Atmos. Chem. Phys.,9,8545–8558,2009 Atmospheric www.atmos-chem-phys.net/9/8545/2009/ Chemistry ©Author(s)2009. Thisworkisdistributedunder theCreativeCommonsAttribution3.0License. and Physics An elevated large-scale dust veil from the Taklimakan Desert: Intercontinental transport and three-dimensional structure as captured by CALIPSO and regional and global models K.Yumimoto1,K.Eguchi1,I.Uno1,T.Takemura1,Z.Liu2,A.Shimizu3,andN.Sugimoto3 1ResearchInstituteforAppliedMechanics,KyushuUniversity,Fukuoka,Japan 2NationalInstituteofAerospace,Hampton,Virginia,USA 3NationalInstituteforEnvironmentalStudies,Tsukuba,Japan Received: 22April2009–PublishedinAtmos. Chem. Phys.Discuss.: 3July2009 Revised: 28October2009–Accepted: 2November2009–Published: 11November2009 Abstract. An intense dust storm occurred during 19– upslopewindalongthehigh,steepmountainsidesoftheTi- 20 May 2007 over the Taklimakan Desert in northwest- betanPlateauandblewlargeamountsofdustintotheair.The ern China. Over the following days, the space-borne lidar updraft lifted the dust particles farther into the upper tropo- CALIOPtrackedanopticallythin,highlyelevated,horizon- sphere(about9kmabovemeansealevel,MSL),wherewest- tally extensive dust veil that was transported intercontinen- erliesaregenerallypresent. Theunusualterrainsurrounding tally over eastern Asia, the Pacific Ocean, North America, the Taklimakan Desert played a key role in the injection of and the Atlantic Ocean. A global aerosol transport model dusttotheuppertropospheretoformthedustveil. (SPRINTARS) simulated the dust veil quite well and pro- vided a three-dimensional view of the intercontinental dust transport.TheSPRINTARSsimulationrevealedthatthedust 1 Introduction veiltraveledat4–10kmaltitudeswithathicknessof1–4km along the isentropic surface between 310 and 340K. The It is well known that Asian dust impacts regional and re- transportspeedwasabout1500km/day. Theestimateddust mote air quality and climate over eastern Asia, the Pacific amount exported to the Pacific was 30.8Gg, of which 65% Ocean, and beyond (Husar et al., 2001; McKendry et al., wasdepositedinthePacificand18%wastransportedtothe 2001;Yuetal.,2003). TheTaklimakanDesert,whichislo- North Atlantic. These results imply that dust veils can fer- catedintheTarimBasinandboundedonthreesidesbyhigh tilize open oceans, add to background dust, and affect the mountains, is one of the largest deserts in the world, cover- radiative budget athigh altitudes through scattering and ab- ing ∼320000km2. It is considered to be a major source of sorption. thedusttransportedintotheNorthPacific. Sunetal.(2001) The injection mechanism that lifts dust particles into the suggested that dust materials from the Taklimakan Desert free atmosphere is important for understanding the forma- can be lifted over 5km and transported over long distances tionofthedustveilandsubsequentlong-rangetransport.We by westerlies. Bory et al. (2003) compared the mineralog- used a regional dust transport model (RC4) to analyze the ical and isotopic characteristics of mineral dust deposits in dustemissionandinjectionoverthesourceregion. TheRC4 northern Greenland and in China and Mongolia, and sug- analysisrevealedthatstrongnortheasterlysurfacewindsas- gested that the Taklimakan Desert is a primary dust source sociated with low pressures invaded the Taklimakan Desert duringspring. Byanalyzingthreeindependentdatasets(Nd throughtheeasterncorridor.Thesewindsthenformedstrong isotopiccompositionmeasurement,back-trajectory,andnu- mericalmodelsimulations),Groussetetal.(2003)suggested thatthedusteventobservedon6March1990intheAlpswas Correspondenceto: K.Yumimoto verylikelytohaveoriginatedintheTaklimakanDesert.Mat- ([email protected]) suki et al. (2003) analyzed aircraft and lidar measurements PublishedbyCopernicusPublicationsonbehalfoftheEuropeanGeosciencesUnion. 