Atmos.Chem.Phys.,14,10931–10960,2014 www.atmos-chem-phys.net/14/10931/2014/ doi:10.5194/acp-14-10931-2014 ©Author(s)2014.CCAttribution3.0License. The BLLAST field experiment: Boundary-Layer Late Afternoon and Sunset Turbulence M.Lothon1,F.Lohou1,D.Pino2,24,F.Couvreux3,E.R.Pardyjak4,J.Reuder5,J.Vilà-GueraudeArellano6, PDurand1,O.Hartogensis6,D.Legain3,P.Augustin7,B.Gioli8,D.H.Lenschow9,I.Faloona10,C.Yagüe11, D.C.Alexander4,W.M.Angevine12,EBargain1,J.Barrié3,E.Bazile3,Y.Bezombes1,E.Blay-Carreras2, A.vandeBoer6,25,J.L.Boichard13,A.Bourdon14,A.Butet14,B.Campistron1,O.deCoster6,J.Cuxart15,A.Dabas3, C.Darbieu1,K.Deboudt7,H.Delbarre7,S.Derrien1,P.Flament7,M.Fourmentin7,A.Garai16,F.Gibert17,A.Graf18, J.Groebner19,F.Guichard3,M.A.Jiménez20,M.Jonassen5,A.vandenKroonenberg21,V.Magliulo26,S.Martin22, D.Martinez15,21,L.Mastrorillo13,A.F.Moene6,F.Molinos15,E.Moulin3,H.P.Pietersen6,B.Piguet3,E.Pique1, C.Román-Cascón11,C.Rufin-Soler23,F.Saïd1,M.Sastre-Marugán11,Y.Seity3,G.J.Steeneveld6,P.Toscano8, O.Traullé3,D.Tzanos3,S.Wacker19,N.Wildmann21,andA.Zaldei8 1Laboratoired’Aérologie,UniversityofToulouse,CNRS,France 2AppliedPhysicsDepartment,BarcelonaTechUPC,Barcelona,Spain 3CNRM-GAME(UMR3589,Météo-FranceandCNRS),Toulouse,France 4UniversityofUtah,SaltLakeCity,Utah,USA 5GeophysicalInstitute,UniversityofBergen,Bergen,Norway 6MeteorologyandAirQualitySection,WageningenUniversity,Wageningen,theNetherlands 7LaboratoiredePhysiqueetChimieAtmosphériques,UniversitéduLittoralCôted’Opale,Dunkerque,France 8InstituteofBiometeorology–NationalResearchCouncil(IBIMET-CNR),Florence,Italy 9NationalCenterforAtmosphericResearch,Boulder,Colorado,USA 10Land,AirandWaterResources,UCDavis,California,USA 11Dpt.GeofísicayMeteorología,UniversidadComplutensedeMadrid,FacultadCienciasFísicas,Madrid,Spain 12CIRES,UniversityofColorado,andNOAAESRL,Boulder,ColoradoUSA 13SEDOO,OMP,Toulouse,France 14ServicedesAvionsFrançaisInstrumentéspourlaRechercheenEnvironnement,CNRS-CNES-Météo-France, Francazal,France 15DepartamentdeFisica,UniversitatdelesIllesBalears,PalmadeMallorca,Spain 16MechanicalandAerospaceEngineering,UniversityofCalifornia,SanDiego,California,USA 17LaboratoiredeMétéorologieDynamique,EcolePolytechnique,Palaiseau,France 18InstitutfürBio-undGeowissenschaften,Juelich,Germany 19PMOD-WRC,DavosDorf,Switzerland 20MediterraneanInstituteforAdvancedStudies(UIB-CSIC),Esporles,IllesBalears,Spain 21UniversityofTübingen,Tübingen,Germany 22TechnischeUniversitaetBraunschweig,Braunschweig,Germany 23InstitutdeRecherchesenENvironnementIndustriel(IRENI),Dunkerque,France 24InstitutofSpaceStudiesofCatalonia(IEEC-UPC),Barcelona,Spain 25MeteorologicalInstitute,UniversityofBonn,Bonn,Germany 26InstituteofMediterraneanAgriculturalandForestSystems–NationalResearchCouncil(ISAFOM-CNR),Naples,Italy Correspondenceto:M.Lothon([email protected]) Received:23March2014–PublishedinAtmos.Chem.Phys.Discuss.:29April2014 Revised:8August2014–Accepted:5September2014–Published:16October2014 PublishedbyCopernicusPublicationsonbehalfoftheEuropeanGeosciencesUnion. 10932 M.Lothonetal.:TheBLLASTfieldexperiment Abstract. Due to the major role of the sun in heating the cycle. On a fair weather day, as the sun rises, the surface earth’s surface, the atmospheric planetary boundary layer heatingwarmstheairabove,whichmixesbyturbulentpro- overlandisinherentlymarkedbyadiurnalcycle.Theafter- cesses within an increasingly deep layer, engulfing air from noontransition,theperiodofthedaythatconnectstheday- thefreeatmosphereabove(Stull,1988;Garratt,1992).Con- time dry convective boundary layer to the night-time stable versely,duringthenight,theradiativelycooledsurfacestrat- boundary layer, still has a number of unanswered scientific ifiestheairabove,whichformsastablenocturnalboundary questions.Thisphaseofthediurnalcycleischallengingfrom layer.Bothmiddayandnocturnalperiods,wheninastation- both modelling and observational perspectives: it is transi- ary state, have been relatively successfully modelled, even tory,mostoftheforcingsaresmallornullandtheturbulence ifseveralissuesremainopen(seethereviewsbyAngevine, regimechangesfromfullyconvective,closetohomogeneous 2008;Cuxart,2008;andHolstlagetal.,2013).Morningand andisotropic,towardamoreheterogeneousandintermittent evening transitions remain difficult to observe and model, state. in large part due to their inherent transience. The late af- These issues motivated the BLLAST (Boundary-Layer ternoon transition typically starts from a well-mixed con- Late Afternoon and Sunset Turbulence) field campaign that vective boundary layer (CBL) and transforms to a residual was conducted from 14 June to 8 July 2011 in southern layeroverlyingastably-stratifiedsurfacelayer.Thisevolving France, in an area of complex and heterogeneous terrain. boundarylayerexhibitscomplexcharacteristicssuchastur- A wide range of instrumented platforms including full-size bulence intermittency and enhancement of anisotropy, hori- aircraft, remotely piloted aircraft systems, remote-sensing zontal heterogeneity, rapidly changing conditions and com- instruments, radiosoundings, tethered balloons, surface flux binationsofweakforcingmechanisms. stations and various meteorological towers were deployed TheevolutionofthePBLhasbeenstudiedsincethe1950s. over different surface types. The boundary layer, from the AnextensiveknowledgeofthediurnalevolutionofthePBL earth’s surface to the free troposphere, was probed during and its influence on the pollutant distribution has been ob- the entire day, with a focus and intense observation periods tained since then (Vilà-Guerau de Arellano et al., 2004, thatwereconductedfrommiddayuntilsunset.TheBLLAST 2009; Casso-Torralba et al., 2008). The increasing knowl- field campaign also provided an opportunity to test innova- edgeofPBLprocesseshasbeenbasedontwomaintypesof tivemeasurementsystems,suchasnewminiaturizedsensors, studies: the application of the theoretical concepts of turbu- andanewtechniqueforfrequentradiosoundingsofthelow lence (Batchelor, 1967; Tennekes and Lumley, 1973; Pope, troposphere. 