Atmos.Chem.Phys.Discuss.,15,1523–1571,2015 D www.atmos-chem-phys-discuss.net/15/1523/2015/ isc ACPD u doi:10.5194/acpd-15-1523-2015 s s ©Author(s)2015.CCAttribution3.0License. io 15,1523–1571,2015 n P a p Thisdiscussionpaperis/hasbeenunderreviewforthejournalAtmosphericChemistry e The climatology of r andPhysics(ACP).PleaserefertothecorrespondingfinalpaperinACPifavailable. dust aerosol over the | arabian peninsula The climatology of dust aerosol over the D is c A.Shalabyetal. u arabian peninsula ss io n P TitlePage a A. Shalaby1, B. Rappenglueck2, and E. A. B. Eltahir3 pe r Abstract Introduction 1TheInternationalCentreforTheoreticalPhysics,Trieste,Italy | Conclusions References 2DepartmentofEarthandAtmosphericSciences,UniversityofHouston,Texas77004,USA D 3MassachusettsInstituteofTechnology,Cambridge,Massachusetts,USA is Tables Figures c u s Received:25September2014–Accepted:19December2014–Published:19January2015 s io (cid:74) (cid:73) n Correspondenceto:A.Shalaby([email protected]) P a (cid:74) (cid:73) PublishedbyCopernicusPublicationsonbehalfoftheEuropeanGeosciencesUnion. p e r Back Close | FullScreen/Esc D is c u Printer-friendlyVersion s s io InteractiveDiscussion n P a p e r 1523 | Abstract D is c ACPD u Dust storms are considered to be a natural hazard over the Arabian Peninsula, since s s theyoccurallyearroundwithmaximumintensityandfrequencyinSpringandSummer. io 15,1523–1571,2015 n TheRegionalClimateModelversion4(RegCM4)hasbeenusedtostudytheclimatol- P a 5 ogy of atmospheric dust over the Arabian Peninsula from 1999 to 2012. This relatively pe The climatology of r long simulation period samples the meteorological conditions that determine the cli- dust aerosol over the matology of mineral dust aerosols over the Arabian Peninsula. The modeled Aerosol | arabian peninsula Optical Depth (AOD) has been compared against ground-based observations of three D is Aerosol Robotic Network (AERONET) stations that are distributed over the Arabian c A.Shalabyetal. u s 10 Peninsula and daily space based observations from the Multi-angle Imaging Spec- sio troRadiometer (MISR), the Moderate resolution Imaging SpectroRadimeter (MODIS) n and Ozone Monitoring Instrument (OMI). The large scale atmospheric circulation and Pa TitlePage p the land surface response that lead to dust uplifting have been analyzed. While the e r Abstract Introduction modeledAODshowsthatthedustseasonextendsfromMarchtoAugustwithtwopro- | nouncedmaxima,oneoverthenorthernArabianPeninsulainMarchwithAODequalto Conclusions References 15 D 0.4andoneoverthesouthernArabianPeninsulainJulywithAODequalto0.7,theob- is Tables Figures c servationsshowthatthedustseasonextendsfromApriltoAugustwithtwopronounced u s maxima, one over the northern Arabian Peninsula in April with AOD equal to 0.5 and sio (cid:74) (cid:73) oneoverthesouthernArabianPeninsulainJulywithAODequalto0.5.Inspringahigh n P pressure dominates the Arabian Peninsula and is responsible for advecting dust from a (cid:74) (cid:73) 20 p e southernandwesternpartoftheArabianPeninsulatonorthernandeasternpartofthe r Back Close Peninsula. Also, fast developed cyclones in northern Arabian Peninsula are responsi- | ble for producing strong dust storms over Iraq and Kuwait. However, in summer the FullScreen/Esc D main driver of the surface dust emission is the strong northerly wind (“Shamal”) that is c transport dust from the northern Arabian Peninsula toward south parallel to the Ara- u Printer-friendlyVersion 25 s s bian Gulf. The AERONET shortwave Top of Atmosphere Radiative Forcing (TOARF) io InteractiveDiscussion n andattheBottomofAtmosphereRadiativeForcing(BOARF)havebeenanalyzedand P a comparedwiththemodeleddirectradiativeforcingofmineraldustaerosol.Theannual p e r 1524 | −2 D modeledTOARFandBOARFare−3.3and−12Wm ,respectively.However,thean- is nual observed TOARF and BOARF are significantly different at −10 and −52Wm−2, cu ACPD s