1、Assimilation of Ocean Surface Wind Data by the HY-2B Satellitein GRAPES:Impacts on Analyses and ForecastsJincheng WANG1,2,Xingwei JIANG*3,4,Xueshun SHEN1,2,Youguang ZHANG3,4,Xiaomin WAN1,2,Wei HAN1,2,and Dan WANG1,21CMA Earth System Modeling and Prediction Center,China Meteorological Administration,
2、Beijing 100081,China2National Meteorology Center,China Meteorological Administration,Beijing 100081,China3Southern Marine Science and Engineering Guangdong Laboratory(Guangzhou),Guangzhou 511458,China4National Satellite Ocean Application Service,Beijing 100081,China(Received 25 October 2021;revised
3、7 April 2022;accepted 6 May 2022)ABSTRACTThe ocean surface wind (OSW)data retrieved from microwave scatterometers have high spatial accuracy andrepresent the only wind data assimilated by global numerical models on the ocean surface,thus playing an important role inimproving the forecast skills of g
4、lobal medium-range weather prediction models.To improve the forecast skills of theGlobal/Regional Assimilation and Prediction System Global Forecast System(GRAPES_GFS),the HY-2B OSW data isassimilated into the GRAPES_GFS four-dimensional variational assimilation(4DVAR)system.Then,the impacts of theH
5、Y-2B OSW data assimilation on the analyses and forecasts of GRAPES_GFS are analyzed based on one-monthassimilation cycle experiments.The results show that after assimilating the HY-2B OSW data,the analysis errors of thewind fields in the lower-middle troposphere (1000600 hPa)of the tropics and the s
6、outhern hemisphere (SH)aresignificantly reduced by an average rate of about 5%.The impacts of the HY-2B OSW data assimilation on the analysisfields of wind,geopotential height,and temperature are not solely limited to the boundary layer but also extend throughoutthe entire troposphere after about tw
7、o days of cycling assimilation.Furthermore,assimilating the HY-2B OSW data cansignificantly improve the forecast skill of wind,geopotential height,and temperature in the troposphere of the tropics andSH.Key words:HY-2B,ocean surface wind,4DVAR,GRAPES-GFS,medium-range weather forecastCitation:Wang,J.
8、C.,X.W.Jiang,X.S.Shen,Y.G.Zhang,X.M.Wan,W.Han,and D.Wang,2023:Assimilation ofocean surface wind data by the HY-2B satellite in GRAPES:Impacts on analyses and forecasts.Adv.Atmos.Sci.,40(1),4461,https:/doi.org/10.1007/s00376-022-1349-2.Article Highlights:The impacts of the HY-2B OSW data assimilation
9、(in 4DVAR)on the global analysis fields are not limited to theboundary layer.The OSW data assimilation impacts on analysis can extend to the whole troposphere after about two days of cyclingassimilation and then remain stable.The OSW data provide a significant positive impact on the GRAPES_GFS forec
10、ast skill in the troposphere of the tropicsand the southern hemisphere.1.IntroductionThe Global Data Assimilation System(GDAS)is oneof the core components of the global numerical weather pre-diction(NWP)system,which provides the initial conditionfor integrating an NWP model.Presently,two main catego
11、riesof observational data can be assimilated into the GDAS,1)conventional observational data,including surface observa-tions,upper-air observations,aircraft-based observations,and ship observations,and 2)observations that are gatheredby remote sensing satellites,which are mainly comprised ofthe brig
12、htness temperature data of the polar-orbiting and geo-stationary satellites,the refractivity data of the Global Naviga-tion Satellite System-Radio Occultation (GNSS-RO),andthe cloud-derived wind data retrieved from the polar-orbitingand geostationary satellites.The satellite observations effec-tivel
13、y fill the shortage of conventional observational data*Corresponding author:Xingwei JIANGEmail:ADVANCES IN ATMOSPHERIC SCIENCES,VOL.40,JANUARY 2023,4461 Original Paper Institute of Atmospheric Physics/Chinese Academy of Sciences,and Science Press and Springer-Verlag GmbH Germany,part of Springer Nat