2000; Wyngaard, 2010) to perform numerical simulations Twelvefairweatherdaysdisplayingvariousmeteorologi- ofatmosphericcharacteristics(Lilly,1967;Deardorff,1972; calconditionswereextensivelydocumentedduringthefield Lenschow,1974;Stull,1976;Moeng,1984;Jacobson,2000; experiment.Theboundary-layergrowthvariedfromoneday Pielke,2002;Stensrud,2007),anddetailedfieldobservations to another depending on many contributions including sta- (e.g.Wangara:1967,Kansas:1968orMinnesota:1973,de- bility, advection, subsidence, the state of the previous day’s scribedinHessetal.,1981andKaimalandWyngaard,1990, residuallayer,aswellaslocal,meso-orsynopticscalecon- remain fundamental references). There have been a large ditions. numberofintensivefieldexperimentssincethen,andinad- Ground-based measurements combined with tethered- dition, systematic observations now made at some observa- balloon and airborne observations captured the turbulence toriesallowtheexplorationofthePBLonalong-termbasis decayfromthesurfacethroughoutthewholeboundarylayer aswell:forexample,atLindenberg,Germany(Beyrichand anddocumentedtheevolutionoftheturbulencecharacteris- Engelbart, 2008), Cabauw, the Netherlands (Van Ulden and ticlengthscalesduringthetransitionperiod. Wieringa, 1996; Hurley and Luhar 2009; Baas et al., 2009; Closely integrated with the field experiment, numerical Bosveld et al., 2014) and CIBA, Spain (Yagüe and Cano, studiesarenowunderwaywithacompletehierarchyofmod- 1994),aswellasfluxmonitoringnetworksworldwide. elstosupportthedatainterpretationandimprovethemodel Most PBL studies were previously devoted to investigat- representations. ingthePBLcharacteristicsandtherelevantprocessesduring midday,whenunstableorneutralconditionsusuallyprevail (Kaimaletal.,1976;MahrtandLenschow,1976;Stull,1988; Moeng and Sullivan, 1994; Cuijpers and Holtslag 1998), or 1 Introduction atnightwhenastableatmosphereistypicallyfound(Nieuw- stadt, 1984; Debyshire, 1990; Garratt 1992; Cuxart et al., Atinterfacebetweentheearth’ssurfaceandtheatmosphere, 2000; Poulos et al., 2002 van de Wiel et al., 2003; Mahrt, the planetary boundary layer (PBL) is a critical component 2014). Limited-area and global meteorological models, as of the earth system. It mediates the transfer of heat, mo- wellasairqualitymodelshavelargelybenefitedfromthese mentum, humidity and trace gases between the surface and the atmosphere. The PBL over land has a strong diurnal Atmos.Chem.Phys.,14,10931–10960,2014 www.atmos-chem-phys.net/14/10931/2014/ M.Lothonetal.:TheBLLASTfieldexperiment 10933 investigationsbyintroducingnewprocess-basedparameteri- Thisgeneralmanuscriptthereforeintroducesthedeeperanal- zations. ysesmadeonspecificissuesthataremadeintheotherarti- As early as the late 1970s, though, André et al. (1978) clesofthespecialissue. compared a third-order moment model with ground-based measurements and soundings of the boundary layer during anentirediurnalcycle.Difficultieswerefoundinthenoctur- 2 Addressedissues nalconditionsandduringthelateafternoontransition.Sev- eralrecentstudieshaveattemptedtosimulatetheentirediur- This section reviews the previous studies that are address- nalcyclebothwithlarge-eddysimulation(LES)andsingle- ing the afternoon transition and turbulence decay. We first columnparameterizedmodels(SCM).TheseincludeKumar remind several definitions proposed in the literature for the et al. (2006), Basu et al. (2008) or Svensson et al. (2011), periodandlayersofinterest,theninvestigatethepastresults whomadeuseofrealisticconditionsbasedontheHorizontal on the turbulence decay process and finally discuss the po- ArrayTurbulenceStudy(HATS,Horstetal.,2004),Wangara tential impacts of the transition and benefits from improved andCASES-99campaigns,respectively.Beareetal.(2006) understanding. andEdwardsetal.(2006)comparedsurfaceobservationsat Cardington, UK, with respectively a LES and a SCM from 2.1 “Convective”,“mixed”or“residual”layers? earlyafternoontothenextmorning.Thelateafternoontran- Definitionandscaling sitiondecaywasdelayedintheLESrelativetotheobserva- tions,butalargeimprovementwasfoundwhenassimilating Definitionsoftheafternoontransition(AT)andtheevening the observations. The single-column model had difficulties transition (ET) (and distinctions between them) may vary forcorrectlyrepresentingturbulencediffusionduringtheaf- according to previous studies. In the study by Nadeau et ternoontransition,whichaffectedthemeanprofiles.Mostof al.