The 26th session of the Joint SteeringCommittee (JSC) was held at theEscuela Superior Politecnica del

Litoral (ESPOL) in Guayaquil, Ecuador. A.O’Neill and N. McFarlane attended onbehalf of SPARC. Although an importantaspect of the JSC sessions involves reviewingprogress toward achieving WCRP objec-tives, particularly within the core projects(CliC, CLIVAR, GEWEX, and SPARC),these sessions also provide an ideal oppor-tunity to examine new approaches andideas. An important part of this year’s meet-ing involved consideration of the report ofthe Task Force that was set up at the 25thsession to further development of theWCRP strategic framework entitledCoordinated Observation and Prediction ofthe Earth System (COPES). A ModellingPanel and a Working Group on Observationand Assimilation of the Climate Systemwere also set up to support the COPES andfacilitate coordination of modelling andobservational activities across the WCRP.

A number of special topics were also placedon the agenda for discussion at the JSC ses-sion this year. The topic of AtmosphericChemistry and Climate was among these.SPARC, in collaboration with IGAC, hasdeveloped a high level of expertise andexperience in this area and consequentlywas asked to lead the discussion. To facili-tate discussion, a short report entitled

“Chemistry-Climate Interactions” wassubmitted by the SPARC Co-Chairs forconsideration of the JSC prior to the ses-sion. In his presentation on this topic A.O’Neill emphasized the importance andbreadth of atmospheric chemistry and cli-mate interactions as a WCRP issue, notingamong other things that (a) the linkbetween air pollution and climate is a keyissue for society, (b) chemical processesaffecting atmospheric composition are, ingeneral, coupled and nonlinear. He thenposed the following questions for consider-ation by the JSC in regard to AtmosphericChemistry and Climate (AC&C):

• What are the concrete objectives:chemical weather and climate predic-tion and scientific underpinning with-in the seamless prediction of COPES?

• Besides SPARC, what will other WCRPProjects do?

• Should SPARC continue to take thelead? What about tropospheric chemi-cal modelling?

• How is the link with IGAC working?IGBP view?

• Is progress fast enough?• What should happen now?

The JSC thanked A. O’Neill for his presen-tation on AC&C. In reply, it reaffirmed theimportance of AC&C issues to WCRP’sobjectives and stressed the need for devel-

SPARC Newsletter n° 25

STRATOSPHERIC PROCESSES AND THEIR ROLE IN CLIMATE

A Project of the World Climate Research Programme

July

2005

I C

CONTENTSThe 26th Session of the JSC of WCRP,by A. O’Neill and N. McFarlane...................1

SPARC and the International Polar Year(IPY) 2007-2008,by M. Baldwin et al. ......................................3

Report on GRIPS, byD. Pendlebury and S. Pawson.......................6

Overview of planned coupled chemistry-climate simulations to support upcomingassessments,by V. Eyring et al. .......................................11

Report on the 1st WG 1 expert meeting onthe LAUTLOS campaign,by U. Leiterer et al. .....................................18

Tropical Troposphere-to-StratosphereTransport: A Lagrangian Perspective,by S. Fueglistaler et al. ...............................20

Temperatures, Transport, and Chemistryin the TTL, by I. Folkins.............................23

Stratosphere-Troposphere Coupling,by P. H. Haynes ...........................................27

Assimilation of Stratospheric Meteorological andConstituent Observations: A Review,by R. B. Rood..........................................................31

Update: the SPARC Polar StratosphericCloud Assessment,by K. Carslaw ..............................................38

In Memory of Gèrard Mégie 1946-2004,by M.-L. Chanin .........................................39

Announcement: The New SPARC DataCenter, by M. Geller....................................39

Future Meetings ..........................................40

The 26th Session of the JointSteering Committee of the WCRP

A. O’Neill, University of Reading, UK (alan@met.rdg.ac.uk)N. McFarlane, SPARC IPO, University of Toronto, Canada(Norm.McFarlane@ec.gc.ca)

SPARC_R10 7/27/05 12:49 PM Page 1

oping a roadmap for chemistry-climatemodels, observations and process studies.For this purpose, the JSC proposed theestablishment of a Joint WCRP-IGBP TaskForce (TF) involving WCRP core projectsand working groups and IGBP (IGAC), ledby SPARC and IGAC as the core-organiz-ers. IPCC and possibly IHDP should bekept informed. The immediate task for theTF is to organize a workshop; outcomes ofthis workshop should be the terms of refer-ence and suggestions for the way forward.

In his presentation to the JSC on SPARC, A.O’Neill summarized developments withinthe last year, noting that the Third SPARCGeneral Assembly was very successful andthat there was a smooth transition of theSPARC Office from Paris to Toronto. Themain part of the presentation was devoted tohighlighting a number of scientific and tech-nical issues of concern to SPARC and puttingbefore the JSC a number of questions per-taining to them. The first two of these ques-tions relate to the SPARC theme of chem-istry-climate interactions and to the earlierpresentation and discussion on atmosphericchemistry and climate. They are also moti-vated in part by the current uncertainty inpredictions of the Antarctic ozone mini-mum by chemistry-climate models (e.g. asreported in the paper by Austin et al., 2003):

• What is SPARC’s role in the chemistry-climate initiative?

• Should SPARC’s Chemistry-ClimateModelling Validation project lead toanother AMIP-like experiment, andhow should it be facilitated?

The JSC affirmed that SPARC should play aleading role in the Joint WCRP-IGBP TaskForce. It encouraged SPARC to open discus-sions with PCMDI to determine if it could helpin facilitating an AMIP like experiment forGCMs with well resolved stratospheres and rel-atively comprehensive treatments of chemistry.

Validation of models places heavy relianceon global observational data sets and/oranalyses. Changes in observing systemsmay give rise to spurious apparent changesin physical variables such as temperature,prompting the following question:

• What needs to be done at a “high level”to ensure that the temperature recordderived from satellites is cross-cali-brated between satellites?

The JSC noted that the importance of thiscalibration issue has been recognised byGCOS and the former WCRP satelliteworking group, leading to the “reprocess-ing project” now proposed under WOAP.

Analysis and modelling of patterns of cli-mate change and variability are cross-cut-ting issues with the WCRP and are beingexamined within several of the core pro-jects. These are also key issues within theCOPES framework, and therefore prompta question on how to foster desirable col-laborative activities:

• How should SPARC work with CLI-VAR on “modes of atmospheric vari-ability” and how they will change in achanging climate (and therebyrespond to a COPES priority)?

The JSC strongly encouraged joint SPARCand CLIVAR activities on “modes of atmo-spheric variability” and their change in achanging climate and was pleased that thistopic is included in joint CLIVAR/SPARCsession at the upcoming AMS meeting inJune. The ideas exchanged there could beused to plan a joint SPARC/CLIVARWorkshop on Stratosphere-Tropospherecoupling and modes of variability for early2006, the scope also to be guided by thediscussions at the SPARC SSG meeting.

A number of papers and posters were pre-sented at the Third SPARC GeneralAssembly dealing with key processes in thetropopause transition layer (TTL) usingcombinations of observational and mod-elling approaches. Use of cloud resolvingmodels (CRM) is proving to be an innova-tive and potentially powerful approach tostudying key processes within the TTL.Extensive use and analysis of the perfor-mance of CRMs for tropospheric applica-tions has been carried out in the context ofthe GEWEX Cloud System Study and thissuggests another avenue for fruitful collab-orations within the WCRP:

• Should SPARC partner with GEWEX ona new initiative to exploit cloud resolv-ing models to understand processes inthe tropopause transition layer?

The JSC encouraged partnering of SPARCand GEWEX to develop a new initiative toexploit cloud resolving models to under-stand processes in the tropopause transi-

tion layer. The Pan-GCSS workshop inAthens provides an opportunity to begindevelopment of this initiative.

An outcome of the SPARC AerosolAssessment Project (ASAP) has been thedevelopment of a number of valuable datasets. The potential importance (and uncer-tainty) of the radiative effects of aerosols iswidely known. However, the sensitivity ofradiative transfer codes to the treatment ofaerosols remains a potentially importantsource of uncertainty suggesting the question:

• Should SPARC partner with GEWEXto initiate an intercomparison of howaerosols are treated in radiative trans-fer codes (with IPCC in mind)?

The JSC also encouraged partnering ofSPARC and GEWEX on issues pertaining tothe radiative effects of aerosols. A merger ofaerosol data sets is needed and a compari-son of the treatment of aerosols in radiativetransfer codes should be carried out.

Evaluating the influences of solar vari-ability on climate is important for under-standing climate change. A goal of thejoint SPARC-CAWSES activity on thistopic is to elucidate the effects of solarvariability on atmospheric composition,for example on ozone. However the effectof solar variability on the radiation bud-get of the whole atmosphere and the sur-face and its impact on the oceans andcryosphere is of broader interest and hasbeen regarded as an important factor inmodelling and evaluation of climatechange. The effects of changes in atmo-spheric composition are not addressed inmost GCMs and this may be an impor-tant shortcoming. The question for theJSC prompted by this issue is:

• What should SPARC do and howshould we work with WGCM (withIPCC in mind)?

In regard to solar forcing WGCM hasfocused on the radiative effects of longterm variations in solar forcing at the topof the atmosphere (TOA) in the absenceof directly attributable changes in atmo-spheric composition.

The JSC recommended that SPARC shouldwork with WGCM in updating TOA solarforcing data while continuing to pursue its

2

SPARC_R10 7/27/05 12:49 PM Page 2

3

SPARC and the International Polar Year (IPY) 2007–2008Mark P. Baldwin, Northwest Research Associates, USA (mark@nwra.com)Jessica L. Neu, University of California at Irvine, USA (jessica.neu@noaa.gov)Sheldon Drobot, University of Colorado, USA (sheldon.drobot@colorado.edu)Pablo Canziani, Pontificia Universidad Catolica, Argentina (canziani@ic.fcen.uba.ar)Shigeo Yoden, Kyoto University, Japan (yoden@kugi.kyoto-u.ac.jp)Norman McFarlane, SPARC IPO, University of Toronto, Canada (Norm.McFarlane@ec.gc.ca)

What is IPY?

Nearly 150 years ago, 13 nations joinedforces for the first internationally co-ordi-nated programme of scientific explorationin the polar regions during the InternationalPolar Year 1882–1883. Beyond the advancesin science and geographical exploration, aprincipal legacy of the first IPY was that itset a precedent for international science co-operation. In 1904 the first permanentAntarctic station was established by thegovernment of Argentina at Orcadas/SouthOrkneys to carry out research (includingweather monitoring) and other activities inthe region. This base still continues its sci-entific activities.

Following the successful IPY, theInternational Meteorological Organizationpromoted the second IPY in 1932–1933 toinvestigate the global implications of thenewly discovered jet stream. Some 40nations participated in the second IPY,which heralded advances in meteorology,atmospheric sciences, geomagnetism, andthe “mapping” of ionospheric phenomena.In the years following World War II, manynations increased the number of perma-nent and semi-permanent stations andbases in both polar regions, thus providingmany of the long term weather datasets that

are crucial to our understanding of polarweather and climate processes and theirrelationship with the rest of the world.

Scientists again decided that an interna-tional science year was warranted to utilizethe new technologies of the era. This time,the scope of the effort was global, and 67nations participated in the InternationalGeophysical Year (IGY) in 1957–1958. TheIGY led to an increased level of research inmany disciplines, and the scientific, institu-

tional, and political legacies of the IGYendured for decades, in many cases to thepresent day. IGY helped consolidate theengagement of many countries in sus-tained Antarctic research.

Today, nations around the world are plan-ning for a new International Polar Year in2007–2008. This IPY will be far more thanan anniversary celebration of the IGY orprevious IPYs; it will be a watershed eventand will use today’s powerful research tools

current interest and activities in regard tothe effects of solar forcing variations onatmospheric composition.

Data Assimilation issues are important forSPARC for several reasons. There is now awealth of stratospheric chemical data avail-able from satellites. Stratospheric analysesare now being produced by major opera-tional weather prediction centres and thesehave been used in off line chemical trans-port models. Some of the presentationsgiven in the SPARC General Assemblyhighlighted encouraging developments as

well as challenges associated with issues ofbias, noise, and inaccuracies inherent intransport algorithms. The questions posedto the JSC in this regard were the following:

• Should the SPARC Working Group onData Assimilation (WGDA) combinewith activity in the Working Group onNumerical Experimentation (WGNE)?

• What can the SPARC WGDA do for theWCRP Observations and AssimilationPanel (WOAP)?

The issue of the cooperation between the

SPARC WGDA, WGNE, and WOAP wasleft for later consideration. A variety of fur-ther actions need to be considered. Both ofthe SPARC presentations were very wellreceived by the JSC. The consensus wasthat they were very timely in highlightingkey issues for the WCRP as a whole andposing clear questions for discussion.

Reference:

Austin, J. et al. Uncertainties and assessmentsof chemistry-climate models of the strato-sphere. Atmos. Chem. Phys., 3(1), 1-27, 2003.

SPARC_R10 7/27/05 12:49 PM Page 3

to better understand the polar regions.Automatic observatories, satellite-basedremote sensing, autonomous vehicles, theInternet, and genomics are just a few of theinnovative approaches for studying previ-ously inaccessible realms. IPY 2007–2008will be fundamentally broader than theIGY and past IPYs because it will explicitlyincorporate multidisciplinary and interdis-ciplinary studies, including biological, eco-logical, and social science elements. Such aprogramme will not only add to our scien-tific understanding, but will also assemblea world community of participants withshared ownership in the results.

IPY 2007–2008 will provide a frameworkand impetus to undertake projects thatnormally could not be achieved by any sin-gle nation. It will allow us to think beyondtraditional borders — whether nationalborders or disciplinary constraints —toward a new level of integrated, coopera-tive science. A coordinated internationalapproach maximizes both impact and costeffectiveness, and the international collab-orations begun today will build relation-ships and understanding that will bringlong-term benefits. Within this context,IPY 2007–2008 will seek to galvanize newand innovative observations and researchwhile at the same time building on andenhancing existing relevant initiatives,many of which have been subject to dwin-dling budgets and cancellations in recentyears despite their relevance for long termmonitoring. In addition, there is clearly anopportunity to organize an exciting rangeof educational and outreach activitiesdesigned to excite and engage the public,with a presence in classrooms around theworld and in the media in varied and inno-vative formats.

IPY 2007–2008Organization

IPY 2007–2008 is organized around sixthemes. Its broad goals are:

• To determine the present environmen-tal status of the polar regions by quanti-fying their spatial and temporal vari-ability;

• To quantify, and understand, past andpresent environmental and humanchange in the polar regions in order toimprove predictions;

• To advance our understanding ofpolar–global interactions by studyingteleconnections on all scales;

• To investigate the unknowns at the fron-tiers of science in the polar regions;

• To use the unique vantage point of thepolar regions to develop and enhanceobservatories studying the Earth’s innercore, the Earth’s magnetic field,geospace, the Sun and beyond; and

• To investigate the cultural, historical,and social processes that shape theresilience and sustainability of circum-polar human societies, and to identifytheir unique contributions to global cul-tural diversity and citizenship.

The IPY International Programme Officeis hosted by the British Antarctic Surveyin Cambridge, UK. The office providesplanning, coordination, and guidance to25 international organizations that sup-port IPY, as well as 28 national IPY com-mittees. In January 2005 the officereceived over 900 “Expressions of Intent”from the international research commu-nity, including one EoI from SPARC.SPARC-IPY was recognized by the IPYJoint Committee as having the “potentialto make a major contribution to the IPY,”and is expected to “clearly contribute tosignificant international collaboration.”SPARC was invited to submit a full pro-posal, due in September 2005. While theIPY office will coordinate the researchefforts, it does not fund the research.Funding is left to national and interna-tional funding agencies.

Although IPY 2007–2008 is orientedtoward the polar surface environment, italso emphasizes connections to otherregions as well as the solid Earth below andthe atmosphere above.

Connection BetweenPolar Climate and the

Stratosphere

There is a strong dynamical connectionbetween the circulation of the high-lati-tude stratosphere, and surface weather andclimate. In particular, stratospheric windanomalies tend to progress downward tothe lowermost stratosphere (near 10 km),and then induce changes to the ArcticOscillation (AO) pattern, which is similarto the NAO (North Atlantic Oscillation).

Our understanding of the mechanisms isadvancing, but it is still incomplete. Overthe Arctic, the phase of the AO affects sur-face winds, temperature, sea-ice motion,and ice extent. There appears to be a mem-ory in summer sea-ice of the previous win-ter’s AO, and part of the observed thinningof sea-ice can be attributed to long-termchanges in the AO.

In the Southern Hemisphere, observationsand models show that the springtimestratospheric ozone hole has not only beenlinked to cooling of the lower stratosphereand strengthening of the circumpolarwinds within the stratospheric polar vor-tex, but that these changes have inducedsurface circulation and temperaturechanges over Antarctica — changes lastingwell into the summer. During spring andsummer, the lower stratosphere at south-ern mid-latitudes has in turn undergonechanges linked to the changes overAntarctica. Although chlorofluorocarbons(CFCs, now banned by international agree-ment) are largely responsible for currentozone depletion, increasing concentrationsof greenhouse gases like carbon dioxideand methane may delay future ozonerecovery.

As greenhouse gases increase, the circula-tion of the stratosphere will likely beaffected. But we are not able to predictwhether high-latitude stratospheric windswill become stronger or weaker a fewdecades from now. Such a trend — posi-tive or negative — is expected to affect theAO at the Earth’s surface. On climate-change timescales, stratospheric effects arepotentially large, and understanding howthe stratosphere will change, as well as howthe stratosphere and troposphere are cou-pled, will contribute to reducing thatuncertainty.

Trends and variability in the AO — includ-ing changes in the stratospheric circulation— would affect the lifetimes of naturalgreenhouse gases such as methane andN2O, as well as of anthropogenic green-house gases such as CFCs or their replace-ments, since the removal of all of thesegases requires, at least in part, transportthrough the stratosphere. Changes in life-times of greenhouse gases could them-selves result in a forcing of the stratospher-ic circulation. It is uncertain to what degreeozone would be affected, since ozone acts

4

SPARC_R10 7/27/05 12:49 PM Page 4

not only as an ultraviolet filter but in amultifaceted manner as a greenhouse gas.Global, and in particular, polar ozone con-centrations might respond sensitively tothe circulation changes as well as tochanges in ozone destroying trace gasesresulting from the degradation of CFCs,N2O and methane.

Trends and variability in the AO are fur-ther reflected in ecosystem changes, andfeedback from ecosystem changes couldbe manifested in the stratosphere as wellas the troposphere. Here one researchtopic would be to study whether changesin permafrost and wetlands are consistentwith AO signals, investigating the effectsof temperature and precipitation changeson methane fluxes and then estimatinghow these fluxes will change with timebased on AO trends, including the possi-bility of enhanced production of watervapour in the stratosphere due tomethane oxidation.

This question of changes in stratosphericwater vapour by any mechanism isextremely important because it has beenposited that increased stratospheric watervapour due to either large methane releasesor decreased latitudinal temperature gradi-ents may have led to an increase in the fre-quency and thickness of polar stratosphericclouds and drastically changed the radiativebalance in the Arctic during the Eocene(55–40 million years ago), with large feed-backs on high-latitude temperatures. Sinceboth of these mechanisms for increasingstratospheric water vapour may be presentin the climate of the near-future, it is essen-tial to try to determine the magnitude ofthe changes that may take place.

Weather in southern-most countries likeAustralia, New Zealand, Chile, andArgentina is strongly influenced byAntarctic atmospheric processes. EvenSouth Africa and Brazil occasionally suffercold spells of Antarctic origin in winter.The future evolution of climate in thosecountries and over the southern oceans isthus strongly linked to the changes that willtake place in the Antarctic troposphere andstratosphere. Some of the richest fisheriesand marine ecosystems in the world couldsuffer major impacts from changes inAntarctic climate, affecting the currentstate of biogeochemical cycles, marine foodchains and fishing activities.

SPARC’s Contributionto IPY

Because IPY will occur over a short timeperiod, SPARC will focus on details of thepolar stratosphere in a programme called“The structure and evolution of the strato-spheric polar vortices during IPY and itslinks to the troposphere.”

The Antarctic ozone hole is one of the mostrecognized environmental issues of the20th century. Ozone changes in the Arctic,though lesser in magnitude, are equallyimportant. Recent research has shown thatthe evolution of stratospheric ozone istightly coupled to a wide range of process-es acting within and outside the winterpolar vortices. Much of this understandinghas been achieved within the SPARC pro-gramme. The IPY programme offers aunique opportunity for SPARC to assemblea range of scientific expertise to study theAntarctic and Arctic Polar Vortices, the lociof key processes associated with ozonedepletion and its eventual recovery, as wellas contribute towards a better understand-ing of the coupling mechanisms betweenthe troposphere and the stratosphere.

SPARC-IPY will co-ordinate the activitiesof the international SPARC community inrelation to IPY. This co-ordination will bedirected toward both satellite and ground-based experimental campaigns, as well asspecific initiatives promoted by SPARC toincrease understanding of the polar atmo-sphere. The services of the SPARC DataCenter will be made available to facilitateacquisition and archiving of key data thatwill be used for projects or generated bythem during the IPY period.

In addition to coordinating and facilitatingIPY projects within the SPARC communi-ty, SPARC-IPY will promote specific initia-tives directed toward the understanding ofmajor features and processes in the polarmiddle atmosphere during the IPY period.These initiatives will include a range ofresearch activities involving modelling,observations, and analysis, and will includeworkshops and meetings as needed ordesirable to facilitate research and dissemi-nation of results. These efforts will be car-ried out in context of the SPARC Projectcore thematic programmes ofStratospheric Chemistry and Climate,Stratosphere-Troposphere Coupling, and

Detection, Attribution, and Prediction ofStratospheric Changes.

The dynamics, transport and chemistry ofthe polar vortices, as well as of propertiesrelevant to microphysical processes, such asthe formation of polar stratosphericclouds, will be documented as completelyas possible. To achieve this detailed picture,SPARC-IPY will bring together availableresearch and operational satellite data, aswell as ground-based and aircraft data.SPARC-IPY will promote co-ordinatedfield campaigns to enhance the database,and encourage work on data assimilationand intercomparison of assimilateddatasets to yield a unique synthesis of dataon the polar vortices. Weather services car-rying out routine radiosonde andozonesonde measurements would beencouraged to increase the frequency ofthe observations and to store the data withfull resolution.

Observations of a range of variables withinthe stratospheric polar vortex will be used,together with data assimilation, modelsand other analysis techniques to create acoherent and comprehensive picture of thecurrent state of the stratosphere in theArctic and Antarctic, and to elucidate fur-ther the interaction of the polar strato-sphere with the underlying troposphere.

The project will involve the multi-nationalSPARC community and its affiliates, e.g.other WCRP research projects. SPARC-IPYwill co-ordinate relevant field campaignssupported by national and internationalresearch programmes, and will seek, wherepossible, to promote additional campaignswhere there are data gaps that need to befilled. As a core project within the WCRP,the SPARC Project relies on co-ordinatedactivities by agencies and groups within theaffiliated community for the resourcesneeded to carry out observational cam-paigns and research within its thematicprogrammes. SPARC-IPY will functionwithin this general SPARC framework ofco-ordination and collaboration withrelated and linked national and institution-al IPY projects.

5

SPARC_R10 7/27/05 12:49 PM Page 5

Introduction

The tenth and final workshop for theGCM-Reality IntercomparisonProject for SPARC (GRIPS) was

held in Toronto. For about a decade, GRIPShas been the modelling focus for SPARC,and has had the role of evaluating andcomparing different dynamical models ofthe stratosphere from the internationalcommunity. A number of tasks have beendefined, ranging from basic evaluation andvalidation of GCMs, through studies thataimed to understand the realism and limi-tations of processes in the models, tolonger model simulations designed toexamine the consistency of how modelsrespond to changes in forcing. Followingthe tradition of past GRIPS workshops, theprogramme included elements related tothe formal tasks, alongside other presenta-tions, ranging from developments in indi-vidual models to broader scientific ques-tions of relevance to the community.

In order to meet the evolving demands onSPARC, the modelling strategy has beenrevised, to the extent that the Chemistry-Climate Model Validation (CCMVal) pro-ject will become the main focus. CCMValhas a broader mandate than GRIPS, withthe intention of addressing issues of rele-vance to the topics of all three of SPARC’smajor themes (Stratospheric Chemistryand Climate, Stratosphere-TroposphereCoupling, and Detection and Attribution ofStratospheric Change). Many of the issuesare directly related to some of the GRIPStasks in explicit relationships (e.g., modelclimate and the factors that influence it),and indirect relationships (e.g., impact ofparameterizations on climate and climatechange). However, in order to answer thesequestions, SPARC needs to have a muchbroader scope for model evaluation andvalidation with formal tasks on details ofchemistry and radiation. CCMVal willencompass detailed analyses of the chem-istry (e.g., photolysis) and transport, as wellas the dynamics and radiation. To this end,CCMVal will likely involve detailed study ofthe radiative, chemical and dynamical

aspects of current models, not necessarilyrestricting attention to GCMs.