respectively.TheanalysisofobservedandmodeledTOARFagreeswithpreviousstud- s io 15,1523–1571,2015 ies in highlighting the need for more accurate specification of surface albedo over the n P region. Due to the high surface albedo of the central Arabian Peninsula, mineral dust a 5 p e The climatology of aerosols tend to warm the atmosphere in summer (June–August). r dust aerosol over the | arabian peninsula 1 Introduction D is c A.Shalabyetal. u Atmospheric mineral dust is a natural aerosol and is ubiquitous in the Earth’s atmo- s s sphere, despite that it is emitted from hyper-arid, arid and semi-arid regions on the ion globe. About 2000Mt is emitted annually to the atmosphere, 1500Mt is deposited to P TitlePage 10 a the land and 500Mt is deposited onto the ocean surface (Shao et al., 2011). It affects pe r Abstract Introduction radiationbyabsorptionandscatteringthatinturnaffectssurfaceandatmospherictem- | perature, also it acts as ice cloud condensation nuclei (ICCN) that impacts the micro- Conclusions References physics of the clouds and its radiative properties (IPCC, 2013). The transported dust D is Tables Figures also carries nutrients and bacteria, which may affect the marine life and land surface c 15 u s life. For instance, 20Mt of nutrients rich Saharan dust is transported to the Amazon s io (cid:74) (cid:73) basin in South America each year (Koren et al., 2006). Atmospheric mineral dust, es- n P pecially fine dust (smaller than 2.5µm) may cause cardiopulmonary disease and lung a (cid:74) (cid:73) p cancer (Giannadaki et al., 2014). It also affects many of human activities like aviation, e r Back Close real estate construction, agriculture, and water resource management (Stefanski and 20 | Sivakumar, 2009). FullScreen/Esc D Dust particles are emitted from major deserts on the globe (e.g., Sahara, the Ara- is c bianPeninsula,TaklamakanandGobidesertsinChina,AustraliadesertsandAtacama u Printer-friendlyVersion s desertinChile).Themineraldustemissionprocesses(i.e.,saltationandsandblasting) sio InteractiveDiscussion n are determined by meteorological conditions (e.g., atmospheric instability, soil mois- 25 P ture). Large-scale wind systems may carry coarse and fine dust particles horizontally a p e for thousands of kilometers (e.g., Koren et al., 2006) and vertically up 6 to 10km (e.g., r 1525 | Gobbi et al., 2004). Depending on dust particle size and atmospheric conditions dust D is particles may remain airborne between 1.5 to 7.4 days and will then deposit to the c ACPD u s surface by gravitational settling or rain washout (Shao et al., 2011). s io 15,1523–1571,2015 This paper focuses on atmospheric mineral dust over the Arabian Peninsula, a re- n P 5 gion,whichhasbeenlessstudiedoverthelastdecadescomparedtoSaharanregions. a p In an early study Wilkerson (1991) described the onset and evolution of dust storms e The climatology of r overIraqandKuwaitanalyzingsatelliteimagesandvisbilityrecords.Wilkerson’sstudy dust aerosol over the | categorizedthetypesofduststormsaspre-frontal,post-frontalandShamaltypes,the arabian peninsula D latter being a regional northwesterly wind that typically occurs in summer (“Shamal” is c A.Shalabyetal. 10 denotes “northerly wind” in Arabic). Mashat et al. (2008) used meteorological obser- u s vationsfromSaudiArabiatoanalyzethemeteorologicalconditionswhichfavoreddust s io storms and their spatial and seasonal distribution. They found that dust storms most n P TitlePage frequently occur in the eastern part of the Arabian Peninsula in spring and extend a p e towards the southern part of the Arabian Peninsula in summer. Alharbi (2009) and Al- r Abstract Introduction harbi et al. (2013) explored dust storms generation over the Arabian Peninsula, dust 15 | Conclusions References source regions and atmospheric conditions that promote dust storms, which include D large-scale atmospheric instability, high surface winds, and dry, rich dust sources. The is Tables Figures c quantification and characterization of mineral dust aerosol (e.g., dust concentrations u s s profile and mineral dust optical properties) became possible after the installation of io (cid:74) (cid:73) n TheAErosolROboticNETwork(AERONET)inBahrainandSaudiArabiaaround1998. 