8546 K.Yumimotoetal.: Anelevatedlarge-scaledustveilfromtheTaklimakanDesert overcentralJapanandfoundthattheTaklimakanDesertwas An intensive dust storm occurred on 19–20 May 2007 in also an important source of atmospheric background dust. the Taklimakan Desert. Over the following days, CALIOP However, direct observational evidence of intercontinental detectedanopticallythin,highlylofted,extensivedustlayer, transportofAsiandusttoEuropeislacking,sothereislittle whichwastransportedovereasternAsia,thePacificOcean, supportforthesefindings. Inaddition, theupliftingmecha- NorthAmerica,andtheAtlanticOcean. Wecallthistypeof nismofdustparticlesintothehightroposphereovertheTak- dust layer a “dust veil” to emphasize its vertically thin and limakan Desert, which leads to the subsequent long-range horizontally extensive structure. In this study, we compre- transport,requiresmoredetailedexploration. hensively investigated the emission and injection processes Long-range transport of dust particles affects Earth’s ra- overthesourceregion,thesubsequentintercontinentaltrans- diative budget both directly through scattering and absorb- port, and the 3-D structure of the dust veil. We combined ingsolarradiation(BohrenandHuffman,1983;Sokolicand regional and global aerosol transport models, in situ mea- Toon, 1996) and indirectly through changes in cloud phys- surements,andpassiveandactivesatelliteobservations. ical and radiative properties as cloud condensation nuclei Themodelsandobservationdatausedinthisstudyarede- (Twomey, 1977; Sassen, 2002). Through deposition, trans- scribed in Sect. 2. A detailed analysis of the emission and ported dust also acts as a source of nutrients for marine injection processes over the Taklimakan Desert using a re- plankton in upper ocean waters (Martin et al., 1994; Duce gionaldustmodelispresentedinSect.3,alongwithanalysis etal., 1991). Toestimatetheseeffectsaccurately, extensive oftheglobaltransportand3-Dstructureofthedustveil. We measurements of dust transport altitudes, patterns, and life- presentourconclusionsinSect.4. timesarerequired.Passivesatelliteobservations(e.g.theTo- talOzoneMappingSpectrometer,TOMS,AerosolIndexand 2 Numericalmodelsandobservationdatasets Moderate Resolution Imaging Spectroradiometer, MODIS, aerosol optical thickness data) have provided useful infor- 2.1 Numericalmodels mation on dust horizontal transport. However, observation of the three-dimensional (3-D) distribution (particularly the We used global and regional numerical models in our anal- verticalstructure)ofdusttransportusingthesepassivemea- yses. Theregionalmodel(withfinerhorizontalandvertical surementsisdifficult. resolutions)wasusedforadetailedanalysisoftheemission The space-based Cloud-Aerosol Lidar with Orthogonal and injection of the dust veil from the Taklimakan Desert. Polarization (CALIOP) aboard the Cloud-Aerosol Lidar Thesubsequentintercontinentaltransportofthedustveilwas and Infrared Pathfinder Satellite Observation (CALIPSO) analyzedusingtheglobalaerosoltransport-radiationmodel. launchedon28April2006providesnewinformationonthe The Regional Atmospheric Modeling System/Chemical globalverticaldistributionsofaerosolsandclouds(Winkeret Weather Forecast System with four-dimensional variational al., 2007). CALIOPoffersauniqueopportunitytomeasure data assimilation system (RAMS/CFORS-4DVAR) regional dust vertical distributions globally through depolarization dust transport model (RC4; Yumimoto et al., 2007, 2008) ratio measurements. Using CALIOP observations, Liu et wasusedtoanalyzedustveilgenerationoverthesourcere- al.(2008a)presentedastudyofalarge-scaledustplumethat gion. RC4 is based on the successful RAMS/CFORS dust originatedintheSaharanDesertandwastransportedacross model(Unoetal.,2004). TheRC4domainwaslocatedover theNorthAtlanticintotheGulfofMexicointhelowertro- eastern Asia (Fig. 1), with a horizontal resolution of 40km posphere(<7km)in10days. Seasonal3-Ddistributionsof and 55 vertical stretching layers from the surface to 20km airbornedustoverEastAsiahavebeenderivedfromthefirst (a vertical resolution of 140m at the surface and 400m at yearofCALIOPmeasurements(Liuetal.,2008b). Dustdis- the top). To investigate the dust veil origin, dust emissions tributionsandseasonalityareinfluencedsignificantlybythe were limited to those over the Taklimakan Desert. In addi- uniqueorographyoftheTibetanPlateau.Huangetal.