Because of the relevance of many GRIPSactivities to CCMVal, a main aspect of the2005 Workshop was on how any unfinishedtasks may be carried forward. It should bestressed that, even though no formal GRIPStasks involving chemistry-climate modelshave been defined, much discussion at thisand previous workshops has been devotedto such models. As these models reach alevel of maturity that justifies cross-com-parison, it is natural that SPARC’s mod-elling activities should look for suitabletests of these models, just as GRIPS hasexplored the performance of the dynami-cal-radiative components of the full CCMs.

WCRP has led numerous modelling effortswith the Working Group on CoupledModels (WGCM) and the Working Groupon Numerical Experimentation (WGNE).Projects such as AMIP, CMIP among othershave fulfilled the needs of the community,with substantial investment in infrastruc-ture. WCRP’s objective is to unify theseefforts into a common framework under theCoordinated Observation and Prediction ofthe Earth System (COPES), with two panelsthat will report directly to the JointScientific Committee (JSC): the WCRPModelling Panel (WMP) and the WCRPObservations and Assimilation Panel(WOAP). The COPES WMP will consist ofrepresentatives from all WCRP programmes(S. Pawson will represent SPARC) and fromall WCRP working groups.

Model Developments

Several presentations discussed recentdevelopments in climate models. Discussionof these model developments enabledspeakers and audience to share commonexperiences with the various models.

J. Scinocca from CCCma and B. Morelfrom LMDz gave presentations on theongoing development and application ofglobal coupled climate models aimed at

understanding climate change and vari-ability. The focus at CCCma is primarily onimproved representation of physical pro-cesses, but a major undertaking over thepast year has been the set-up and executionof a large number of scenario runs for usein the upcoming IPCC Fourth AssessmentReport. The CCCma global model is usedoperationally to produce seasonal fore-casts, and because the model computes‘weather’ at 20-minute time steps, one canuse the model to say something aboutextreme events and their changing proba-bilities. The LMDz-Reprobus CCM hasbeen evaluated with a 10-year climatologyand compares well with observations.Sensitivity studies for the orographic grav-ity wave forcing have been performed, andimply that the introduction of the strato-sphere has led to an increase in the surfaceAO persistence and predictability.

A. Bushell presented the efforts in extend-ing the Met Office HadGAM1 ‘NewDynamics’ Climate Model. The model haschanged from non-hydrostatic to quasi-hydrostatic, and to a hybrid sigma-pressuregrid, and includes new dynamics (e.g. massflux convection, statistical cloud scheme,prognostic ice microphysics, non-localboundary layer), extra middle atmospherephysics such as methane oxidation (pho-tolysis of water vapour at higher levels),spectral gravity wave parameterization(new dissipation and launch spectra, and atransparent upper boundary, hydrostaticnon-rotating dispersion relation), and newadded input data. The group is also lookingto incorporate satellite data assimilation,increasing predictability at seasonaltimescales and improving process repre-sentation, such as the QBO.

R. Stolarski and M. Gupta reported onresults of the Goddard stratospheric chemi-cal transport model (CTM) and on updatingthe NASA Goddard Coupled Chemistry-Climate model (GEOS-GCM). The credibleperformance of the CTM driven by GEOS-GCM meteorology in simulating the age ofair, the life-cycle of the polar vortex and the

6

Report on GRIPSMarch 14-17, Toronto, Canada

Diane Pendlebury, SPARC IPO, University of Toronto, Canada (diane@atmosp.physics.utoronto.ca)Steven Pawson, NASA Goddard Space Flight Center, USA (pawson@gmao.gsfc.nasa.gov)

SPARC_R10 7/27/05 12:49 PM Page 6

observed ozone trends is the basis for devel-opment of a coupled model. The CTMtransports source molecules, (e.g. CFCs,Halons, methyl bromide, methane, nitrousoxide) with specified surface mixing ratios,and chemical families (e.g. NOx, ClOx).

Additionally, GEOS-GCM simulationsusing different scenarios of SST and CTMproduced ozone distributions that haveshown reasonable behaviour. Chemistryand radiation in GEOS-CCM are presentlycoupled only through stratospheric andmesospheric ozone. Future plans includeinvoking the radiative coupling of chemi-cally modified water vapour between 380Ksurface and top of the model domain,extensive evaluation of the model withUARS and AURA observations, introducinga combined tropospheric-stratospheric-mesospheric photochemical mechanism,and coupling with an ocean-ice model.

StratosphericForecasting

G. Roff presented the study from WGNE onstratospheric prediction. Better skill in thestratosphere is expected since the dynamics isdominated by a quasi-stationary polar vor-tex, unlike the troposphere, which is influ-enced by transient and synoptic scale waves.Therefore, it is to better test our skill in pre-dicting “stratospheric weather” when thepolar vortex is undergoing strong changesover a short period of time, such as suddenwarmings. The results show that stratospher-ic forecasting in the Northern Hemisphere(NH) and Southern Hemisphere (SH) showssimilar characteristics. The stratosphericforecasting performance at six days is compa-rable to three days in the troposphere, butthere exists large variability in the forecastskill at six days, often depending on howactive the planetary waves are and thushow quickly the vortex distorts. Skill isincreased by increasing stratospheric verti-cal resolution and by raising the lid.

T. Hirooka showed results from a studyexamining the predictability of StratosphericSudden Warmings (SSWs) in the NHinferred from ensemble forecast data. Twocase studies (Mukougawa and Hirooka(2004, Mon. Wea. Rev.), Mukougawa et al.(2005, GRL, submitted)) found predictabili-ty times of two weeks to one month for SSWevents in December of 1998 and 2001. It wasfound that different ensemble members

showed a high sensitivity of the predictionskill to the initial conditions. This case studylooks at SSWs in 2002/03 and 2003/04. It wasfound that the predictability of SSWs inthese cases was approximately two to threeweeks, and that the predictability is depen-dent on occurrences of SSW events.

While stratospheric forecasting has beenperipheral to the main aims of GRIPS, sev-eral presentations have addressed it at var-ious workshops. This issue has beenaddressed by the WGNE group, but withthe reorganization of SPARC’s modellingactivities and the attempts by the WCRP tounify their modelling work throughCOPES, the topic remains relevant toSPARC. Discussions of stratospheric fore-casting were thus based on this premise. Itwas noted that various groups are nowusing the same models for analysis/fore-casting as for climate studies. This meansthat study of model performance whenconstrained by atmospheric data will likelyemerge as an important diagnostic for theclimate models. Such activities will likelybe coordinated through SPARC’s dataassimilation group, in conjunction withCCMVal and the WCRP-COPES panels.

Polar Vortices,Warmings andAnnular Modes

G. Roff presented the results from the com-parison of polar vortices between modelsand analyses (GRIPS task 1i). The NMCdataset indicates that the typical character-istics for the SH polar vortex are; a rapidand deep onset throughout the depth ofthe atmosphere, a high correlation betweenits size and the maximum wind speed, it isgenerally polar centred and symmetric,and during its demise there is gradualdecay from aloft. All the models tend tocapture the main features of the vortex, butthe vertical extent is greatly affected bymodel characteristics (e.g. those modelswith a sponge layer are forced to close thepolar vortex near the top levels, and thosewith no gravity wave parameterizationtend to have very strong, deep vortices).Elliptical diagnostics show that models thatdo not extend high enough have polar vor-tices that are artificially curtailed aloft.

L. Polvani presented a new climatology ofSSWs. From observations we can deter-

mine their frequency, amplitude, type anddistribution, and from models we can learnabout their dynamical behaviour. Using theWMO definition (easterly winds at 60Nand 10hPa), SSWs can be classified intotwo main groups according to their evolu-tion: vortex displacement and vortex split-ting. In the NCAR/NCEP and ERA-40analyses, it was found that about three SSWevents occurred every four years, and thatvortex displacements accounted for twothirds of these events. These numbersshould serve as benchmarks for models.Other results are that the largest variabilityis in January-March, splitting events areconcentrated in January and February, andthere is little evidence of trends in thenumber of warmings. He also noted thatthe Baldwin and Dunkerton picture ofdownward propagation of the AO signalmight be misleading in that the weak vor-tex events are actually SSWs, and the strongvortex events are really non-events.

Aspects of the circulation that have receivedmuch attention are the annular modes inthe two hemispheres. In a study using theMRI Chemical GCM, Y. Kuroda examinedthe impacts of solar variability on the annu-lar mode in the SH and compared themodel response to observations. Regardingthe Northern Annular Mode (NAM), animportant question remains: what drivesthe daily variability of the NAM? There is no‘NAM tendency’ equation. A. Haklanderpresented an analysis of mid-latitude strato-spheric wind variations, using the zonal-mean momentum equation, to investigatechanges in normalized cross-covariances ofeddy forcing terms with the zonal meanwind tendency. Using the TransformedEulerian Mean formulation, the resultingdaily absolute angular momentum changesyield both a change in relative and planetaryangular momentum, each of similar magni-tude. There is a downward propagation oflow-frequency variations of the wind ten-dency not visible in the resolved eddy forc-ing terms, suggesting that gravity wave dragmay play a role in the downward propaga-tion of variations in the upper stratosphereand lower mesosphere.

Kodera (2002, 2003) found that the signalassociated with the NAO in winter extendsto hemispheric scale and into the upperstratosphere in High Solar (HS) years (K-phenomenon). Ogi et al. (2003) also foundthat the signal associated with the winter

7

SPARC_R10 7/27/05 12:49 PM Page 7

NAO tends to persist until next summer inHS years (O-phenomenon). K. Koderapresented a study using ERA-40 data, todetermine whether the Southern AnnularMode (SAM) exhibits similar behaviour. Acorrelation analysis was performed on theOctober-November mean SAM index sep-arately on HS and LS years. Both phenom-ena were found in SAM, and SAM is verypersistent in HS years. Observations, andmodel runs with and without chemistrysuggest that ozone plays a key role for per-sistence of SAM signal. The high extensionof the SAM signal toward the upper strato-sphere in late winter (K-phenomenon) isimportant factor for the driving of ozoneto polar-lower stratosphere for the persis-tence of the AM (O-phenomenon).

J. Perlwitz presented a study on the impactof stratospheric climate change on the tro-posphere by stratosphere-tropospheredynamical coupling. The key processes instratosphere-troposphere dynamical cou-pling are the upward propagation of plane-tary Rossby waves from the troposphere tothe stratosphere, and the absorption of waveactivity in the stratosphere which changesthe mean flow, which in turn changes theregion of strongest interaction of the wavesdue to these changes in the basic state. Thezonal mean flow perturbation progressesdownward and poleward. Wave activity mayalso be reflected back down into the tropo-sphere such that the structure of tropo-spheric waves is modified, but there is littleeffect on the zonal mean.

To study the downward propagation of theNorthern Annular Mode (NAM), the lead-ing EOFs were calculated of the daily zonalmean height, and the time-lagged correla-tion coefficients for DJF relative to 10 hPa.It was found that the downward progres-sion of the NAM was captured in all mod-els studied, except in the GISS model, butthat the relationship between the strato-spheric and near-surface NAM fields isstronger in models than in reanalysis.Models also show a stronger persistence ofNAM in the stratosphere.

Discussion on stratospheric variability andits coupling to the troposphere was lively.While early GRIPS tasks studied warmingsand interannual variability of the coupledtroposphere-stratosphere system, thesemodel evaluations were limited by the lackof long runs. Since it is now becoming pos-

sible to run multi-decadal simulations (evenwith interactive chemistry), it is timely torevisit these questions. Reviving the earlyGRIPS Tasks, in the context of examing 20-50 year simulations with present models, isthus possible. The presentations by G. Roffand J. Perlwitz, as well as many ideas raisedin discussion, pointed to the continuedvalue of such evaluations, for both funda-mental understanding of GCMs (or CCMs)and for insight into how changes in the cir-culation into the 21st century may beimpacted by the baseline circulation statis-tics of the models. Such work is encouragedin the GRIPS-CCMVal transition period.

Gravity Wave Drag

A realistic simulation of the climate of themiddle atmosphere requires the transfer ofangular momentum by unresolved gravitywaves (GW). Without a gravity wave drag(GWD) parameterization scheme mostGCMs will not produce many of the basicfeatures of the middle atmosphere (e.g., coldsummer mesopause, zonal wind reversal ofthe mesosphere, QBO). A number of param-eterizations are currently used in GCMs,with the fundamental difference betweenthem being how the waves “break” anddeposit their momentum to the backgroundflow. However, due to the lack of observa-tional evidence, the parameters of a GWDparameterization are typically “tuned” toobtain a reasonable mean climate.

While gravity wave properties at sourcelevel are the greatest uncertainty, typicallythey are specified so that a reasonable mid-dle atmosphere results from numericalsimulations. However, this feature doesnot allow for changes of source propertiesin different climate change scenarios.Using the WACCM, F. Sassi tested threeGW generation schemes: the base case(with a zonally uniform GW source thatoperates continuously with an intermit-tency factor), the Charron and Manzini(2002) scheme for GW production due tofrontal development, and the Beres (2004)scheme for production of GW by convec-tion. The results show that model consis-tent GW schemes are in general preferableto ad hoc solutions. Frontogenesis andconvection introduce realistic seasonaland spatial variability, however, theseschemes introduce a new level of tuningthat can be arbitrary and model depen-dent, and needs further exploration.

C. McLandress addressed the question ofhow the differences in the parameteriza-tion of wave breaking actually affect cli-mate simulations. Gravity wave dissipationcan occur in two ways: nonlinear dissipa-tion, and critical level (CL) dissipation. It isthe nonlinear dissipation that is differentin each of the GWD parameterizations.Using the standard settings from Hines,Warner-McInytre (WM) and Alexander-Dunkerton (AD) will produce large differ-ences in the GCM responses. However, bymodifying the saturation threshold forWM and AD, it is possible for all threeparametrizations to deposit their momen-tum at the same height. The wind responseis nearly identical. This suggests that it isheight at which the GWs are dissipated,and not the details of the nonlinear dissi-pation mechanisms, that is the crucial fac-tor in determining the GCM response.

T. Shaw addressed the question of spuriousdownward influence of the mesosphere onthe stratosphere due to using GWDparametrizations that are not constrainedby momentum conservation, either inprinciple or in the way in which they areimplemented. Momentum conservationimplies that GWD induced downwelling(and heating) at a given altitude dependsonly on the gravity wave momentum fluxthrough that altitude (Haynes et al., 1991),such that GWD feedbacks from changes inthe zonal wind above a given level arerestricted to the region above. Therefore,zonal wind changes above a level cannotaffect the circulation below it via GWDfeedbacks. To respect the momentum con-straint and avoid spurious downwardinfluence, any nonzero parameterizedmomentum flux at the model lid must bedeposited in the model domain. Dynamicalfeedbacks from parameterizations could befalsely interpreted as stratospheric andmesospheric effects on climate.

S. Pawson showed runs from the GEOS-4GCM that were capable of producing aQBO. The AD GWD parametrization wasused and the control run produced a rea-sonable SAO and a low-frequency oscilla-tion in the tropical stratosphere, with someresemblance to the QBO with a period ofabout 20 months. GWD by a “convectivespectrum” in the AD scheme does thework, however, the spectral parametersneed to be quite different from those rec-ommended by Alexander and Rosenlof

8

SPARC_R10 7/27/05 12:49 PM Page 8

(AR). It was found that in order to producethe QBO shorter wavelengths (100 km vs.4000 km) were used than in the standardAR scheme, a much narrower spectralwidth was needed, and a stronger momen-tum flux at launch level was needed. It wasalso found that the QBO period increasedwith slightly weaker GW forcing and therewas a better downward extent of the QBO,especially in the easterly anomalies.Increasing the vertical resolution also led totighter shear zones and perhaps betterdownward propagation into the lowestpart of the stratosphere.

Issues in Chemistry-Climate Modelling

R. Stolarski gave a presentation on the needfor, and the problems with producing, high-quality, long-term datasets. Long-termdatasets, such as the total ozone data fromTOMS and SBUV, enable us to test how pro-cesses act together to give decadal-scaleresponses. The ultimate prediction test isour hindcast capability so that decadal scalehindcasts compared to long-term data setsprovide additional evaluations of modelsbeyond process-oriented evaluation.

A long-term dataset is comprised of datafrom many satellites, and each instrumenthas calibration and drift issues. It is neces-sary to estimate trend uncertainty due toinstrument drift and to remove known sys-tematic errors from long-term datasets.However, residual errors could remain thatlead to drift uncertainty. In addition, theintroduction of a new instrument into adataset introduces a new uncertainty, so it isimportant to have enough overlap to prop-erly calibrate instruments and improveuncertainties. Ground stations are useful inthat they may be calibrated as needed, butstatistical uncertainty dominates the trenduncertainty, and there are still drift uncer-tainties in individual ground stations.Using a combination of both satellite andground-based measurements is better.

Extratropical ozone builds up inwinter/spring due to transport from thetropics, and decreases in late-spring/sum-mer due to photochemistry (more so inNH than in SH). It tends to be that yearsthat are high/low in spring are high/low insummer. T. Shepherd showed that the cor-relation coefficient (of detrended data)between different months from 35° to 60°

shows a remarkable seasonal persistence ofozone anomalies until fall. In both hemi-spheres, 35°-80° springtime ozone is a goodpredictor for both 35°-60° and 60°-80°summertime ozone. In fact it is a betterpredictor of 60°-80° summertime ozonethan is 60°-80° springtime ozone. Once thevortex breaks down, the correlationsbecome coherent throughout the extra-tropics (Fioletov and Shepherd, 2005). TheSH midlatitude ozone variability seems tobe slaved to NH variability. This memorycould be in tropical zonal winds, i.e. the“flywheel” (Scott and Haynes, 1998).

V. Fomichev presented the result from anongoing study to diagnose the impact ofincreasing greenhouse gas (GHG) concen-trations in the Canadian MiddleAtmosphere Model (CMAM). A series ofmulti-year runs has been performed withdouble CO2, with and without interactivechemistry, and with sea surface tempera-tures (SSTs) prescribed for both 1xCO2 and2xCO2 climate. In response to CO2 dou-bling, the middle atmosphere cools ~10Kwith maximum impact near thestratopause. The ozone radiative feedback(through both solar and IR heating)reduces the CO2 induced cooling by up to~4.5K. The main impacts of SST changeson the middle atmosphere are a higher andwarmer tropopause, cooling just above thetropopause (in both the tropics and extra-tropics), and a decrease in ozone near thetropopause, both possibly related to the“tropopause shift”. However, the winterpolar regions can be highly variable, solong runs are required to increase confi-dence in the results.

On behalf of A. Scaife and N. Butchart, A.Bushell presented the results from theLevel 4 GRIPS task on how climate changeaffects the Brewer-Dobson circulation. Thetroposphere-stratosphere mass exchangedetermines the lifetimes of key trace gases,and transports ozone down into the tropo-sphere. Therefore, changes in the mass cir-culation will affect both ozone and climatepredictions. The Unified Model predictsthat an increase in GHGs will cause a sys-tematic increase in troposphere-strato-sphere exchange (Butchart and Scaife,2001), and this study is an attempt to assessthe robustness of this predicted change inmass exchange and the role of planetarywave driving using 14 different model runsfrom 11 different groups.

9

Table 1: The GRIPS tasks, leaders and stages of completion.

Level 1 Tasks :1a Documentation S. Pawson1b Climatology S. Pawson1c Troposphere-Stratosphere K. Kodera

Connection1d Sudden Warmings W. Lahoz complete1e Travelling Waves/Tides K. Hamilton complete1f Tropical Waves T. Horinouchi complete1g Stratosphere-Troposphere -

Exchange1h Spatial Wavenumber J. Koshyk Complete1i Polar Vortices G. Roff ending, report this meeting1j Transport -

Level 2 Tasks :2a Radiation Scheme Comparison P. Forster, V. Fomichev, First stage almost

and Validation U. Langematz complete2b Inferred GWD S. Pawson complete2c Impacts of Mesospheric Drag S. Beagley, B. Boville complete2d GWD Evalution C. McLandress complete

Level 3 Tasks :3a PINMIP : Impact of Mt. G. Stenchikov, ongoing

Pinatubo aerosols A. Robock3b Response to solar forcing K. Kodera, Matthes complete

anomalies3c Response to ozone trends U. Langematz near completion3d Response to CO2 change -

Level 4: Tasks :4a Changes in residual circulation A. Bushnell, A. Scaife, report this meeting

due to climate change N. Butchart

SPARC_R10 7/27/05 12:49 PM Page 9

The results of the study are that all modelshave a Brewer Dobson circulation withupwelling in a “tropical pipe” extending toabout ±30° (annual mean), and the pipe isdisplaced by 10°-20° toward the summerpole. However, at 70 hPa, the upwellingmass flux varies by up to a factor of 2between models. The maximum upwellingis in DJF and is often 50% larger than inJJA. All models confirm a positive trend inthe mass flux across the tropopause due toclimate change. This occurs throughoutthe year and is consistent with an increasein planetary wave driving at upper levels.The increase is about 2% per decade. Thisfeedback could amount to a ~20% decreasein the lifetime of N2O and CFCs, for exam-ple, by 2100 and is not included in stan-dard climate models.

J. Austin presented a study on the age of airand the meridional circulation in a GFDLcoupled chemistry-climate model. Two 21-year runs were used: one with 1960 forc-ings and one with 1980 forcings. In time-slice simulations water vapour increased by4% per decade from 1960 to 2000, entirelyconsistent with CH4 oxidation. The tropicalupwelling increased at the equivalent rateof 1.7% per decade. The decrease isapproximately balanced by a 2.7% perdecade decrease in the age of air. Modelresults are in reasonable agreement withobservations for tropical upwelling, butunderpredict age of air by over 30%,implying that there is too much mixing. Asummary of GFDL’s strategy for chem-istry-climate evolution was also outlined.

J. de Grandpré presented a 20 year runfrom the CMAM including heteroge-neous chemistry, interactive ozone, andcurrent chlorine loading. The findingsindicate that CMAM has a realistic repre-sentation of the residual circulation, butthat the homogeneity of temperature, andthat of long-lived constituents is toostrong in the NH winter period. Ozoneflux from the stratosphere to the tropo-sphere has an upper limit of 750 Tg/year,and methane and nitrogen loss in thestratosphere is 13.6 Tg/year and 16.1Tg/year respectively.

Discussion of chemistry-climate simula-tions tackled several aspects. There is aneed to balance between studies thataddress the mechanisms of interaction andthe necessity of producing multi-decadal

integrations that span the period of around1950-2100, in order to meet requirementsfor the forthcoming ozone assessment. Keyscientific questions for CCM evaluation arethe detection, attribution and prediction oftrends in atmospheric ozone in the interac-tive chemistry-climate environment, thefuture recovery of stratospheric ozonewithin the context of human-induced andnatural variabilities, the effects of changingchemical composition on the coupled tro-pospheric-stratospheric dynamics andradiation budget and vice-versa, and theeffects of natural perturbations (e.g. vol-canic) on the interactive chemistry-climateenvironment. All of these are priorities inSPARC research.

Cross-evaluation of multiple CCMs hasnot progressed substantially since the pre-vious ozone assessment, and this is a taskthat will be undertaken within CCMVal.Many aspects of the validation may drawfrom GRIPS studies, but chemical evalua-tion will require different diagnostics and adifferent set of expertise than has beenavailable for GRIPS.

In many ways, interpretation of chemistry-climate prediction studies is not a fullymature field. The discussions showed thatthe various research groups have differentlevels of accountability to funding agenciesand, on this basis, have different prioritiesfor model experiments. While the baselinescientific questions are essentially welldefined, neither the strategy for numericalexperimentation nor the analysis methodsare well established. This means that morebasic research is required, which may con-tradict the perceived requirement that allgroups run their simulations under identi-cal, controlled conditions. While there wasconsensus that emission scenarios foranthropogenic “climate” and “chemistry”gases will be adopted from established sce-narios (from 3-D climate models and 2-Dchemistry models), there is less agreementon issues such as: which scenarios would beused; what boundary conditions should beused (e.g., future sea-surface tempera-tures); whether or not to impose solarcycles in incoming irradiance. The treat-ment of aerosols remains quite primitive,with models run with low aerosol loadingat the moment, although this is a veryimportant climate issue. In the predictioncontext, it is necessary to understand howmajor eruptions at different times will

affect climate response and the detection ofozone changes. The role of bromine gaseswas also the focus of interesting discussion,motivated by the cameo appearance of R.Salawich at the meeting. This discussionespecially underscored the importance ofscientific experimentation, in order to bet-ter understand the fundamental issues inchemistry-climate change.

