observed and modeled air-sea flux in the northwest ... · sebas

1
Observed and Modeled Air-Sea Flux in the Northwest Tropical Atlan<c Introduc<on A surface mooring site, called the Northwest Tropical Atlan<c Sta<on (NTAS) has been maintained with NOAA support since 2001. Surface meteorological variables observed on the NTAS buoy are withheld from models and made available as independent reference data to explore air-sea interac<on processes, validate remote sensing products, anchor hybrid air-sea flux fields, and inves<gate model performance. Methods The buoy meteorological measurements are a con<guous 11 year data set at 1 minute intervals from mooring deployments between April 2001 and March 2012. For comparison to models, the buoy data were averaged to daily values and adjusted to standard heights. Air-Sea fluxes were es<mated using the COARE 3.0 bulk algorithm (Fairall 1996; 2003), which takes into account air stability. The buoy fluxes were used to describe the seasonal cycle, to iden<fy the presence of the ITCZ, and to diagnose the strengths and weaknesses of three reanalysis model products: ERA-Interim (European Center for Medium-Range Weather Forecasts), MERRA-2 (Na<onal Aeronau<cs and Space Administra<on). NCEP-2 (Na<onal Centers for Environmental Predic<ons) Conclusions Na<ve fluxes from ERA, NCEP and MERRA showed large differences in individual flux components and net heat flux (Qnet) rela<ve to the buoy. Using a common bulk flux algorithm brought the turbulent flux components closer to the buoy values and reduced the Qnet discrepancies to about 10 W/m 2 . Radia<ve fluxes from the reanalysis products were inconsistent in reproducing the observed annual cycle. ERA performed well, MERRA had persistent biases, and NCEP showed large discrepancies. A regional assessment comparing remote sensing products to the models indicated that seasonal variability in cloud cover (as diagnosed from longwave radia<on) is not well reproduced in some of the models, leading to the radia<ve flux discrepancies. Precipita<on showed rela<vely good agreement among observa<ons, models and remote sensing products. Thus, to the extent that radia<on differences were due to clouds, precipita<ng clouds did not appear to be the main culprit. Sebas<en Bigorre, Albert J. Plueddemann and Robert A. Weller Woods Hole Oceanographic Ins<tu<on Woods Hole, MA 02543 USA Results – Buoy flux vs. models Comparing the na<ve fluxes (as received in the reanalysis models) to the buoy fluxes showed significant differences (larger than the buoy errors) in several flux components. The eleven-year mean buoy Qnet showed an ocean heat gain of 42 W/m 2 , while NCEP and MERRA showed heat loss of a few W/m 2 . Applying the COARE algorithm to the surface meteorological variables from the models brought the turbulent flux components closer to the buoy values and reduced the Qnet discrepancies to about 10 W/m 2 . Influence of the ITCZ Trade winds from the Northern and Southern hemispheres converge at the Intertropical Convergence Zone (ITCZ), which can be iden<fied as an east-west precipita<on band migra<ng seasonally north and south. The ITCZ has a mean loca<on near 6 N in the Atlan<c. Cloud cover associated with the ITCZ influences the regional radia<ve forcing. The seasonal cycle is the dominant signal in heat flux at the NTAS site, but there are notable impacts from the northernmost excursion of the ITCZ in the fall (Sep-Nov): Wind speeds drop, precipita<on peaks, and a dis<nct “shoulder” is seen in longwave radia<on. The NTAS site is located at 51°W, 15°N, roughly 1,000 km east of the lesser An<lles, in the North Atlan<c trade wind regime. The NTAS buoy is oueifed with ASIMET meteorological instrumenta<on (Hosom et al. 1995), measuring ten variables that enable fluxes of momentum, heat and moisture to be computed. Moorings are replaced annually. Calibra<on, intercomparison, and characteriza<on of errors (e.g. island heat effect, flow distor<on) improve measurement accuracy. Heat flux component errors are 2-4 W/m 2 , while the error in net heat flux is about 8 W/m 2 (Colbo and Weller (2009) and Bigorre et al. (2013). Data Availability Time series data of surface meteorology for individual deployments are available from hfp://uop.whoi.edu/currentprojects/NTAS/ntasarchive.html Concatenated, mul<-year <me series of meteorological data and fluxes are available from hfp://uop.whoi.edu/ReferenceDataSets/ntasreference.html Eleven-year average net heat flux from the NTAS buoy compared to (lej) the na<ve fluxes from the models and (right) fluxes computed using the COARE algorithm and surface meteorological variables from the models. Error bars of +/- 8 W/m 2 are shown for the buoy flux. The majority of the discrepancy in net heat flux ajer applying the COARE algorithm was found in the radia<ve fluxes. Shortwave radia<on (SWR): ERA in good agreement with buoy; MERRA overes<mates ocean heat gain; NCEP underes<mates ocean heat gain, especially in winter. Longwave radia<on (LWR): ERA in good agreement with buoy; MERRA shows a persistent low bias; NCEP is a poor match to the observed annual cycle. The annual cycle of precipita<on was reproduced reasonably well by all three reanalysis products, although with notable differences in amplitude, especially near ITCZ . ERA and NCEP are high compared to the buoy while MERRA is similar. There is a good agreement between buoy measurements and satellite remote sensing products for precipita<on (Tropical Rainfall Measurement Mission; TRMM), shortwave and longwave radia<on (Clouds and Earth’s Radiant Energy System; CERES). Eleven-year mean annual cycle of (lej) precipita<on, (middle) shortwave radia<on, (right) longwave radia<on at the NTAS site from the buoy compared to three reanalysis models. Remote sensing observa<ons from (lej) TRMM (middle and right) CERES are also shown. The <me period influenced by the ITCZ is shaded. Results – Satellite flux vs. models Satellite remote sensing (e.g. CERES) and model products allowed regional paferns of the annual cycle to be examined: MERRA shows a pafern of LWR variability comparable to that of CERES, although with a notable low bias. NCEP radia<on paferns are different in outer tropics in winter (similar to discrepancy with NTAS buoy), and near Equator in summer. Precipita<on is well represented in models, although it is enhanced near South America in NCEP . Contours of precipita<on rate (mm/d) in the topical Atlan<c for (lej) January and (right) October. These are 17-year monthly composites from blended satellite and rain gauge products (Chiang et al., 2002). The NTAS site is indicated by the green dot. Acknowledgment This research was supported by the Na<onal Oceanic and Atmospheric Administra<on (NOAA), Climate Program Office (CPO FundRef number (100007298), Ocean Observing and Monitoring Division, through the Coopera<ve Ins<tute for the North Atlan<c Region (CINAR) under Coopera<ve Agreement NA14OAR4320158. NTAS is part of the OceanSITES network (www.oceansites.org Hovmoller diagrams for: net longwave radia<on (W/m 2 , lej), precipita<on rate (mm/d, right) showing the annual cycle vs. la<tude for eleven-year means. Solid lines indicate the maximum value for each month. A dashed line is shown at the la<tude of the NTAS mooring. Based on longitudinal averages from 51W to 44W. Maps of net longwave (lej) and shortwave (right) radia<on (W/m 2 ), for winter and summer, using data from 2001 to 2012.