(2011),theATstartsassoonasthesurfacesensibleheat thenumericalsimulationsquotedaboveareabletoreproduce flux begins to decrease and ends when it becomes negative. themulti-layeringthatoccursintheeveningandthegenera- Grimsdell and Angevine (2002) have used different subjec- tionofanocturnaljet,butthetransitiontimingremainshard tive criteria based on UHF wind profiler measurements, in to catch for several important variables (including surface order to analyse the behaviour of the CBL top (estimated fluxes, mean wind and temperature, and friction velocity). from the reflectivity) with respect to the depth of the layer In addition, most of the simulations described above could with a significant amount of turbulence (estimated from the only be compared with surface measurements of fluxes and spectral width). In their study, the AT start is defined as (i) turbulence and with vertical profiles of mean variables, but thetimewhentheverticalstructureclosetothetopbeginsto rarelywithturbulenceobservationsuptothePBLtop. “decouple”ortheturbulencestartstodecayatthetopor(ii) There are still relatively few observational studies dedi- thetimewhentheCBLtopstartstodescend.(i)or(ii)were catedtothetransitoryprocessesinthecloud-freeorshallow- considered on distinct days, depending on the behaviour of convective PBL, e.g. Grant (1997) (in Cardington, UK), theCBLtop:(i)wasusedforcaseswithan“inversionlayer Brazel et al. (2005) (Phoenix Air Flow Experiment), Fer- separation”and(ii)forcaseswithadescentoftheCBLtop. nando et al. (2004), Fernando et al. (2013) (The Phoenix Defined as such, the AT usually lasts several hours. Grims- Evening Transition Flow Experiment). Also notable are the dellandAngevine(2002)foundthatthetransitionwasgrad- LIFT/FLATLAND experiment (Cohn et al., 2002) in the ualandnotsudden,fromaCBL-topperspective.TheETis plains of Illinois, LITFASS (Beyrich et al., 2006) over het- usuallydefinedastheperiodoftimefromzerosurfacesensi- erogeneous surface in Germany, and CASES-99 (Poulos et bleheatfluxtoawell-establishednocturnalstablelayer,with al., 2002) in Kansas for the study of the nocturnal stable quasi-steadydepth. boundary layer. Without being specifically dedicated to the InthecontextoftheATandET,thedefinitionsofthesur- afternoon and evening transitions, these observational cam- facelayer,themixedlayer(andCBL),theresiduallayerand paignswerethebasisofkeystudiesonthelateafternoonor the nocturnal stable boundary layer have to be carefully re- eveningtransitions. visited. Theresultsbasedonthepreviouslymentionedcampaigns Criteria typically used to define the depth of the CBL and on numerical experiments revealed some key issues of during midday are, among others, the depth of well-mixed thelateafternoontransition,whichwerechosenastheguide- scalars, the depth of significant turbulence, the depth of in- linefortheBoundary-LayerLateAfternoonandSunsetTur- creasing relative humidity, the height of the capping inver- bulence (BLLAST) project. In the following section, we sion or of minimum buoyancy flux (Angevine et al., 1994; presentinmoredetailtheissuesraisedbytheafternoontran- MoengandSullivan,1994;Seibertetal.,2000;ZhuandAl- sition,basedonthebackgroundofpreviousstudies.Section brecht, 2002; Brooks and Fowler, 2011). These criteria all 3describesindetailtheexperimentalset-upandstrategythat find approximately the same depth in a well-defined CBL, were chosen to address those issues, and Sect. 4 points out but they start to evolve differently during the AT and may thepotentialoftheBLLASTdatasettobringsomeanswers. separatefromeachotherasobserved,e.g.byGrimsdelland www.atmos-chem-phys.net/14/10931/2014/ Atmos.Chem.Phys.,14,10931–10960,2014 10934 M.Lothonetal.:TheBLLASTfieldexperiment Angevine(2002):thedepthoftheCBLmaydecrease,while timescale t∗=Zi/w∗, where Zi is the CBL depth, and w∗ theresidualinversionremainslevelorevolvesonitsownde- istheconvectivevelocityscale(Deardorff,1970;Willisand pendingonadvectionandsubsidence. Deardorff,1976).Inthiscontext,thepowercoefficientnisa In unstable conditions, the surface layer is mainly gov- functionofτf/t∗. erned by shear and buoyancy, and the outer layer above is Recently, Nadeau et al. (2011) considered a realistic de- governedbybuoyancy.Consequently,duringtheday,incon- crease of the surface sensible heat flux, based on observa- vective conditions, most of the boundary-layer processes in tions of the LITFASS-2003 experiment (Beyrich and Men- the outer layer can typically be scaled based on the surface gelkamp,2006).TheyshowedthattheTKEdecayphasecan buoyancyfluxandtheboundary-layerheight(Deardorffscal- be separated in two stages: first, a slow decay during the ing,Deardorff,1970;WillisandDeadorff,1976).Inthesur- AT followed by a rapid collapse of turbulence during the face layer, the Monin–Obukhov similarity theory (MOST, ET. Also Nadeau et al. (2011) were able to model the de- MoninandObukhov,1954)hasbeenwidelyused.Bothscal- cay observed in the surface layer with a model based on a ings are the basis for robust parameterizations in bulk and mixed-layerparameterization,ratherthanonasurface-based mesoscalemodels.However,duringtheafternoontransition, parameterization.BasedontheCASES-99dataset,Rizzaet thesurfacebuoyancyfluxdecreasestowardzero,andthein- al.(2013)performedaLESstudyofthedecayphasewhose fluenceofothercompetingprocessesasradiation,advection, resultscorroboratethefindingsofNadeauetal.