the Bay ofBengal, starting from the north east coast of India.

Fig. 4a,b. Seasonal mean amplitude of the diurnal harmonic in brightness temperature (K) for DJF (a) and JJA(b). From Yang and Slingo (2001).

a

b

145


Fig. 5a-c. Phase of the diurnal harmonic of brightness temperature over (a) Bay of Bengal in JJA, (b) MaritimeContinent in DJF, and (c) Mexican coast in JJA. Phase is given in terms of the local time of the maximum. Basedon Yang and Slingo (2001).

a

b

c

146


The inferred propagation speed is between 15and 20 ms–1, in line with those of coherentdisturbances that were observed propagatingrapidly southwards over the Bay of Bengal duringthe recent Joint Air-Sea Monsoon InteractionExperiment (JASMINE; Webster et al., 2002).These propagating disturbances generatesignificant diurnal variations in precipitation overthe Bay of Bengal, suggesting that these waveshave a fairly deep structure. In fact, the inferredpropagation speed is consistent with that ex-pected for a diurnally generated gravity wavewhose equivalent depth is 40 m and whose spatialscale is wavenumber 20.

These lines of constant phase can also be seenaround the islands of Indonesia, particularlyduring northern winter (fig. 5b), and suggest acomplex system of diurnally varying convectionover the Maritime Continent. Here the inferredpropagation speed away from the islands isslower than over the Bay of Bengal case, beingtypically nearer 10ms–1. Again the signal isevident for many hundreds of kilometres, forexample, north-eastwards of New Guinea. Sim-ilarly the lines of constant phase spreading awayfrom the Mexican coast in northern summer(fig. 5c) also suggest a slower propagation speedthan seen in fig. 5a for the Bay of Bengal. If thesignal is associated with a gravity wave, thenthe slower speed is possibly indicative of ashallower wave, more typical of a sea/land bree-ze effect.

Over the complex system of islands whichmake up the Maritime Continent, these coherentsignals in the diurnal cycle provide compellingevidence of important land-sea breeze and grav-ity wave effects, which may play a crucial rolein the heat and moisture budget of this key regionfor the tropical and global circulation (see discus-sion in Section 2). The presence of these signalshas been confirmed in an analysis of data fromthe Tropical Oceans-Global Atmosphere CoupledOcean-Atmosphere Response Experiment(TOGA-COARE) by Liberti et al. (2001), whoshowed that the signal of the diurnal cycle overNew Guinea could be detected in the wind andcloud fields up to 600 km from the coast. Again,the average speed of propagation of the featureswas around 15ms−1 in agreement with the resultsof Yang and Slingo (2001).

The results of the studies by Neale and Slingo(2003) and Yang and Slingo (2001) point to theneed for a parametrization of the effects of land-sea breezes and, at a more general level, gravitywaves for triggering convection and providingsub-gridscale surface wind variability that mightdrive stronger turbulent fluxes. Specifically, land/sea breezes and gravity waves generated by thediurnal cycle in convection over the islands maybe a crucial part of the energy and hydrologicalbudgets of coastal regions and especially aroundlarge island complexes. A parametrization of theeffects of land/sea breezes is currently beingdeveloped, based on the theoretical ideas of Liuand Moncrieff (1996), and will be implementedin the Met Office Unified Model.

Normally gravity waves are hard to observe,but the results of Yang and Slingo (2001), inwhich the source of the waves has a regularperiodicity, have provided important evidence oftheir existence and interaction with convection.It is possible that horizontally propagating gravitywaves, generated by convection, may be wide-spread and therefore important for organizingconvection on larger scales. It is clear fromsatellite imagery that tropical convection posses-ses a considerable degree of self-organization,some of which is associated with theoreticalequatorial wave structures (Wheeler and Kiladis,1999). Mapes (1993) has provided a paradigmin which gravity waves or buoyancy bores,generated by convection, could give rise to, ashe called it, ‘gregarious’ convection. The poten-tial importance of gravity waves in providing theupscale energy transfer at high wavenumbershas recently been pointed out by Koshyk andHamilton (2001), based on a high resolution(N270; 0.33° × 0.4°) GCM simulation. How theeffects of unresolved gravity waves should beincluded in climate models is unclear, but theuse of a stochastic forcing field may be ap-propriate.