The lack of agreement on these issues issomewhat contradictory to the need forSPARC to contribute to the nextWMO/UNEP ozone assessment. A majorpoint raised was the time needed for groupsto complete useful climate runs, and thefact that IPCC scenarios were not set intime for groups to respond. While the time-line allowed for 2-D runs to be performed,it will not be adequate for most groups tocomplete full 3-D CCM runs. The need tolead the science in defining acceptable sce-narios, rather than lag behind many groupsrunning different scenarios is desirable.This implies the need to quickly agree onscenarios for the upcoming assessements.Even then, some groups are already wellinto their simulations and these groupshave not coordinated their scenarios.

Discussion on GRIPSTasks and Other

IssuesThe GRIPS tasks are in various stages ofcompletion; some tasks can be brought toclosure, and others should carry forwardinto CCMVal. A summary of the tasks andtheir status is given in Table 1.

Plans to update the model evaluation (Tasks1a,b) will essentially be superseded byCCMVal. Studies of troposphere-strato-sphere connections are ongoing and will bepart of the transition from GRIPS toCCMVal, and although task 1d (SuddenWarmings) was completed, it may be revived.Level-2 tasks, aimed at understanding theparameterizations used in models and theirimpacts on the simulated climate, are mostlycomplete. P. Forster has almost completed thecomparison stage of the radiation codes(Task 2a) and the results were presented atthe SPARC General Assembly 2004.

Level-3 tasks are aimed at understandinghow the climate models respond to differ-ent physical forcing mechanisms. The PIN-MIP experiments (Task 3a) will likely

10

SPARC_R10 7/27/05 12:49 PM Page 10

evolve into CCMVal tasks, along with task3d. The solar forcing task (3b) is completeand will be reported on in this meeting.Substantial progress has also been madewith the “1980 – 2000” ozone project, andit is almost complete (U. Langematz couldunfortunately not attend the workshop).GRIPS had one Level-4 task, an examina-tion of the changes in residual circulationin a changing climate, which was discussed.

While GRIPS is now ending, many of theremaining questions in “meteorological”modelling of the middle atmosphere willcarry forward into CCMVal, where they willbe complemented by questions of relevanceto transport, chemistry and radiation.CCMVal should draw from the experiencesof GRIPS, but will face new challenges. Oneof these concerns data management. Becauseof the possibilities for longer model runs,with chemical as well as meteorological out-put, CCMVal will require more resources.The SPARC Data Centre has already madesome plans to cope with the amount of datathat will be necessary to handle for some ofthe planned projects. Another challenge forCCMVal will be community involvement, sothat the tasks do not fall on certain individu-als. CCMVal is taking an “open conference”

approach to many of its activites, but therewill remain a need for focused, workshop-style activities within SPARC. The key will beto effectively combine COPES and CCMValactivities and procure funding.

Acknowledgements

We are grateful to David Sankey (U.Toronto) for organizing the meeting and toNorm McFarlane (SPARC Office) and TedShepherd (U. Toronto) for their invaluablecontributions to its success.

References

Beres, J. H., 2004: Gravity wave generation by a

three-dimensional thermal forcing. J. Atmos.

Sci., 61, pp. 1805–1815.

Butchart, N. and A. A. Scaife, 2001: Removal of

chlorofluorocarbons by increased mass exchange

between the stratosphere and troposphere in a

changing climate, Nature, 410, 799-802.

Charron M. and E. Manzini, 2002: Gravity

waves from fronts: parameterization and mid-

dle atmosphere response in a general circula-

tion model. J. Atmos. Sci., Vol. 59, No. 5, pp.

923–941.

Fioletov V. E. and T. G. Shepherd, (2005),

Summertime total ozone variations over middle

and polar latitudes, Geophys. Res. Lett., 32,

L04807, doi:10.1029/2004GL022080.

Haynes, P. H., Marks, C. J., McIntyre, M. E.,

Shepherd, T. G., and Shine, K. P., 1991: On the

“downward control” of extratropical diabatic

circulations by eddy-induced mean zonal

forces. J. Atmos. Sci., 48, pp. 651–678.

Mukougawa, H. and T. Hirooka, 2004:

Predictability of stratospheric sudden warming:

a case study for 1998/99 winter. Mon. Wea. Rev.,

132, pp. 1764–1776.

Kodera K., and Y. Kuroda, Dynamical response

to the solar cycle, J. Geophys. Res., 107 (D24),

4749, doi:10.1029/2002JD002224, 2002.

Ogi M., Y. Tachibana, and K. Yamazaki, 2003:

Impact of the wintertime North Atlantic

Oscillation (NAO) on the summertime atmo-

spheric circulation, Geophys. Res. Lett., 30 (13),

1704, doi:10.1029/2003GL017280.

Scott, R. K., and Haynes, P. H., 1998: Internal

interannual variability of the extratropical

stratospheric circulation: the low-latitude fly-

wheel. Quart. J. Roy. Met. Soc., 124, 2149-2173. 11

Overview of planned coupled chemistry-climate simulationsto support upcoming ozone and climate assessments

Veronika Eyring, DLR-Institut für Physik der Atmosphäre, Germany (Veronika.Eyring@dlr.de)Douglas E. Kinnison, Atmospheric Chemistry Division, National Center for Atmospheric Research, Boulder,USA (dkin@ucar.edu)Theodore G. Shepherd, University of Toronto, Canada (tgs@atmosp.physics.utoronto.ca)On behalf of the CCM Validation Activity for SPARC (CCMVal)

IntroductionSPARC has established a new validationactivity, CCMVal, for coupled chemistry-climate models (CCMs). The activity isbased on the framework developed at theSPARC workshop on process-orientedCCM validation held in Grainau, Germanyin November 2003 (Eyring et al., 2004,2005) and draws upon the experienceswithin the SPARC GCM-RealityIntercomparison Project (GRIPS) (See theReport on GRIPS in this issue). As moreclimate models include chemical compo-nents, the time has arrived for formal com-parisons of these coupled chemistry-cli-

mate models. Within SPARC, this newactivity will be one of the supporting “pil-lars” of the integrated themes.

The goal of the new activity is to improveunderstanding of CCMs and their underlyingGCMs (General Circulation Models) throughprocess-oriented validation. One outcome ofthis effort is expected to be improvements inhow well CCMs represent physical, chemical,and dynamical processes. In addition, this effortwill focus on understanding the ability ofCCMs to reproduce past trends and variabilityand providing predictions from ensembles oflong model runs. Achieving these goals will

involve comparing CCM constituent distribu-tions with (robust) relationships between con-stituent variables as found in observations.This effort is both a model-model and model-data comparison exercise.At the Grainau work-shop, a set of key diagnostics was definedfor evaluating CCM performance withrespect to radiation, dynamics, transport,and stratospheric chemistry and microphysics(http://www.pa.op.dlr.de/CCMVal/). Thisapproach allows modellers to decide (based ontheir own priorities and resources) whichdiagnostics to examine in any particular area.The CCMVal activity will help coordinate andorganize CCM model efforts around the world.

SPARC_R10 7/27/05 12:49 PM Page 11

12

Table 1: Main features of current coupled chemistry-climate models (CCMs). CCMs are listed alphabetically. The horizontal resolution is given in eitherdegrees latitude x longitude (grid point models), or as T21, T30, etc., which are the resolution in spectral models corresponding to triangular truncation ofthe spectral domain with 21, 30, etc., wavenumbers, respectively. All CCMs have a comprehensive range of chemical reactions except the UMUCAM model,which has parameterized ozone chemistry. The coupling between chemistry and dynamics is represented in all models, but to different degrees. All modelsinclude orographic gravity wave drag schemes (O-GWD); most models additionally include non-orographic gravity wave drag schemes (NonO-GWD).

Model Horizontal No. Vertical Group and location References Contactsresolution Levels/ Upper

Boundary

AMTRAC 2˚x 2.5˚ 48 / 0.0017 hPa GFDL, USA Anderson et al. (2004); J. AustinAustin (2002)

CCSR/ NIES T21 30 / 0.06 hPa NIES, Tokyo, Japan Nagashima et al. (2002); H. Akiyoshi, T.Takigawa et al. (1999) Nagashima, M. Takahashi

CMAM T32 or T47 65 / 0.0006 hPa MSC, University of Beagley et al. (1997); T.G. ShepherdToronto and York de Grandpré et al. (2000)University, Canada

E39/C T30 39 / 10 hPa DLR Oberpfaffenhofen, Dameris et al. (2005) M. Dameris, V. Eyring,Germany V. Grewe, M. Ponater

ECHAM5 T42 39 / 0.01 hPa MPI Mainz, MPI Hamburg, Jöckel et al. (2004); C. Brühl, M. Giorgetta,/ MESSy DLR Oberpfaffenhofen, Roeckner et al. (2003); P. Jöckel, E. Manzini, B. Steil

Germany Sander et al. (2004)

FUBCMAM T21 34 / 0.0068 hPa FU Berlin, MPI Mainz, Langematz, et al. (2005) U. LangematzGermany

GCCM T42 18 / 2.5 hPa Univ. of Oslo, Norway; Wong et al. (2004) M. Gauss, I. Isaksen SUNY Albany, USA

GEOS CCM 2˚ x 2.5˚ 55 / 80km NASA/GSFC, USA In preparation A. Douglass, P.A. Newman,S. Pawson, R. Stolarski

GISS 4˚ x 5˚ 23 / 0.002 hPa NASA GISS, Schmidt et al. (2005a) D. Rind, D. Shindell New York, USA

HAMMONIA T31 67 / 2.10-7 hPa MPI Hamburg, Germany Schmidt et al. (2005b) G. Brasseur, M. Giorgetta,H. Schmidt,

LMDREPRO 2.5˚ x 3.75˚ 50 / 0.07 hPa IPSL, France In preparation S. Bekki, D. Hauglustaine,L. Jourdain

MAECHAM T30 39 / 0.01 hPa MPI Mainz, MPI Hamburg, Manzini et al. (2003); C. Brühl, M. Giorgetta,/CHEM Germany Steil et al. (2003) E. Manzini, B. Steil

MRI T42 68/0.01 hPa MRI, Tsukuba, Japan Shibata and Deushi (2005); K. ShibataShibata et al. (2005)

SOCOL T30 39 / 0.01 hPa PMOD/WRC and ETHZ, Egorova et al. (2004) E. RozanovSwitzerland

ULAQ 10˚ x 20˚ 26 / 0.04 hPa University of L’Aquila, Pitari et al. (2002) E. Manzini, G. Pitari Italy

UMETRAC 2.5˚ x 3.75˚ 64 / 0.01 hPa UK Met Office, UK Austin (2002); Austin G. Bodeker, N. Butchart,NIWA Lauder (NZ) and Butchart (2003) H. Struthers

UM SLIMCAT 2.5˚ x 3.75˚ 64 / 0.01 hPa University of Leeds, UK Tian and Chipperfield (2005) M.P. Chipperfield, W. Tian

UMUCAM 2.5˚ x 3.75˚ 58 / 0.1 hPa University of Cambridge, Braesicke and Pyle P. Braesicke, J.A. Pyle UK (2003 and 2004)

WACCM3 2˚ x 2.5˚ 66 / 140 km NCAR, USA Sassi et al. (2005) B. Boville, R. Garcia,A. Gettelman, D. Kinnison,D. Marsh

SPARC_R10 7/27/05 12:49 PM Page 12

In this way,the CCM community can provide themaximum amount of useful scientific informa-tion for WMO/UNEP and IPCC assessments.

As a first step, the CCM community hasdefined two reference simulations and a setof model forcings to support the upcomingWMO/UNEP Scientific Assessment ofOzone Depletion. The forcings are definedby natural and anthropogenic emissionsbased on existing scenarios, on atmospher-ic observations, and on the Kyoto andMontreal Protocols and Amendments. Inthe following sections, we describe currentmodels and proposed model simulations,and discuss several special issues related tothe use of CCMs.

Participating models

During the last few years, a number of newCCMs have been developed, which signifi-cantly deepens the pool of available mod-els. In comparison with the models used insupport of the last WMO/UNEP ozoneassessment (Austin et al., 2003; WMO,2003), current CCMs generally haveimproved representations of physical pro-cesses, and modelling groups have greatercomputational resources. Table 1 gives anoverview of current coupled-chemistry cli-mate models around the world.

Multi-modelsimulations to

support the upcomingWMO/UNEPassessment

CCMs represent both natural dynamicalvariability and the dynamical response toforcings such as sea surface temperatures(SSTs). As a result, a meaningful comparisonof different CCM results requires a properanalysis of statistical significance and a care-ful representation of natural and anthro-pogenic forcings. To address these issues, aset of questions has been set up for the com-munity to decide on possible model simula-tions and forcings. The draft was opened fordiscussion within the CCM community (seeTable 1) and with several other experts.

The proposed scenarios were developed toaddress the following key questions outlinedby the WMO/UNEP Steering Committee tobe of significance to the upcoming assess-

ment: (1) How well do we understand theobserved changes in stratospheric ozone(polar and extra-polar) over the past fewdecades during which stratospheric climateand constituents (including halogens, nitro-gen oxides, water, and methane) were chang-ing? (2) What does our best understanding ofthe climate and halogens, as well as thechanging stratospheric composition, portendfor the future? (3) Given this understanding,what options do we have for influencing thefuture state of the stratospheric ozone layer?

In order to address questions (1) and (2),two reference simulations (REF) have beenproposed.

Reproduce the past:Reference simulation 1 (REF1),Core time period 1980 to 2004

REF 1 is designed to reproduce the well-observed period of the last 25 years duringwhich ozone depletion is well recorded, andallows a more detailed investigation of therole of natural variability and other atmo-spheric changes important for ozone balanceand trends. This transient simulationincludes all anthropogenic and natural forc-ings based on changes in trace gases, solarvariability, volcanic eruptions, quasi-biennialoscillation (QBO), and sea surface tempera-tures (SSTs). SSTs in this run are based onobservations. Depending on computerresources some model groups might be ableto start earlier. We highly recommend report-ing results for REF1 between 1960 and 2004to examine model variability. Forcings for thesimulation and a detailed description can bedownloaded from the CCMVal website(http://www.pa.op.dlr.de/CCMVal/Forcings/CCMVal_Forcings.html). They are definedfor the time period 1950 to 2004.

SSTs in REF1 are prescribed as monthlymeans following the global sea ice and seasurface temperature (HadISST1) data setprovided by the UK Met Office HadleyCentre (Rayner et al., 2003). This data set isbased on blended satellite and in situobservations.

Both chemical and direct radiative effects ofenhanced stratospheric aerosol abundancefrom large volcanic eruptions are consid-ered in REF1. The three major volcaniceruptions (Agung, 1963; El Chichon, 1982;Pinatubo, 1991) are taken into account, i.e.,additional heating rates and sulfate aerosol

densities are prescribed on the basis ofmodel estimates and measurements,respectively. A climatology of sulfate surfacearea density (SAD) based on monthly zonalmeans derived from various satellite datasets between 1979 and 1999 has been pro-vided by David Considine (NASA LangleyResearch Center, USA). Details on how torepresent the sulfate SAD before 1979 aredescribed on the CCMVal web site.

The QBO is generally described by zonalwind profiles measured at the equator. Whilethe QBO is an internal mode of atmospher-ic variability and not a “forcing” in the usualsense, at the present time most models donot exhibit a QBO. This leads to an under-estimation of ozone variability, and compro-mises the comparison with observations.While some of the models internally gener-ate a QBO, for the others it has been agreedto assimilate observed tropical winds.Assimilation of the zonal wind in the QBOdomain can add the QBO to the system, thusproviding, for example, its effects on trans-port and chemistry. Radiosonde data fromCanton Island (1953-1967), Gan/Maledives(1967-1975) and Singapore (1976-2000)have been used to develop a time series ofmeasured monthly mean winds at the equa-tor (Naujokat, 1986; Labitzke et al., 2002).This data set covers the lower stratosphereup to 10 hPa. Based on rocket wind mea-surements near 8 degree latitude, the QBOdata set has been vertically extended to 3hPa. The software package to assimilate theQBO by a linear relaxation method (alsoknown as “nudging”) as well as the winddata sets have been provided by MarcoGiorgetta (MPI Hamburg, Germany).

The influence of the 11-year solar cycle onphotolysis rates is parameterized accordingto the intensity of the 10.7 cm radiation ofthe sun (which is a proxy to the phase ofthe given solar cycle). The spectral distri-bution of changes in the observed extra-terrestrial flux is based on investigationspresented by Lean et al. (1997) (seehttp://www.drao.nrc.ca/icarus/www/sol_home.shtml for details).

Making predictions:Reference simulation 2 (REF2),Core time period 1980 to 2025

REF 2 is an internally consistent simulationfrom the past into the future. The proposedtransient simulation uses the IPCC SRES

13

SPARC_R10 7/27/05 12:49 PM Page 13

scenario A1B(medium) (IPCC, 2000). REF 2only includes anthropogenic forcings; natu-ral forcings such as solar variability are notconsidered, and the QBO is not externallyforced (neither in the past, nor in the future).Sulfate surface area density is consistent withREF1 through 1999. Sulfate surface areadensities beyond 1999 will be fixed at 1999conditions (volcanically clean conditions).Changes in halogens will be prescribed fol-lowing the Ab scenario (WMO, 2003; Table4B-2). SSTs in this run are based on coupledatmosphere-ocean model-derived SSTs.Depending on computer resources somemodel groups might be able to run longerand/or start earlier. We recommend report-ing results for REF2 until 2050. The forcingson the website are defined through 2100.

Fully coupled atmosphere-ocean CCMs thatextend to the middle atmosphere andinclude coupled chemistry, will use theirinternally calculated SSTs. CCMs driven bySSTs and sea ice distributions from theunderlying IPCC coupled-ocean model simu-lation could use the model consistent SSTs.One constraint is to make the SST datasetconsistent with the SRES greenhouse gas(GHG) scenario A1B (medium). All otherCCM groups will run with the same SSTs,provided by a single IPCC coupled-oceanmodel simulation. These simulations havegood spatial resolution, so the data-setsshould be suitable for all the CCMs partici-pating in the WMO/UNEP assessment.

Sensitivity simulations

Scenarios for sensitivity experiments toaddress question (3) will be defined later.Possible sensitivity experiments could be:

SCN 1 (REF 1 with enhanced Bry): An addi-tional simulation is being developed to rep-resent the known lower stratospheric deficitin modelled inorganic bromine abundance.This simulation will be identical to REF 1,with the exception of including source gasabundances that will increase the strato-spheric burden of Bry. Details of this simula-tion will be made available shortly.

SCN 2 (REF 2 with natural forcings): Asensitivity simulation has been definedsimilar to REF1, with the inclusion of solarvariability, volcanic activity, and the QBOin the past. Future forcings include arepeating solar cycle and QBO under vol-canically clean aerosol conditions. SSTs are

based on REF2. Greenhouse gases andhalogens will be the same as in REF2.

A summary of the proposed CCMVal ref-erence and sensitivity simulations is givenin Table 2.

A web site containing descriptions of themodel simulations, as well as relevant forc-ings (past and present), can be found athttp://www.pa.op.dlr.de/CCMVal/Forcings/CCMVal_Forcings.html. The forcings forthe specified simulations may be down-loaded from this website.

Discussion

In the effort to select the simulations in Table2, several issues arose that required a specialaction or decision. A brief perspective onthese issues is presented in this section.

Greenhouse Gases and Halogens

Historical and Future Trends: It has beenagreed that the simulations to represent thepast (REF1) should not stop in the year2000, but should be extended until 2004.Between the years 2000 and 2004 new mea-surements of trace gases from the SCIA-MACHY, MIPAS, AURA, ACE and ODINsatellite instruments became available. Inaddition, new data from existing satelliteinstruments, such as TOMS, GOME, andHALOE, are also available for CCM inter-comparisons. In response, StephenMontzka (NOAA Climate Monitoring andDiagnostics Laboratory, USA) has offeredto update the datasets of halogen and othergreenhouse gas observations to 2004.Datasets of sea surface temperatures up to2004 are available on the Hadley Centreweb site (see http://www.hadobs.org/).

For the prediction simulation (REF 2) thecommunity agreed to run with the GHGscenario SRES A1B(medium) with thehalogen scenario Ab from WMO (2003).However, Paul Fraser (CSIRO, Australia)mentioned that the SRES reference sce-nario A1B is very unlikely to be realistic forCH4 over the next 20 years. A1B requiresCH4 to increase from 1760 ppb in 2000 to2026 ppb in 2020, i.e., with a growth rate of13-14 ppb per year. This growth rate hasnot been observed since the late 1970s. Incontrast, the growth rate in the Southernand Northern Hemispheres for the pastfive years has been less than 1 ppb per year,

with the current globally averaged concen-tration at 1750 ppb (2004).

In the proposed simulation REF2, the CCMsare driven by SSTs and sea ice distributionsfrom coupled ocean-atmosphere modelsimulations using IPCC SRES GHG scenar-ios. To be consistent with the IPCC simula-tions, the GHG scenarios must be the sameas in the coupled ocean-atmosphere modelsimulations. Therefore, it has been decidedthat CH4 emissions during this phase of theassessment process will not be changed fromthe IPCC SRES GHG scenarios for REF2.

Inorganic Bromine Deficit: Model repre-sentations of inorganic bromine radicals inthe lower stratosphere and comparisonswith observations have recently been docu-mented (see WMO, 2003, Chapters 1 and 2;Salawitch et al., 2005 and references with-in). Results from these comparisons strong-ly suggest that models greatly underesti-mate the total inorganic bromine (Bry) inthis region (up to 6 pptv). Furthermore, it isclear that using time-dependent boundaryconditions as prescribed in the Ab scenario(WMO 2003) will not correct the modelledBry distribution. It is believed that this dis-crepancy occurs because very short-lived(VSL) bromine-containing source gases arenot included in the models. Incorporatingthese species into CCMs will require under-standing of the magnitude and geographicdistribution of the sources of these VSLgases and their loss processes in the atmo-sphere. Based on input from RossSalawitch (Jet Propulsion Laboratory,USA), Martyn Chipperfield (University ofLeeds, UK), and Stephen Montzka, andconsidering the time constraints of includ-ing VSL species and related processes intoCCMs, it was decided that no attemptshould be made to address the Bry deficit inthe reference simulations (REF1 andREF2). This means that the reference CCMexperiments will necessarily underestimatestratospheric Bry by about 25%. This willimpact their ability to reproduce, for exam-ple, polar ozone loss quantitatively, andpredict the future ozone changes; thiscaveat needs to be remembered when ana-lyzing the results. However, a sensitivitysimulation is being developed to examinethis issue (SCN1). Quantification of theeffect of enhanced Bry on ozone trends forthe WMO/UNEP 2006 ozone assessmentwill most likely be done with 2D and 3Dchemical-transport models.

14

SPARC_R10 7/27/05 12:49 PM Page 14

QBO and Solar Variability

Solar activity, as well as the QBO, has astrong influence on ozone variability. SomeCCMs with high horizontal and vertical res-olutions are able to internally generate aQBO. However, the majority of CCMs donot generate a QBO. Consequently thesemodels simulate permanent tropical easter-lies instead of a QBO. As the QBO is impor-tant for wave propagation and interactionwith high latitudes, the latter CCMs there-fore have a known deficit which wouldaffect both the means and variabilities oftrace gas distributions. Therefore, part ofthe community felt that QBO and solarvariability should also be included in futureyears in the reference simulations to the year2025 and, therefore, suggested using SCN2as the reference simulation instead of REF2.

However, others have reservations aboutincluding a QBO and solar cycle in thefuture, since these are not anthropogenicforcings and, hence, cannot be predicted.In the case of the QBO, which is an internalmode of atmospheric variability and not a“forcing” at all, the amplitude and phase ofthe QBO will have no connection to theprognostic model variables or the SSTsand, of course, will not respond to climatechanges.

The obvious way to address the opposingviews is to encourage groups to run bothsimulations, REF2 and SCN2. However, dueto limited time and computer resources it isnot very likely that all, or even most, groupscan afford to run both the REF2 and SCN2

simulations. Therefore, a decision wasmade to make REF1 and REF2 the highestpriority and encourage groups to run SCN2in addition if resources allow.

Sea Surface Temperatures

One of the most critical issues in the dis-cussion was the design of the future simu-lation REF2. The problem is how to extendSST observations into the future withoutintroducing a discontinuity at the present-to-future transition.

One possibility would be to add time-evolving anomalies to the observed SSTsthat are specified for REF1. However, thecommunity sees at least two problems withthis approach. First, the patterns and tem-poral variability are changed, depending onthe shortcomings of the coupled system.Second, the ice distribution in the SSTobservational dataset is not the same as inthe model. This is especially problematic inregions where the ice cover disagrees signif-icantly between model and observations.

We can avoid these problems if the one-simulation-for-all design is abandoned infavour of a design including two separatesimulations. The first would be for theobserved period (REF1), for which we canassess the degree to which observed strato-spheric dynamics and chemistry are repro-duced. The second would be an internallyconsistent simulation from the past to thefuture (REF2). With this approach, fullycoupled atmosphere-ocean CCMs with theatmosphere extending to the middle atmo-

sphere and with coupled chemistry, willuse their internally calculated SSTs inREF2, whereas all other CCMs will usemodelled SSTs from a coupled atmo-sphere-ocean simulation for the full timeperiod (1980 to 2025 or longer). One con-straint is to make the external SST datasetconsistent with the GHG scenario A1B.