(2011). entrainmentorwindshearbecomerelativelymoreimportant. In both laboratory experiments and numerical studies, So neither the convective scaling, nor the MOST-based sta- suchasthosementionedabove,thedecayoftheturbulentki- bleboundary-layerscaling,arevalid.Itisthereforenecessary neticenergyisfoundtodependontheformulationofthede- toexplorethevalidityofconvectiveandstablescalings,and creaseinthesurface–atmosphereenergyexchanges(e.g.ei- how to represent the transition using non-dimensional anal- ther expressed as prescribed surface sensible heat fluxes or ysis or new scalings. In this context, van Driel and Jonker surfacetemperature),butwithnoconsensusontheexactre- (2011), based on an idealized LES and 0-D model study of lationshipbetweentheforcingandthepowerlaw. a non-stationary PBL, suggest considering the time it takes On the observational side, Fitzjarrald et al. (2004) pro- fortheenergytotravelfromthesurfaceuptothetopofthe vided aircraft measurements of the turbulence decay within boundary layer. McNaughton et al. (2007), Sorbjan (2010, the PBL, and revealed a sharper and more systematic de- 2012) and Kumar et al. (2006) also proposed new scalings cay of the wind vertical velocity relative to the horizontal that could be tested in the context of transitory phases, like components. Most of the other previous observational stud- the local Richardson number and Nieuwstadt scalings. A ieshavefocusedonthedecayoftheTKEinthesurfacelayer questionthatisstillpoorlyunderstoodisthefollowing:how (e.g. Fernando et al., 2004; Brazel et al., 2005), with little longdoestheCBLremainquasi-stationaryduringtheAT,or, quantificationofhowturbulenceisdecayingintheupperlev- equivalently,forhowlongdoestheconvectivescalingapply els,andhowthedifferentlevelsinteractwitheachother. asthesurfacefluxdecreases? 2.2.2 Theevolutionoflengthscales 2.2 Turbulencedecayprocess Characteristic scales of turbulence are relevant for under- 2.2.1 Turbulencekineticenergy(TKE)decay standingandquantifyingPBLprocessesandtheirrepresen- tation in meteorological models. Various length scales can Several authors have previously studied the transition beconsideredtocharacterizeturbulenceprocesses,withdif- regimes of turbulence with laboratory experiments ferent ways to estimate them including the wavelength of (e.g. Monin and Yaglom, 1975; Cole and Fernando, the energy spectrum peak (energy production), the integral 1998). The first LES study of the decaying atmospheric scale(energy-containingeddies)orotherscalesdefinedwith convective mixed layer was performed by Nieuwstadt and a weighted integral of the spectrum, and also the buoyancy Brost (1986). The authors analyzed an academic case of a length scale, the Ozmidov scale (that is the scale where the sheared, clear mixed layer, in which turbulence decayed as buoyancy forces affecting the vertical momentum are equal a result of a sudden shut-off of the upward surface sensible to the inertial forces; Fernando, 1991), etc. During midday, heat flux. In both the LES simulations and the laboratory thoseareoftenproportional(LenschowandStankov,1986), experiments, the turbulent kinetic energy is found to decay but this is not expected to remain valid in the late after- followingapowerlawt−noftimet. noon.Aspartsoftheboundarylayerbecomestablystratified, Later, Sorbjan (1997) considered a gradual change of the the buoyancy length and Ozmidov scale (Fernando, 1991), heatfluxwithtime,inresponsetothedecreasingoftheeleva- etc., becomerelevant. Forthe Phoenix AirflowExperiment, tionofthesun.Theevolutionofthedecayingshearedmixed the observations of Pardyjak (2001) indicate that these two layer was shown to be governed by two timescales: the ex- scalesdecreasequitelinearlyinthehoursfollowingET. ternal (or “forcing”) timescale τ – that is the timescale of Indeed,thereisalackofagreementintheevolutionofthe f thegraduallychangingoftheheatflux–andtheconvective vertical velocity characteristic length scale during the late Atmos.Chem.Phys.,14,10931–10960,2014 www.atmos-chem-phys.net/14/10931/2014/ M.Lothonetal.:TheBLLASTfieldexperiment 10935 afternoon transition, partly due to the difficulty of address- 2.2.3 Competinginfluences:“theunforcedtransition” ingtheissue,bothwithnumericalstudiesandobservations. Vertical motions up to 1m s−1 extending horizontally over Thedecayofturbulenceandtheevolutionofthecharacteris- several km have been observed, weaker but of larger scale tic length scales need to be related to the relevant forcing than the midday eddies (Aupetit, 1989). Possible explana- mechanisms, not only to the rate of surface buoyancy de- tions for those include growth of boundary-layer scales, or crease,butalsotocompetitiveforcesorprocessesgenerated surfacevariabilityandorographythatcaninducemesoscale byclouds,entrainment,radiativeprocesses,shearandadvec- circulations. tion. Angevine (2008) suggests the term of “unforced tran- ByusingLES,NieuwstadtandBrost(1986)foundthatthe sition”,becausethoseprocessesareusuallyweakduringthe lengthscaleoftheverticalvelocityspectrumpeakremained laterpartoftheAT,butallmaycomeintoplay. constant during the decay process. The study by Sorbjan ThefollowingquestionsareraisedbytheATandETperi- (1997)mentionedpreviouslyreflectedthatsmalleddieshad ods: atendencytodecayearlierthanlargeeddies.Consequently, – HowdoesentrainmentevolveduringtheAT?Whatisits organized convection persisted in the decaying mixed layer role in the afternoon transition? Nieuwstadt and Brost even when the buoyancy flux at the surface became nega- (1986) suggested that large eddies are still active for tive,andanocturnalinversionwasbeingdevelopednearthe sometimeindrivingentrainmentatthetopoftheresid- earth’ssurface.Theseresultswerelaterconfirmedbythedi- ual layer, in spite of the decoupling from the surface. rectnumericalsimulationofShawandBarnard(2002). ThiswascorroboratednumericallybyPinoetal.