Although the focus of this paper is on scaleinteractions in time, the above discussion alsopoints out that scale interactions in space maybe crucial for capturing the organized nature oftropical convection. Indeed, to achieve the correctscale interactions on diurnal to seasonal time-scales may well require addressing the issues ofscale interactions in space also.

147


4. The Madden Julian Oscillation

The Madden Julian Oscillation (MJO) do-minates tropical variability on sub-seasonal time-scales and is known to have global influences.Yet a complete understanding of the processesinvolved in the initiation and maintenance of theMJO and an adequate simulation of it by GCMshas continued to be elusive. The results fromAMIP I showed a wide range of skill in simul-ating the MJO (Slingo et al., 1996), with eventhe most skillful models being unable to representthe large-scale organization of convection as-sociated with the MJO (Sperber et al., 1997). Thesituation does not appear to have improved muchsince AMIP I. The results of a more limitedintercomparison by the CLIVAR Asian/Australian Monsoon Panel (Wu et al., 2002), hasshown that models are still unable to reproducethe observed concentration of power at the 40-50 day timescale.

The importance of acquiring an understand-ing of the MJO and of improving our ability tosimulate it cannot be over-emphasized. The MJOis known to be intimately related to active/ breakcycles of the Australian and Asian Monsoonsand can affect weather over the Western U.S. andpossibly elsewhere. Recent research by Waliseret al. (2003) has shown that the MJO offerspotential to provide extended predictability upto 15-20 days in the tropics, which, if real-izable, would have enormous economic benefits,particularly for crop management and health. Forthe coupled ocean-atmosphere system, associat-ed westerly wind events generate ocean Kelvinwaves, which may significantly modify theevolution and amplitude of El Niño, as was thecase in 1997 (e.g., McPhaden and Yu, 1999).Because of the influence of the MJO on equa-torial ocean dynamics, the large interannualvariability in the activity of the MJO has impli-cations for the predictability of the coupledocean-atmosphere system (Slingo et al., 1999).

Recent research has pointed to possible ave-nues that might lead to improvements in thesimulation of the MJO. Firstly, Inness et al.(2001) have shown that the simulation of theMJO is significantly improved in an atmosphericversion of the UM when the vertical resolutionis increased. Secondly, there is good evidence

that the MJO is, at least to some extent, a coupledocean-atmosphere mode. Several observationallybased studies (e.g., Woolnough et al., 2000) haveshown that the atmosphere appears to force theocean in a systematic way during the passage ofthe MJO. In turn, an idealized modeling studyby Woolnough et al. (2001) has demonstratedthat these intraseasonal SST anomalies can forcethe organisation of tropical convection in a man-ner that favours longer timescales (~ 60 days),typical of the MJO.

4.1. Vertical resolution, tri-modal convection and the Madden Julian Oscillation

Experiments using the UM with two differentvertical resolutions (19 and 30 levels) haveshown significant differences in the amount ofvariability in the tropical upper troposphericzonal wind component associated with the MJO(fig. 6; Inness et al. 2001). Most of the extralevels were placed in the middle and upper tropo-sphere, decreasing the layer thickness in mid-troposphere from 100 hPa to 50 hPa, and givinga much better representation of the temperatureand humidity structure around the freezing level.The results suggested a change in the temporalorganization of convection which was inves-tigated further using an aqua-planet version ofthe UM. The advantages of using an aqua-planetare that the homogeneity of the set-up allows usto obtain a large sample of convective events overwarm SSTs, and the removal of the land areasexcludes circulations forced by land-sea con-trasts. This means that convective events indifferent geographical locations are subject tothe same large-scale forcing, giving a cleanercomparison.

Many conceptual models of tropical convec-tion are based on a bimodal cloud distribution,emphasizing shallow «trade-wind» or boundarylayer cumuli and deep cumulonimbi. However,TOGA COARE results have shown the domi-nance of cumulus congestus clouds, and point toa tri-modal cloud distribution in which the freez-ing level inversion is the key (Johnson et al.,1999). The results from the aqua-planet exper-iments, described in detail in Inness et al. (2001),show that when the vertical resolution is in-

148


creased in the UM, the spectrum of tropical cloudtypes changes from a bimodal distribution to atri-modal distribution with a third peak in themid-troposphere, near the melting level. As-sociated with periods when these mid-levelcongestus clouds are dominant, the detrainmentfrom these clouds significantly moistens the mid-troposphere (fig. 7a,b). In comparison, the 19-level version of the model shows no evidence ofa tri-modal distribution in convection and no suchmoistening events.