(2008) tion,tounderstandvisuallyhowdustparticleswereinjected investigated the vertical structures of Asian dust based on intohighaltitudesovertheTarimBasin, wealsoperformed CALIOP observations during the Pacific Dust Experiment a simplified wind-driven tracer simulation with wind fields (PACDEX). However, it is difficult to capture detailed 3-D obtained from the RAMS/CFORS simulation. In the tracer structuresanddailyvariationsusingCALIOPmeasurements simulation,dusttracerswereemittedfromthesurfacewhen alonebecauseofthelargelongitudinalintervalbetweentwo thesurfacewindspeedbecamegreaterthan6.5m/sandwere consecutive CALIPSO orbits (∼1000km at mid-latitudes). thentransportedbythemodeledwindfields. Thenumberof Comprehensive studies are required of each dust source to emitted tracers was proportional to the cube of the surface understand its role in Asian dust generation and transport. windspeed. Thethresholdwindspeedandthirdpower-law Recently, Uno et al. (2008, 2009), Generoso et al. (2008), relation were based on the results of both surface observa- andEguchietal.(2009)performedintegratedanalysesusing tions(e.g.KurosakiandMikami,2003)andnumericalmod- CALIOP measurements and numerical model simulations, els(e.g.TegenandFung,1994;Takemuraetal.,2005). providing more detailed observations of the 3-D structures ofAsianandSaharandusttransport. Atmos. Chem. Phys.,9,8545–8558,2009 www.atmos-chem-phys.net/9/8545/2009/ Figure 1 K.Yumimotoetal.: Anelevatedlarge-scaledustveilfromtheTaklimakanDesert 8547 Fig.1.TheRC4modelingdomainandtopographyoftheTaklimakanDesert. The global Spectral Radiation-Transport Model for CALIOP data was used for cloud layer detections. Our Aerosol Species (SPRINTARS; Takemura et al., 2005) was method of CALIOP data processing was identical to those used in this study. SPRINTARS includes explicit estimates ofUnoetal.(2008,2009),Haraetal.(2009),andEguchiet of the direct, first indirect, and second indirect effects of al.(2009). aerosols. ThehorizontalresolutionwasT106,with56layers inasigmacoordinate. Tofocusonthetransportofthedust 2.2.2 NIESlidarnetwork veil from the Taklimakan Desert, emissions from all other regions(e.g.theGobiDesert,Mongolia,orInnerMongolia) The National Institute for Environmental Studies (NIES) wereneglected. lidar network (Shimizu et al., 2008) consists of 21 stan- Both the regional and global model simulations were dard lidar observation sites distributed over Japan, Korea, nudged by 2.5◦×2.5◦ National Centers for Environmen- China, Mongolia, and Thailand. It provides vertical pro- tal Prediction/Nation al Center for Atmospheric Research files of aerosols and clouds with a high temporal resolution (NCEP/NCAR) reanalysis data, with a time interval of 6h. (15min). Observation results are displayed in real time at The reanalysis data were also used for the meteorological http://soramame.taiki.go.jp/dss/kosa/andhavebeenusedfor boundaryconditions ofRAMS. inverse modeling of Asian dust in combination with a nu- merical model (Yumimoto et al., 2007, 2008). Shimizu et 2.2 Observationaldata al. (2004) have provided a detailed description of the NIES lidardataprocessing. 2.2.1 CALIOP 2.2.3 Otherobservations The space-based lidar CALIOP, which was launched on 28 April 2006 aboard CALIPSO, provides vertical distribu- We also used Aerosol Index (AI) measurements from the tionsofaerosolandcloudsonaglobalscale(Winkeretal., Ozone Monitoring Instrument (OMI) to examine the hori- 42 2007). For comparison with the model simulations, we re- zontal distributions of dust loading over the source region. trievedverticalprofilesofthedustextinctioncoefficientfrom TheOMIAIprovidesasemi-quantitativeestimateofcolum- the total attenuated backscatter contained in the Level 1B nar aerosol loading in a given pixel. A positive AI value CALIOP