(2006), Pinoetal.(2006)haveshownthatthecharacteristiclength butstillneedstobeconfirmedbyobservationsandfur- scale, based on a weighted integral of the energy spectrum, therstudy.Canutetal.(2012)withaLES,foundanin- has a different evolution during the decay. They found that creaseintheentrainmentrateinthelateafternoon.The the characteristic length scales increase with time, for all evolution of entrainment has to be linked to the evolu- variables but the vertical velocity, for which the scale re- tion of scales. Van Heerwaarden et al. (2009) and Lo- mainedalmostconstant.Basedontethered-balloonobserva- houetal.(2010)haveshownhowentrainmentcanhave tions, Grant (1997) showed that the peak of the vertical ve- impactdowntothesurface,withsignaturesonevapora- locityspectrashiftstosmallerlengthscalesduringtheETin tionorintegralscales,respectively.Thus,theevolution thesurfacelayer,andremainssteadyabove. of the entrainment process needs to be linked with the WiththeTKEdecayitself,theevolutionofthecharacter- evolutionoflengthscalesthroughouttheentiredepthof istic length scales has been one of the main questions ad- theboundarylayer. dressedinthepaststudiesontheafternoontransition.How- ever, the scale issue remains unclear and only partly under- – Whatistheinfluenceofradiationinthedecayprocess? stood.Athoroughinvestigationofwhetherthescalesinthe Since the surface buoyancy flux is weak, radiation di- mixed (and then residual) layer really increase or decrease vergencecanmakeasignificantcontributionduringthis is necessary. In addition, it must be understood whether the period, both at the surface and at the top of the mixed characteristic length scales decrease in the surface layer as layer(Steeneveldetal.,2010). the nocturnal boundary layer starts to build, as stated by KaimalandFinnigan(1994). – Whatistheroleofland-useandsurfaceheterogeneityin Anotherimportantrelatedquestionistheanisotropyofthe theevolutionofturbulenceintensityandscales?Howdo turbulence.Fitzjarraldetal.(2004)withfluxtowersandair- theheatstorageinthegroundorvegetationcanopyand craftmeasurementsandPinoetal.(2006)bymeansofLES radiative long-wave and short-wave components come showed that the turbulence does not relax to an isotropic into play? Pardyjak and Fernando (2009) and Nadeau state during the decay process. Contrarily, Monin and Ya- et al. (2011) have studied the turbulence decay in the glom(1975)foundinlaboratoryexperimentsthattheturbu- surfacelayeroverseveraltypesofsurfaceandproposed lencemaintainstheinitialisotropyduringthedecay.Lothon a simple model for the decay in the convective surface etal.(2006)havefoundwithmiddaylidarobservationsinthe layer. But the role of surface heterogeneity on the dy- CBLthattheratiobetweenlongitudinal(i.e.alongthesam- namics of the decaying CBL has still not been suffi- plingdirection)andtransverse(i.e.perpendiculartothesam- cientlyaddressed. pling direction) vertical velocity integral scales was smaller thanitwouldbeinisotropicturbulence,i.e.theturbulenceis – How do the processes of the AT and ET interact “squashed”.ThesurfacelayerdatafromPardyjak(2001)also with the flow reversal that occurs in mountainous or indicated that vertical turbulence was damped and isotropy coastal areas, forced by mesoscale pressure and tem- rapidlyincreased.However,itremainsunclearhowsquashed perature gradients? Recently, the TRANSFLEX (The itremainslateranduntilsunset. PhoenixEveningtransitionFlowExperiment;Fernando etal.,2013)andMATERHORN(FernandoandPardy- jak, 2013) experiments addressed the issue of the flow www.atmos-chem-phys.net/14/10931/2014/ Atmos.Chem.Phys.,14,10931–10960,2014 10936 M.Lothonetal.:TheBLLASTfieldexperiment reversal over mountain slopes during the evening tran- flows(Soleretal.,2002)andofthenocturnaljet(Bantaetal., sition. With tethered-balloon observations and tracers 2003), and proposing explanations for turbulence intermit- along the slopes, Fernando et al. (2013) showed the tency(vandeWieletal.,2002a,b;Sunetal.,2003;Cuxart complexityoftheflowadjustment,withthegeneration and Jiménez, 2007). CASES-99 also nicely documents the of multiple fronts in the flow reversal process. The AT eveningtransition.Lundquistetal.(2003)forexamplerevis- and ET in complex terrain need to be specifically ad- ited the explanations and occurrence of inertial oscillations. dressed,sincetheyprecedetheshiftingofavalleywind However,theroleoftheATandETinsettingauspiciousor circulation,orseabreeze. unfavourableconditionsfortheappearanceofthenocturnal jet and occurrence of turbulence intermittency still needs to 2.3 Potentialimpacts befurtheraddressed. Finally, the AT and ET may have important impacts on the transport,mixinganddistributionoftracespecies,theset-up 3 TheBLLASTfieldexperiment ofanocturnaljet,oronthedaytimegrowthofthefollowing- dayPBL. The issues presented above motivated several research groups(listedinTable1)toplanandexecuteadedicatedfield Whatistheimpactofthistransitionon experimentthatfocusedontheafternoonandeveningtransi- thetransportofscalarspecies? tions, with a dense array of complementary observations in timeandspacefromthemid-afternoontothenight. During the evening transition, Acevedo and Fitzjarrald TheBLLASTfieldcampaigntookplaceinearlysummer, (2001)reportedoccurrencesofspecifichumidityjumps,and from 14 June to 8 July 2011 in France. The site is called dropsinsurfacetemperature,accompaniedbyanabruptde- “PlateaudeLannemezan”,aplateauofabout200km2area,a cayinwindvelocity.Similarly,Mahrtetal.