Observational studies have shown that,during the suppressed phase of the MJO, tropicalconvection is dominated by clouds that terminatearound the stable layer at the 0°C level (Johnsonet al., 1999), and that these clouds provide asource of moisture to the mid-troposphere (Linand Johnson, 1996). Inness et al. (2001) arguedthat the development of a stable layer around thetropical freezing level, which is frequently ob-

served over the tropical oceans, acts to reinforcethe transition from the enhanced convectivephase to the suppressed phase of the MJO. Sub-sequently, the moistening of the mid-tropo-sphere during the suppressed phase acts to rein-force the transition back to the active phase. Thisis consistent with the ‘recharge-discharge’ theoryfor the MJO proposed by Bladé and Hartmann(1993) in which the MJO timescale may be setby the time it takes for the moist static energy tobuild up following the decay of the previousconvective event. It may be that the rechargingof the moist static energy is achieved in part bythe injection of moisture into the mid-troposphereby the cumulus congestus clouds that dominateduring the suppressed phase of the MJO.

The appearance of these congestus clouds hasbeen postulated as the reason for the improve-ment in the simulation of the MJO in the 30-levelversion of the UM. This is shown to be partly

Fig. 6. Time-series of an index of MJO activity based on the 20-100 day filtered variance of 200 hPa zonal windin the Tropics. Bold line - ECMWF re-analysis data; thin solid line - L30 GCM AMIPII integration; dotted line -L19 GCM AMIPII integration. From Inness et al. (2001).

149


due to improved resolution of the freezing leveland of the convective processes occurring at thislevel. However, the results also suggest thatconvection and cloud microphysics schemesmust be able to represent cumulus congestusclouds which, being neither shallow nor deepcumulus as well as often weakly precipitating,tend not to be explicitly represented in currentschemes.

The results of Inness et al. (2001) haveemphasized the importance of vertical resolution,in line with the recent study of Tompkins andEmanuel (2000), as well as the need to properlyrepresent the tri-modal structure of tropicalconvection. The importance of the cumuluscongestus stage of tropical convection is beingstressed here as a potentially important ingredientfor the MJO. This means that vertical resolution

Fig. 7a,b. Time-height cross-sections of the increment to specific humidity due to the convective parametrization(g/kg/day) for a 90-day period from the aqua-planet integrations, averaged over a 3 × 3 grid-point box centred onthe equator. a) L30 aqua-planet; b) L19 aqua-planet. From Inness et al. (2001).

a

b

150


in the free troposphere must be adequate to resolvethe formation of the freezing level inversion andthe cooling associated with melting precipitation.

4.2. Coupling with the upper ocean: bringing together the diurnal cycle and the MJO

Following TOGA-COARE, it has becomeincreasingly apparent that coupling between theocean and atmosphere may be important for awide range of timescales, not only those asso-

ciated with seasonal to interannual variations.High frequency observations of the variations inSST, taken by the WHOI IMET (Woods HoleOceanographic Institution Improved Meteor-ological) buoy during TOGA-COARE (Wellerand Anderson, 1996), have shown pronounceddiurnal variations in skin temperature in excessof 1 K. Slower variations on intraseasonal time-scales, related to the Madden-Julian Oscillation(MJO), are also evident.

A particularly noteworthy aspect of thesebuoy observations is the fact that the diurnal

Fig. 8a-d. Lag correlations between observed OLR (convection) and surface fields: a) Sea Surface Temperature(SST); b) shortwave radiation; c) zonal wind stress, and d) latent heat flux. Negative lags indicate that the convectionlags the surface field, positive lags indicates that the convection leads the surface fields. The sign convention issuch that positive correlations indicate that enhanced convection (a negative OLR anomaly) is correlated with anegative SST anomaly, reduced shortwave radiation at the surface, enhanced evaporation or an easterly windstress anomaly. From Woolnough et al. (2000).

a b

c d

151


variations in SST only occur during suppressedphases of the MJO, when the winds are light andthe net heat flux into the ocean is large. Further-more, Johnson et al. (1999) showed that cumuluscongestus clouds are most prevalent during lightwind conditions in the presence of a strongdiurnal cycle in SST. These clouds occur mostfrequently in the late afternoon, with a behaviourthat resembles more closely the diurnal cycle inland convection, suggesting that they may betriggered by the diurnal cycle in SST. This is astrong indication that coupling with the upperocean may be important even on diurnal time-scales. Although this hypothesis has yet to betested in fully coupled models, it is worth notinghere that, although Inness et al. (2001) were ableto show that increasing the vertical resolution ofthe model improved the subseasonal convectiveorganisation, the observed trimodal distributionin convection was only just starting to be cap-tured, suggesting that these diurnal variations inSST may be important.