20 Pa (cid:74) (cid:73) Smirnovetal.(2002)usedaoneyeardatarecord(July1998–July1999)oftheBahrain p e siteintheArabianGulftodeducetheclimatologyofaerosolopticalproperties(i.e.,the r Back Close AerosolOpticalDepth(AOD)andtheÅngströmparameter).In2004anintensivemea- | FullScreen/Esc surement campaign was held in the United Arab Emirates (UAE), which led to the D installation of various AERONET stations. Among them only the Mezaira site is fully is 25 c u Printer-friendlyVersion operationaluptodate.Thiscampaigncharacterizedthenatureofatmosphericaerosol s s over the UAE and validated satellite aerosol products over a bright surface such as io InteractiveDiscussion n desert (Reid et al., 2005; Eck et al., 2008). Kim et al. (2011) studied AERONET data P a records from North Africa and the Arabian Peninsula and derived the seasonal behav- p e r 1526 | ior of mineral dust aerosol optical properties in these regions. This study showed that D is theArabianPeninsuladustismoreabsorbingintheshortwaverangethantheSaharan c ACPD u s dust does. García et al. (2012) analyzed most of the AERONET stations on the globe s io 15,1523–1571,2015 includingtheArabianPeninsulaandfoundthatatmosphericaerosolsoverhighsurface n P 5 albedo regions, as deserts, lead to a warming of the Earth’s atmosphere. a p Using a regional climate model offers the advantage to investigate meso-scale phe- e The climatology of r nomena such as surface dust emission processes, transportation, deposition and its dust aerosol over the | radiative impact on the regional climate at a higher spatial resolution (e.g., ≤50km) arabian peninsula D than global models with coarse spatial resolution (e.g., ≥100km). The Regional Cli- is c A.Shalabyetal. 10 mate Model Version 3 (RegCM3) (Pal et al., 2007) and RegCM4 (Giorgi et al., 2012) u s havebeenusedtoinvestigatemineraldustaerosolsanditsradiativeimpactonregions s io in North and West Africa (Zakey et al., 2006; Solmon, 2008; Konarè et al., 2008; Na- n P TitlePage pat et al., 2012; Steiner et al., 2014), but studies on the Middle East area are rare. a p e Marcella and Eltahir (2010, 2011) used RegCM3 for the first time over the Northern r Abstract Introduction Arabian Peninsula, to explore the impact of dust on climate and found that the im- 15 | Conclusions References plementation of subgrid wind variability and dust aerosol lateral boundary conditions D couldenhancethedustsimulation.NazrulIslamandAlmazroui(2012)usedRegCM4, is Tables Figures c showed that the direct radiative effect of mineral dust results in a decrease of surface u s s temperatureandincreaseinprecipitationovertheArabianPeninsulainthewetseason io (cid:74) (cid:73) n (November–April). 20 Pa (cid:74) (cid:73) The first objective of this study is to define and identify the dust season by means of p e available observations (e.g. through AERONET and satellite data bases) and regional r Back Close climate modeling. As the uncertainties of aerosol radiative forcing are large (IPCC, | FullScreen/Esc 2013), more investigation on the sources of such uncertainties, especially on regional D scales,isneeded.Thus,thesecondobjectiveofthisstudyistoestimatetheBottomOf is 25 c u Printer-friendlyVersion AtmosphereRadiativeForcing(BOARF)andtheTopofAtmosphereRadiativeForcing s s (TOARF) of mineral dust aerosols using AERONET products and the model. These io InteractiveDiscussion n objectives integrate observations and modeling to obtain a better assessment of the P a atmospheric behavior of mineral dust aerosol. p e r 1527 | 2 Methods D is c ACPD u 2.1 Observational datasets s s io 15,1523–1571,2015 n The observation of atmospheric dust from space has become