(1999)observed fewkilometrefromthePyreneanfoothills(Fig.1),andabout thatthelatentheatfluxduringeveningeventsdecreasedmore 45km from the highest peaks of the Spanish border. The slowlythanthestrengthofturbulenceandtheboundary-layer surface is covered by heterogeneous vegetation: grasslands, depth. This led to the significant moistening of the surface meadows,cropsandforest(Fig.2).Thecampaigncombined layer.ThiswasalsorecentlyreportedbyBoninetal.(2013) in situ measurements from towers, balloons and airplanes withunmannedaerialsystems. withground-basedremotesensing.Themeasurementswere Recent studies (Vilà-Guerau de Arellano et al., 2004; intensifiedduringtheATondayswithfavourableconditions Casso-Torralba et al., 2008) have shown that morning and (discussedlaterinthetext),calledintensiveobservationpe- afternoon transition are also important for the exchange of riods(IOPs). species.Inearlymorning,whenhighentrainmentrateshave Twosites(hereafter“sites1and2”)containedmostofthe beenobserved,theremainingpollutantsoftheresiduallayer ground-basedinstrumentsandwerethefocusofflightoper- areintroducedintheshallowboundarylayer,thusincreasing ations.Thereweretwomainobservationalstrategies,which ordecreasingtheirconcentration.Intheevening,theresidual focused on (1) vertical structure and (2) spatial heterogene- partoverlyingthestablelayercanbeincorporatedinthefree ity.Athirdsupportingsite(site3)wasinstrumentedtoallow troposphere,sothatwatervapourandchemicalcomponents the estimation of the 3-D wind circulation, advection terms emitted at the surface and diluted into the convective layer andspatialvariabilityatthesub-mesoscale. duringthedaycanbeintroducedinthefreeatmosphereand In the following, we first describe the observations made transportedatlargerscale,andinseverallayers(Bantaetal., continuouslyduringthefieldexperiment,andsecond,those 1998;Berkowitzetal.,1998). specifically made during the IOPs. The last subsections presenttheforecastmodelsupportduringthefieldcampaign, HowdotheATandETinteractwiththe educationalaspects,andtheavailabledataset. appearanceofthenocturnaljet? 3.1 Continuousobservations Mahrt (1981; 1999) pointed out that the evolution of the stress divergence during evening transitions increased the 3.1.1 Boundary-layerprofiling ageostrophicwind,andledtothedevelopmentofalow-level jet (wind speed maximum), accompanied by decoupling of Several remote-sensing instruments were deployed during theflowjustabovethesurface. BLLASToverthe3sitesforcontinuousmonitoringoftheat- ThelargenumberofstudiesoriginatingfromtheCASES- mosphere.Verticalprofilingofthewindfrom10mto16km 97, CASES-99 and SABLES-98 experiments (Cuxart et al., a.g.l. was accomplished at site 1 with a combination of so- 2000, Poulos et al., 2002) provide a comprehensive docu- dar (from 10m to 300ma.g.l.), ultra-high frequency (UHF) mentationofthestableandverystableboundarylayersand radar(from200mto3000ma.g.l.)andveryhighfrequency theirturbulenceregimes(vandeWieletal.,2003;Sunetal., (VHF)radar(from1.5kmto16kma.g.l.)profilers.Boththe 2012), giving a better understanding of nocturnal drainage UHFandthesodarprofilingsystemscanalsomeasuresome Atmos.Chem.Phys.,14,10931–10960,2014 www.atmos-chem-phys.net/14/10931/2014/ M.Lothonetal.:TheBLLASTfieldexperiment 10937 Table1.GroupsinvolvedintheBLLASTcampaignandtheinstrumentationtheyimplemented. Country,group Instrumentation France,LA Windprofilers,surfacestation,tetheredballoon,radiosounding France,CNRM-GAME Windprofiler,lidar,ceilometer,scintillometer,surfacestation, turbulentprobeundertetheredballoon,frequentradiosounding France,SAFIRE PiperAztecaircraft France,LPCA Sodar,surfacestation,SMPSandcascadeimpactor France,LMD Dopplerlidar theNetherlands,MAQ Sodar,scintillometer,surfacestation USA,UtahUniv. Surfacestation,tetheredballoon USA,UCDavis Radiosoundings Italy,CNR SkyArrowaircraft Spain,Univ.Comp.deMadrid Microbarometers Spain,UniversitatdelesIllesBalears Surfacestation,soilmeasurements Norway,Univ.Bergen SUMORPAS,surfacestation Germany,Univ.Tübingen MASCRPAS Germany,Univ.Braunschweig M2AVRPAS Germany,Univ.Lipp Octo-copterRPAS Germany,Univ.Heidelberg SiriusIRPAS Germany,Univ.Bremen BUSCARPAS,Funjet1RPAS,Funjet2RPAS Switzerland,PMOD-WRC Radiationsensors UK,Univ.Reading SensorsonSUMORPAS characteristics of atmospheric turbulence (the turbulent en- high-frequencysensorsmeasuringturbulenceproperties.All ergy dissipation rate can be estimated with a UHF profiler, eddy-covariance sensing systems were mounted at heights and the temperature structure coefficient with a sodar). The that ensure that the instruments were in the constant flux UHFprofileralsoestimatestheheightofthemixedlayer,or layer (above ∼3–5 times the