Recent research has attempted to assess theimportance of the intraseasonal SST variationsseen in TOGA-COARE for determining thetransient characteristics and spatial organizationof tropical convection associated with the MJO.Woolnough et al. (2000) showed that, for theIndian Ocean and West Pacific, a coherent rela-tionship exists between convection, surfacefluxes and SST in which the SSTs are warmerthan normal about 10 days prior to the maximumin convective activity (fig. 8a-d). This warmingis associated with increased solar radiation,reduced surface evaporation and light winds.Following the convective maximum, the SSTscool due to reduced solar radiation and enhancedevaporation associated with stronger winds. Akey requirement for the observed phase rela-tionship between the latent heat flux, winds andconvection is the presence of a westerly basic state.

In a related study, Woolnough et al. (2001)used the observed SST perturbations to form thebasis of a series of experiments with the aqua-planet version of the UM to investigate, firstly,the organisation of tropical convection by theseintraseasonal anomalies, and, secondly, how thisorganisation depends on the temporal behaviourof these SST anomalies. Their results showed thatintraseasonal SST anomalies could potentially

organise convection in a manner that favoursthe longer timescales (~ 60 days), typical of theobserved MJO, and which produces a phase re-lationship between the convection and SST,consistent with the observed structure over theIndian and West Pacific Oceans.

There is convincing evidence, therefore, thatthe MJO may require coupling with the ocean inorder to simulate it correctly. Preliminary studieswith coupled models appear to support thisconclusion (e.g., Waliser et al., 1999). Recentexperiments with the coupled version of the UM(HadCM3) have shown that the organization andpropagation characteristics of the MJO areimproved in comparison with the results fromthe atmosphere-only version, HadAM3 (fig. 9a-c; Inness et al., 2003; Inness and Slingo, 2003).Whereas the atmosphere-only model has a pre-dominantly standing oscillation in the con-vection, the coupled model produces a more real-istic eastward propagating signal. As fig. 9a-cshows, this is associated with coherent variationsin SST; these show a similar phase relationshipwith convection as in observations (fig. 8a-d),with warmer SSTs preceding the maximum inconvection by between 5 and 10 days.

Whilst coupled models are capable of pro-ducing the correct relationship between con-vection and SST on intraseasonal timescales,supporting the notion that the MJO is at least inpart a coupled phenomenon, these models stillunderestimate the activity of the MJO (Innesset al., 2003; Inness and Slingo, 2003). Althoughthe results from HadCM3 have shown that thecorrect phase relationship between convectionand SST can be simulated, the magnitude of theSST perturbations is smaller than observed. Thisis despite changes in the surface fluxes that aresimilar to the observed. This suggests that therepresentation of the upper layers of the oceanin HadCM3 may not be designed to respondrealistically to subseasonal variations in windsand fluxes. Most coupled climate models havea relatively coarse vertical resolution in theupper ocean, typically of the order of 10 m,which is therefore not sufficiently high toresolve the detailed structure of the observedmixed layer.

Lukas and Lindstrom (1991), for example,have emphasized the importance of salinity

152


gradients within the mixed layer for modifyingthe coupling between the ocean and atmosphere.The large freshwater flux during the active phaseof the MJO, for example, can set up a saltstratified barrier layer so that a shallow mixedlayer forms which can respond rapidly to fluxvariations, such as the diurnal cycle in solarradiation. The TOGA COARE observationssuggest that the mixed layer depth is indeedhighly variable, and that the barrier layer is oftenas thin as a few metres during light wind con-ditions (e.g., Anderson et al., 1996; Inall et al.,1998 and Feng et al., 2000). The presence ofthis barrier layer can potentially provide muchstronger local coupling in the warm pool regionthan is currently found in coupled models whichdo not resolve the detailed structure of the warmpool upper ocean. This suggests that coupledmodels may need a more detailed representationof the mixed layer if the correct coupling between

the ocean and atmosphere on sub-seasonal time-scales is to be achieved.