There has also been a debate on whether ornot the model simulations should use thesame set of SSTs for future years in REF2.Obviously, if different SSTs are used, theforced low frequency variability could bequite different between the simulations. Oneof the biggest uncertainties is the predictabil-ity of the decadal timescale and the separa-tion of internal from externally-forced vari-ability in the models. However, the focus ofthe future simulation is not a model-modelintercomparison. Rather we would like toprovide the best available prediction of thefuture. REF2 is a simulation that focuses onconsistency and that follows the IPCC simu-lations. Essentially we are asking that mod-elling groups make their best prediction.Therefore, it is not necessary to have consis-tent SSTs. In fact, by applying different SSTs,the change in climate and its variability areeffectively included in the simulation. (Touse a common set of SSTs would certainlyunderestimate the uncertainty in future cli-mate predictions, and any error in those SSTpredictions would lead to a bias in the modelpredictions.)

Finally, an agreement was reached that atleast a subset of groups will run with thesame SST forcings, whereas others will use

15

Table 2: Summary of proposed CCMVal simulations.

Scenario Period Trace Gas Halogens SSTs Background Solar QBO Enhanced & Volcanic Aerosol Variability BrOY

REF1 1980-2004 OBS OBS OBS OBS OBS OBS or If possible GHG used for used for 2002 HadISST1 Surface Area MAVER internally

1960 to WMO/UNEP WMO/UNEP Density data set, generated2004 2002 runs. runs. data (SAD) observed

Extended until provided by flux2004 David Considine

REF2 1980-2025 OBS + OBS + Ab Modelled OBS / SAD - Only -If possible A1B(medium) scenario from SSTs from 1999 internallyuntil 2050. WMO/UNEP generated

2002

SCN1 1980-2004 OBS OBS OBS OBS OBS OBS or Includedused for internally Based on

WMO/UNEP generated Salawitch et al.2002 runs [2005]

SCN2 1980-2025 OBS + A1B OBS + Ab 2002 Modelled OBS / SAD OBS OBS / -(medium) scenario from SSTs from 1999 repeating in in future or

WMO/UNEP future internally2002 generated

SPARC_R10 7/27/05 12:49 PM Page 15

internally calculated SSTs or model-con-sistent SSTs. This will allow us to addressboth views.

Summary and Outlook

CCM Modelling groups are encouraged torun the proposed reference simulations withthe same forcings. In order to facilitate theset-up of the reference simulations, CCMValhas established a website where the forcingsfor the simulations can be downloaded(http://www.pa.op.dlr.de/CCMVal/Forcings/CCMVal_Forcings.html). This web site wasdeveloped to serve the needs of the CCMcommunity, and encourage consistency ofanthropogenic and natural forcings in futuremodel-model and model-observation inter-comparisons. Any updates as well as detailedexplanation and further discussion will beplaced on this website.

We encourage the groups to run both sim-ulations, REF1 and REF2. If a model onlyprovides REF2 it will be more difficult toassess the model’s ability to simulate realis-tic trends and variability. Changes on thedecadal timescale are not necessarily partof the secular trend. It is quite probablethat some of the changes are due to low fre-quency variability that is likely to be unpre-dictable if the source is internal. It is thenpossible that some of the differencesbetween the deterministic model predic-tions will be attributed to unpredictabilityand not to differences in the fundamentalforcings and responses of the models. Forthese reasons, we encourage groups to runensembles. Depending on computerresources, a subset of groups might also beable to carry out sensitivity simulations.Especially if the prediction simulation onlycovers the short-term prediction (e.g. until2025), it would be very useful to see howthe prediction changes if a solar cycle andthe QBO are included. If you are interestedin this topic, please run the sensitivity sim-ulation SCN2.

In agreement with the experts in this field, ithas been decided that the enhanced strato-spheric bromine scenario should not beincluded in the reference simulation (REF1and REF2). Enhancing the inorganicbromine reservoir increases BrO, a reactionpartner for anthropogenically derived ClO,above that found in the standard simulationin the first few kilometres of the strato-sphere. The sensitivity of ozone to enhanced

bromine in the lowermost stratosphere willlikely depend on details of the model simu-lation of ClO just above the tropopause.Due to the inherent three-dimensionalnature of accurately simulating ClO andBrO near the tropopause, it is hoped thatone or more of the CCMs will carry outsimulation SCN1. It is also expected that 2Dand 3D chemical transport model (CTM)simulations will be relied upon to furtherassess the sensitivity of ozone trends tobromine in the lowermost stratosphere forthe next UNEP/WMO ozone assessment.

CCMVal will provide a list of model re-commendations that will be placed on thewebsite. We encourage groups to check theCCMVal forcing website for recommenda-tions concerning the model set-up and thevariables that should be stored in order toallow for sophisticated intercomparisons ofchemistry, transport, dynamics and radia-tion within the CCM.

A detailed intercomparison of CCM resultsand observations has successfully started.Model results from 10 European modelgroups that are participating in theEuropean Integrated Project SCOUT-O3and one model group from outside Europe(CCSR/NIES) have been obtained. Thefirst phase of the intercomparison will bebased on existing runs. With the exceptionof total column ozone, only transientmodel simulations for the time period1980 to 1999 will be compared (no timeslice experiments). We would like toencourage other model groups to join inthe intercomparison and to send data fromexisting runs. As soon as the results of theCCMVal simulations with equal forcingsbecome available, the intercomparisonsand analyses will be repeated. It will beinteresting to see how the results and inter-pretation change when runs with equalforcings are compared. CCMVal is stilllooking for volunteers from around theworld to assist with the intercomparison. Ifyou are interested in a certain diagnostic orscientific topics, please contact us.

A second CCMVal workshop will be heldfrom October 17 to 19, 2005 at the NationalCenter for Atmospheric Research in Boulder,Colorado (http://www.pa.op.dlr.de/work-shops/CCMVal2005/). The 2005 Chemistry-Climate Modelling Workshop will focus onprogress in chemistry-climate modellingand process-oriented validation of CCMs.

The aims of the workshop are to identifynear-term and long-term goals withinthe validation architecture and to coordi-nate activities among the participatingmodelling groups. In addition we willdiscuss how CCM results can support theWMO/UNEP Scientific Assessment ofOzone Depletion 2006 and other upcom-ing assessments. We encourage the par-ticipation of global modellers as well asscientists who make atmospheric obser-vations that are relevant for model evalu-ation.

Acknowledgements

We wish to thank the community for a live-ly and fruitful discussion and for the excel-lent cooperation. Special thanks go toByron Boville, Christoph Brühl, NealButchart, Martyn Chipperfield, DavidConsidine, Martin Dameris, David Fahey,Rolando Garcia, Marco Giorgetta, ElisaManzini, Jerry Meehl, Stephen Montzka, A.Ravishankara and Ross Salawitch whohelped us formulate the reference simula-tions and putting together the CCMValforcing website.

We would also like to thank H. Akiyoshi, J.Austin, S. Bekki, G. Bodeker, P. Braesicke,N. Harris, D. Hauglustaine, A. Gettelman,I. Isaksen, M. Gauss, U. Langematz, E.Manzini, T. Nagashima, P. Newman, S.Pawson, G. Pitari, D. Rind, E. Rozanov, K.Shibata, D. Shindell, R. Stolarski, H.Struthers, M. Takahashi and J. Pyle for gen-eral comments.

References

Austin, J. et al., 2003: Uncertainties and assess-

ments of chemistry-climate models of the

stratosphere, Atmos. Chem. Phys., 3, 1-27.

Anderson, J. L. et al., 2004: The new GFDL glob-

al atmosphere and land model AM2/LM2:

Evaluation with prescribed SST simulations, J.

Climate, in press.

Austin, J., 2002: A three-dimensional coupled

chemistry-climate model simulation of past

stratospheric trends, J. Atmos. Sci., 59, 218-232.

Austin, J. and N. Butchart, 2003: Coupled chem-

istry-climate model simulation for the period

1980 to 2020: ozone depletion and the start of

ozone recovery, Q. J. R. Meteorol. Soc., 129,

3,225-3,249.

16

SPARC_R10 7/27/05 12:49 PM Page 16

Beagley, S.R. et al., 1997: Radiative-dynamical

climatology of the first-generation Canadian

Middle Atmosphere Model, Atmos.-Ocean, 35,

293-331.

Braesicke, P. and J. A. Pyle, 2003: Changing

ozone and changing circulation: Possible feed-

backs?, Geophys. Res. Lett., 30(2), 1059,

doi:10.1029/2002GL015973.

Braesicke, P. and J. A. Pyle, 2004: Sensitivity of

dynamics and ozone to different representa-

tions of SSTs in the Unified Model, Q. J. R.

Meteorol. Soc., 130, 2,033-2,046.

Dameris, M. et al., 2005: Long-term changes and

variability in a transient simulation with a chem-

istry-climate model employing realistic forcings,

Atmos. Chem. Phys. Discuss., 5, 2297-2353.

de Grandpré, J. et al., 2000: Ozone climatology

using interactive chemistry: Results from the

Canadian Middle Atmosphere Model, J.

Geophys. Res., 105, 26,475-26,491.

Egorova, T. et al., 2005: Chemistry-climate

model SOCOL: a validation of the present-day

climatology, Atmos. Chem. Phys. Discuss., 5,

509-555.

Eyring, V. et al., 2004: Comprehensive Summary

on the Workshop on “Process-Oriented

Validation of Coupled Chemistry-Climate

Models”, SPARC Newsletter No. 23, p. 5-11,

http://www.atmosp.physics.utoronto.ca/SPARC

/News23/23_Eyring.html.

Eyring, V. et al., 2005: A strategy for process-ori-

ented validation of coupled chemistry-climate

models, Bull. Am. Meteorol. Soc., in press.

IPCC, 2000: Emission Scenarios. A Special

Report of IPCC Working Group III, Cambridge,

University Press, Cambridge, UK.

Jöckel, P. et al., 2005: Technical Note: The

Modular Earth Submodel System (MESSy) - a

new approach towards Earth System Modeling,

Atmos. Chem. Phys., 5, 433-444.

Labitzke, K. et al. 2002: The Berlin stratospheric

data series. Meteorological Institute, Free University

of Berlin, CD-ROM.

Langematz, U. et al., 2005: Chemical effects in

11-year solar cycle simulations with the Freie

Universitaet Berlin Climate Middle Atmosphere

Model (FUB-CMAM-CHEM), Geophys. Res.

Lett., in press.

Lean, J. et al., 1997: Detection and parameteri-

zation of variations in solar mid and near

ultrviolet radiation (200 to 400 nm). J. Geophys.

Res., 102, 29,939-29,956.

Manzini, E. et al., 2003: A new interactive chem-

istry climate model. 2: Sensitivity of the middle

atmosphere to ozone depletion and increase in

greenhouse gases: implications for recent

stratospheric cooling, J. Geophys. Res.,

108(D14), 4429, doi:10.1029/2002JD002977.

Nagashima, T. et al., 2002: Future development

of the ozone layer calculated by a general circu-

lation model with fully interactive chemistry,

Geophys. Res. Lett., 29 (8), 1162, doi:

10.1029/2001GL014026.

Naujokat, B., 1986: An update of the observed

quasi-biennial oscillation of the stratospheric

winds over the tropics. J. Atmos. Sci., 43, 1873-

1877.

Pawson, S. et al., 2000: The GCM-Reality Inter-

comparison Project for SPARC: Scientific Issues

and Initial Results, Bull. Am. Meteorol. Soc., 81,

781-796.

Pitari, G. et al., 2002: Feedback of future climate

and sulfur emission changes an stratospheric

aerosols and ozone, J. Atmos. Sci., 59(3),

414–440.

Rayner, N. A. et al., 2003: Global analyses of sea

surface temperature, sea ice, and night marine

air temperature since the late nineteenth centu-

ry, J. Geophys. Res., 108, No. D14, 4407

10.1029/2002JD002670.

Roeckner E. et al., 2003: The atmospheric gen-

eral circulation model ECHAM5, Part 1, MPI

Report, No. 349, ISSN 0937-1060.

Salawitch, R. J. et al., 2005: Sensitivity of Ozone

to Bromine in the Lower Stratosphere, Geophys.

Res. Lett., in press.

Sander, R. et al., 2005: Technical Note: The new

comprehensive atmospheric chemistry module

MECCA, Atmos. Chem. Phys., 5, 445-450.

Sassi, F. et al., 2005: The effects of interactive

ozone chemistry on simulations of the middle

atmosphere, Geophys. Res. Lett., submitted.

Schmidt, G. A. et al., 2005a: Present day atmo-

spheric simulations using GISS ModelE:

Comparison to in-situ, satellite and reanalysis

data, J. Climate, in press.

Schmidt, H. et al., 2005b: The HAMMONIA

Chemistry Climate Model: Sensitivity of the

Mesopause Region to the 11-year Solar Cycle

and CO2 Doubling, J. Climate, submitted.

Shibata, K and M. Deushi, 2005: Partitioning

between resolved wave forcing and unresolved

gravity wave forcing to the quasi-biennial oscil-

lation as revealed with a coupled chemistry-cli-

mate model, Geophys. Res. Lett., in press.

Shibata, K. et al., 2005: Development of an MRI

chemical transport model for the study of

stratospheric chemistry, Papers in Geophysics

and Meteorology, 55, 75-118, in press.

Steil, B. et al., 2003: A new interactive chemistry

climate model. 1: Present day climatology and

interannual variability of the middle atmo-

sphere using the model and 9 years of

HALOE/UARS data, J. Geophys. Res., 108(D9),

4290,doi:10.1029/2002JD002971.

Takigawa, M. et al., 1999: Simulation of ozone

and other chemical species using a Center for

Climate Systems Research/National Institute for

Environmental Studies atmospheric GCM with

coupled stratospheric chemistry, J. Geophys.

Res., 104, 14,003-14,018.

Tian, W. and M.P. Chipperfield, 2005: A New

coupled chemistry-climate model for the

stratosphere: The importance of coupling for

future O3-climate predictions, Q. J. Roy. Met.

Soc., 131, 281-303.

WMO, 2003: Scientific Assessment of Ozone

Depletion: 2002, Global Ozone Research and

Monitoring Project - Report No. 47, 498 pp, Geneva.

Wong, S. et al., 2004: A global climate-chemistry

model study of present-day tropospheric chemistry

and radiative forcing from changes in tropospheric

O3 since the preindustrial period, J. Geophys.

Res., 109, No. D11,doi:10.1029/2003JD003998.

17

SPARC_R10 7/27/05 12:49 PM Page 17

The LAUTLOS field campaign washosted by the FMI Arctic ResearchCentre, Sodankylä, and was assisted

by Vaisala. It was successfully conducted inJanuary and February of 2004. The pur-pose of LAUTLOS-WAVVAP (LAPBIATUpper Troposphere Lower StratosphereWater Vapour Validation Project) is thecomparison and validation of the world’sbest hygrometers that are usable asresearch-type radiosondes for precisewater vapour measurements in the tropo-sphere and stratosphere (up to 10 hPa),and to improve and validate thesehygrometers and radiosondes. The instru-ments used in the study included theMeteolabor Snow White hygrometer, theNOAA frostpoint hygrometer, the CAOFlash Lyman-alpha hygrometer, theLindenberg FN sonde, and Vaisala’s latestRS92 GPS-version. The aim is to define anoptimal working range in temperature,water vapour mixing ratio, relative humid-ity, and pressure for each of the participat-ing hygrometers/radiosondes. In additionto the balloon-borne instruments, theUniversity of Bern operated its groundbased 22 GHz microwave instrument,MIAWARA, at Sodankylä to obtain watervapour profiles from approximately 25 to70 km. A further microwave radiometerwas operated from a LearJet of the SwissAir Force to obtain water vapour profilesclose to the balloon locations. The fieldcampaign consisted of 30 balloon flightscarrying integrated payloads.

The WG1 expert meeting was sponsored byCOST 723 (see home page http://www.cost723.org/meetings/wg1_2) and co-sponsored bySPARC. At the meeting, the principal inves-tigators reported on the status of the dataarchiving and the evaluation process. Theparticipants discussed the results obtained

during the campaign, the details of thecomparison procedures, and possiblefuture publications. The LAUTLOS dataarchive consists of vertical profiles from 9different systems.

Figure 1 (see colour insert I) shows a com-parison of relative humidity measured bydifferent humidity sondes during balloonascent. As seen from the plot, the sondes

are generally in good agreement with eachother and do not reveal significant biases.The temperature profile taken by theVaisala RS-80 temperature sensor is alsoshown on the plot in order to demonstratethe sondes relative performance at differ-ent temperatures.

During the LAUTLOS campaign, 11 watervapour profiles were obtained using the

18

Report on the 1st WG 1 Expert Meeting on theLAUTLOS Campaign at Lindenberg

August 24-27, 2004.Ulrich Leiterer, German Weather Service, Meteorological Observatory Lindenberg, Germany (ulrich.leiterer@dwd.de)Vladimir Yuskov, Central Aerological Observatory Moscow, Russia (vladimir@caomsk.mipt.ru)Roland Neuber, Alfred Wegener Institute for Polar and Marine Research Potsdam, Germany (neuber@awi-potsdam.de)Paul Ruppert, Meteolabor AG, Switzerland (paul.ruppert@meteolabor.ch)Ari Paukkunen, Vaisala Oyj, Helsinki, Finland (Ari.Paukkunen@vaisala.com)Esko Kyrö, Arctic Research Centre (FMI/ARC), Sodankylä, Finland (esko.kyro@fmi.fi)Beat Duber, Institute of Applied Physics, University of Bern, Switzerland

2726252423222120191817161514131211109

Hei

ght

(km

)

8.07.57.06.56.05.55.04.54.03.53.0

mean VMR profile OUTSIDE the vortex (11.02, 15.02, 16.02)mean VMR profile INSIDE the vortex (30.02, 6.04, 24.02, 25.02)

Water vapour mixing ratio (ppmv)

FLASH-B Sodankylä (67.4N, 26.6E)

Figure 2: Mean water vapour profiles inside (thick blue) and outside (thick black) the polar vortex cal-

culated using the measurements made by the FLASH-B hygrometer during the LAUTLOS campaign

(January 29 – February 27). Thin blue and black lines on the plot indicate the profiles obtained inside

the vortex (flights on 30.01, 6.02, 24.02, 25.02.2004) and outside the vortex (flights on 11.02, 15.02,

16.02.2004).

SPARC_R10 7/27/05 12:49 PM Page 18

FLASH-B instrument; among them fourinside the polar vortex, four on the edge ofthe vortex, and three outside the vortex.The polar vortex during this time period(February 2004) appeared to be relativelyweak, with generally warm temperatures.Therefore it is assumed that no PSCsformed for the duration of the campaign.

Figure 2 shows the calculated mean watervapour profile both inside and outside thepolar vortex using the FLASH-B measure-ments, which appear to be in a good agree-ment with those obtained by the NOAAfrost point hygrometer (see Figure 3). Theprofiles obtained inside the vortex clearlyshow higher water vapour values than thosetaken outside the vortex. This is caused bythe fact that the water vapour mixing ratioin the stratosphere increases with heightthrough the oxidation of methane. Thus,downwelling air motion in the vortex (dia-batic descent) leads to higher water vapourcontent inside the vortex, compared to thatoutside the vortex at the same altitude. Thedifference reaches 1.4 ppmv at 25 km alti-tude. Since the FLASH-B water vapourmeasurements are in good agreement withthe other independent sensors, the calculat-ed mean water vapour profiles can be con-sidered as a reference for the Arctic strato-sphere over Sodankylä in February with themoderate-strength vortex.

The water vapour profiles obtained by theFLASH-B and NOAA frostpoint hygrome-ters at the edge of polar vortex on 17February are shown in Figure 3. The pro-files indicate some filamentary structure,which can be explained by differentialadvection of air masses originating from

inside and outside the vortex. This is con-firmed by a calculation of potential vortic-ity from a back-trajectory analysis per-formed using ECMWF operational data.The laminae are captured well by bothhygrometers.

The stratospheric water vapour mixingratio inside, outside, and at the edge of thepolar vortex has been accurately measured.The large dry bias in Arctic stratosphericwater vapour typically found in modelsimplies the need for future regular mea-

surements of water vapour in the polarstratosphere to allow for validation andimprovement of climate models. Thenumerous comparisons are now inprogress and results are due by May 2005.All data are available in FMIARC ftp-serv-er for the participants. The LAUTLOSdatabase will be free for all interested scien-tists after May 31, 2005.

It was proposed that the 2nd LAUTLOSdata evaluation meeting be held in Finlandfrom August 29 to September 03, 2005.

19

uu

uu

uu

uu

uuuu

uu

uu

u

uu

uu

uu

uu

u

uuu

uu

uu

u

uu

uu

uu

u

uu

uu

u

uuuuu

uu

u

uu

uuu

u

uu

u

u

u

u

u

u

u

u

Height (km

)Potential vorticity (m²s-¹kg-¹K 10-6)

Potential vorticity

NOAA frostpointFLASH-B

0 20 40 60 80 100 120 140 160 180 200650

600

550

500

450

400

350

3002 3 4 5 6 7 8

25

24

2322

2120191817161514131211109

u

Water vapour mixing ratio (ppmv)

Pote

ntia

l tem

pera

ture

(K)

Figure 3: Vertical water vapour profiles measured by FLASH-B (blue) and NOAA (black) hygrometers

at the edge of the polar vortex on 17 February, and the potential vorticity profile obtained from back-

trajectory analysis. Potential temperature and estimated height are shown on the vertical axis.

Laminae at approximately 22 km and 18 km seen in the potential vorticity profile are also captured by

both the FLASH-B and NOAA frostpoint hygrometers.

SPARC_R10 7/27/05 12:49 PM Page 19

Introduction

A growing body of observations and theo-retical considerations suggests that thetransition from the troposphere to thestratosphere is gradual, rather than a rela-tively sharp discontinuity at thetropopause. In the tropics, it has been sug-gested that a tropical tropopause layer(TTL) spans the transition region from theconvectively dominated overturning circu-lation of the Hadley cell to the region ofslow upwelling (primarily wave-driven) ofthe lower stratospheric Brewer-Dobsoncirculation. Also see the article by IanFolkins in this issue.

The transport of chemical tracers throughthe TTL is an important part of the globalclimate system. Air enters the stratospheric‘overworld’ primarily in the tropics (Holtonet al.,1995), and therefore processes in theTTL play a significant role in determiningtimescales for transport into the strato-sphere of chemical species with tropospher-ic sources. For various reasons, among themthe problem that current global-scale mod-els cannot resolve individual convective cellspenetrating the TTL, the modelling ofchemical tracer transport in this region pro-vides a formidable challenge. We haverecently carried out a number of experi-ments with trajectory calculations based on3-dimensional wind fields from global-scaleatmospheric models and meteorologicalanalysis datasets in order to characterize thecirculation in the TTL as represented inthese models and datasets. In particular, wefocused on identifying typical pathways oftropical troposphere-to-stratosphere trans-port (TST), and their implications forstratospheric water vapour concentrations.

Here, we provide an overview of some of the

results reported by Hatsushika andYamazaki (2003) (henceforth HY03),Bonazzola and Haynes (2004) (henceforthBH04), Fueglistaler et al. (2004, 2005)(henceforth FWP04, FBHP05) andFueglistaler and Haynes (2005) (henceforthFH05). HY03 used an atmospheric generalcirculation model with a resolution ofT42L50, while the other experiments useddata from the European Center forMedium-range Weather Forecasts(ECMWF). BH04 and FWP04 used opera-tional analyses with resolutions of T106/L31and T106/L50, and T511L60, respectively,and FBHP05 and FH05 used the ERA-40reanalysis data with resolution T159/L60.

Circulation in the TTLand tropical TST

An intriguing result found in all experi-ments is that TST-trajectories enter theTTL predominantly over the westernPacific. This reflects the models’enhanced vertical transport due to con-vection over the western Pacific warmpool, and may have important implica-tions, particularly for the transport ofshort-lived species into the stratosphere.We note, however, that this result needscareful interpretation given that none ofthe models explicitly resolves individualconvective cells. An integration of theseresults from a global perspective withresults obtained from mesoscale cloudresolving simulations is therefore animportant and challenging task. Itremains to be seen whether the latter leadto substantial changes of the pictureobtained from the global-scale models.

Within the TTL itself, horizontal advec-tion by the upper-level monsoonal anticy-

clones and the equatorial easterlies playsan important role in determining the pathof TST trajectories. A small fraction ofTST trajectories could be identified astravelling with the northern subtropical jetaround the globe (the southern subtropi-cal jet appears to be more detached fromthe circulation governing TST). Typically,TST trajectories were found to travel sev-eral thousand kilometers horizontallyfrom their point of entry into the TTL tothe point where they encountered theirminimum temperatures. Experiments todetermine average residence times in theTTL show a peak of about 13 days for achange of 10 K in potential temperature at360 K, which is approximately the level ofzero net radiative heating derived fromradiative transfer calculations (Gettelmanet al., 2004).