Upload: others

Post on 30-Sep-2020

0 views

Category:

Documents


0 download

TRANSCRIPT

Page 1: Observed and Modeled Air-Sea Flux in the Northwest ... · Sebas

ObservedandModeledAir-SeaFluxintheNorthwestTropicalAtlan<c

Introduc<onAsurfacemooringsite,calledtheNorthwestTropicalAtlan<cSta<on(NTAS)hasbeenmaintainedwithNOAAsupportsince2001.SurfacemeteorologicalvariablesobservedontheNTASbuoyarewithheldfrommodelsandmadeavailableas independent reference data to explore air-sea interac<on processes, validate remote sensing products, anchorhybridair-seafluxfields,andinves<gatemodelperformance.

MethodsThe buoy meteorological measurements are a con<guous 11 year data set at 1 minute intervals from mooringdeploymentsbetweenApril2001andMarch2012.Forcomparisontomodels, thebuoydatawereaveragedtodailyvalues and adjusted to standard heights. Air-Sea fluxeswere es<mated using the COARE 3.0 bulk algorithm (Fairall1996;2003),whichtakesintoaccountairstability.

Thebuoyfluxeswereused todescribe the seasonal cycle, to iden<fy thepresenceof the ITCZ,and todiagnose thestrengthsandweaknessesofthreereanalysismodelproducts:

•  ERA-Interim(EuropeanCenterforMedium-RangeWeatherForecasts),•  MERRA-2(Na<onalAeronau<csandSpaceAdministra<on).

•  NCEP-2(Na<onalCentersforEnvironmentalPredic<ons)

Conclusions•  Na<vefluxesfromERA,NCEPandMERRAshowedlargedifferencesinindividualfluxcomponents

andnetheatflux(Qnet)rela<vetothebuoy.•  Usingacommonbulkfluxalgorithmbroughttheturbulentfluxcomponentsclosertothebuoy

valuesandreducedtheQnetdiscrepanciestoabout10W/m2.•  Radia<vefluxesfromthereanalysisproductswereinconsistentinreproducingtheobservedannual

cycle.ERAperformedwell,MERRAhadpersistentbiases,andNCEPshowedlargediscrepancies.•  Aregionalassessmentcomparingremotesensingproductstothemodelsindicatedthatseasonal

variabilityincloudcover(asdiagnosedfromlongwaveradia<on)isnotwellreproducedinsomeofthemodels,leadingtotheradia<vefluxdiscrepancies.

•  Precipita<onshowedrela<velygoodagreementamongobserva<ons,modelsandremotesensingproducts.Thus,totheextentthatradia<ondifferenceswereduetoclouds,precipita<ngcloudsdidnotappeartobethemainculprit.

Sebas<enBigorre,AlbertJ.PlueddemannandRobertA.WellerWoodsHoleOceanographicIns<tu<onWoodsHole,MA02543USA

Results–Buoyfluxvs.modelsComparing thena<vefluxes (as received in the reanalysismodels) to thebuoyfluxes showed significantdifferences(largerthanthebuoyerrors)inseveralfluxcomponents.Theeleven-yearmeanbuoyQnetshowedanoceanheatgainof42W/m2,whileNCEPandMERRAshowedheat lossofa fewW/m2.ApplyingtheCOAREalgorithmtothesurfacemeteorological variables from the models brought the turbulent flux components closer to the buoy values andreducedtheQnetdiscrepanciestoabout10W/m2.

InfluenceoftheITCZTrade winds from the Northern and Southern hemispheres converge at the Intertropical Convergence Zone (ITCZ),whichcanbeiden<fiedasaneast-westprecipita<onbandmigra<ngseasonallynorthandsouth.TheITCZhasameanloca<onnear6NintheAtlan<c.CloudcoverassociatedwiththeITCZinfluencestheregionalradia<veforcing.

The seasonal cycle is the dominant signal in heat flux at the NTAS site, but there are notable impacts from thenorthernmost excursion of the ITCZ in the fall (Sep-Nov): Wind speeds drop, precipita<on peaks, and a dis<nct“shoulder”isseeninlongwaveradia<on.

TheNTASsiteislocatedat51°W,15°N,roughly1,000kmeastofthelesserAn<lles,intheNorthAtlan<ctradewindregime.TheNTASbuoyisoueifedwithASIMETmeteorologicalinstrumenta<on(Hosometal.1995),measuringtenvariablesthatenablefluxesofmomentum,heatandmoisturetobecomputed.Mooringsarereplacedannually.Calibra<on,intercomparison,andcharacteriza<onoferrors(e.g.islandheateffect,flowdistor<on)improvemeasurementaccuracy.Heatfluxcomponenterrorsare2-4W/m2,whiletheerrorinnetheatfluxisabout8W/m2(ColboandWeller(2009)andBigorreetal.(2013).