data (ver. 2.01), using Fernald’s inversion (Fer- indicatesthepresenceofUV-absorbingaerosols(Prosperoet nald, 1984) by setting the lidar ratio S1=50sr (Shimizu et al.,2002). TheWorldMeteorologicalOrganization(WMO) al., 2004). Then, the retrieved vertical profiles were aver- SYNOP data (wind speed, wind direction, visibility, and aged to the resolution of the CALIOP Level 2 data (5km). weather) were also used to investigate the emission of dust Thecloud-aerosoldiscrimination(CAD)indexintheLevel2 from the Taklimakan Desert. Shao et al. (2003) found the www.atmos-chem-phys.net/9/8545/2009/ Atmos. Chem. Phys.,9,8545–8558,2009 8548 K.Yumimotoetal.: Anelevatedlarge-scaledustveilfromtheTaklimakanDesert 20Na30)N40N50N60N 192018 M43AY182123 M06AY172137 M34AY162241 CM5A7ALY-6IP152S25O6 M2- 55dAuYs1t4 2-e3640x M4ti8nAcY-t3i1o32n37 5 cM1-o62AeYf. (1-/1km;1C 1l2oA57gD6 M 2 s≥1Ac Y1a0l1e012)080 M44AY102095 M12AY 082296 M16AY073300 M39AY063315 11M0205eight (km)7AY b) 60N H 0 50N heig4ht (km; abo8ve sea lev1e2l) 60N 40N 0521 50N 0522 30N 0523 40N 20N 0524 30N c) 40N50N60N 192018 M43AY182123 M06AY172137 M34AY162241 M57AY152256 M25AY0154223650 M48AY132375 M16AY0526 112576 M21AY112T08r0a 0M4j5e4A2cY7to1r02y09 A5 M12AY05025828082296 M16AY07303050 2M399AY063315 M07AY000555323109 0055323100N053301N40N50N60N 206N0E30N 20N 11205Height (km) 0 90E -6RC4 d-5ust ex-4tinctio-3n co1e92-f012.8 (M413A/kYm-118;2 1l2Ro3 MgH06 AiscYce12a≥1071l2eE113)70 M3, 4A-3Y51≤62T24≤1 M-517A1Y1155220S56E PM-265RAYIN1T423A60 -RM458SAY d1u32s137t-85 4e0M1x6AtYincti-o3n c1o12e57f-6. 2 M21(151AR0/YkWH1mi1c2;0e8 0l≥ oM41g41A 0Ys,c1 -a023l095e5 ≤)M1T2A1≤Y2-01W1 082296 M16AY0973030W0 M39AY063315 M07AY 60W 20N30N40N50N60N Fig.2.Three-dimensionalanalysisofintercontinentaltransportofthedustveil.(a)Cross-sectionsoftheCALIOPdustextinctioncoefficient with the cloudy regions shaded in gray (CAD>100). (b) HYSPLIT forward and backward trajectories started at 17:00UTC on 24 May ◦ ◦ at36–38 Nand145–146 E7000mabovesealevelonthe24May-162157CALIPSOorbit. Thecolorsrepresenttrajectoryheights. (c) Cross-sectionsofthemodel-simulateddustextinctioncoefficient. Thefirstfourcross-sectionsovereasternAsiaduring21–24Maywere simulatedtheRC4model,whiletherestwerederivedfromtheSPRINTARSmodel. Grayshadedareasrepresentregionsthatsatisfythe condition of heterogeneous ice formation for mineral dust particles: relative humidity with respect to ice (RHice)>110%, and −35◦C< temperature(T)<−11◦C.ThecolorscalerangesforCALIOP,RC4,andSPRINTARSarenotconsistent. followingempiricalrelationshipsbetweenvisibilityanddust 3 Resultsanddiscussion concentrations by fitting near-surface total suspended parti- cle(TSP)measurementstovisibilities: Figure 2a shows transects of the CALIOP dust extinction C =3802.29V−0.85 forV <3.5 (1) coefficient during 21–30 May 2007. The CALIOP tracked SYNOP SYNOP SYNOP a very thin, highly elevated, large-scale dust veil from the C =exp(−0.11V +7.62)forV >3.5 SYNOP SYNOP SYNOP Taklimakan Desert to the North Atlantic Ocean (yellowish- where C is the dust concentration (µg/m3), and greenfeaturesathighaltitudesintheverticaltransectsofthe SYNOP V is the SYNOP visibility (km). Because C is CALIOP extinction coefficients). In the following subsec- SYNOP SYNOP based on an empirical equation, it is sometimes difficult to tions, we discuss emission and injection processes over the estimatetheabsoluteconcentrationlevel. TarimBasinfrom17to22Maywhenthedustveilformed. Atmos. Chem. Phys.,9,8545–8558,2009 www.atmos-chem-phys.net/9/8545/2009/ K.Yumimotoetal.: Anelevatedlarge-scaledustveilfromtheTaklimakanDesert 8549 a) May 17 b) May 18 L1 L1 50N 50N 40N A R 40N T 30N 30N 70E 80E 90E 100E 110E 120E 70E 80E 90E 100E 110E 120E c) May 19 d) May 20 L2 50N L1 50N 40N 40N 30N 30N 70E 80E 90E 100E 110E 120E 70E 80E 90E 100E 110E 120E e) May 21 f) May 22 L2 50N 50N L2 40N 40N 30N 30N 70E 80E 90E 100E 110E 120E 70E 80E 90E 100E 110E 120E 0 1 2 3 4 5 12 m/s OMI AI (surface u-v) Fig.3.OMIAI(color),theRC4simulateddailyaveragedsurfacewind(bluevectors),andthetopography(browncontours).Thelocationsof