The results of BH04 and FWP04 indicatethat during boreal summer the distancetravelled by the particles is shorter thanduring boreal winter, but that their ascentrate is faster during boreal winter, asexpected from the annual cycle of thestrength of the stratospheric Brewer-Dobson circulation. These results supportthe notion of Holton and Gettelman (2001)that horizontal advection is a crucial aspectof tropical TST. Our results, however, alsoemphasize the role of zonal inhomo-geneities of vertical transport. Firstly, theconvection over the western Pacific warmpool is the predominant source for air inthe TTL, and eventually the stratosphere.Secondly, within the TTL this region ischaracterized by enhanced upwelling andan associated cold temperature anomaly.Matsuno (1966) and Gill (1980) showedthat these planetary-scale features can beunderstood as a stationary, planetary-scalewave response due to localized heating in

20

Tropical Troposphere-to-Stratosphere Transport:A Lagrangian Perspective

Stefan Fueglistaler, University of Washington, Seattle, USA (stefan@atmos.washington.edu)M. Bonazzola, European Centre for Medium-range Weather Forecasts, Reading, UK (Marine.Bonazzola@ecmwf.int)H. Hatsushika, Environmental Science Research Laboratory, Abiko, Japan (hatusika@criepi.denken.or.jp)Peter H. Haynes, Dept. of Applied Mathematics and Theoretical Physics, Cambridge, UK (phh@damtp.cam.ac.uk)Thomas Peter, Institute for Atmosphere and Climate, ETH Zürich, Switzerland (Thomas.Peter@ethz.ch)Heini Wernli, Institut für Physik der Atmosphäre, Mainz, Germany (wernli@uni-mainz.de)K. Yamazaki, Hokkaido University, Sapporo, Japan (yamazaki@ees.hokudai.ac.jp)

SPARC_R10 7/27/05 12:49 PM Page 20

the troposphere as a result of convectionover the western Pacific warm pool.

Figure 1 (see colour insert I) provides aschematic of tropical TST, emphasizing therole of the Western Pacific warm pool regionand the upper-level anticyclones. Duringboreal summer, the Indian/Southeast-Asianmonsoon is found to play a similar role,although it is much less symmetric aroundthe equator as the convection, which plays animportant role in determining the flux intothe TTL and the pattern of horizontal circu-lation within the TTL itself, is shifted signif-icantly to the north of the equator.

Experiments addressing interannual vari-ability show that the largest modificationsof the circulation patterns of tropical TSToccur due to El-Nino/Southern Oscillation(ENSO) events. In particular, during El-Nino, convection is more homogeneousover the entire tropical Pacific, which inturn induces a homogenization of temper-atures and circulation within the TTL.

Figure 2 (see colour insert II) shows the

horizontal wind and temperature fields at90 hPa from ERA-40 data during borealwinter for a strong La-Nina, a strong El-Nino situation, and the climatologicalmean state for the period 1979-2001. Inaddition, the figure shows the spatial den-sity distribution where trajectories risingfrom the troposphere to the stratosphereare found to enter the TTL (red contourlines).

Stratospheric watervapour

Although it has been widely accepted sincethe seminal work of Brewer (1949) that theextremely low temperatures at the tropicaltropopause constrain the water vapour fluxinto the stratosphere to a few parts per mil-lion by volume (ppmv), the details of thedehydration processes of tropical TSTremain controversial. We therefore pursuetwo goals with the trajectory calculations.Firstly, we want to quantify the relevance ofthe previously discussed large-scale spatialstructure of temperature and circulation in

the TTL for water vapour mixing ratio ofair entering the stratosphere (H2O).Secondly, we want to quantify the degree towhich observations of stratospheric watervapour concentrations can be explained bymodel calculations that greatly simplifycloud microphysics and mesoscale process-es, but take into account the 3-dimensionaltemperature history of tropical TST asresolved by global scale models. Of particu-lar interest in this context is the minimumtemperature experienced by an air parcel asit ascends from the troposphere to thestratosphere. FBHP05 termed this point the‘Lagrangian cold point’, to be contrastedwith the traditional cold point tropopause.

In all experiments we found that the prev-iously discussed circulation in the TTL issuch that tropical TST trajectories effi-ciently sample the regions of lowest tem-peratures. Correspondingly, the spatialdensity distribution of the Lagrangian coldpoints as shown in Figure 2 (black contourlines) is highest in regions of lowest tem-peratures. These zonal temperatureanomalies substantially lower H2O com-pared to what might be expected fromzonal mean tropical tropopause tempera-tures. Therefore we conclude that the large-scale 4-dimensional structure of tempera-tures and circulation in the TTL is crucialfor understanding H2O. FBHP05 showedthat predictions of H2O based on the satu-ration mixing ratio at the Lagrangian coldpoint of TST-trajectories agree very wellwith a broad range of observations, with aremaining moist bias of about 0.2 ppmv, orabout 5% of H2O.

Figure 3 reproduces some results shown byFBHP05 for mean entry mixing ratios ofH2O over selected periods, and the climato-logical mean annual cycle for the period1992-2002. Note, for example, that theannual cycle of the model calculationsbased on the 3-dimensional temperaturehistory of TST (Figure 3, cyan line) signifi-cantly improves the agreement with obser-vations (Figure 3, black line) comparedwith a prediction based on tropical zonalmean cold point temperatures (Figure 3b,grey line). Remaining differences are dis-cussed in detail by FBHP05, note for exam-ple that the obvious phase shift of approx-imately 1 month between the model pre-diction (cyan) and observation (black) islikely due to known problems of the strato-spheric circulation in ERA-40 data.

211992-2001, 400K

[H2O

] (p

pmv)

[H2O

] (p

pmv)

2

3

4

5(a) (b)

2

3

4

5

6

7

8

9

J F M A M J J A S O N D

Figure 3: Comparison of model predictions and observations of stratospheric water vapour, taken fromFBHP05. (a): Model predictions for concentration of water vapour [H2O]e (blue, with (left) and with-out (right) taking into account the seasonal cycle of the strength of the stratospheric Brewer-Dobson cir-culation) and tropical (30°S-30°N) mean water vapour mixing ratio at 400 K potential temperature(cyan) for the periods 1979-2001, 1989-1992 and 1992-2001. Observations (black) of [H2O]e are fromMichelsen et al. (2000) (1) and Engel et al. (1996) (2), and (3) is the tropical mean water vapour mix-ing ratio at 400 K from HALOE. (b): Climatological mean annual cycle of model prediction for [H2O]e(blue) and tropical mean at 400 K (cyan), HALOE at 380 K (black, dash-dot) and at 400 K (black,solid), and, for reference, the saturation mixing ratio of zonal mean (10°S-10°N) cold point tropopausetemperatures based on ERA-40 data.

SPARC_R10 7/27/05 12:49 PM Page 21

FH05 further showed that for the period1995-2002 their model predictions canexplain observed interannual variability ofwater vapour concentrations in the tropicallower stratosphere to within measurementuncertainties. Largest modulations of H2Oare found to result from the temperatureperturbation around the tropicaltropopause due to the Quasi-BiennialOscillation (QBO) and due to strong El-Nino situations. Based on simulations ofidealized ENSO situations, HY03 predictedthat La-Nina situations should be drierthan El-Nino. This was partly confirmed byFH05 in their reconstruction of H2O forthe period 1979-2001. However, the inter-ference of ENSO with other processes thataffect tropical tropopause temperatures,most notably the QBO, makes it difficult toclassify unambiguously the effect of ENSO,particularly La-Nina situations, on H2O.

Summary

The good agreement of model predictionswith observations of stratospheric watervapour suggests that the trajectory calcula-tions, and the assumption of dehydrationto the saturation mixing ratio of theLagrangian cold point, capture the mainprocesses controlling H2O. Thus, the syn-optic-scale temperature history of tropicalTST apparently can explain H2O very well,and smaller scale effects such as over-shooting convection apparently need notbe invoked at first order. Consequently, itappears that current global-scale modelsmay do a better job in representing tropi-cal TST than might have been expected.That notwithstanding, a comprehensivetheory explaining the spatio-temporal cir-

culation and temperature structure of theTTL that encompasses all scales, fromindividual convective cells to the plane-tary-scale stationary wave pattern, and therole of the stratospheric Brewer-Dobsoncirculation, remains an important andambitious task.

References

Brewer, A. W., Evidence for a world circula-tion provided by the measurements of heli-um and water vapour distribution in thestratosphere, Q. J. R. Meteorol. Soc., 75, 351- 363, 1949.

Bonazzola, M., and P. H. Haynes, A trajec-tory-based study of the tropical tropopauseregion, J. Geophys. Res., 109,doi:10.1029/2003JD004356, 2004.

Engel, A., C. Schiller, U. Schmidt, R.Borchers, H. Ovarlez, J. Ovarlez, The totalhydrogen budget in the Arctic winterstratosphere during the European ArcticStratospheric Ozone Experiment, J.Geophys. Res., 101, 14495—14504, 1996.

Fueglistaler, S., Wernli, H., Peter, T., Tropicaltroposphere-to-stratosphere transport inferredfrom trajectory calculations, J. Geophys.Res., 109, doi:10.1029/2003JD004069, 2004.

Fueglistaler, S., M. Bonazzola, P. H. Haynes, T.Peter, Stratospheric water vapour predictedfrom the Lagrangian temperature history ofair entering the stratosphere in the tropics, J.Geophys. Res., 110, doi:10.1029/2004JD005516(in press), 2005.

Fueglistaler, S., and P. H. Haynes, Control

of interannual and longer-term variabilityof stratospheric water vapour, manuscriptsubmitted to J. Geophys. Res., 2005.

Gettelman, A., P. M. de F. Forster, M.Fujiwara, Q. Fu, H. Vömel, L. K. Gohar, C.Johanson, M.Ammerman, Radiation balanceof the tropical tropopause layer, J. Geophys.Res., 109, doi:10.1029/2003JD004190, 2004.

Gill, A. E., Some simple solutions for heatinduced tropical circulation, Q. J. R.Meteorol. Soc., 106, 447 - 462, 1980.

Hatsushika, H., Yamazaki, K., Stratosphericdrain over Indonesia and dehydrationwithin the tropical tropopause layer diag-nosed by air parcel trajectories, J. Geophys.Res., doi:10.1029/2002JD002986, 2003.

Holton, J. R., P. H. Haynes, M. E. McIntyre,A. R. Douglass, R. B. Rood, and L. Pfister,Stratosphere-troposphere exchange, Rev.Geophys., 33, 403 - 440, 1995.

Holton, J. R., A. Gettelman, Horizontaltransport and the dehydration of thestratosphere, Geophys. Res. Lett., 28, 2799 -2802, 2001.

Matsuno, T., Quasi-geostrophic motions inthe equatorial area, J. Meteorol. Soc. Jpn, 44,25 - 43, 1966.

Michelsen, H. A., F. W. Irion, G. L.Manney, G. C. Toon, M. R. Gunson,Features and trends in Atmospheric TraceMolecule Spectroscopy (ATMOS) version3 stratospheric water vapor and methanemeasurements, J. Geophys. Res., 105, 22713- 22724, 2000.

22

SPARC_R10 7/27/05 12:49 PM Page 22

The Tropical Tropopause Layer (TTL)has been recently introduced(Highwood and Hoskins, 1998,

Folkins et al., 1999) as a layer whose prop-erties are intermediate between the tropo-sphere and stratosphere. If examined with afine enough spatial scale, any atmosphericproperty will vary continuously in goingfrom one atmospheric layer to the next.One could therefore invoke the existenceof a transition layer between any two adja-cent atmospheric layers. This article willattempt to explain what is unique about theTTL, and why it provides conceptual advan-tages in thinking about some atmosphericscience problems.

The issue of how to define the TTL is notsettled. Here, the base of the TTL will bedefined as the Level of Zero clear sky radia-tive Heating (LZH), which occurs near 15 km.The transition from radiative cooling(below 15 km) to radiative heating (above15 km) is driven by the combination of arapid decrease in water vapour mixingrations (such that longwave cooling fromwater vapour is negligible above 15 km),and also by the onset of extremely coldtemperatures, which suppress longwaveemission from CO2 and O3. From radiativeconsiderations, an air parcel which detrainsfrom a deep convective clouds above theLZH will rise upward across isentropic sur-faces into the stratosphere. This definitionof the base of the TTL therefore seems to bethe definition most relevant to strato-sphere-troposphere exchange (STE). Airparcels that detrain into the TTL have someprobability of influencing the chemicalcomposition of the stratosphere, whilethose which detrain below the TTL havevery little. It should be kept in mind, how-ever, that air parcels moving horizontallynear 15 km will roughly follow an isentrop-ic surface that can undulate above andbelow 15 km, so that the base of the TTL isnot a material surface. Perhaps the mostreasonable definition of the base of the TTLis the height at which an air parcel detrain-ing from a deep convective cloud firstattains a 50% likelihood of ascending intothe stratosphere. This height probablyoccurs near the LZH because radiativeheating is the most dominant diabatic pro-cess in the clear sky atmosphere.

It is more difficult to define the top of theTTL. A useful conceptual definition is that itis the height at which the upward convectivemass flux becomes small in comparison tothe Brewer Dobson mass flux. That is, thatthe convective outflow that feeds the base ofthe Brewer Dobson circulation has essen-tially become exhausted. Unfortunately, it isintrinsically difficult to diagnose the highaltitude tail of the convective detrainmentprofile from observations. Measurements ofozone and other chemical species suggestthat undiluted convective outflow can occuras high as 17 km (Folkins et al., 2002a, Tucket al., 2004), which would place the top ofthe TTL near the climatological height ofthe cold point tropopause. This is also aconvenient definition for TTL dehydration.However, deep convective overshootingthrough the tropical tropopause, followedby mixing with ambient stratospheric air,provides a mechanism for convectivedetrainment above 17 km whose existence isdifficult to detect from chemical tracers.

It has been argued that the input of signif-icant amounts of overshooting tropo-spheric air above the tropical tropopause isunlikely, since it would undermine the sea-sonal variation in CO2 propagating upwardfrom the tropopause (Boering et al., 1995).Model simulations suggest, however, thatthis is not true in all circumstances(Sherwood and Dessler, 2003).

As defined here, the TTL is a consequenceof the geometry of the flow in the uppertropical troposphere. In order to feed theBrewer-Dobson circulation, there must besome convective detrainment above thealtitude at which the background meanflow changes direction. Since the Brewer-Dobson circulation is about 100 timessmaller than the Hadley circulation, onewould anticipate that roughly 1% of themass flux from tropical deep convectiondetrains into the TTL.

Figure 1 gives an overview of the mass fluxdivergences in the upper tropical tropo-sphere associated with convective outflow(�c), radiative cooling (�r), and evaporativecooling (�e). These were obtained using asimple one-dimensional model of the trop-ical atmosphere constrained by observed

temperature and moisture profiles (Folkinsand Martin, 2005). To first order, radiativeconvergence balances convective diver-gence. In other words, convective outflowis the mass source required to supply adownwardly increasing subsiding radiativemass flux (or to supply an upwardlyincreasing ascending radiative mass flux inthe TTL).

The magnitude of the evaporative drivendownward mass flux increases below 13 km,giving rise to a convergence, which partiallybalances the divergence from convection.Due to the extremely cold temperatures,saturated water vapour mixing ratios in theTTL are very low, typically less that 10 ppmv.At such low mixing ratios, heat releasesassociated with changes in phase of waterare typically smaller than those due toradiative heating or cooling, so that from athermodynamic perspective, air parcels inthe TTL can be effectively treated as dry. Inthe TTL, one should not ordinarily have toworry about the role of evaporating icecrystal in driving vertical motions.

Figure 1 shows that the rate at which deepconvection injects air into the upper tropi-cal troposphere reaches a maximum of0.4/day near 12.5 km. The convectivereplacement time �r can be defined as�r=1/�c. This replacement time varies from2.5 days at the peak of deep outflow mode,to about two weeks at the base of the TTL.Within the TTL, estimations of �c and �r

using simple models become highly uncer-tain. This is due to a variety of reasons. Thedivergences shown in Figure 1 were calcu-lated by assuming that the mass budget ofthe tropics (here defined as 20°S-20°N)could be treated in isolation from theextratropics. This assumption is probablyvalid at most heights. In the TTL, however,it is likely that the rate of mass transferbetween the tropics and extratropics iscomparable with the convective diver-gence. The curve labelled �ex in Figure 1 isan estimate of the rate at which tropical airis exported from the tropics to the extra-tropics, using assimilated meteorologicaldata from the Goddard Earth ObservingSystem (GEOS) of the NASA DataAssimilation Office. In principle, diagnos-tic approaches to estimating the strength of

23

Temperatures, Transport, and Chemistry in the TTLIan Folkins, Dalhousie University, Halifax, Canada (Ian.Folkins@dal.ca)

SPARC_R10 7/27/05 12:49 PM Page 23

the convective divergence in the TTLshould take this mass transport pathwayinto account. In addition, the radiativedivergence shown in Figure 1 was calculat-ed by assuming clear sky conditions. Clearsky heating rates in the TTL are very small.The relative importance of cloud radiativeeffects can therefore be expected to bemuch higher in the TTL than in the rest ofthe tropical troposphere.

The ideas behind the TTL are not entirelynew. Until recently, however, there had beena widely held implicit assumption thatmuch of the air rising in deep convectiveclouds detrained near, or just below, thetropical tropopause. This view partly arosefrom the fact that stratiform anvil cloudsfrom deep convection frequently extend tothe cold point tropopause (~17 km).Inferring mass outflow from the presenceof ice crystals is, however, problematic. Icecrystals may arise from vertical motionsassociated with gravity waves generated byconvection, rather than by the updraftsthemselves. In addition, stratiform anvilscan extend from 10 km to 17 km. There isno reason to assume a priori that outflow ispreferentially distributed near cloud top.

Temperature profiles from radiosondesalso frequently give the impression of astrong convective influence extending upto the cold point tropopause. Deep convec-tion is often associated with a cold andwell-defined cold point tropopause, withlayers near the tropopause where the lapserate approaches the dry adiabatic value.The strong influence of deep convection ofTTL temperatures does not necessarilyimply rapid mass outflow rates in the TTL,however, since the weakness of radiativeheating rates in the TTL implies longradiative damping timescales (Hartmannet al., 2001). Temperature anomalies in theTTL arising from deep convection can betherefore expected to be very persistent.

The notion that most of the air transport-ed upward in deep convective cloudsreaches the tropical tropopause also givesrise to an apparent thermodynamic para-dox. In the absence of mixing, the level ofneutral buoyancy of an air parcel risingupward inside a convective updraft occursat the height at which the equivalentpotential temperature of the air parcel isequal to the potential temperature of thebackground atmosphere. (This statement

makes a number of assumptions, amongthem that water vapour concentrations aresufficiently low that their effect on densitycan be ignored, and that the presence ofcondensate does not have an appreciableeffect on the density of an air parcel.)While the potential temperature at the coldpoint tropopause is rarely lower than 365 K,near surface air parcels with �e > 365 K areextremely rare (Folkins and Martin, 2005).On thermodynamic grounds, one wouldtherefore also expect convective detrain-ment near the cold point tropopause to beextremely rare.

There have been a number of attempts toreconcile the idea of significant convectivedetrainment near the tropical tropopausewith its apparent thermodynamic implau-sibility. One mechanism that would allowair parcels to detrain at a higher potentialtemperature surface is convective over-shooting and irreversible mixing. It hasalso been pointed out that the potentialtemperature of the cold point tropopauseroughly corresponds to the saturatedequivalent potential temperature of thewarmest tropical sea surface temperatures,so that air parcels with �e ~ 365 K can in

fact be generated near the surface(Chimonas and Rossi, 1987). Another sug-gestion is that the freezing of lofted icecould act as a source of heating that wouldallow air parcels to detrain near the tropi-cal tropopause (Zipser, 2003). Such esti-mates are, however, quite sensitive toassumptions on the amount of lofted ice,and the efficiency with which rising airparcels retain this additional heat.

Both simple diagnostic models, as well asin situ chemical evidence, indicate that out-flow from deep convective clouds rangesfrom 10 km to 17 km. This suggests thatapproaches that attempt to relate a meandetrainment potential temperature to amean boundary layer �e may be misguided.Instead, it may be more appropriate toattempt to find some relationship betweenthe shape of the deep convective detrain-ment profile in the upper troposphere withthe probability distribution of boundarylayer �e in actively convecting regions(Folkins and Martin, 2005).

The mean profile of any chemical tracer inthe tropics should reflect an approximatebalance between the competing tendencies

24

9

10

11

12

13

14

15

16

17

-0.4 -0.3 -0.2 -0.1 0 0.1 0.2 0.3 0.4Divergence (1/day)

-0.4 -0.3 -0.2 -0.1 0 0.1 0.2 0.3 0.4Divergence (1/day)

9

10

11

12

13

14

15

16

17

Heig

ht (k

m)

9

10

11

12

13

14

15

16

17

Height (km)

δe

δex

TTL

δcδr

Figure 1: Tropical mean (20°S – 20°N) mass flux divergences calculated from a one-dimensional modelof the tropical atmosphere (Folkins and Martin, 2005). δc refers to the convective divergence, δr to theradiative divergence, δe to the evaporative divergence, and δex to the divergence associated with exportof tropical air into the extratropics. The vertical arrows indicate that the clear sky mass flux is upwardabove 15 km. The top of the TTL, near 17 km, has been drawn with a dashed line to indicate that itsheight is uncertain.

SPARC_R10 7/27/05 12:49 PM Page 24

of in situ chemistry, vertical advection,convective detrainment, as well as otherpossible influences such as STE. In thevicinity of the 12.5 km �c maximum, chem-ical tracers with upper tropospheric life-time longer than a week should be main-tained near their boundary layer values bythe strength of deep convective detrain-ment. However, as the timescale for con-vective replacement increases toward thecold-point tropopause, the mean concen-tration of a chemical tracer can be expect-ed to become increasingly determined byin situ chemistry.

Figure 2 (see colour insert II) shows ozoneclimatologies at ten tropical locations fromthe SHADOZ campaign (Thompson et al.,2003). Most of these profiles exhibit theclassic “S-shape”. The minimum near 12.5 kmis probably due to the strong convective out-flow at this altitude. Ozone mixing ratios atlocations characterized by maritime deepconvection are smaller than those with con-tinental deep convection. The two dashedlines are ozone profiles calculated from amodel forced by the convective divergenceprofile shown in Figure 1 (Folkins andMartin, 2005). In one of the profiles, themean ozone mixing ratio in air detrainingfrom convective clouds is assumed to be20 ppbv, while in the other, it is assumed tobe 30 ppbv. In the model, the increase inozone mixing ratios above 13 km is due tothe increase in the convective replacementtimescale, which allows ozone to approachits steady state mixing ratio. The increase inozone above 13 km is not driven by anincrease in the rate of in situ chemical pro-duction from O2 photolysis. This source ofozone is insignificant below 16.5 km.

Figure 3 show profiles of CO in the tropicsobtained from various field campaigns.Unfortunately, there are far fewer measure-ments of CO than O3 in the tropics, especial-ly between DC8 (~11 km) and ER-2 flightaltitudes (15 km to 20 km). None of the COprofiles shown in Figure 3 contain enoughdata to constitute a climatology. CO is pri-marily destroyed in the upper troposphere byOH attack. Since the sources of CO are main-ly at the surface, one would expect deep con-vection to maintain high concentrations ofCO near the 12.5 km convective outflowmaximum, and decrease toward thetropopause as the convective source weak-ened. This expectation is broadly consistentwith the CO profiles shown in Figure 3.

One would also expect tracers, which areinfluenced by STE to exhibit a vertical gra-dient in the upper tropical troposphere. Ithas been noted that the onset of a decreasein CFC-11 sometimes occurs below thetropical tropopause (Tuck, 1997). This wasattributed to the existence of a “standingreserve” of stratospheric air in the uppertropical troposphere. The existence of sucha reservoir of stratospheric air in the TTL isconsistent with Figure 1, which suggeststhat the timescales for sideways tropical-extratropical exchange and deep convec-tion in the TTL may be comparable. Inthese cases, the vertical gradient in the trac-er concentration is not necessarily causedby a change in the magnitude of STE withheight, but by a decrease in convective out-flow, which gives any stratospheric input alonger chemical signature.