This section has emphasized the potential roleof an interactive upper ocean in determiningtropical variability on diurnal to intraseasonaltimescales, and in particular for simulating theMJO. It is suggested that the ocean plays anintegral part in the scale interactions in the tropicson diurnal to intraseasonal timescales. The basicparadigm is as follows. The diurnal cycle in SSTleads to a triggering of convection in associationwith the enhanced skin temperatures. In turnthese cumulus congestus clouds moisten the freetroposphere and hence precondition the atmos-phere for deep convection. This preconditioningmay set the timescale for the next active phaseof the MJO and thus influence the intraseasonalorganization of convection. Furthermore, therectification of the diurnal SST variations tolonger timescales may be a key factor in setting

Fig. 9a-c. Lag correlations between precipitation atevery longitude and an index of MJO activity at 90°E,based on the 20-100 day filtered 200 hPa velocitypotential, from: a) a version of the coupled ocean-atmosphere model, HadCM3, and b) the equivalentatmosphere-only model, HadAM3. c) Shows thesimulated lag correlations between the precipi-tation and SST at every longitude (as in fig. 7a) fromHadCM3. From Inness et al. (2003).

ca

b

153


up the intraseasonal variations in the bulk SST.Scale interactions lie at the heart of this paradigm,with the suggestion that the diurnal variationsduring the quiescent phase of the MJO may becrucial for setting the timescale of the MJO. Ifthis paradigm is correct, then models will needto consider a complete upper ocean/atmospheresystem in which the structure of the upper oceanis adequately resolved.

5. Concluding remarks

This review paper has used examples ofcurrent research into systematic errors in climatemodels to demonstrate the importance of scaleinteractions on diurnal, intraseasonal and sea-sonal timescales for the mean and variability ofthe tropical climate system. It has enabled someconclusions to be drawn about possible processesthat may need to be represented, and some recom-mendations to be made regarding model improve-ments. It has been shown that the MaritimeContinent heat source is a major driver of theglobal circulation but yet is poorly representedin GCMs. This complex system of islands, withits pronounced diurnal cycle in convection, givesrise to extensive sea/land breeze circulations thatmay critically influence the energy budget andhydrological cycle and hence influence the meanclimate of that region. Indeed, Neale and Slingo(2003) showed that even minor changes to thediurnal cycle projected significantly on to theseasonal mean climate emphasizing the im-portance of scale interactions on diurnal to sea-sonal timescales.

Results from an atmospheric model with vary-ing vertical resolution have demonstrated thatmoistening of the free troposphere by cumuluscongestus clouds, which form during thesuppressed phase of the MJO, may be crucial forconvective preconditioning in readiness for thenext active phase. It has been shown that thisdominant cloud type is not represented in models,which generally fail to capture the observed tri-modal distribution of convection. The importanceof having adequate vertical resolution to representthe freezing level inversion has been emphasized.

Observations have shown that the diurnalcycle in SST is large during suppressed or light

wind conditions in the tropics and may be atrigger for cumulus congestus clouds. SSTs alsovary coherently with the MJO in such a manneras to suggest that they are an important com-ponent of the eastward propagation and timescaleof the MJO. Both the diurnal and intraseasonalvariations in SST involve detailed changes in thesalinity and temperature structure of the mixedlayer that cannot be adequately represented incurrent coupled models, with the suggestion thatmore detailed modeling of mixed layer processesin coupled models may improve the simulationof the MJO. The diurnal to sub-seasonal va-riations in SST and their potential importancefor organized tropical convection also indicatethat the use of atmosphere-only models to studyatmospheric variability may be misleading. Tomake progress in improving atmospheric simu-lations and predictions, even on daily to weeklytimescales, it may be desirable to develop anatmosphere/upper ocean modeling system, whichwould enable these potentially important coup-lings to operate.

At a broader level, it could be argued thatfeedbacks between the ocean and atmosphere onsubseasonal timescales may be important for theregulation of warm pool SSTs. There has beenconsiderable discussion of why SST rarelyexceeds 30°C and why the Western Pacific andIndian Ocean warm pool has a relatively uniformSST between 28°C and 30°C. The thermostathypothesis of Ramanathan and Collins (1991)was followed by extensive debates on the roleof large-scale atmospheric dynamics, eva-poration and ocean dynamics. However, as notedby Webster et al. (1996), all these studies wereconducted using monthly mean data andtherefore excluded the potential influence ofcoupled processes on diurnal to intraseasonaltimescales. A better understanding of the ad-justment timescales for the atmosphere andupper-ocean, which lie at the heart of theregulation of warm pool SSTs, probably requiresa fuller appreciation of the scale interactionsbetween diurnal, intraseasonal and seasonaltimescales in the fully coupled atmosphere-oceansystem. We now have substantial evidence fromobservational programmes, such as TOGA-COARE, that variability in the tropics on dif-ferent space and time scales is intimately linked,

154


and that the fully interactive coupling betweenthe atmosphere and the ocean at all scales maybe important for determining the mean climateand its low frequency variability.