lowpressureregions(circledcharacterL)andassociatedcoldfronts(solidblacklines),whichwereestimatedfromthepressureatmeansea levelinthe2.5◦×2.5◦NCEP/NCARreanalysisproduct,theMODIStruecolorimage(alsoseeFig.6),andtheRC4resultsarealsoshown. ◦ ◦ ◦ ◦ TheSYNOPobservationstationsarealsoshownin(a)asblacksolidcircles:Alar(A:81.1 E,40.5 N),Tazhong(T:83.7 E,39.0 N),and ◦ ◦ Ruoqiang(R:88.2 E,39.0 N). Wethenpresentresultsonthe3-Dstructureoftheintercon- from the RC4 regional model simulation with distributions tinentaldustveiltransport. of OMI AI during 17–22 May. Figure 4 presents daily- averaged cross-sections of wind fields for the regions indi- 3.1 Emissionandinjectionofthedustveil catedbythegreenrectanglesinFig.3. Forvisualizationof theinjectionprocess,resultsfromthesimplifiedwind-driven Theemissionandinjectionofdustintotheairoverthesource tracer simulations are also shown. The tracer distributions regionwereveryimportantfortheformationofthedustveil agree quite closely with those of the modeled dust extinc- and its subsequent long-range transport. We performed a tion coefficients. Figure 5 shows the simulated RC4 and comprehensive analysis to investigate these processes using reported SYNOP surface wind speeds and directions at the multiple measurement datasets and analysis tools. Figure 3 three SYNOP stations Alar, Tazhong, and Ruoqiang, which shows the meteorological conditions over the Tarim Basin werelocatedrespectivelyinthenorthern,middle,andeastern www.atmos-chem-phys.net/9/8545/2009/ Atmos. Chem. Phys.,9,8545–8558,2009 8550 K.Yumimotoetal.: Anelevatedlarge-scaledustveilfromtheTaklimakanDesert regionsoftheTaklimakanDesert(markedbylettersA,T,and This dust storm appeared to be weaker than the first one. RinFig.3). DustconcentrationsderivedfromSYNOPvisi- SYNOP stations observed higher visibilities over the basin bilitiesandRC4surfacedustconcentrationsarealsoshown. (≥6km). On 17 May, a low pressure region (L ) was in northern Aokietal.(2005)classifiedthemesoscalecoldwindthat 1 Mongolia, as shown in Fig. 3a. At this time, strong north- producesduststormsintheTarimBasinintothreepatterns. ernwindswereproducedoverthenorthernsideoftheTarim The wind that produced the dust veil studied here may be Basin,buttheairinsidethebasinwasstillcalm. Thetracer identified as Pattern 1. This pattern is characterized by an simulation showed that only a small amount of dust tracers easterlymesoscalesurfacewind,whichisseparatedfromthe were emitted from the surface to the basin (Fig. 4a). The synoptic-scalecoldwesterly. Thesurfacecoldwindchanges threeSYNOPstationsreportedweaksurfacewinds(≤3m/s; itsdirectiontowestwardaftergoingaroundtheeasternside notshown). of the Tian Shan range (compare Figs. 5 and 3 of Aoki et On18May(Figs.3band4b),coldnortherlysurfacewinds al., 2005). Eguchi et al. (2009) reported another case of associated with the low pressure L became stronger over long-rangetransportofTaklimakandustoccurringduring8– 1 northwestern China. Forced by the Tibetan Plateau, these 10May2007thatfitPattern1.Thehighandsteepmountain- winds changed direction and blew into the basin along the sideterrainoftheTibetanPlateauonthesouthernsideofthe corridorontheeasternside. Meanwhile,acoldfrontformed basinplaysakeyroleinformingthestrongupdraftthatcar- over Mongolia, as indicated by the black line in Fig. 3b. riesdust particlesinto theuppertroposphere forlong-range OMI AI levels were still low over the desert. In the tracer transportbywesterlies. simulation(Fig.4b),strongsurfacewindsbegantosweepa To investigate the evolution of dust veil transport in the significantamountofdustintotheair,andthetopofthedust uppertroposphere(9–12km)duringthefirstfewdays,Fig.6 plumereachedtheheightoftheTibetanPlateau. AtSYNOP shows a day-by-day overview of the RC4 simulated dust stations, both the wind speed and direction changed sud- aerosol optical depth (AOD), wind field at ∼8km above denly, and a strong northeasterly wind formed in the basin, mean