DataAvailabilityTimeseriesdataofsurfacemeteorologyforindividualdeploymentsareavailablefromhfp://uop.whoi.edu/currentprojects/NTAS/ntasarchive.html

Concatenated,mul<-year<meseriesofmeteorologicaldataandfluxesareavailablefromhfp://uop.whoi.edu/ReferenceDataSets/ntasreference.html

Eleven-yearaveragenetheatfluxfromtheNTASbuoycomparedto(lej)thena<vefluxesfromthemodelsand(right)fluxescomputedusingtheCOAREalgorithmandsurfacemeteorologicalvariablesfromthemodels.Errorbarsof+/-8

W/m2areshownforthebuoyflux.

ThemajorityofthediscrepancyinnetheatfluxajerapplyingtheCOAREalgorithmwasfoundintheradia<vefluxes.

•  Shortwave radia<on (SWR): ERA in good agreement with buoy; MERRA overes<mates ocean heat gain; NCEPunderes<matesoceanheatgain,especiallyinwinter.

•  Longwaveradia<on(LWR): ERAingoodagreementwithbuoy;MERRAshowsapersistentlowbias;NCEPisapoormatchtotheobservedannualcycle.

The annual cycle of precipita<on was reproduced reasonably well by all three reanalysis products, although withnotabledifferencesinamplitude,especiallynearITCZ.ERAandNCEParehighcomparedtothebuoywhileMERRAissimilar.There is a good agreement between buoy measurements and satellite remote sensing products for precipita<on(TropicalRainfallMeasurementMission;TRMM),shortwaveandlongwaveradia<on(CloudsandEarth’sRadiantEnergySystem;CERES).

Eleven-yearmeanannualcycleof(lej)precipita<on,(middle)shortwaveradia<on,(right)longwaveradia<onattheNTASsitefromthebuoycomparedtothreereanalysismodels.Remotesensingobserva<onsfrom(lej)TRMM

(middleandright)CERESarealsoshown.The<meperiodinfluencedbytheITCZisshaded.

Results–Satellitefluxvs.modelsSatelliteremotesensing(e.g.CERES)andmodelproductsallowedregionalpafernsoftheannualcycletobeexamined:MERRAshowsapafernofLWRvariabilitycomparabletothatofCERES,althoughwithanotablelowbias.NCEPradia<onpafernsaredifferentinoutertropicsinwinter(similartodiscrepancywithNTASbuoy),andnearEquatorinsummer.Precipita<oniswellrepresentedinmodels,althoughitisenhancednearSouthAmericainNCEP.

Contoursofprecipita<onrate(mm/d)inthetopicalAtlan<cfor(lej)Januaryand(right)October.Theseare17-yearmonthlycompositesfromblendedsatelliteandraingaugeproducts(Chiangetal.,2002).TheNTAS

siteisindicatedbythegreendot.

AcknowledgmentThisresearchwassupportedbytheNa<onalOceanicandAtmosphericAdministra<on(NOAA),ClimateProgramOffice(CPOFundRefnumber(100007298),OceanObservingandMonitoringDivision,throughtheCoopera<veIns<tutefortheNorthAtlan<cRegion(CINAR)underCoopera<veAgreementNA14OAR4320158.NTASispartoftheOceanSITESnetwork(www.oceansites.org

Hovmollerdiagrams for: net longwaveradia<on (W/m2, lej),precipita<onrate (mm/d, right) showing theannualcycle vs. la<tude for eleven-yearmeans. Solid lines indicate themaximumvalue for eachmonth.Adashed line isshownatthela<tudeoftheNTASmooring.Basedonlongitudinalaveragesfrom51Wto44W.

Mapsofnetlongwave(lej)andshortwave(right)radia<on(W/m2),forwinterandsummer,usingdatafrom2001to2012.