Acknowledgements

J. M. Slingo acknowledges the support of theNERC U.K. Universities’ Global AtmosphericModelling Programme (UGAMP). P.M. Innesswas supported through the EC Project on in-terannual and decadal climate variability scaleinteractions experiments (SINTEX: ENV4-CT98- 0714). R.B. Neale acknowledges thesupport of the Met Office through grant Met1b/2601. S.J. Woolnough and G.-Y. Yang acknow-ledge the support of NERC through grants GR3/10798 and GR3/12541.

REFERENCES

ANDERSON, S.P., R.A. WELLER and R.B. LUKAS (1996):Surface buoyancy forcing and the mixed layer of theWestern Pacific warm pool: observations and 1D modelresults, J. Climate, 9, 3056-3085.

BLADÉ, I. and D.L. HARTMANN (1993): Tropical intraseasonaloscillations in a simple nonlinear model, J. Atmos. Sci.,50, 2922-2939.

FENG, M., R. LUKAS, P. HACKER, R.A. WELLER and S.P.ANDERSON (2000): Upper-ocean heat and salt balancesin the western equatorial Pacific in response to theintraseasonal oscillation during TOGA COARE, J.Climate, 13, 2409-2427.

INALL, M.E., K.J. RICHARDS and G. ELDIN (1998): thechanging structure of the upper ocean in the westernequatorial Pacific during TOGA COARE, J. Geophys.Res., 103, 21385-21400.

INNESS, P.M. and J.M. SLINGO (2003): Simulation of the MJOin a coupled GCM. I: comparison with observations andan atmosphere-only GCM, J. Climate, 16, 345-364.

INNESS, P.M., J.M. SLINGO, S.J. WOOLNOUGH, R.B. NEALEand V.D. POPE (2001): Organization of tropical con-vection in a GCM with varying vertical resolution:implications for the simulation of the Madden-JulianOscillation, Climate Dyn., 17, 777-793.

INNESS, P.M., J.M. SLINGO, E. GUILYARDI and J. COLE(2003): Simulation of the MJO in a coupled GCM. II:the role of the basic state, J. Climate, 16, 365-382.

JOHNSON, R.H., T.M. RICKENBACH, S.A. RUTLEDGE, P.E.CIESIELSKI and W.H. SCHUBERT (1999): Trimodalcharacteristics of tropical convection, J. Climate, 12,2397-2418.

KOSHYK, J.N. and K. HAMILTON (2001): The horizontalkinetic energy spectrum, and the spectral budget

simulated by a high-resolution Troposphere-Strato-sphere-Mesosphere GCM, J. Atmos. Sci., 58, 329-348.

LIBERTI, G.L., F. CHÉRUY and M. DESBOIS (2001): Landeffect on the diurnal cycle of clouds over the TOGACOARE area, as observed from GMS IR Data, Mon.Weather Rev., 129, 1500-1517.

LIN, X. and R.H. JOHNSON (1996): Heating, moistening andrainfall over the western Pacific warm pool during TOGACOARE, J. Atmos. Sci., 53, 3367-3383.

LIU, C. and M.W. MONCRIEFF (1996): A numerical study ofthe effects of ambient flow and shear on density currents,Mon. Weather Rev., 124, 2282-2303.

LUKAS, R. and E. LINDSTROM (1991): The mixed layer ofthe western equatorial Pacific Ocean, J. Geophys. Res.,96, 3343-3357.

MAPES, B.E. (1993): Gregarious tropical convection, J.Atmos. Sci., 50, 2026-2037.

MCPHADEN, M.J. and X. YU (1999): Equatorial waves andthe 1997-1998 El Niño, Geophys. Res. Letters, 26,2961-2964.

NEALE, R.B. and J.M. SLINGO (2003): Understanding theclimate of the Maritime Continent and its influence onthe global circulation, J. Climate, 16, 834-848.