sea level (MSL), center locations of the low pres- consistentwiththeRC4simulatedwindfields(Fig.5). sures, and cloud distribution observed by MODIS/AQUA On19May(Figs.3cand4c),strongwindblastedtoward overeasternAsia. LongitudinalverticaltransectsoftheRC4 the southern side of the basin and produced upslope winds dustextinctioncoefficientsandpotentialtemperaturearealso (about0.2m/sintheverticaldirection). Thewindblowndust shown.NotethattheRC4simulationshowninFig.6allowed tracerswerethencarrieduptofreeatmospherebytheupward onlydustemissionfromtheTaklimakanDesert. windalongthenorthernslopeoftheTibetanPlateau. Atthis On21May(Fig.6a),dustwasinjectedintotheuppertro- point, the OMI AI became larger over the basin. SYNOP posphere, reaching an altitude of 9kmm.s.l. at 04:00UTC stations, except the northern station at Alar, reported strong (alsoseeFig.4e). At∼19:00UTC,CALIOPpassedoverthe surfacewinds(>5m/s)andlowvisibilities(<3km). Partic- GobiDesertanddetectedtwodensedustlayers(Figs.7aand ularlyatRuoqiang,considerablylowvisibility(<1km)per- 2a: 21May-190843path). Onedustlayerwasnearthesur- sistedfor15h. face,whiletheotherwasatanaltitudeof6–10km. TheRC4 On20May(Figs.3dand4d),thestrongsurfacewindcon- simulation (Figs. 7b and 2c: 21 May-190843 path), which tinuedtoinjectdusttracersintotheatmosphere.TheOMIAI onlyconsidereddustemissionsfromtheTaklimakanDesert, indicated that dense dust spread over the entire basin. The reproducedonlyonedustlayerathigheraltitudes,indicating upliftedparticlesreachedanaltitudeof∼9kmandwerecap- thattheupperdustlayerobservedbyCALIOPoriginatedin turedwithinthe320–340Kpotentialtemperaturezone.They the Taklimakan Desert. The lower layer should have been weresubsequentlytransportedeastwardbythestrongwest- generatedbyothersources(verylikelytheGobiDesert). erlywind(>20m/s)andstartedtheirintercontinentaltravel. On 22 May, the eastward transport of the dust increased SYNOPstationsreportedthatstrongsurfacewinds(>5m/s) in speed within the westerly (upper panel of Fig. 6b; also persistedoverthebasinonthisday. Followingthelowpres- see the OMI AI distribution in Fig. 3f). The vertical cross- sureL ,anotherlowpressureregion(L )appearednorthof section (lower panel in Fig. 6b) shows that the dust loading 1 2 thebasin. was transported across the Altun Shan range in the north- On 21 May (Figs. 3e and 4e), the OMI AI showed that easternTibetanPlateau(Fig.1)withathinstructurewithina mostofthedustloadinghadbeentransportedeastwardoutof potentialtemperaturezoneof320–340K.CALIOP(Fig.2a: thebasin.AtTazhong(center)andRuoqiang(east),thewind 22 May-181306 path) also detected a thin dust layer at 6– directionchangedquicklyfromeasterlytowesterly.SYNOP 12kmonthisday. reported rain at all three stations, indicating a widespread On 23 May (Fig. 6c), the low pressure region L was 2 raineventwithinthebasin. deepening and forming a cold front over the Gobi Desert. On 22 May (Figs. 3f and 4f), OMI AI indicated that The modeled AOT horizontal distribution (upper panels in the dust has been transported to Mongolia and had become Fig. 6c) shows that the dust veil was transported in front weaker. Meanwhile, another dust storm occurred in the of the L cold front extending in the north–south direction 2 basin, associated with the second low pressure region (L ). throughthemeanderingwesterly. Theverticalcross-section 2 Atmos. Chem. Phys.,9,8545–8558,2009 www.atmos-chem-phys.net/9/8545/2009/ K.Yumimotoetal.: Anelevatedlarge-scaledustveilfromtheTaklimakanDesert 8551 Fig.4. Averagedcross-sectionsofwindfieldsfortheregionsindicatedbygreenrectanglesinFig.3. Greendotsrepresentdistributionsof dusttracersfromthesimplewind-driventracersimulation.Coloredcontoursdepictthedaily-averagedwindspeedintheeast-westdirection. Blackcontoursrepresentthedaily-averagedpotentialtemperatures. (lower panel in Fig. 6c) shows that the lower part of the On 24 May (Fig. 6d), the dust veil was transported far- dustremainedbehindtheL coldfrontnear115◦E,whereas therbetweenL andL alongthewesterly,withitsfrontpart 2 2 1 the upper part (i.e. the dust veil) was transported beyond reachingJapan. Thecross-sectionshowedthatthedustveil the cold front. This indicates that the dust veil in the up- traveledat6–11kmaltitudewithintheisentropicsurfaceof per troposphere could travel faster than lower dust clouds potentialtemperature320–340K.Inthefollowingdays,the (e.g. the Gobi dust, which was transported typically at al- dustveilcontinueditslongjourneytoNorthAmericaandthe titudes<5km). Eguchietal.