height of the local roughness ofanystrongverticalgradientsintheatmosphere(Angevine elements), except the instruments mounted at the forest site etal.,1994;Héoetal.,2003). where this was not possible. The first aim of those stations In addition, another UHF profiler and a sodar were de- was to provide a thorough description of the surface fluxes ployed at sites 2 and 3, respectively (Fig. 1), to build a tri- intheheterogeneouslandscapeofBLLASTarea,whileair- angle of wind profilers, allowing the estimation of the 3-D borne and scintillometer measurements give access to inte- windatthescaleoftheplateau. grated estimates. Beyond this, most of the surface stations Lidarswerealsoextensivelyutilizedinthecampaign.Two wereimplementedwithotherdedicatedobjectives: backscatter lidars, deployed at sites 1 and 2, monitored the – At ss1 (at site 1) (Fig. 3a), two masts equipped for aerosol vertical structure continuously. They provided esti- measuring all the terms of the surface energy balance mationsoftheboundary-layertopanddepthofaerosollay- wereinstalledinagrassandawheatfield,respectively. ers.ADopplerlidarwasalsooperatedatsite1,andprovided A third station with a sonic anemometer and a water profilesoftheverticalwindatabout5stimeinterval. vapourandCO fastsensorwaslocatedattheedgebe- 2 A ceilometer at site 1 supplied the cloud-base height. A tweenbothfields.Measurementsfromthesestationsare fullskycamerawascollocatedwiththeceilometerandpro- being used to investigate the Monin–Obukhov similar- videdaqualitativemonitoringofthecloudcoverwithanim- itytheoryoveraheterogeneousterrainbyusingaflux- ageoftheentireskyeveryminute. footprintmodel(vandeBoeretal.,2013). 3.1.2 Surface-layermeasurementson – Thess2(atsite1)(Fig.3d)wascomposedoftwo10m variouslandscapes towers 20m apart. The first tower was equipped with six sonic anemometers (at 0.85, 1.12, 2.23, 3.23, 5.27 During the BLLAST experiment, seven surface sites, here- and 8.22 m) and nine fast-response fine-wire thermo- after denoted “ss1” to “ss7”, were instrumented above var- couples(at0.019,0.131,0.191,0.569,1.12,2.23,3.23, ious vegetation types and for different objectives (Figs. 2 5.27,8.22m).Thesecondtowerhad6long-waveradi- and 3). The sites characteristics (altitude, vegetation type ation sensors installed at the same heights as the sonic and height), the measured variables and the sensors used anemometers.Theaimofthisset-upwastoinvestigate arelistedinTablesA1andA2oftheAppendix.Inaddition near-surfacelong-waveradiationandbuoyancyfluxdi- to classical meteorological measurements, all the sites had vergence, and the delay between the surface flux sign www.atmos-chem-phys.net/14/10931/2014/ Atmos.Chem.Phys.,14,10931–10960,2014 10938 M.Lothonetal.:TheBLLASTfieldexperiment Figure1.Experimentalarea.Thesmallframeatthetop-leftcornersituatestheBLLASTexperimentarea(bluesquare)atthelargerscaleof thecountry.Thelargeblueovaldelimitstheexplorationareaofthemannedaircraft,andthesmallerpurplecircleindicatesthetemporary restricted area (TRA) for the operations of the remotely piloted aircraft systems (RPASs). The orange dotted triangle locates the profiler network,andthegreenlinesrepresentthepathsofthetwolargeaperturescintillometers.Instruments(otherthansurfacestations)deployed overthethreesitesareschematizedontherightsideofthefigure. changeandthetemperaturegradientsignchange(Blay- surface. At the top of the tower, a high-resolution IR Carreras et al., 2014b), as well as the formation of camera (1Hz image frequency of a 45◦×34◦ field of extremely shallow flows (Manins and Sawford, 1979; view) pointed either toward the ss2, or toward the ss3 Mahrtetal.,2001). (Garaietal.,2013). – Thess3(atsite1)(Fig.3e)focusedonasmall-scale(a – At site 2, eddy-covariance stations sampled three con- few meters) surface heterogeneity study (Cuxart et al., tiguouslargeareas(about1–2kmlong)withrelatively 2014). A flat surface (150m×150m), covered with a homogeneous vegetation: forest (ss5) (Fig. 3c), maize mixofbaresoil,smallbushes,grassandsmallpuddles, (ss6)andmoor(ss7).Thesitewasspecificallydevoted whichconstitutedaveryheterogeneoussurface,hadits to the study of the role of surface heterogeneity. The soil characteristics (temperature, humidity) extensively turbulence characteristics and decay over the different mapped.Theverticalairtemperatureprofileinthefirst vegetationcoverswillbecomparedtakingintoaccount 1.5mandtheenergyfluxeswerealsomonitored. the local circulations which may develop between the fieldsduringthisphaseoftheday. Three high-resolution micro-barometers were also de- ployed at ss3, at each vertex of a triangle with 150m Forconsistency,uniformdataprocessingwascarriedoutfor side length, 1m a.g.l. These high-precision digital in- alleddy-covariancestationsmentionedabove. strumentscandetectverysmallpressureperturbations, In addition to the previous measurements, three scintil- of the order of 0.1Pa, at 2Hz sampling frequency. lometers were used. They measured the structure parameter The objective was to study the small-scale static pres- ofrefractiveindexandtemperatureaveragedalongthepath sure fluctuations produced in the atmospheric bound- betweenthetransmitterandthereceiver(Moeneetal.,2009). ary layer due to turbulent motions or the propagation Therefore,andwiththehelpofMOST,theyprovideaninte- of waves of different