The base of the TTL denotes a change in thenature of the clear sky water vapour budget.In most of the clear sky tropical troposphere,the occurrence of supersaturation withrespect to ice is inhibited by large-scaledescent. In the TTL, however, air parcelsexperience colder temperatures as theyascend toward the cold point tropopause.Supersaturation with respect to ice should bequite frequent, provided ice depositionnuclei are sufficiently rare. Measurementshave, in fact, sug-gested an increasein the frequency ofclear sky supersat-uration at the baseof the TTL(Folkins et al.,2002b). Thesem e a s u r e m e n t swould tend to sup-port the view thatd e h y d r a t i o noccurs throughoutthe TTL ratherthan simply in thevicinity of the coldpoint tropopause.U n a m b i g u o u sdetection of clearsky supersatura-tion with respectto ice remains asignificant experi-mental challenge,however, becausein regions wherethe relative humid-

ity approaches 100%, it is possible to observesupersaturation with respect to ice simplydue to the existence of random (and possiblybias) errors in measurements of watervapour mixing ratio and temperature.

The concept of the TTL was partly moti-vated by modelling studies indicating thatconvective control of tropical temperaturesdid not extend as high as the cold pointtropopause (Thuburn and Craig, 2002). Itis clear that the base of the TTL is associat-ed with a change in the nature of radiative-convective equilibrium. Figure 4 (seecolour insert III) shows the difference inDecember to February (DJF) temperaturesat 14 tropical radiosondes locations from aDJF climatology obtained by averagingover all 14 locations. Below 14 -15 km,locations characterized by persistent deepconvection tend to be anomalously warm,while those with less convection tend to beanomalously cold.

One of the ways in which convectioninfluences the large scale temperatureprofile is by the emission of gravity waves.These gravity waves give rise to irre-versible vertical motions in the back-ground atmosphere, which diminish thecontrast in density between the atmo-sphere and the convective plumes

25

0 25 50 75 100 125 150 175 200

CO (ppbv)

0

2

4

6

8

10

12

14

16

18

Height (km

)

0

2

4

6

8

10

12

14

16

18

Hei

ght (

km)

PEM West APEM West BPEM TropicsTrace ASTRATASHOE

Figure 3: CO climatologies obtained by averaging over CO measurements

(15°S – 15°N) from a variety of aircraft field campaigns.

SPARC_R10 7/27/05 12:49 PM Page 25

(Bretherton and Smolarkiewicz, 1989). Awarm positively buoyant plume would beexpected to give rise to downwardmotions in the background atmosphere,while a negatively buoyant plume (over-shooting updraft or downdraft) wouldgive rise to upward motions. The associa-tion between convection and warmanomalies below the TTL would suggestthat convective plumes are on averagepositively buoyant below 15 km, while theassociation between convection and coldanomalies in the TTL would suggest thatconvective plumes in this region are nega-tively buoyant.

The coincidence of the LZH with theheight at which the relationship betweenconvective frequency and temperaturechanges sign is coincident with the viewthat deep convection exerts a dominantcontrol on temperature at these altitudes.Below the TTL, the induced downwardmotions from deep convection increasetropical temperatures away from radiativeequilibrium, and give rise to radiative cool-ing, which within the TTL, the inducedupward motions from deep convectiondecrease tropical temperatures away fromradiative equilibrium, and give rise toradiative heating.

There have been other explanations putforward for the association between coldtemperatures and deep convection in theTTL. One possibility is that it is due to thedeep convection injection of cold, nega-tively buoyant air which irreversibly mixeswith the ambient warmer air (Sherwood etal., 2003, Kuang and Bretherton, 2004).

In the lower tropical stratosphere,upwelling is strongest during the Decemberto Febraury (DJF) season and weakest dur-ing the June to August (JJA) season. This isassociated with a seasonal temperaturecycle, in which temperatures are coldestduring the DJF season and warmest duringthe JJA season. Figure 5 (see colour insertIII) shows the difference between the JJAand DJF seasonal temperature climatolo-gies at 14 radiosonde locations. The ampli-tude of the lower stratospheric seasonalcycle peaks near 18 km. At most locations,the onset of this seasonal cycle occurs near15 km, so that the stratospheric seasonalcycle extends to the base of the TTL. Below15 km, most tropical locations are warmerduring the DJF season.

Predictions of the future evolution of trop-ical cold point temperatures are made dif-ficult both by uncertainty in the mecha-nism by which tropical deep convectioninfluences TTL temperatures, and alsobecause the relative roles played by theHadley and Brewer-Dobson circulations incontrolling TTL temperatures have not yetbeen established. It is important to deter-mine the nature of this control, however,because the Hadley and Brewer Dobsoncirculations can be expected to change inresponse to future changes in CO2. Byinfluencing tropical cold point tempera-tures, these changes can be expected toinfluence stratospheric water vapour.

Ozone plays an important role in the radia-tive budget of the TTL. Its climatologicalprofile will be influenced by changes inboth the shape of the convective detrain-ment profile, and by changes in strato-spheric upwelling. It would therefore bedesirable to investigate the future evolutionof TTL temperatures using models inwhich the dynamics is self-consistentlycoupled with radiation and chemistry.

References

Boering, K. A., et al., Measurements of strato-

spheric carbon dioxide and water vapor at

northern midlatitudes: Implications for tropo-

sphere-to-stratosphere transport, Geophys. Res.

Lett., 22, 2737-2740, 1995.

Bretherton, C. S., and P. K. Smolarkiewicz,

Gravity waves, compensating subsidence and

detrainment around culmuls clouds, J. Atmos.

Sci., 46, 740-759, 1989.

Chimonas, G., and R. Rossi, The relationship

between tropopause potential temperature and

the buoyant energy of storm air, J. Atmos.

Sci.,44, 2902-2911, 1987.

Folkins, I., M. Lowewenstein, J. Podolske, S.

Oltmans, and M. Proffitt, A barrier to vertical

mixing at 14 km in the tropics: Evidence from

ozonesondes and aircraft measurements, J.

Geophys. Res.,104, 22095-22102, 1999.

Folkins, I., C. Braun, A. M. Thompson, and J. C.

Witte, Tropical ozone as an indicator of deep

convection, J. Geophys. Res., 107(D13),

10.1029/2001JD001178, 2002a.

Folkins, I., E. J. Hintsa, K. K. Kelly, and E. M.

Weinstock, A simple explanation for the rela-

tive humidity increase between 11 and 15 km in

the tropics, J. Geophys. Res., 107,

10.1029/2002JD002185, 2002b.

Folkins, I., and R. Martin, The vertical structure

of tropical convection, and its impact on the

budgets of water vapor and ozone, to be pub-

lished in J. Atmos. Sci., 2005.

Hartmann, D. L., J. R. Holton, and Q. Fu, The

heat balance of the tropical tropopause, cirrus,

and stratospheric dehydration, Geophys. Res.

Lett., 28, 1969-1972, 2001.

Highwood, E. J., and B. J. Hoskins, The tropical

tropopause, Q. J. R. Meteorol. Soc., 124, 1579-

1604, 1998.

Kuang, Z., and C. S. Bretherton, Convective

influence of the heat balance of the tropical

tropopause layer: A cloud-resolving model

study. J. Atmos. Sci., 61, 2919-2927, 2004.

Sherwood, S. C., and A. E. Dessler, Convective

mixing near the tropical tropopause: Insights

from seasonal variations, J. Atmos. Sci., 60,

1847-1856, 2003.

Sherwood, S. C., T. Horinouchi, and H. A.

Zeleznik, Convective impact on temperatures

observed near the tropical tropopause, J. Atmos.

Sci., 60, 1847-1856, 2003.

Thompson, A. M., et al., Southern Hemisphere

Additional Ozonesondes (SHADOZ) 1998-

2000 tropical ozone climatology 1. Comparison

with Total Ozone Mapping Spectrometer

(TOMS) and ground-based measurements. J.

Geophys. Res., 108, 8238, 10.1029/2001JD

000967, 2002.

Thuburn, J., and G. C. Craig, On the temperature

structure of the tropical sub-stratosphere, J.

Geophys. Res., 107, 10.1029/2001JD000448, 2002.

Tuck, A. F., et al., The Brewer Dobson circula-

tion in the light of high altitude in situ aircraft

observations. Q. J. R. Meteorol. Soc., 124, 1-70,

1997.

Tuck, A. F., et al., Horizontal variability 1-2 km

below the tropical tropopause, J. Geophys. Res.,

109, 10.1029/2003JD003942, 2004.

Zipser, E. J., Some views on “hot towers” after 50

years of tropical field programmes and two years

of TRMM data. Meteorological Monographs, 29,

49-58, 2003.

26

SPARC_R10 7/27/05 12:49 PM Page 26

Introduction

It is widely accepted that the tropo-sphere has a strong dynamical effecton the stratosphere, primarily through

the upward propagation of waves, bothlow-frequency large-scale Rossby waves(‘planetary waves’) and high-frequencyinertia-gravity waves. Understanding ofthis effect is based on simple theories ofwave propagation (including the well-known Charney-Drazin criterion for ver-tical Rossby wave propagation), experi-ments in many different types of numeri-cal models, plus observational indicatorssuch as differences in stratospheric circu-lation between summer and winter, andbetween the hemispheres. An importantaspect of this effect is that in the strato-sphere there is a two-way interactionbetween waves and mean flow. Breaking ordissipating waves exert a systematic meanforce G that changes the mean flow. Themean flow, on the other hand, affects thepropagation, breaking and dissipation ofwaves and hence itself affects G . The two-way interaction can lead, for example, tosensitivity to initial conditions, or to inter-nal dynamical variability. Yoden et al.(2002) and Haynes (2005) review somethese issues.

Nonetheless, it is still not yet the case thatour understanding of the dynamical effectof the troposphere on the stratosphere canbe said to be complete. For example, forevents such as stratospheric sudden warm-ings (including the unexpected SouthernHemisphere sudden warming ofSeptember 2002), in which the stratospher-ic circulation is highly perturbed, it doesnot seem possible to identify unambigu-ously anomalously large tropospheric waveforcing as the cause. (See for examplepapers in the recent special issue of Journalof Atmospheric Sciences, volume 62, num-ber 3.) One of the reasons may be that theview of the troposphere having a one-waydynamical effect on the stratosphere is seri-ously limited. There are plenty of reasonswhy the coupling should be two-way. Thelarge-scale extratropical dynamics of theatmosphere is inherently non-local in bothhorizontal and vertical. Changes in the

stratosphere must inevitably affect the tro-posphere and vice versa – the key question,of course, is how much?

Stratospheric causeand tropospheric

effect?

The dynamical effect of the stratosphere onthe troposphere is now a major researchactivity. (See previous SPARC newsletterarticles by Gillett et al. 2003, Hartmann2004.) Whilst there has for some time beensignificant evidence from numerical modelstudies, early examples being Boville(1984) and Kodera et al. (1990), thatimposed perturbations to the stratospherelead to changes in the tropospheric circula-tion and that the mechanism for commu-nication of these changes is dynamical,much of the recent heightened interest inthis topic has arisen from studies of theNorthern Hemisphere (NH) and SouthernHemisphere (SH) ‘annular modes’ (here-after NAM and SAM). Methods such asEOF analysis have identified these as dom-

inant signals in variability in both tropo-sphere (e.g. Thompson and Wallace 2000and references therein) and stratosphere(e.g. Baldwin and Dunkerton 1999). Inboth troposphere and stratosphere, theannular modes are associated with varia-tion in the strength and position of the jet(the mid-latitude jet in the troposphere,the polar night jet in the stratosphere).However the underlying dynamical charac-ter of the modes is somewhat different inthe two cases.

In the troposphere the annular mode vari-ability is believed to arise from two-wayinteraction between baroclinic eddies andthe tropospheric mid-latitude jet (e.g.Robinson 1991, Feldstein and Lee 1998,Hartmann and Lo 1998). In the strato-sphere, on the other hand, the annular vari-ability is a manifestation of the variation inthe strength of the polar-night jet, driven bythe wave force G . A significant ingredient isthe variability in tropospheric wave forcing,although as noted earlier, there is animportant role for two-way interactionbetween waves and mean flow here too.

27

Stratosphere-Troposphere Dynamical CouplingPeter Haynes, Department of Applied Mathematics and Theoretical Physics,Cambridge, UK (phh@damtp.cam.ac.uk)

HIGH LOW HIGH-LOW

ZON

AL W

IND

WAV

E 1+

2+3 Al

titud

e (k

m)

20N 40N 60N 80N 20N 40N 60N 80N 20N 40N 60N 80N

25

20

15

10

5

25

20

15

10

5

40

70100

300500

900

40

70100

300500

900

Pres

sure

(hPa

)

Figure 1: (From Hartmann et al. 2000.) Composites for periods of high and low NAM index and their

difference (left, centre and right, respectively) in longitudinal wind (top) and Eliassen-Palm flux, which

indicates wave propagation and transport of westward momentum, and its divergence, which indicates

eastward wave force. Positive contours are grey, negative contours black, with negative regions shaded.

The Eliassen-Palm flux is calculated only for longitudinal wavenumbers 1, 2 and 3. In the ‘high’ phase

wave fluxes tend to be directed equatorward within the troposphere and to converge in the subtropical

troposphere, whereas in the ‘low’ phase wave fluxes tend to be directed upward from troposphere to

stratosphere and to converge, implying an anomalous westward wave force, in the mid- and high lati-

tude stratosphere. (Copyright (2000) National Academy of Sciences, USA. Reproduced with permission.)

SPARC_R10 7/27/05 12:49 PM Page 27

What has caused widespread interest is thatthere seem to be strong correlations betweenthe annular mode variability in the tropo-sphere and that in the stratosphere. Forexample, in NH winter there is significantcorrelation between the annular mode indexdefined on the basis of the surface pressurefield and the stratospheric circulation, sothat when there is a strong pole-to-equatorpressure gradient at the surface, indicatingstrong eastward surface flow, there are alsostrong eastward winds throughout the mid-latitude troposphere and in the mid-to-highlatitude stratosphere (Thompson andWallace, 2000). There is correspondingorganisation in the wave fluxes indicatingpropagation and momentum transport,both in the troposphere (Limpasuvan andHartmann 2000) (as is expected from theaccepted mechanism for the variability) andalso in the stratosphere (Hartmann et al.2000) (see Figure 1) implying correspondingdifferences in the wave force G .

Baldwin and Dunkerton (1999, 2001) haveshown that the vertical structure of NAMvariation in NH winter typically shows adownward propagation from middlestratosphere to troposphere and the corre-sponding figure from their 2001 paper isnow (quite justifiably) de rigeur in any sci-

entific talk in this area. Does this pictureimply a direct effect, with some delay, ofanomalies in the stratospheric circulationin mid-stratosphere on the troposphere?

The answer is ‘not necessarily’. The equato-rial quasi-biennial oscillation (QBO) is aneducational example. The QBO is mani-fested by downward propagating anoma-lies in zonal wind, however it was pointedout by Plumb (1977) that, at least in simpledynamical models of the QBO, there is infact no downward propagation of informa-tion. The evolution of the flow below somegiven level is completely independent ofany changes that are made above that level.The distinction in elementary wave theorybetween phase propagation on the onehand, and group propagation on the other,with only the latter unambiguously associ-ated with propagation of information, iswell-known. The downward propagationof wind anomalies in the QBO can be seenas a sort of phase propagation.

Simple dynamical models of the equato-rial QBO of the type studied by Plumb(1977) are not necessarily relevant to theextratropical stratosphere, in particularbecause rotation is omitted and because aWKBJ approximation is used in calculat-

ing the structure of the wavesfor given mean flow. HoweverPlumb and Semeniuk (2003)have shown in a simple wave-mean model of the extratropi-cal stratosphere (that doesincorporate rotation and uses aweaker approximation for thespatial structure of the waves)that forcing at low levels cangive rise in the stratosphereabove to downward propagat-ing structures similar to thoseobserved by Baldwin andDunkerton. Therefore, suchstructures do not necessarilyimply downward propagationof information.

There is similar uncertainty ininterpretation of the correlationsbetween anomalies of zonalvelocities and wave fluxes of thetype shown by Hartmann et al.(2000), (Figure 1), for example.While these correlations stronglysuggest dynamical connectionsbetween troposphere and strato-

sphere involving two-way interactionsbetween mean flow and waves, it is impos-sible to tell from the correlations alonewhether there is an effect of the strato-sphere on the troposphere, or the tropo-sphere on the stratosphere (or indeedwhether it makes sense only to think of thetroposphere-stratosphere system as intrin-sically coupled, with the whole idea of theeffect of one component on the other asintrinsically flawed).

Notwithstanding the above, there are nowmany examples of simulations in numericalmodels where changes in the tropospherehave been shown to result from imposedchanges in the stratosphere. In these casesdownward propagation of information isdifficult to discount. Some relevant exam-ples include imposed perturbations to theupper stratosphere (Kodera et al. 1990, Gray2003) as a simple representation of solarcycle effects, changes to stratospheric radia-tive equilibrum temperature profiles in asimplified general circulation model(Polvani and Kushner 2002, Kushner andPolvani 2004), changes to SouthernHemisphere stratospheric ozone in a gener-al circulation model (Gillett and Thompson2003) and changes to stratospheric initialconditions in a numerical weather predic-

28

TWO−WAY INTERACTIONBETWEEN WAVES/EDDIES

AND MEAN FLOW

TWO−WAY INTERACTIONBETWEEN WAVES/EDDIES

AND MEAN FLOWSYNOPTIC−SCALE EDDIESMEAN FLOW

MEAN FLOW LARGE−SCALE ROSSBY WAVES

DYNAMICS OF MEANCIRCULATION

UPWARD AND DOWNWARDWAVE PROPAGATION

TROPOSPHERE

STRATOSPHERE

(low−frequency annular variability)

(internal variability, sensitivity to external forcing)

Figure 2: Schematic diagram indicating the role of different aspects of the dynamics in the dynamical mechanismsdiscussed in the text. Note that ‘dynamics of mean circulation’ includes non-local PV inversion (or equivalently theshort-time effect of the meridional circulation) and the effect of the meridional circulation on longer time scales,including the ‘downward control’ limit.

SPARC_R10 7/27/05 12:49 PM Page 28

tion model (Charlton et al. 2004). Theresults of Scott and Polvani (2004), obtainedin a simplied dynamical model with adamped troposphere, show nicely that thewave flux from troposphere to stratospherecannot be considered to be set by the tropo-spheric wave forcing alone and that thestratosphere can to some extent determinehow much wave flux it accepts.

Dynamicalmechanisms

There are several possible dynamical mech-anisms by which the stratosphere mightaffect the troposphere. There are twoaspects to this: (a) how information mightbe communicated in the vertical and (b)why the tropospheric response might belarger than expected. Figure 2 is a schemat-ic diagram indicating some of the pointsdiscussed below.

Taking (a) first, one mechanism is via thenon-local inversion operator that deter-mines quantities such as velocity or tem-perature from the potential vorticity (PV)distribution. (The non-locality arises fromthe rapid propagation of inertio-gravitywaves required to maintain a state ofgeostrophic and hydrostatic balance.) Anychange in the PV distribution in the lowerstratosphere will inevitably give rise tochanges in wind and temperature in thetroposphere. Hartley et al. (1998),Ambaum and Hoskins (2002) and Black(2002) show explicit calculations to illus-trate this point. This might account forsome of the lower part of the height-timeplots shown by Baldwin and Dunkerton(1999, 2001). If the restriction is made tozonal mean fields, then the vertical non-locality of PV inversion is precisely equiv-alent to a statement that a wave-inducedforce localised to the stratosphere will,through the instantaneous induced merid-ional circulation, give rise to an accelera-tion in the troposphere below and many ofthe papers on this topic have chosen thelatter description. An important refine-ment is that on longer timescales, in thepresence of radiative damping, the merid-ional circulation tends to be narrower anddeeper below (Haynes et al. 1991, Holtonet al. 1995) (tending to a ‘downward con-trol’ response in the steady state limit)potentially allowing an enhanced tropo-spheric response to a stratospheric waveforce.

A second distinct mechanism for commu-nication in the vertical is via Rossby wavepropagation. The propagation of Rossbywaves out of the troposphere might besensitive to by variation in the ‘refractive’properties of the lower stratospheric flow(Hartmann et al., 2000, see alsoLimpasuvan and Hartman 2000), orindeed there might be downward reflec-tion of Rossby waves from higher in thestratosphere (e.g. Perlwitz and Harnik2003, 2004).

A third mechanism for downward propa-gation of information might be through atwo-way interaction between waves andmean flow. The results of Plumb andSemeniuk (2003) discussed earlier do notcompletely rule out the possibility thatthere can be real downward propagation ofinformation through this interaction.Recent investigation (Steven Hardiman,personal communication) using the samemodel as Plumb and Semeniuk has shownno evidence of any distinct downwardpropagation of the effect of imposed upperlevel stratospheric perturbations through asuch a mechanism when stratosphericRossby wave amplitudes are weak. Whenstratospheric wave amplitudes are large,however, the dynamics of waves and meanflow is highly nonlinear and imposedupper level perturbations can have signifi-cant effects at lower levels.

Turning to (b), the two-way interactionbetween baroclinic waves and mean flow inthe troposphere that gives rise to annularmode variability may also serve as an‘amplifier’ for external forcing (includingdynamical forcing from the stratosphere)(e.g. Hartmann et al. 2000). This may allowthe tropospheric response to be signifi-cantly larger than might be expected fromzonal-mean dynamics, for example.

Putting (a) and (b) together Song andRobinson (2004) report numerical experi-ments showing the effect of imposedstratospheric perturbations on the tropo-sphere and the effect arises through adownward penetrating response in themean circulation communicates the effectof stratospheric wave forcing to the tropo-sphere, where the response is amplified bythe eddy (i.e. wave) feedbacks associatedwith annular variability. They name theirmechanism ‘downward control with eddyfeedback’. However, having proposed this

mechanism (which to this author at leastseemed interesting and plausible) Song andRobinson then present further numericalexperiments that show that it cannot be thefull explanation for the effect of strato-spheric perturbations on tropospheric cir-culation that they observe. In particularthey show that the effect on the tropo-spheric circulation is much weaker whenthe Rossby waves in the stratosphere areartificially damped. The conclusion istherefore that the Rossby waves play a sig-nificant role in downward communicationof information.

Other strong evidence that the strato-sphere plays an active, rather than a pas-sive, role in tropospheric variations associ-ated with the NAM comes from observedand modelled changes on the time scale ofthe NAM variations. Baldwin et al. (2003)have shown that this time scale is signifi-cantly longer at times of the year (NHwinter, SH spring) when there is strongRossby wave propagation into the strato-sphere. Correspondingly, artificial sup-pression of stratospheric variability inmodel simulations reduces the time scaleof the tropospheric NAM (Norton 2003).The dynamical mechanism here is likely tobe that, when there is significant flux ofRossby waves into the stratosphere, theflow in the stratosphere acts as an integra-tor (and hence low-pass-filter) of this flux(or rather the variability in this flux), sincestratospheric damping times are relativelylong. Any stratospheric effect on the tro-posphere will therefore tend to increasethe time scales of the variability. Whenthere is little flux of Rossby waves into thestratosphere (in summer, or in SH mid-winter) the effect is absent.

Conclusion

The possibility of significant stratosphericeffects on the troposphere has implicationsfor many aspects of month-to-month andyear-to-year variability and systematicchange in the tropospheric circulation. Itsuggests possible mechanisms for explain-ing apparent signals in the troposphericcirculation of the solar cycle, inputs of vol-canic aerosol to the stratosphere, and theequatorial QBO. It also strengthens the linkbetween between possible climate changein the troposphere and changes in thestratosphere, due to ozone depletion orincreasing greenhouse gases. Whilst care is

29

SPARC_R10 7/27/05 12:49 PM Page 29

needed in interpreting observations asimplying real downward influence ordownward propagation, there is convinc-ing evidence from numerical model simu-lations that changes in the stratosphere cansometimes lead to significant effects in thetroposphere.

The dynamical mechanisms required toexplain these effects are still being investi-gated. It seems clear that the two-way inter-action between synoptic-scale waves/eddiesand mean flow in the troposphere that givesrise to ‘annular variability’ is important,notwithstanding the fact that the nature ofannular variability is still being vigorouslydebated (e.g. Cash et al. 2005). The two-wayinteraction between waves and mean flowin the stratosphere also seems likely to berelevant, though the relative roles of waves,mean flow, and coupling between them isnot yet clear.

Further clarification of these dynamicalmechanisms and their role in the realatmosphere will most likely come fromcareful studies in a sequence of numericalmodels. Much has already been learnedfrom simplified models that include thelarge-scale dynamics plus highly simplifiedrepresentations of processes such as radia-tion, and more work with these models, aswell as with sophisticated general circula-tion models, is surely needed to resolvesome of the remaining uncertainties.

An important general point that has beenrevived by the recent interest in tropo-sphere-stratosphere coupling is that,whether or not one is interested in thedynamical details, the fact is that the cou-pled system exhibits strong (dynamical)internal variability and that any attempt toexplore correlations between one part ofthe atmosphere and the other, or to predictfuture changes, needs to take this intoaccount. Such studies therefore need to uselong integrations or large ensembles,requiring significant computationalresources. There is understandable pressureto make as rapid progress as possible withcoupled chemical-climate simulations,which also requires significant computa-tional resources, but there is still much tolearn about the variability and predictabilityof the coupled physics and dynamics of tro-posphere-stratosphere system without cou-pling to chemistry, and this should not beoverlooked.