(2009)reportedaTaklimakan NorthAtlantic,asdiscussedinthefollowingsubsection. dustlayerthatwastransportedat∼10kmandcaughtupwith a Gobi dust layer generated 5 days earlier, forming a two- layereddustdistributionovertheeasternNorthPacific. www.atmos-chem-phys.net/9/8545/2009/ Atmos. Chem. Phys.,9,8545–8558,2009 Figure 5 8552 K.Yumimotoetal.: Anelevatedlarge-scaledustveilfromtheTaklimakanDesert 3.2 Intercontinentaltransport As shown in Fig. 2a, CALIOP captured the intercontinen- tal transport of the dust veil from the Taklimakan Desert to the North Atlantic Ocean during 20–31 May 2007. Corre- sponding vertical transects of the model-simulated dust ex- tinction coefficients are presented in Fig. 2c. In that figure, fourtransectsovereasternAsiaduring21–24Maywerede- rivedfromRC4,whiletherestwerefromSPRINTARS.To- gether, the two models tracked the dust veil transcontinen- tal transport quite well. In addition, we performed a pre- liminarySPRINTARSsimulationinwhichthedustemission of the Taklimakan Desert was eliminated and other aerosol sources (sulfate, sea salt, carbonaceous particles, and dust from non-Taklimakan Desert sources) were included. The non-Taklimakan dust simulation could not capture the dust veil measured by CALIOP (not shown). This indicates that dust from other sources (i.e. the Gobi Desert) and non-dust aerosols could not be carried to such a high altitude, and hence did not contribute to the dust veil formation. Fig- ure 2b shows forward and backward trajectories by the Hy- bridSingleParticleLagrangianIntegratedTrajectory(HYS- PLIT) Model (Draxler and Hess, 1998). The trajectories showthelongjourneyofthedustveilfromtheTaklimakan Desert, over the North Pacific and North America, and into theNorthAtlanticOcean. The two panels in Fig. 8b present vertical cross-sections ofthedustextinctioncoefficientsmeasuredbyCALIOPand simulated by the RC4 model along the CALIPSO track on 23 May in Fig. 8a. The NIES lidar at Nagasaki (129.98◦E, 32.94◦N)alsodetecteda1-km-thickdustveilpassingover- headat6–7kmaltitudeduringthenightof23–24May(left panel in Fig. 8c). For comparison, the RC4 simulated dust extinction coefficient at Nagasaki is also presented in the right panel of Fig. 8c. The height, thickness, and passage timeofthedustmeasuredbytheNIESlidarwereconsistent withthosemeasuredbyCALIOP.TheRC4modelcaptured similar,butconsiderablyunderestimated,dustveilcharacter- istics. Figure 9a shows a longitudinal cross-section of the SPRINTARS dust extinction coefficient and potential tem- peraturealongtheHYSPLITtrajectoryfromtheTaklimakan DeserttotheNorthAtlanticOcean. Figure9b(lowerpanels) compares averaged vertical profiles of dust extinction coef- ficients along the CALIOP orbit paths at ∼145◦E (eastern Asia),5◦W(thedateline),and70◦W(AtlanticOcean). The verticalprofilesarenormalizedbythemaximumvalues. The SPRINTARS dust extinction coefficient (Fig. 9a) showsthatthedustveilformedovertheTaklimakanDesert Fig.5. TimeseriesoftheSYNOP-observedandRC4-s4im6u latedsur- was transported eastward at 5–10km along the isentropic facewindspeedanddirectionduring18–21MayatAlar,Tazhong, surfaceof310–340KovereasternAsia.CALIOPobserveda andRuoqiang(seeFig.3). Dustconcentrationsderivedfromvisi- thindustlayerof<1kmon24May(theleftpanelofFig.9b; bilitiesandraineventsareshownassymbols. TheRC4simulated seealsoFig.2). However,themodelswereunabletocapture surfacedustconcentrationsarealsoshown. the fine structure of the dust layer due to their low vertical resolution. ComparedtotheCALIOPandRC4profiles,the