very vital to the under- P a standing of this phenomenon. Space-borne observations provide global information p e The climatology of of the spatial distribution (horizontal and vertical) and the temporal evolution of the r 5 dust aerosol over the AerosolOpticalDepth(AOD).However,comparisonofmodeleddust(AOD)withsatel- | arabian peninsula litemeasurementsmaybeproblematic,sincesatelliteobservationsapplyassumptions D about the nature of the aerosol, which is different from approaches in numerical mod- isc A.Shalabyetal. u els (Woodward, 2001). On the other hand, ground-based observation is needed to s s provide more details of atmospheric dust characteristics such as high frequency mea- io 10 n surements of near ground concentration (e.g., PM10) and AOD. Also ground-based P TitlePage a p measurements include physical and chemical analysis of aerosols, such as its geo- e r Abstract Introduction metrical shape, size distribution and chemical composition. The ground-based mea- | surements are also used to validate the space-borne observations. For instance, the Conclusions References AOD from AERONET stations has been used to validate satellite AOD products (Mar- D 15 is Tables Figures tonchik et al., 2004; Abdou et al., 2005). In the following subsections we will briefly c u s discuss the characteristics of AERONET and satellite datasets. s io (cid:74) (cid:73) n 2.1.1 AERONET Pa (cid:74) (cid:73) p e The AErosol RObotic NETwork (AERONET) provides valuable measurements of at- r Back Close mospheric aerosol optical properties. AERONET stations comprise an automatic sun | 20 FullScreen/Esc photometer.Ithasitsownalgorithmtoevaluatedataqualityandinstrumentfunctional- D ity.AERONETmeasuresAODfordifferentwavelengths,fromnearIR(1064nm)tonear isc u Printer-friendlyVersion UV(340nm).AERONETsunphotometerprovidesinformationaboutaerosolsizedistri- s s bution,aerosolradiativeforcing,andaerosolshape(spherical/non-spherical).Detailed ion InteractiveDiscussion description of the instrument and its function can be found in Holben et al. (1998). P 25 a p e r 1528 | The measurements pass multilevel quality assurance (QA): level 1 without cloud D is screening, level 1.5 has cloud screened but may be without final calibration. Level 2 c ACPD u s has cloud screened and quality assured calibrations. The AOD accuracy of calibrated s io 15,1523–1571,2015 AERONET station is wavelength dependent and varies from root mean square error n P 5 of ±0.012 (UV band) to root mean square error of ±0.006 (IR band) at overhead sun a p (airmass=1) (Schmid et al., 1999). e The climatology of r The AERONET stations are widely used over the globe. However there are only few dust aerosol over the | AERONETstationsintheArabianPeninsulawithrelativelylongmeasurementrecords. arabian peninsula D AmongthemistheSolar-VillagestationinSaudiArabia,theKuwaitUniversitystationin is c A.Shalabyetal. 10 KuwaitandtheMezairasiteintheUnitedArabEmirates(UAE).Fordetailedinformation u s about the site and their locations see Table 1 and Fig. 1. s io Each station has its unique features; the Kuwait station is downwind of major dust n P TitlePage sourcesinSouthernIraq.Solar-VillageislocatedinthecenterofSaudiArabiaanditis a p e atarelativelyhighaltitude(764ma.s.l.)comparedtoothersites,whichcouldreflectthe r Abstract Introduction transportationofdusttohighaltitudes.MezairainUnitedArabEmirate(UAE)isinland 15 | Conclusions References site and could be considered a receptor site for dust coming from southern Arabian D Peninsula. is Tables Figures c u s 2.1.2 MISR sio (cid:74) (cid:73) n P Most satellite instruments looks down to the earth (0 degrees Nadir angle), or toward a (cid:74) (cid:73) p the edge of the planet and receive the reflected sunlight. The Earth’s surface, clouds e 20 r Back Close and aerosols reflect sunlight in different direction, which requires an instrument that | accounts for such different reflected angle. The