POPE, V.D., M.L. GALLANI, P.R. ROWNTREE and R.A.STRATTON (1999): The impact of new physical para-metrizations in the Hadley Centre Climate Model-HadAM3, Climate Dyn., 16, 123-146.

RAMANATHAN, V. and W. COLLINS (1991): Thermodynamicregulation of ocean warming by cirrus clouds deducedfrom observations of the 1987 El Niño, Nature, 351,27-32.

SALBY, M.L., H.H. HENDON, K. WOODBERRY and K.TANAKA (1991): Analysis of global cloud imageryfrom multiple satellites, Bull. Am. Meteorol. Soc., 72,467-480.

SLINGO, J.M., K.R. SPERBER and 22 others (1996):Intraseasonal oscillations in 15 atmospheric generalcirculation models: results from an AMIP diagnosticsubproject, Climate Dyn., 12, 325-357.

SLINGO, J.M., D.P. ROWELL, K.R. SPERBER and F. NORTLEY(1999): On the predictability of the interannualbehaviour of the Madden-Julian Oscillation and itsrelationship with El Niño, Q. J. R. Meteorol. Soc., 125,583-609.

SPERBER, K.R., J.M. SLINGO, P.M. INNESS and W.K-M. LAU(1997): On the maintenance and initiation of theintraseasonal oscillation in the NCEP/NCAR reanalysisand the GLA and UKMO AMIP simulations, ClimateDyn., 13, 769-795.

TOMPKINS, A.M. and K.A. EMANUEL (2000): The verticalresolution sensitivity of simulated cequilibrium tropicaltemperature and water vapour profiles, Q. J. R. Meteorol.Soc., 126, 1219-1238.

WALISER, D.E., K.M. LAU and J-H. KIM (1999): Theinfluence of coupled SSTs on the Madden-JulianOscillation: A model perturbation experiment, J. Atmos.Sci., 56, 333-358.

WALISER, D.E., K.M. LAU, W. STERN and C. JONES (2003):Potential predictability of the Madden-Julian Oscillation,Bull. Amer. Meteorol Soc., 84, 33-50.

WEBSTER, P.J., C.A. CLAYSON and J.A. CURRY (1996):

155


Clouds, radiation and the diurnal cycle of sea surfacetemperature in the tropical Western Pacific Ocean,J. Climate, 9, 1712-1730.

WEBSTER, P.J., E.F. BRADLEY, C.E. FAIRALL, J.S. GODFREY,P. HACKER, R.A. HOUZE JR., R. LUKAS, Y. SERRA , J.M.HUMMAN, D.M. LAWRENCE, C.A. RUSSELL, M.N. RYAN,K. SAHAMI and P. ZUIDEMA (2002): The JASMINE pilotstudy, Bull. Amer. Meteorol Soc. , 83 (11), 1603-1630.

WELLER, R.A. and S.P. ANDERSON (1996): Surface meteo-rology and air-sea fluxes in the western equatorial Pacificwarm pool during the TOGA Coupled Ocean-AtmosphereResponse Experiment, J. Climate, 9, 1959-1990.

WHEELER, M. and G.N. KILADIS (1999): Convectivelycoupled equatorial waves: analysis of clouds andtemperature in the wavenumber-frequency domain,J. Atmos. Sci., 56, 374-399.

WOOLNOUGH, S.J., J.M. SLINGO and B.J. HOSKINS (2000):

The relationship between convection and sea surfacetemperature on intraseasonal timescales, J. Climate, 13,2086-2104.

WOOLNOUGH, S.J., J.M. SLINGO and B.J. HOSKINS (2001):The organisation of tropical convection by intraseasonalsea surface temperature anomalies, Q. J. R. Meteorol.Soc., 127, 887-907

WU, M-L., S. SCHUBERT, I.-S. KANG and D. WALISER (2002):Forced and free intraseasonal variability over theSouth Asian Monsoon region simulated by 10 AGCMs,J. Climate, 15, 2862-2880.

XIE, P and P. ARKIN (1996): Analyses of global monthlyprecipitation using gauge observations, satelliteestimates and numerical model predictions, J. Climate,9, 840-858.

YANG, G-Y. and J. M. SLINGO (2001): The diurnal cycle inthe tropics, Mon. Weath. Rev., 129, 784-801.

scale interactions on diurnal to seasonal timescales and

Documents