References

Ambaum, M. H. P., and B. J. Hoskins, 2002. The

NAO troposphere-stratosphere connection. J.

Climate, 15:1969–78.

Baldwin M. P., and T. J. Dunkerton, 1999.

Propagation of the Arctic Oscillation from the

stratosphere to the troposphere. J. Geophys. Res.,

104:30937-46.

Baldwin M. P., and T. J. Dunkerton, 2001.

Stratospheric harbingers of anomalous weather

regimes. Science 294:581-84.

Baldwin M. P., D. B. Stephenson, D. W. J.

Thompson, T. J. Dunkerton, A. J. Charlton, and

A. O’Neill, 2003. Stratospheric memory and

skill of extended-range weather forecasts.

Science, 301:636-40.

Black R. X., 2002. Stratospheric forcing of sur-

face climate in the Arctic Oscillation. J. Climate,

15:268–77.

Boville B. A., 1984. The influence of the polar

night jet on the tropospheric circulation in a

GCM. J. Atmos. Sci., 41: 1132–1142.

Cash B. A., P. J. Kushner, and G. K. Vallis, 2005.

Zonal asymmetries, teleconnections and annu-

lar patterns in a GCM. J. Atmos. Sci., 62:

207–219.

Charlton A. J., A. O’Neill, W. A. Lahoz, and A. C.

Massacand, 2004. Sensitivity of tropospheric

forecasts to stratospheric initial conditions.

Quart. J. Roy. Met. Soc., 130: 1771-1792.

Feldstein S., and S. Lee, 1998. Is the atmospher-

ic zonal index driven by an eddy feedback? J.

Atmos. Sci., 55:3077-86.

Gillett N. P., M. P. Baldwin, D. W. J. Thompson,

E. F. Shuckburgh, W. A. Norton, and J. L. Neu,

2003. Report on the SPARC workshop on the

role of the stratosphere in tropospheric climate.

SPARC Newsletter 21: 7-9.

Gillett, N. P., and D. W. J. Thompson, 2003.

Simulation of recent Southern Hemisphere cli-

mate change. Science, 302: 273-275.

Gray, L. J., 2003. The influence of the equatorial

upper stratosphere on stratospheric sudden

warmings. Geophys. Res. Lett., 30: 1166.

Hartley, D. E., J. T. Villarin, R. X. Black, and C. A.

Davis, 1998. A new perspective on the dynami-

cal link between the stratosphere and tropo-

sphere. Nature, 391:471–74.

Hartmann, D. L., 2004. The stratosphere in the

climate system. SPARC Newsletter 22: 15-18.

Hartmann, D. L., and F. Lo, 1998. Wave-driven

zonal flow vacillation in the Southern

Hemisphere. J. Atmos. Sci., 55:1303-15.

Hartmann, D. L., J. M. Wallace, V. Limpasuvan,

D. W. J. Thompson, and J. R. Holton, 2000. Can

ozone depletion and global warming interact to

produce rapid climate change?. Proc. Nat. Acad.

Sci. (Wash. DC), 97:1412–17.

Haynes, P.H., 2005. Stratospheric dynamics.

Ann. Rev. Fluid Mech., 37:263-293.

Haynes, P. H., C. J. Marks, M. E. McIntyre, T. G.

Shepherd, and K. P. Shine, 1991. On the “down-

ward control” of extratropical diabatic circula-

tions by eddy-induced mean zonal forces. J.

Atmos. Sci., 48: 651–78.

Holton, J. R., P. H. Haynes, and M. E. McIntyre,

A. R. Douglass, R. B. Rood, and L. Pfister, 1995.

Stratosphere–troposphere exchange. Revs.

Geophys., 33:403–39.

Kodera, K., K. Yamazaki, K. Chiba, and K.

Shibata, 1990. Downward propagation of upper

stratospheric mean zonal wind perturbation to

the troposphere. Geophys. Res. Lett., 17:1263-66.

Kushner, P. J., and L. M. Polvani, 2004.

Stratosphere-troposphere coupling in a relative-

ly simple AGCM: the role of eddies. J. Climate,

17: 629-639.

Limpasuvan, V., and D. L. Hartmann, 2000.

Wave-maintained annular modes of climate

variability. J. Climate, 13:4414-29.

Norton, W. A., 2003. Sensitivity of northern

hemisphere surface climate to simulation of the

stratospheric polar vortex. Geophys. Res. Lett.

30:1627.

Perlwitz, J., and N. Harnik, 2003. Observational

evidence of a stratospheric influence on the tro-

posphere by planetary wave reflection. J.

Climate, 16:3011-26.

Perlwitz, J., and N. Harnik, 2004. Downward

coupling between the stratosphere and the tro-

posphere: the relative roles of wave and zonal

mean processes. J. Climate, 17:4902-4909.

30

SPARC_R10 7/27/05 12:49 PM Page 30

Plumb, R. A., 1977. The interaction of two inter-

nal waves with the mean flow: implications for

the theory of the quasi-biennial oscillation. J.

Atmos. Sci., 34:1847–58.

Plumb, R. A., and K. Semeniuk, 2003.

Downward migration of extratropical zonal

wind anomalies. J. Geophys. Res., 108: 4223,

doi:10.1029/2002JD002773.

Polvani, L. M., and P. J. Kushner, 2002.

Tropospheric response to stratospheric pertur-

bations in a relatively simple general circulation

model. Geophys. Res. Lett,. 29: 1114,

doi:10.1029/2001GL014284.

Robinson, W. A., 1991. The dynamics of the

zonal index in a simple model of the atmo-

sphere. Tellus, 43A:295–305.

Scott, R. K., and L. M. Polvani, 2004.

Stratospheric control of upward wave flux near

the tropopause. Geophys. Res. Lett.,

31:doi:10.1029/2003GL017965.

Song, Y., and W. A. Robinson, 2004. Dynamical

mechanisms of stratospheric influences on the

troposphere. J. Atmos. Sci., 61:1711-25.

Thompson, D. W. J., and J. M. Wallace, 2000.

Annular modes in the extratropical circulation.

Part I: Month-to-month variability. J. Climate,

13:1000-16.

Yoden, S., M. Taguchi, and Y. Naito, 2002.

Numerical studies on time variations of the tro-

posphere-stratosphere coupled system. J. Met.

Soc. Japan, 80:811–830.

31

Assimilation of Stratospheric Meteorological andConstituent Observations: A Review

Richard B. Rood, NASA Goddard Space Flight Center, USA (Richard.B.Rood@nasa.gov)

Introduction

The assimilation of stratosphericobservations has been the focus ofseveral research groups in the past fif-

teen years. The use of products from theassimilation of meteorological data is nowwidespread. As new data types become avail-able researchers are anxious to try assimila-tion experiments. This article will reviewprogress in the last 3-5 years and evaluate thatprogress in terms of the underlying geophys-ical robustness of data assimilation systems.

A dictionary definition of assimilation is:to incorporate or absorb, for instance, intothe mind or the prevailing culture. ForEarth science, assimilation is the incorpo-ration of observational information into aphysical model. Or more specifically, dataassimilation is the objective melding ofobserved information with model-predict-ed information.

Assimilation rigorously combines statisticalmodelling with physical modelling. Daley(1991) is the standard text on data assimila-tion. Cohn (1997) explores the theory of dataassimilation, and its foundation in estimationtheory. Swinbank et al. (2003) is a collectionof tutorial lectures on data assimilation.Assimilation is difficult to do well, easy to dopoorly, and its role in Earth science is expand-ing and sometimes controversial.

In the atmosphere the predominantobservations that are assimilated are tem-peratures, or radiances that represent the

temperature. These observations comefrom radiosondes, aircraft and satellites.Wind data are also assimilated, and are ofcrucial importance to the quality of theassimilation. As a function of altitude theamount of wind data available for assimila-tion drops dramatically above thetropopause. Hence, the state of the strato-sphere is determined primarily by tempera-ture observations and information propa-gated from the surface and the troposphere.A good tropospheric assimilation is requiredfor a good stratospheric assimilation.

A number of constituents are also of inter-est. Water vapour has been routinely assim-ilated in the troposphere for many yearsand is important to the specification of tro-pospheric physics. More recently, severalcentres and groups have undertaken theassimilation of ozone. There have also beena number of experiments addressing inter-acting chemical species. Both ozone assim-ilation and chemical assimilation arereviewed in Swinbank et al. (2003).

There are a number of factors that motivatethe desire to assimilate geophysical parame-ters. The list below is specifically for ozone,but it is representative of the motivation ofother geophysical parameters as well.

Motivations for OzoneAssimilation

• Mapping: There are spatial and tempo-ral gaps in the ozone observing system.A basic goal of ozone assimilation is toprovide vertically resolved global mapsof ozone.

• Short-term ozone forecasting: There isinterest in providing operational ozoneforecasts in order to predict the fluctua-tions of ultraviolet radiation at the sur-face of the earth (Long et al., 1996).

• Chemical constraints: Ozone is impor-tant in many chemical cycles. Assimilationof ozone into a chemistry model providesconstraints on other observed con-stituents and helps to provide estimatesof unobserved constituents.

Table 1: Simulation Framework: General Circulation Model, or forecast model is associated with errors:

(ed, ep) = (discretization error, parametrization error) and (eb, ev) = (bias error, variability error). The

size of ‘e’ in each case represents the size of the error associated with the process in each row. The derived

products are likely to be physically consistent, but have significant errors; i.e. the theory-based con-

straints are met.

Boundary Conditions Emissions, SST, … e

Representative Equations DA/Dt=P-LA-(n/H)A+q/H e

Discrete/Parametrize (An+�t-An)/�t = …. (ed, ep)

Theory/Constraints ∂ug/∂z= -(∂T/∂y)R/(Hf0) Scale Analysis

Primary Products (i.e. A) T, u, v, F, H2O, O3 (eb, ev)

Derived Products (F(A)) Potential Vorticity, v*,w*, … Consistent

SPARC_R10 7/27/05 12:49 PM Page 31

• Unified ozone data sets: There are sev-eral sources of ozone data with signifi-cant differences in spatial and temporalcharacteristics as well as their expectederror characteristics. Data assimilationprovides a potential strategy for com-bining these data sets into a unified dataset.

• Tropospheric ozone: Most of the ozoneis in the stratosphere, and troposphericozone is sensitive to surface emission ofpollutants. Therefore, the challenges ofobtaining accurate tropospheric ozonemeasurements from space are signifi-cant. The combination of observationswith the meteorological informationprovided by the model offers one of thebetter approaches available to obtainglobal estimates of tropospheric ozone.

• Improvement of wind analysis: Thephotochemical time scale for ozone islong compared with transporttimescales in the lower stratosphere andupper troposphere. Therefore ozonemeasurements contain informationabout the wind field that might beobtained in multi-variate assimilation.

• Radiative transfer: Ozone is importantin both longwave and shortwave radia-tive transfer. Therefore accurate repre-sentation of ozone is important in theradiative transfer calculations needed toextract (retrieve) information frommany satellite instruments. In addition,accurate representation of ozone has thepotential to impact the quality of thetemperature analysis in multi-variateassimilation.

• Observing system monitoring: Ozoneassimilation offers an effective way tocharacterize instrument performance rel-ative to other sources of ozone observa-tions as well as the stability of measure-ments over the lifetime of an instrument.

• Retrieval of ozone:Ozone assimilation offersthe possibility of pro-viding more accurateinitial guesses for ozoneretrieval algorithms thanare currently available.

• Assimilation research:Ozone (constituent)assimilation can be pro-ductively approached asa univariate linear prob-lem. Therefore it is agood framework forinvestigating assimila-tion science; for exam-ple, the impact of flowdependent covariancefunctions.

• Model and observation validation:Ozone assimilation provides severalapproaches to contribute to the valida-tion of models and observations.

Some of the goals mentioned above can bemeaningfully addressed with the currentstate of the art, others cannot. It is straight-forward to produce global maps of totalcolumn ozone which can be used in, forinstance, radiative transfer calculations.The use of ozone measurements to provideconstraints on other reactive species is anapplication that has been explored since the1980’s (see Jackman et al., 1987) and mod-ern data assimilation techniques potentiallyadvance this field. The impact of ozoneassimilation on the meteorological analysisof temperature and wind, and henceimprovement of the weather forecast, isalso possible. The most straightforwardimpact would be on the temperature analy-sis in the stratosphere. The improvement ofthe wind analysis is a more difficult chal-lenge and confounded by the fact that

where improve-ments in the windanalyses are mostneeded, the trop-ics, the ozone gra-dients are relative-ly weak. The use ofozone assimilationto monitor instru-ment performanceand to character-ize new observingsystems is currentlypossible and pro-ductive (see Stajner

et al., 2004). The improvement of retrievalsusing assimilation techniques to provideozone first guess fields that are representativeof the specific environmental conditions isalso an active research topic. The goal of pro-ducing unified ozone data sets from severalinstruments is of little value until bias can becorrectly accommodated in data assimila-tion. This final topic will be discussed morefully below.

In order to discuss why assimilation is suc-cessful in addressing some of these goalsand less successful in others, the underly-ing physical foundation of assimilation willbe considered. This will be based on thediscussion of modelling and assimilationand comparison of the attributes of theconcepts of performing simulation (mod-elling) and assimilation. Following this,expository examples will be presented. Anunderlying tenet is that by comparingresults from a simulation to results from anassimilation using the same model, theresearcher is investigating cause and effectin a controlled experiment.

Simulation andAssimilation

In order to provide an overarching back-ground for thinking about model simula-tion, it is useful to consider the elements ofthe modelling, or simulation, frameworkdescribed in Table 1. In this framework aresix major ingredients. The first is the speci-fication of boundary and initial conditions(e.g. topography and sea surface tempera-tures for an atmospheric model). Boundaryconditions are generally prescribed fromexternal sources of information.

32

(A, U)

Grid Point (i,j)

Figure 1: Discretization of resolved transport and the choice of whereto represent information impacts the physics, such as conservation,scale analysis limits, and stability (Rood, 1987). Grids may beorthogonal, uniform area, adaptive, or unstructured.

Grid Point (i,j)

Grid Point (i,j+1) Grid Point (i+1,j+1)

Grid Point (i+1,j)

(U)

(U)

(A)(U) (U)

Figure 2: Another choice of where to represent information on the grid.

SPARC_R10 7/27/05 12:49 PM Page 32

The next three items in the table are inti-mately related. They are the representativeequations, the discrete or parameterizedequations, and the constraints drawn fromtheory. The representative equations arethe analytic form of forces or factors thatare considered important in the represen-tation of the evolution of a set of parame-ters. All of the analytic expressions used inatmospheric modelling are approxima-tions; therefore, even the equations themodeller is trying to solve have a priorierrors. That is, the equations are scaledfrom some more complete representation,and only terms that are expected to beimportant are included in the analyticexpressions (see Holton, 2004). The solu-tion is, therefore, a balance amongst com-peting forces and tendencies. Most com-monly, the analytic equations are a set ofnonlinear partial differential equations.

The discrete or parameterized equationsarise because it is generally not possible tosolve the analytic equations in closed form.The strategy used by scientists is to developa discrete representation of the equationswhich are then solved using numericaltechniques. These solutions are, at best, dis-crete estimates to solutions of the analyticequations. The discretization and parame-terization of the analytic equations intro-duce a large source of error. This introducesanother level of balancing in the model;namely, these errors are generally managedthrough a subjective balancing process thatkeeps the numerical solution from runningaway to obviously incorrect estimates.

While all of the terms in the analytic equa-tion are potentially important, there areconditions or times when there is a domi-nant balance between, for instance, twoterms. An example of this is thermal windbalance in the middle latitudes of the strato-sphere (see Holton, 2004). It is these bal-ances, generally at the extremes of spatialand temporal scales, which provide the con-straints drawn from theory. If the modellerimplements discrete methods which consis-tently represent the relationship betweenthe analytic equations and the constraintsdrawn from theory, then the modellermaintains a substantive scientific basis forthe interpretation of model results.

The last two items in Table 1 represent theproducts that are drawn from the model.These are divided into two types: primary

products and derived products. The prima-ry products are variables such as wind, tem-perature, water, ozone – parameters that aremost often, explicitly modelled; that is, anequation is written for them. The derivedproducts are often functional relationshipsbetween the primary products; for instance,potential vorticity (Holton, 2004). A com-mon derived product is the balance, or thebudget, of the different terms of the dis-cretized equations. The budget is then stud-ied, explicitly, on how the balance is main-tained and how this compares with budgetsderived directly from observations. In gen-eral, the primary products can be directlyevaluated with observations and errors ofbias and variability estimated. If attentionhas been paid in the discretization of theanalytic equations to honour the theoreticalconstraints, then the derived products willbehave consistently with the primary prod-ucts and theory. They will have errors ofbias and variability, but they will behave ina way that supports scientific scrutiny.

In order to explore the elements of themodelling framework described above, thefollowing continuity equation for a con-stituent, A, will be posed as the representa-tive equation. The continuity equation rep-resents the conservation of mass for a con-stituent and is an archetypical equation ofgeophysical models. Brasseur and Solomon(1986) and Dessler (2000) provide goodbackgrounds for understanding atmo-spheric chemistry and transport. The con-tinuity equation for A is:

∂A = – �•UA + M + P – LA – n A – q (1)∂t H H

where A is some constituent, U is velocity(“resolved” transport, “advection”), M is“Mixing” (“unresolved” transport, param-eterization), P is production, L is loss, n is“deposition velocity”, q is emission, H isrepresentative length scale for n and q, t istime, and � is the gradient operator.

Attention will be focused on the discretiza-tion of the resolved advective transport.Figures 1 and 2 illustrate the basic concepts.On the left of the figure a mesh has beenlaid down to cover the spatial domain ofinterest. In this case it is a rectangular mesh.The mesh does not have to be rectangular,uniform, or orthogonal. In fact the meshcan be unstructured or can be built to adaptto the features that are being modelled. Thechoice of the mesh is determined by themodeller and depends upon the diagnosticand prognostic applications of the model(see Randall, 2000). The choice of mesh canalso be determined by the computationaladvantages that might be realized.

Points can be prescribed to determine loca-tion with the mesh. In Figure 1 both theadvective velocity and the constituent areprescribed at the centre of the cell. In Figure2, the velocities are prescribed at the centreof the cell edges, and the constituent is pre-scribed in the centre of the cell. There areno hard and fast rules about where theparameters are prescribed, but small differ-ences in their prescription can have a hugeimpact on the quality of the estimated solu-tion to the equation, i.e. the simulation. Theprescription directly impacts the ability ofthe model to represent conservation prop-erties and to provide the link between theanalytic equations and the theoretical con-straints (see Rood, 1987; Lin and Rood,1996; Lin, 2004). In addition, the prescrip-tion is strongly related to the stability of thenumerical method; that is, the ability torepresent any credible estimate at all.

The use of numerical techniques to repre-sent the partial differential equations thatrepresent the model physics is a straight-forward way to develop a model. There aremany approaches to discretization of thedynamical equations that govern geophysi-cal processes (Jacobson, 1998; Randall,2000). One approach that has been recent-ly adopted by several modelling centres is

33

Table 2: Assimilation Framework: O is the “observation” operator, Pf is forecast model error covariance,R is the observation error covariance, and x is the innovation. Generally resolved, predicted variablesare assimilated into the models.

Emissions, SST, … e Boundary Conditions

DA/Dt=P-LA-(n/H)A+q/H e (OPfOT+R)x=A0-OAf

(An+�t-An)/�t=… e Discrete/Error Modelling

∂ug/∂z=-(∂T/∂y)R/(Hf0) Scale Analysis Constraints on Increments

Ai=T, u, v, F, H2O, O3 (eb, ev) (eb, ev) reduced

Pot. Vorticity, v*, w*, … Consistent Inconsistent

SPARC_R10 7/27/05 12:50 PM Page 33

described in Lin (2004). In this approachthe cells are treated as finite volumes andpiecewise continuous functions are fitlocally to the cells. These piecewise contin-uous functions are then integrated aroundthe volume to yield the forces acting on thevolume. This method, which was derivedwith physical consistency as a requirementfor the scheme, has proven to have numer-ous scientific advantages. The scheme usesthe philosophy that if the physics are prop-erly represented, then the accuracy of thescheme can be robustly built on a physicalfoundation.

Table 2 shows elements of an assimilationframework that parallels the modelling ele-ments in Table 1. The concept of boundaryconditions remains the same; that is, somespecified information at the spatial andtemporal domain edges. Of particularnote, the motivation for doing data assimi-lation is often to provide the initial condi-tions for predictive forecasts.

Data assimilation adds an additional forc-ing to the representative equations of thephysical model; namely, information fromthe observations. This forcing is formallyadded through a correction to the modelthat is calculated, for example, by (seeStajner et al., 2001):

(OPƒOT + R)x = Ao – OAƒ (2)

The terms in the equation are as follows: Ao

are observations of the constituent, Aƒ aremodel forecast (simulated estimates of theconstituent, often called the first guess), Ois the observation operator, Pƒ is the errorcovariance function of the forecast, R is theerror covariance function of the observa-tions, x is the innovation that represents theobservation-based correction to the model,and ( )T is the matrix transform operation.The observation operator, O, is a functionthat maps the parameter to be assimilatedinto observation space.

The error covariance functions Pƒ and Rrepresent the errors of the informationfrom the forecast model and the informa-tion from the observations, respectively.This explicitly shows that data assimilationis the error-weighted combination ofinformation from two primary sources.From first principles, the error covariancefunctions are prohibitive to calculate.Stajner et al. (2001) show a method for

estimating the error covariances in anozone assimilation system.

Parallel to the elements in the simulationframework (Table 1), discrete numericalmethods are needed to estimate the errorsas well as to solve the matrix equations inEquation (2). Addressing physical con-straints from theory is a matter of bothimportance and difficulty. Often, for exam-ple, it is assumed that the increments ofdifferent parameters that are used to cor-rect the model are in some sort of physicalbalance. For instance, wind and tempera-ture increments might be expected to be ingeostrophic balance. However, in general,the data insertion process acts like an addi-tional forcing term in the equation, andcontributes a significant portion of thebudget. This explicitly alters the physicalbalance defined by the representative equa-tions of the model. Therefore, there is noreason to expect that the correct geophysi-cal balances are represented in an assimi-lated data product. This is contrary to theprevailing notion that the model andobservations are ‘consistent’ with oneanother after assimilation.

The final two elements in Table 2 are,again, the products. In a good assimilationthe primary products, most often the prog-nostic variables, are well estimated. That is,both the bias errors and the variance errorsare reduced relative to the model simula-tion. However, the derived products arelikely to be physically inconsistent becauseof the nature of the corrective forcingadded by the observations. These errors areoften found to be larger than those in self-determining model simulations. This is ofgreat consequence as many users look todata assimilation to provide estimates ofunobserved or derived quantities. Molod etal. (1996) and Kistler et al. (2001) providediscussions on the characteristics of theerrors associated with primary and derivedproducts in data assimilation systems.

As suggested earlier, the specification offorecast and model error covariances andtheir evolution with time is a difficultproblem. In order to get a handle on theseproblems it is generally assumed that theobservational errors and model errors areunbiased over some suitable period oftime, e.g. the length of the forecast betweentimes of data insertion. It is also assumedthat the errors are in a Gaussian distribu-

tion. The majority of assimilation theory isdeveloped based on these assumptions,which are, in fact, not valid assumptions. Inparticular, when the observations arebiased, there would be the expectation thatthe actual balance of geophysical terms isdifferent from the balance determined bythe assimilation. Furthermore, since thebiases will have spatial and temporal vari-ability, the balances determined by theassimilation are quite complex. Aside frombiases between the observations and themodel prediction, there are biases betweendifferent observation systems for the sameparameters. These biases are potentiallycorrectible if there is a known standard ofaccuracy defined by a particular observingsystem. However, the problem of bias is adifficult one to address and perhaps thegreatest challenge facing assimilation (see,Dee and da Silva, 1998).

As a final general consideration, there aremany time scales represented by the repre-sentative equations of the model. Some ofthese time scales represent balances that areachieved almost instantly between differentvariables. Other time scales are long (e.g.the general circulation), which will deter-mine the distribution of long-lived traceconstituents. It is possible in assimilationto produce a very accurate representationof the observed state variables, and thosevariables which are balanced on fast timescales. On the other hand, improved esti-mates in the state variables are found, atleast sometimes, to be associated withdegraded estimates of those features deter-mined by long time scales. Conceptually,this can be thought of as the impact of biaspropagating through the physical model.With the assumption that the observationsare fundamentally accurate, this indicateserrors in the specification of the physicsthat demand further research.

Transportapplications: Whathave we learned?

Rood et al. (1989) first used winds andtemperatures from a meteorologicalassimilation to study stratospheric trans-port. Since that time there have been pro-ductive studies of both tropospheric andstratospheric transport. However, a num-ber of barriers have been met in recentyears, and the question arises - has a wall

34

SPARC_R10 7/27/05 12:50 PM Page 34

been reached where foundational ele-ments of data assimilation are limiting theability to do quantitative transport appli-cations? Stohl et al. (2004; and referencestherein) provide an overview of some ofthe limits that need to be considered intransport applications.