Multiangle Imaging Spectroradiometer FullScreen/Esc D (MISR) is a unique instrument flown in the space since late 1998. It has nine cameras is c corresponding to nine view angles, the middle one pointing toward the nadir, four of u Printer-friendlyVersion s 25 theminaforwarddirectionwith(26.1,45.6,60.0and70.0◦,respectively)relativetothe sio InteractiveDiscussion nadir camera and the other four in the rearward direction with (−26.1, −45.6, −60.0 n P and −70.0◦ respectively). Each direction measures four individual wavelength (443, a p e 555, 670 and 865nm). Therefore, MISR has 36 channels (Diner et al., 1998). MISR r 1529 | has been used in many fields including studies on, clouds (Marchand et al., 2010), D is aerosols (Martonchik et al., 2004; Abdou et al., 2005; Marey et al., 2011) and Earth c ACPD u s surface (Pinty et al., 2011). For our purpose, we will focus on the aerosol facilities of s io 15,1523–1571,2015 MISR. The MISR global aerosol retrieval is used to obtain AOD values to characterize n P 5 thetypesofaerosolbasedontheirphysicalandopticalproperties,andaerosolparticle a p shape (spherical or non-spherical). The aerosol retrieval algorithm strategy has many e The climatology of r steps.First,itutilizesalookuptablethatcontainsasuiteofanaturalaerosoltypesand dust aerosol over the | calculatedaerosolopticalpropertiesbyaradiativetransfermodel.Second,theretrieval arabian peninsula D of theaerosol overdark surfaceslike theocean dependson red andnear infrared(IR) is 10 channels (Martonchik et al.,1998). The most difficult retrieval is above bright surface cu A.Shalabyetal. s like deserts. This latter retrieval needs special treatment and is therefore associated s io with large uncertainties (Diner, 1998; Martonchik, 1998; Abdou et al., 2005). n P TitlePage In this work we concentrate only on the MISR AOD dataset and its comparison with a p e the model output. The MISR data used in our study spans 6 years from 2006 to 2012. r Abstract Introduction ◦ ◦ The data resolution is 0.5 ×0.5 . Data has been retrieved from the Giovanni website 15 | Conclusions References (http://Giovanni.gsfc.nasa.gov/giovanni). D is Tables Figures c 2.1.3 MODIS (Deepblue) u s s io (cid:74) (cid:73) The Moderate Resolution Imaging Spectroradiometer (MODIS) instrument is aboard n P the NASA EOS (Earth Observing System) Terra and Aqua satellites and began trans- a (cid:74) (cid:73) p mitting date in 2000. The instruments have high spatial resolution (10km resolu- e 20 r Back Close tion) and almost global coverage. Like MISR, MODIS aerosol retrievals are based on | a lookup table procedure in which the satellite measured radiances are matched to FullScreen/Esc D pre-calculated values in the lookup table. The values of the aerosol properties used to is c create the calculated radiances are retrieved (Remer et al., 2005; Abdou et al., 2005). u Printer-friendlyVersion s 25 The MODIS data set spans from January 2008 to December 2011. Data has been sio InteractiveDiscussion retrieved from the Giovanni website (http://Giovanni.gsfc.nasa.gov/giovanni). n P a p e r 1530 | 2.1.4 OMI D is c ACPD u TheOzoneMonitoringInstrument(OMI)hasbeenorbitingtheEarthononeoftheEOS s s mission “Aura spacecraft” since July 2004. OMI is a high spatial resolution (13km× io 15,1523–1571,2015 n 24km) ground pixel size ultraviolet/visible (UV/VIS) backscatter spectrometer (Levelt P a 5 et al., 2006). The OMI aerosol retrieval algorithm is the same as for the Total Ozone pe The climatology of r Mapping spectrometer (TOMS) near-UV method of aerosol absorption sensing from dust aerosol over the space (Torres et al., 2005). The accuracy of the OMI retrieval of aerosol optical depth | arabian peninsula is around 30% relative to AERONET measurements (Torres et al., 2005). D is The extinction AOD at 500nm (near-UV) has been selected from the OMI product c A.Shalabyetal. u s 10 from Giova◦nni◦site portal (http://Giovanni.gsfc.nasa.gov/giovanni). From