types (Viana et al., 2009, 2010; gratedmeasurementofsurfacefluxesovertheheterogeneous Sastreetal.,2012;Románetal.,2014). regionssampledbythesetofsurfacestations.Adoublebeam laserscintillometerwitha110mpathlengthwasdeployedat – Thess4iscomposedofthe60mtower(Fig.3b)which ss1(Hartogensisetal.,2002)andtwolargeaperturescintil- isapermanentplatformattheCentredeRecherchesAt- lometerswithpathlengthsof3and4kmwereaimedtoward mosphériques(CRA).Itprovidesyear-roundfluxmea- thenorthandthesouth-east,respectively(Fig.2). surements and a vertical profile of turbulence close to Atmos.Chem.Phys.,14,10931–10960,2014 www.atmos-chem-phys.net/14/10931/2014/ M.Lothonetal.:TheBLLASTfieldexperiment 10939 Figure2.Satelliteview(fromGoogleEarth)ofthearea,showingtheinstrumentedsitelocations.Surfacesitesovervariousvegetationare notedss1toss7:(ss1)wheat,grassandedge;(ss2)prairies;(ss3)micro-scalesurfaceheterogeneities;(ss4)60mtower;(ss5)forest;(ss6) maize;and(ss7)moor.Thelightyellowlinesrepresentthepathsofthetwolargeaperturescintillometersandtheorangecircleindicatesthe limitoftheTRA. Finally, for the purpose of characterizing aerosol optical lar radiation on surface–atmosphere interaction plays a ma- properties and studying aerosol effects on the evolution of jor role. Some IOP days were conducted the day following theboundarylayer,aerosolssizedistributionwasmonitored arainepisodewhenthemorningwascloudyandconditions atsite1,byuseofaground-basedScanningMobilityParti- cleared up by midday. Over the 3.5 weeks of the field cam- cleSizer(SMPS;range10nm–1µm)andanopticalcounter paign,therewere12dayswithfavourableconditions(corre- (OPC; range 0.3–20µm). For sulfates analysis, a proxy for spondingto12IOPs). secondary aerosols formation, aerosols were also collected During the IOPs, two manned aircraft, remotely piloted at 12m height, using a three-stage cascade impactor, with aircraft systems (RPASs), tethered and ascending balloons, cut-offdiametersof10µm,100and30nm. andinsituaerosolmeasurementswereoperatedintensively. Figure 4 illustrates the observational strategy utilized dur- 3.2 Intensiveobservationperiods(IOPs) ingBLLASTIOPsandTable2summarizestheoperationfor eachIOP. Forthejointoperationsofballoons,airplanesandRPASs, Observationswereintensifiedunderfair-weatherconditions, a temporary restricted airspace (TRA) was issued and acti- withmostlydryconvectionduringtheday,andclearskyor vated daily from 05:00 to 21:30UTC (note that 05:00UTC fairweathercumuliduringtheafternoonandeveningtransi- is 07:00LT). The TRA covered an area of 4km radius in- tions.Thesecharacteristicscorrespondtoanti-cycloniccon- cluding sites 1, 2 and 3 with an upper limit of 1.6kma.g.l ditions (mountain–plain breeze regime), post-frontal condi- (see Figs. 1 and 2). While activated, only the two manned tions,orweak-pressuregradientconditions.Thesesituations BLLASTresearchaircraftwereallowedtoentertheTRA.In arenotspecifictotheATandETstudiesbuttypicalforcon- these cases all RPASs and tethered-balloon operations were vectiveboundary-layerstudiesforwhichtheinfluenceofso- www.atmos-chem-phys.net/14/10931/2014/ Atmos.Chem.Phys.,14,10931–10960,2014 10940 M.Lothonetal.:TheBLLASTfieldexperiment Figure3.ExamplesofsurfacesitesduringBLLAST:(a)oneofthetowersatthess1overthewheat,(b)ss4withthe60mtower,(c)ss5over theforest,(d)ss2overtheprairies,and(e)ss3overthemicro-scaleheterogeneoussurfacewiththess460mtowerbehindandtheOctocopter flyingaround.Authorsofthepictures:(a),(d)PatrickDumas;(b),(c)SolèneDerrien;(e)DanielGrenouillet. limited to low-level flights, ensuring at least 150m vertical reuseafteritwasrecovered.Areal-timemodelpredictedthe separationbetweenthelowestflightlevelofthemannedair- landingareaandaidedinthedecisionofwhentocuttheline craftandthehighestRPAS. thatreleasedtheprobeandthesmallerballoon.Thetimein- terval between two soundings was between 60 and 90min. 3.2.1 Balloons Atotalof62soundingsweremadewiththistechnique,with 80% probe retrieval rate (Table 2). Additionally, a few ra- Radiosoundings diosondeswerelaunchedsimultaneouslyatthethreesitesto estimatethedivergenceatthespatialscaleoftheplateauon IOPs6,7and11. A total of 67 standard MODEM and GRAW radiosondes were launched from site 1 during the IOP days at least 4 times per day at 06:00, 12:00, 18:00 and 24:00UTC, and Tetheredballoons assimilated by the Météo-France forecast operational mod- els (Table 2). At site 2, a new technique was used for fre- Three tethered balloons (one at site 1 and two at site 2) quent soundings of the lower troposphere only, during the operated during all the IOP days (except IOP 4, Table 2), AT(Legainetal.,2013).Twoballoons,withdifferentsizes, from early afternoon to just after sunset. One balloon was attachedtothesameVaisalaprobe,werereleased.Thelarger equippedwithanewlydevelopedturbulenceprobe,operated balloonallowedascentuptoabout2kmheightatwhichtime atsite1(Canutetal.,2014).Thisprobewascomposedofa the probe and the smaller balloon were separated from the sonicanemometer(GillWindsonic3-D),whoseoscillations largerballoon.Thesmallerballoonbroughttheprobesafely angles were measured by an inertial platform, and a plat- to the ground. A package protecting the probe allowed its inum fine wire in a radiation shield for fast air temperature Atmos.Chem.Phys.,14,10931–10960,2014 www.atmos-chem-phys.net/14/10931/2014/