In transport applications, winds and tem-peratures are taken from a meteorologicalassimilation and used as input to a con-stituent transport model (CTM). Theresultant distributions of trace con-stituents are then compared with observa-tions. The constituent observations aretelling indicators of atmospheric motionson all time scales. Further, there is a wealthof very high quality constituent observa-tions from many observational platforms.Rigorous quantitative Earth science hasbeen significantly advanced by comparisonof constituent observations and modelestimates. Overall, it is found that themeteorological analyses do a very good jobof representing variability associated withsynoptic and planetary waves. This hasbeen invaluable in accounting for dynami-cal variability, and allowing the evaluationof constituents from multiple observation-al platforms. On the other hand, those geo-physical parameters that rely on the repre-sentation of the general circulation, forinstance the lifetime of long-lived con-stituents, are poorly represented.

Returning to the concepts introduced inthe previous section, when the primaryproducts of assimilation (e.g. winds andtemperature) dominate the variability,then the variability is well represented.When the derived products (e.g. the resid-ual circulation) are responsible for thevariability, then the variability is poorlyrepresented. This is found to be consistent-ly true. Transport calculations from allassimilation systems are found to have toomuch mixing between the tropics and themiddle latitudes. In the tropics, where thetemperature observations do not stronglydetermine the winds, the quality of thetransport degrades relative to the middlelatitudes. In the troposphere, if the trans-port features are associated with convec-tion or surface exchanges, i.e. the quantitiesderived from the constrained physicalparameterizations, then they are notrobustly represented.

By the nature of data assimilation, improve-

ments to the system provide better estimatesto the variables that are being assimilated.Simmons et al. (2003) show impressiveimprovement of the representation of windsin the tropics in the products from theEuropean Centre for Medium-rangeWeather Forecasts. However, a number ofconference papers have shown that theseimprovements have not extended to the rep-resentation of transport features associatedwith the general circulation; the derivedparameters associated with long time scalesfor adjustment are degraded. Ruhnke et al.(2003) were amongst the first to demon-strate this through the transport of ozone,where overestimates of lower stratosphericozone increased with newer versions of thedata assimilation system.

Douglass et al. (2003) and Schoeberl et al.(2003) each provide detailed studies thatexpose some of the foundational short-comings of the physical consistency of dataassimilation. In their studies they investi-gate the transport and mixing of atmo-spheric constituents in the upper tropo-sphere and the lower stratosphere. Figure 3is from Douglass et al. (2003) and showsozone probability distribution functions intwo latitude bands from four experimentsusing a constituent transport model. Inthree of these experiments (Panels B, D,and E), winds and temperatures are takenfrom an assimilation system. In panel C,results are from an experiment using windsfrom a general circulation model; that is, afree-running model without assimilation.Panel A shows ozonesonde observations;the sondes reflect similar distributions inthe two latitude bands. In all of the numer-

ical experiments, the means in the two lat-itude bands are displaced from each other,unlike the observations. In the three exper-iments using winds from different dataassimilation systems (DAS), the half-widthof the distributions is much too wide.

There are a number of points to be made inthis figure. First, the winds from the assim-ilation system in Panel B and the model inPanel C both use the finite-volume dynam-ics of Lin (2004). Therefore, these experi-ments are side-by-side comparisons thatshow the impact of inserting data into themodel. Aside from developing a bias, theassimilation system shows much moremixing. As Douglass et al. show, the instan-taneous representation of the wind is bet-ter in the assimilation, but the transport isworse. This is attributed to the fact thatthere are consistent biases in the modelprediction of the tropical winds and thecorrection added by the data insertioncauses spurious mixing. Tan et al. [2004]investigate the dynamical mechanisms ofthe mixing in the tropics and the subtrop-ics. Second, the assimilation systems usedfor Panels D and E have a different assimi-lation model, and their representation oftransport is worse than that from thefinite-volume model. This improvement isattributed to the fact that the finite-volumemodel represents the physics of the atmo-sphere better, in particular, the representa-tion of the vertical velocity. Third, theresults in Panel B show significantimprovement compared to the olderassimilation systems used in Panels D andE. Older assimilation systems have enoughdeficiencies that scientists have shied away

35

shaded 15°S-26°SBOLD 15°S-6°N CTMFVDAS

SONDESCTMFVGCM CTMTERRA CTMTRMM

0 25 50 25 2550 50 7575 505025 25DUDUDU

DAS -driven GCM-driven

frac

tion

0.2A B C D E

0.1

0.0

DUDU

Figure 3: PDFs of total ozone, observations and chemical transport model (CTM). PDFs from DAS-driven show displaced means and spreads that are too wide, whereas GCM-driven PDFs have displacedmeans with a better half-width. This shows that there is too much tropical-extratropical mixing inDAS. From Douglass et al. (JGR, 2003).

SPARC_R10 7/27/05 12:50 PM Page 35

from doing tropical transport studies. Thisexample demonstrates both the improve-ments that have been gained in recentyears, and indicates that the use of windsfrom assimilation in transport studiesmight have fundamental limitations.

Figure 4 (see colour insert IV) is fromSchoeberl et al. (2003). The Schoeberl et al.(2003) study is similar in spirit to theDouglass et al. study, but uses Lagrangiantrajectories instead of Eulerian advectionschemes. This allows Schoeberl et al. (2003)to address, directly, whether or not the spu-rious mixing revealed in the Douglass et al.(2003) paper is related to the advectionscheme. In this figure the results from twocompletely independent assimilation sys-tems are used; UKMO (United KingdomMet Office) and DAO (Data AssimilationOffice). The DAO system uses the finite-volume dynamical core (labelled DAO) andthe finite-volume GCM (labelled GCM).Vertical winds are also calculated two ways,diabatically using the heating rate informa-tion from the assimilation system, kinemat-ically, through continuity, using the hori-zontal winds from the assimilation.

The figure shows the impact of the methodof calculating the vertical wind using thediabatic information. When the diabaticinformation is used there is much lesstransport in the vertical. While this is gen-erally in better agreement with observa-tions and theory, the diabatic winds nolonger satisfy mass continuity with the hor-izontal winds. This points to a self-limitingaspect of using diabatic winds in Euleriancalculations such as the ones of Douglass etal. (2003). The Schoeberl et al. calculationsalso show that even with the diabatic verti-cal winds, there remains significant hori-zontal mixing, which is compressed alongisentropic surfaces. The final panel showsthat for the simulation, the free-runningmodel, there is much less dispersion, whichis in better agreement with both observa-tions and theory. Schoeberl et al. attributethe excess dispersion in the assimilationsystems to noise that is introduced by datainsertion. (They also note that the finite-volume dynamics is much improved rela-tive to previous generation models.)

These two studies point to the fact thatdata insertion impacts the physics thatmaintains the balances in the conservationequations of momentum, heat, and mass.

Both bias and the generation of noise havean impact. Both problems are difficult toaddress, with the problem of bias havingfundamental issues of tractability. Again,while the data assimilation system doesindeed provide better estimates of the pri-mary variables, as the impact of data inser-tion is adjusted through the physics repre-sented in the model, the derived parame-ters are often degraded. (Lait (2002) pro-vides an interesting exposition of subtleartifacts related to biases between differentradiosonde instruments.) One conclusionis that while there may be greater discrep-ancies in the absolute, day-to-day represen-tation of constituents with free-runningmodels, the consistent representation ofthe underlying physics allows more robuststudy of transport mechanisms and thosefeatures in the constituent data which aredirectly related to dynamics.

Ozone Assimilation

The last five years have seen the publicationof a number of papers on the assimilationof ozone and other constituents. Thesestudies show constant improvement in thestate of the art. They demonstrate theimportance of having both total columnobservations as well as high-resolutionprofile information. Because of thisprogress, products from ozone assimilationare on the verge of being geophysicallyinteresting. Applications, for example,include improvement of radiative transfer,monitoring of instruments, and providinginformation useful for estimating tropo-spheric ozone.

Figure 5 (see colour insert IV) shows anexample from an assimilation of ozonedata (Wargan et al. 2005). In this exam-ple, there are two satellite instruments,the Solar Backscattered Ultraviolet/2(SBUV) and the Michelson Interferometerfor Passive Atmospheric Sounding(MIPAS). SBUV is a nadir sounder andmeasures very thick layers with the verti-cal information in the middle and upperstratosphere. MIPAS is a limb sounderwith much finer vertical resolution andmeasurements extending into the lowerstratosphere. SBUV also measures totalcolumn ozone, which is assimilated in allexperiments. The results from threeassimilation experiments are shownthrough comparison with an ozonesondeprofile. Ozonesondes were not assimilat-

ed into the systems, therefore, these dataprovide an independent measure of per-formance.

There are several attributes to be noted inFigure 5. The quality of the MIPAS-only (+SBUV total column) assimilation is thebest of those presented. This suggests thatthe vertical resolution of MIPAS instru-ment has a large impact. Even though theMIPAS observations are assimilated onlyabove 70 hPa, the assimilation captures theessence of the structure of the ozone pro-file down to 300 hPa. This indicates thatthe model information in the lower strato-sphere and upper troposphere is geophysi-cally meaningful. Further, the model iseffectively distributing information in thehorizontal between the satellite profiles.The comparison with the SBUV-only (+SBUV total column) assimilation showsthat the thick-layered information of theSBUV observations, even in combinationwith the model information, does not rep-resent the ozone peak very well. Thisimpacts the quality of the lower strato-spheric analysis as the column is adjustedto represent the constraints of the totalozone observations. Finally, from firstprinciples, the combined SBUV andMIPAS assimilation might be expected tobe the best since it has the maximumamount of information. This is not foundto be the case, and suggests that the weightsof the various error covariances and theuse of the observations can be improved.The optimal balance of nadir and limbobservations is not straightforward, andthese experiments reveal the challenges thatneed to be addressed when multiple types ofinstruments are used in data assimilation.

Summary

In the past 3-5 years there has beennotable improvement in the representa-tion by data assimilation of the primaryparameters that describe the physical stateof the stratosphere, i.e. wind and temper-ature. There has also been a significantimprovement in the state of the art ofozone assimilation. Derived parameters,such as the residual circulation, have notseen similar improvements. This isattributed to insertion of data into themodel during the data assimilation cycle.Both bias and noise will have a responsethat is realized through changes in thephysical balance of the model being used

36

SPARC_R10 7/27/05 12:50 PM Page 36

for assimilation. As primary variables arepushed closer and closer to observations,this physical response often pushes thederived parameters further from reality. Aproblem that has not been discussed is thedifficult problem of gravity waves andgravity wave dissipation. The data inser-tion process acts as an additional sourceof gravity waves, and hence, there are bothnear field and far field impacts.

The question is raised as to whether or notwe have reached a limit in the transportand climate problems that can beaddressed with assimilated data. Clearly,we have reached a point where the lack ofphysical consistency is impacting the abil-ity to study the problems at hand. This istrue not only in transport studies, but inany studies that require a consistent,closed budget. Furthermore, because ofthe exquisite sensitivity of data assimila-tion systems to the input observations,long-term trends strongly reflect changesin the observing system. Improvement inthe quality of assimilated data systems ismost likely to follow the use of new obser-vations, the reduction of bias, and thedevelopment of bias correction tech-niques. The reduction of bias, and the keyto development of assimilated data setsfor use in chemistry and climate, willdepend on the development of improvedphysical parameterizations. Theseimprovements should include the abilityto directly predict and constrain newobservations of the physical processes.Based on recent experience, the details ofthe assimilation method, i.e. the statisticalanalysis algorithm, and the improvementof the representation of error covarianceare secondary to addressing the problemsof bias and physics.

References

Brasseur, G., and S. Solomon, Aeronomy of the

Middle Atmosphere, D. Reidel Publishing

Company, 452 pp, 1986.

Cohn, S. E., An introduction to estimation the-

ory, J. Met. Soc. Japan, 75 (1B), 257-288, 1997.

Daley, R., Atmospheric Data Analysis,

Cambridge University Press, 457 pp, 1991.

Dee, D. P., and A. da Silva, Data assimilation in

the presence of forecast bias, Q. J. Roy. Meteor.

Soc., 124 (545), 269-295 Part A, 1998.

Dessler, A. E., The Chemistry and Physics of

Stratospheric Ozone, Academic Press, 214 pp, 2000.

Douglass, A. R., M. R. Schoeberl, R. B. Rood,

and S. Pawson, Evaluation of Transport in the

Lower Tropical Stratosphere in a Global

Chemistry and Transport Model, J. Geophys.

Res., 108, 4259, 2003.

Holton, J. R., An Introduction to Dynamic

Meteorology, Elsevier Academic Press, 535 pp,

2004.

Jacobson, M. Z., Fundamentals of Atmospheric

Modeling, Cambridge University Press, 672 pp,

1998 - (2nd Ed, to appear, 2005).

Jackman C. H., P. D. Guthrie, and J. A. Kaye, An

intercomparison of nitrogen-containing species

in Nimbus 7 LIMS and SAMS data, J. Geophys.

Res., 92, 995-1008, 1987.

Kistler R. et al., The NCEP-NCAR 50-year

Reanalysis: Monthly means CD-ROM and doc-

umentation, Bull. Amer. Meteor. Soc. 82, 247-

267, 2001.

Lait, L. R., Systematic differences between

radiosonde measurements, Geophys. Res. Lett.,

29, 1382, 2002.

Lin, S.-J., and Rood, R. B., Multidimensional

Flux-Form Semi-Lagrangian Transport Schemes,

Mon. Wea. Rev., 124, 2046-2070, 1996.

Lin, S. J., A “vertically Lagrangian” finite-vol-

ume dynamical core for global models, Mon.

Wea. Rev., 132, 2293-2307, 2004.

Long, C. S., A. J. Miller, H. T. Lee, J. D. Wild, R. C.

Przywarty, and D. Hufford, Ultraviolet index

forecasts issued by the National Weather Service,

Bull. Amer. Meteorol. Soc., 77, 729-748, 1996.

Molod, A., H. M. Helfand, and L. L. Takacs, The

climatology of parameterized physical processes

in the GEOS-1 GCM and their impact on the

GEOS-1 data assimilation system, J. Climate, 9,

764-785, 1996.

Randall, D. A. (Ed.), General Circulation Model

Development: Past, Present, and Future,

Academic Press, 807 pp, 2000.

Rood, R. B., Numerical advection algorithms and

their role in atmospheric transport and chem-

istry models, Rev. Geophys., 25, 71-100, 1987.

Rood, R. B., D. J. Allen, W. Baker, D. Lamich, and

J. A. Kaye, The Use of Assimilated Stratospheric

Data in Constituent Transport Calculations, J.

Atmos. Sci., 46, 687-701, 1989

Ruhnke, R., C. Weiss; T. Reddmann, and W.

Kouker, A comparison of multi-annual CTM

calculations using operational and ERA-40

ECMWF analyses, Abstract European

Geophysical Society, Nice, 2003.

Schoeberl, M. R., A. R. Douglass, Z. Zhu, and S.

Pawson, A comparison of the lower stratospher-

ic age-spectra derived from a general circulation

model and two data assimilation systems, J.

Geophys. Res., 108, Art. No. 4113, 2003

Simmons, A., Representation of the

Stratosphere in ECMWF Operations and ERA-

40, in ECMWF/SPARC Workshop on Modelling

and Assimilation for the Stratosphere and

Tropopause. 23-26 June, 2003.

http://www.ecmwf.int/publications/library/do/

references/list/17123

Stajner, I., L. P. Riishojgaard, and R. B. Rood,

The GEOS ozone data assimilation system:

Specification of error statistics, Q. J. R. Meteorol.

Soc., 127, 1069-1094, 2001.

Stajner, I., N. Winslow, R. B. Rood, and S.

Pawson, Monitoring of Observation Errors in

the Assimilation of Satellite Ozone Data, J.

Geophys. Res., 109, Art. No. D06309, 2004.

Stohl, A., O. R. Cooper, and P. James, A

Cautionary Note on the Use of Meteorological

Analysis Fields for Quantifying Atmospheric

Mixing, J. Atmos. Sci., 61, 1446-1453. 2004.

Swinbank, R., and W. A. Lahoz (Eds.), Data

Assimilation for the Earth System, NATO Science

Series: IV: Earth and Environmental Sciences,

26, Kluwer, 388 p, 2003.

Tan, W-W., M.A. Geller, S. Pawson and A. da

Silva, A case study of excessive subtropical

transport in the stratosphere of a data assimila-

tion system, J. Geophys. Res., 109, 2004.

Wargan, K., I. Stajner, S. Pawson, R. B. Rood,

and W. –W. Tan, Assimilation of Ozone Data

from the Michelson Interferometer for Passive

Atmospheric Sounding, Quart. J. Royal. Met.

Soc., to appear, 2005.

37

SPARC_R10 7/27/05 12:50 PM Page 37

The SPARC Polar Stratospheric Cloud Assessment

The SPARC Polar Stratospheric Cloud Assessment (SPA) has kicked off with a chapter scoping meeting at the Coolfont (Spa) resort in WestVirginia, USA from 12-13 May. The meeting was kindly hosted by our local organiser Mike Fromm of the US Naval Research Laboratory (NRL).

The aim of SPA is to assess our understanding of polar stratospheric clouds (PSCs). The motivation for the assessment is that there remaingenuine gaps in our understanding of PSC distribution, formation and long term change that are important to stratospheric chemistry.Our understanding is patchy and specific rather than global and integrated and there has been a tendency to undertake detailed processstudies rather than integrative studies. There is also no consensus on how to describe PSCs and denitrification in global models, whichmeans we are unable to reliably predict changes that might occur in a future stratosphere. An important factor that limits our under-standing is the lack of a large-scale consistent and evaluated dataset for model testing, the creation of which is one aim of SPA.

The purpose of the meeting was to agree on the organisation of the chapters and distribute writing tasks. We also managed to create an additionalchapter on meteorological processes and agreed to produce a “Twenty Questions and Answers About PSCs” document for the stratospheric com-munity, along the lines of David Fahey’s excellent WMO Ozone Assessment pamphlet. If written well, we believe that such a document will allowthe stratospheric community to better comprehend the rather acronym burdened world of PSCs.

The chapters are as follows:

1. PSC processes (Niels Larsen)This chapter will set the scene and also serve to pull the whole assessment together.

2. Temperatures and meteorological diagnostics for PSC studies (Gloria Manney and Steve Eckermann)This chapter was seen as a very important addition to SPA and was created during the meeting. Much of the analysis of PSC occurrenceand the interpretation of datasets in terms of microphysical quantities requires accurate meteorological analyses. We also know thatmesoscale processes play an important role in forming solid PSCs. The aim of this chapter will be to assess how accurate the meteoro-logical analyses need to be and what effect uncertainties have on our interpretation of PSC measurements.

3. PSC detection and discrimination (Beiping Luo and Mike Fromm)The overall aim of this chapter is to define the range of PSC properties that can be detected by different instruments. Statements havepreviously been made about the frequency of occurrence of various cloud types that may have more to do with instrument detectionthresholds than with actual PSC existence, particularly when it comes to the existence of a few large solid particles. This chapter will pro-vide a consistent way of converting PSC signatures into ranges of microphysical properties, where possible.

4. PSC observations and their interpretation (Terry Deshler and Lamont Poole)This is the chapter that most closely resembles a PSC climatology. Climatologies have been produced in the past. The aim here is to refinethem to provide information about different types of PSC.

5. Denitrification and dehydration observations (Michelle Santee and Gerald Nedoluha)The occurrence of denitrification and dehydration is closely related to that of PSCs. This chapter will mirror chapter 4 by examining thesatellite and in situ record of denitrification and dehydration.

6. Using models to assess our understanding of PSCs (Katja Drdla and Ken Carslaw)How good are the different PSC models that are used? Models range from those that include, as far as possible, all the microphysical pro-cesses to those that parametrise PSC processes for use in large scale models. This chapter will assess these models, including denitrifica-tion and dehydration, for a few well defined cases based on the data from chapters 4 and 5. An important aspect will be the extent towhich laboratory measurements of PSC formation, models and field observations are consistent.

7. PSCs in a changing stratosphere (Markus Rex and Richard Bevilacqua)The aim of this chapter is to understand what controls long term changes and interannual variability in PSCs. It will also explore sim-ple ways for global models such as CCMs to test whether they have the right physics to correctly predict long term changes that are rel-evant to ozone depletion studies.

SPA has ambitious aims that go beyond a straightforward review of understanding. A 10 month period was felt to be essential for all par-ticipants to make substantial progress, particularly in cases where large datasets need to be assembled and analysed. Our next meeting isplanned for early spring 2006 in Europe.

Co-chairs: Ken Carslaw, University of Leeds, UK (carslaw@env.leeds.ac.uk)Katja Drdla, Ames Research Centre, USA (Katja.Drdla-1@nasa.gov)

Meeting participants: Chapter lead authors (Niels Larsen, Gloria Manney, Terry Deshler, Lamont Poole, Beiping Luo, Mike Fromm,Michelle Santee, Gerald Nedoluha, Richard Bevilacqua) and coauthors (Harald Flentje, John Remedios, Christiane Voigt, HideakiNakajima, Jerome Alfred, Tony Strawa)

38

Update

SPARC_R10 7/27/05 12:50 PM Page 38

Gérard Mégie died at the age of 58,following an extended illnesswhich did not prevent him from

assuming his duties as Chairman of CNRSuntil the last minute. He left a feeling ofemptiness in the numerous national,European and international organisationswhere he was active.

Gérard Mégie was well known in theSPARC community. The bulk of hisresearch dealt with the development oforiginal methods for measuring atmo-spheric variables by lidar, and the mod-elling of the natural variability of ozoneand how it is influenced by human activi-ties. He participated in the implementationof many different means for observing theatmosphere from the ground and fromother platforms (airplane, balloon, satel-lite) and played a leading role in setting upthe Network for Detection of StratosphericChanges (NDSC).

Besides his increasing responsibilities in thenational research community, he chairedthe International Ozone Committee of the

International Council for Science (ICSU)from 1988 to 1996, the Earth ObservationCommittee of the European Space Agency(ESA) from 1994 to 1999, and theStratosphere Scientific Committee of theEuropean Commission (1989-2004). He co-chaired the International ScientificCommittee of the Montreal Protocol for theprotection of the ozone layer and theWMO-UNEP Scientific Assessment ofOzone Depletion. He became a member of

the European Research Advisory Board(EURAB) of the European Commission in2001, and a member of the EuropeanResearch Council Expert Group (ERCEG)of the European Union in 2003.

Gérard Mégie was the author of more than240 scientific publications and two bookson stratospheric ozone: ‘Stratosphère etCouche d’Ozone’ (Editions Masson) in1991 and ‘Ozone, l’Equilibre Rompu’(Presses du CNRS) in 1989. He was also thescientific coordinator of two reports of theFrench Academy of Sciences: ‘Ozone etPropriétés Oxydantes de la Troposphère’(1993), and ‘Ozone Stratosphérique’(1998) published by Editions Lavoisier.

Gérard Mégie was an internationally rec-ognized scientist as well as a highly culti-vated person, a humanist and a teacher,appreciated by everyone for his open-mindedness, his intellectual disciplineand his keen analytical ability. We misshim greatly.

Marie-Lise Chanin 39

In Memory of Gérard Mégie 1946-2004

The New SPARC Data CenterMarvin Geller, Stony Brook, New York, USA (marvin.geller@sunysb.edu)

Dr. Stefan Liess from Stony Brook University has been serving as the SPARC DataCenter Scientist at http://www.sparc.sunysb.edu since December 2004. Stefan Liesswill continue the work of his predecessors Petra Udelhofen and Xuelong Zhou inworking with the SPARC projects to acquire newly available data as needed, as well asto maintain archives of data that have been used in past SPARC projects so that theymay be acquired as needed by the world community, and also be available for updat-ing should the SPARC projects desire to do so. Subject to funding, plans for the nearfuture include a modernization of hardware facilities in order to accommodate thegrowing demand for data storage. This demand arises from several sources, e.g. froman ongoing collection of high-resolution radiosonde data from 93 U.S. stations, aplanned chemistry-climate model intercomparison project (CCMVal), and to pro-vide storage space for SPARC-IPY data. The implementation of online plotting inorder to ease data access and decrease data transfer is also being planned. Please con-tact Stefan Liess at stefan.liess@stonybrook.edu for any questions regarding theSPARC Data Center. The SPARC Data Center has been funded by NASA grants toStony Brook University with Marvin Geller as Principal Investigator. A new proposalis being prepared for submission to NASA for continued support.

SPARC Data Center Website: http://www.sparc.sunysb.edu

Announcement

SPARC_R10 7/27/05 12:50 PM Page 39

W. Randel (USA), T.G. Shepherd (Canada). Stratosphere-Troposphere Dynamical Coupling: M. Baldwin (USA), S. Yoden (Japan)