that data we sio selected 2 ×2 box around the three AERONET stations (see Fig. 1) for comparison n purposes with the observations. The time span of this data is from January 2008 to Pa TitlePage p December 2011. e r Abstract Introduction | 2.2 Model description and experimental design Conclusions References D is Tables Figures 15 2.2.1 General model description cu s s The International Centre for Theoretical Physics (ICTP) Regional Climate Model io (cid:74) (cid:73) n (RegCM) was built upon the National Center for Atmospheric Research (NCAR) P a (cid:74) (cid:73) Mesoscale Model version 4 (MM4) (Giorgi and Bates, 1989; Giorgi et al., 1993a, b). p e The Regional Climate Model version 4 (RegCM4) is the second major development r Back Close oftheRegCMcoreafterRegCM3(Paletal.,2007).Thecodingstructureiscompletely | 20 FullScreen/Esc changed.IthasbecometotallyFORTRAN90compliantandmodularstructuredandits D parallelizationandmemorymanagementhasbecomemoreefficient.TheRegCM4has isc u Printer-friendlyVersion morephysicsoptions,whichincludetheCommunityLandsurfaceModel(CLM3.5)land s s surfaceparameterization,theTiedtkeconvectionscheme,theUniversityofWashington ion InteractiveDiscussion (UW)planetaryboundarylayer(PBL)scheme,andtheRapidRadiativeTransferModel P 25 a p (RRTM) (Giorgi et al., 2012). e r 1531 | RegCM4 is an online climate-chemistry model and has an online gas-phase chem- D is istry scheme (CBMZ) (Shalaby et al., 2012). It has various aerosol components such c ACPD u s as,foursizebindust,twosizebinseasalt,sulphate,blackcarbonandorganiccarbon. s io 15,1523–1571,2015 Sulphate, black carbon and organic carbon (Solmon et al., 2006), as well as dust and n P 5 sea-salt are radiatively active (Zakey et al., 2006, 2008). a p e The climatology of r 2.2.2 Dust parameterization dust aerosol over the | arabian peninsula Sand particles are affected by many forces that determine their fate. Dust particles D havethreedynamicmodes:(a)saltation:smallparticlesmovebyjumpinglikeleap-frog, isc A.Shalabyetal. u Onceliftedbywinditwilldriftdownwindandreturntohitthegroundagainandtransfer s s energy and momentum to other soil aggregates (soil particles). (b) Creeping: large io 10 n dustparticlescannotbeliftedintotheair,butwilljustmoveandslideontheground.(c) P TitlePage a p Suspension:iftheupwarddraftisstrongenoughtocompensatethegravitationalforce e r Abstract Introduction of the dust particles, dust will remain airborne and be transported by the wind over | longer distances until its gravitational force overcome the uplifting force. It is believed Conclusions References that saltation is the main mechanism for surface dust emission (Shao et al., 1993; D 15 is Tables Figures Marticorena and Bergametti, 1995). c u s FollowingMarticorenaandBergametti(1995)andAlfaroandGomes(2001),acom- s io (cid:74) (cid:73) plex dust emission scheme has been implemented in RegCM3 (Zakey et al., 2006). n P This emission scheme is based on parameterization of soil aggregate saltation and a (cid:74) (cid:73) p sandblasting processes. According to this scheme, a critical parameter for the dust e 20 r Back Close saltationprocessisthethresholdfrictionvelocityu∗,whichisafunctionofparticlesize t | D (Eq. 1), such that, u∗ represents an ideal minimum threshold friction velocity, f is FullScreen/Esc p ts eff acorrectionfactoraccountingfortheeffectofsurfaceroughnessandfw isafactorthat Dis accounts for the effect of soil moisture content on the threshold friction velocity. The cu Printer-friendlyVersion s 25 particle size is determined by the land surface soil texture. Calculating the threshold sio InteractiveDiscussion frictionvelocityisrequiredtocalculatethehorizontaldustflux(dH )(Eq.2),suchthat, n F P E is the ratio of erodible to total surface, dS is the relative surface of soil aggregate a rel p e of diameter D to the total aggregate surface and R(D ) is the ratio of the threshold r p p 1532 |