Atmos. Chem. Phys.,9,8545–8558,2009 www.atmos-chem-phys.net/9/8545/2009/ K.Yumimotoetal.: Anelevatedlarge-scaledustveilfromtheTaklimakanDesert 8553 a) 21 May 2007, 04UTC b) 22 May 2007, 04UTC 0521-190843 0522-181306 60N 60N L1 L2 L1 50N 50N L2 L0 40N 40N 30N 30N 20N 20N 70E 80E 90E 100E 110E 120E 130E 140E 150E 160E 170E 70E 80E 90E 100E 110E 120E 130E 140E 150E 160E 170E Height (km)1050 80E 90E 100E 110E 120E 130E 140E 150E 000..105RC4 dusr ext. coef. (1/km) RHice≥110, -35≤T≤-11 Height (km)1500 80E 90E 100E 110E 120E 130E 140E 150E 000..105RC4 dusr ext. coef. (1/km) RHice≥110, -35≤T≤-11 c) 23 May 2007, 04UTC d) 24 May 2007, 04UTC 0523-171734 0524-162157 60N 60N L2 50N L2 L1 50N L1 40N 40N 30N 30N 20N 20N 70E 80E 90E 100E 110E 120E 130E 140E 150E 160E 170E 70E 80E 90E 100E 110E 120E 130E 140E 150E 160E 170E Height (km)1500 00..00525RC4 dusr ext. coef. (1/km) RHice≥110, -35≤T≤-11 Height (km)1050 00..00525RC4 dusr ext. coef. (1/km) RHice≥110, -35≤T≤-11 80E 90E 100E 110E 120E 130E 140E 150E 0 80E 90E 100E 110E 120E 130E 140E 150E 0 Fig.6.Day-by-dayoverviewofthedailyevolutionofdustveiltransportfrom21to24May.Theupperpanelsshowhorizontaldistributions ofthemodeledaerosolopticaldepth(AOD)(redcontours)andmodeledwind(greenvectors)attheRC425thlayer(∼8kmm.s.l.).Pinklines showtheCALIPSOorbitpaths. TheMODIS/Aquared-green-bluecolorcompositecloudimagesarealsoshown,alongwiththelocations oflowpressureregionsandcoldfrontsatthesurface. Lowerpanelsshowcross-sections(alongtheorangelinesintheuppertwopanels) oftheRC4simulateddustextinctioncoefficient(colorcontour), potentialtemperature(blackline), andregionsthatsatisfythecondition ofheterogeneousiceformationformineraldustparticles: relativehumiditywithrespecttoice(RHice)>110%,and−35◦C<temperature (T)<−11◦C(grayshading). centerofthedustlayersimulatedbySPRINTARSwastrans- Fig.2).TheHYSPLITforwardtrajectory(Fig.2b)presented ported at a lower altitude. This may reflect the fact that the dustpassingoverthewesternPacificathigherlatitudesand global model with a coarser resolution could not properly altitudes than both the CALIOP observation and SPRINT- representtherapidlychangingsurfaceelevationandcompli- ARS simulation. Comparison of vertical profiles near the catedterrainoverthesourceandsurroundingregions,which dateline(centerpanelofFig.9b)indicatesthattheSPRINT- playsanimportantroleintheinjectionofdustparticlesinto ARSsimulationcapturedthedustveilquiteclosely. the upper troposphere (see Sect. 3.1). However, SPRINT- The dust veil reached the west coast of North America ARSsimulateddustaltitudesreasonablycloselyinareasfar on28–29Mayat4–10km(seealsoFig.2: 28May-110044 from the source region (see the middle and right panels of and29May-100512orbits). Thedustveilwasentrainedinto Fig.9b). thelowertroposphereandmighthaveimpactedthelocalair TheSPRINTARSsimulationshowsthatthedustveiltrav- quality over North America. On 31 May (Fig. 2: 31 May- elled at 4–9km in altitude with a thickness of 1–4km, and 063507orbit),bothCALIOPandSPRINTARSdetectedthe horizontally at 30–40◦N over the Pacific Ocean (also see dustveilovertheNorthAtlanticOceanverticallyat2–6km www.atmos-chem-phys.net/9/8545/2009/ Atmos. Chem. Phys.,9,8545–8558,2009 8554 Figure 7 K.Yumimotoetal.: Anelevatedlarge-scaledustveilfromtheTaklimakanDesert Figure 8 Fig.7. Cross-sectionsoftheCA LIOPandRC4dustextinctioncoefficientsalongthe21May-190843orbitpath. Thecolorscalerangesfor theCALIOPandRC4dataarenotconsistent. 48 Fig. 8. Comparison of measured and simulated dust extinction coefficients. (a) CALIPSO orbit path and location of the Nagasaki lidar site; (b) cross-sectionsoftheCALIOPmeasuredandRC4simulateddustextinctioncoefficientsalongtheCALIPSOorbitshownin(a); (c)time-heightcross-sectionsofthemeasuredNIESlidarandsimulatedRC4dustextinctioncoefficientsattheNagasakilidarsite. Black counterlinesintherightpanelsrepresenttheRC4potentialtemperature. ThecolorscalerangesfortheobservationsandRC4dataarenot consistent. Atmos. Chem. Phys.,9,8545–8558,2009 www.atmos-chem-phys.net/9/8545/2009/ 49