14、ure 2023over the ocean and thus significantly improve the skill ofglobal numerical weather forecasts.Satellite observationsare playing an increasingly important role in promoting thedevelopment of todays NWP(Cardinali,2009).Despite a growing amount of satellite observations,themost collected variabl
15、e by satellite observations is the bright-ness temperature(mass field).The number of wind(motionfield)observations is relatively lower over the ocean,withonly the atmospheric motion vectors(AMVs)data availablein cloud regions (Feng and Wang,2019).Becausegeostrophic balance between the motion field a
16、nd the massfield is not satisfied in tropical regions,the existing dataassimilation algorithm is incapable of deriving the high preci-sion motion field from the observational data of the massfield,which limits the accuracy of the initial condition poten-tially leading to poor forecasting skill for t
17、yphoon and otherdisastrous weather events in the tropics.To remedy the lackof wind observations over the ocean,the microwave scat-terometer was developed to observe the ocean surface wind(OSW)(Figa-Saldaa and Stoffelen,2000;Figa-Saldaa etal.,2002).Owing to the continuous improvement of theOSW retrie
18、val algorithm,the accuracy of the OSW data isalso dramatically enhanced(Hersbach et al.,2007;Wang etal.,2015;De Kloe et al.,2017;Lin et al.,2017a;Wang etal.,2017).The main NWP operational centers across theworld have incorporated the assimilation of the OSW datainto operation,which greatly improved
19、the forecast skills ofglobal NWP(Ayina et al.,2006;Candy and Keogh,2006;Hersbach,2010;Bi et al.,2011;Laloyaux et al.,2016;DeChiara et al.,2016,2017;Duan et al.,2017).The OSW dataassimilated in the GDAS is mainly collected by theadvanced scatterometer(ASCAT)onboard the Meteorologi-cal Operational Sat
20、ellite-A/B and the quick scatterometer(QuickSCAT)and the Ku-band scatterometer mountedonboard the HY-2A satellite(HSCAT-A)(Bi et al.,2011;Jiang et al.,2012;De Chiara et al.,2016;Lin et al.,2017b).The HY-2B satellite is the second ocean dynamic envi-ronment monitoring satellite following the HY-2A sa
21、tellite(Jiang et al.,2013).It was successfully launched on 25 Octo-ber 2018,equipped with a Ku-band rotating fan-beam scan-ning scatterometer (HSCAT-B).The HSCAT-B OSWretrieval algorithm is designed based on the mature retrievalalgorithm and quality control algorithm of the HY-2A scat-terometer(HSCA
22、T-A)(Jiang and Lin,2009;Wang et al.,2015,2017;Lin et al.,2017a).The evaluation shows thatthe OSW of HSCAT-B is clearly more accurate than that ofHSCAT-A and has similar precision to that of ASCAT-A/B(Wang et al.,2020).Data retrieved from ASCAT-A and BOSW have displayed good accuracy when assessed wi
23、th drop-windsonde and buoy winds(Chou et al.,2013;Lin et al.,2015).Many operational centers have assimilated ASCAT-A and B OSW data into global numerical weather predictionsystems operationally (e.g.,ECMWF,NCEP),and it hasmade a significant contribution to improving forecast skill(Bi et al.,2011;De
24、Chiara et al.,2016;Lin et al.,2017b).These assessment results lay a solid foundation for theassimilation and application of HSCAT-B OSW in globalnumerical weather prediction.Meanwhile,compared withthe two small observational swaths(550 km)of ASCAT,the observational swath of HSCAT-B is as large as 18
25、00 km,which covers 90%of the global ocean area.As a result,theamount of retrieved OSW data is more than 1.5 times thatof ASCAT given the same spatial resolution.So far,studies on the assimilation of the HSCAT-BOSW and its impact on the initial analysis and forecast ofNWP model are lacking.Therefore,
26、to investigate the optimalscheme for the processing and assimilation of the HSCAT-B OSW data and improve the forecast performance of theGlobal/Regional Assimilation and Prediction System(GRAPES)over oceanic regions,this study designed and car-ried out a one-month HSCAT-B OSW data assimilation exper-
27、iment based on the GRAPES global four-dimensional varia-tional assimilation system(4DVAR),and evaluated the influ-ence of the HSCAT-B OSW data on the forecast skills ofGRAPES Global Forecast System (GRAPES_GFS).Ourresults are expected to offer useful insight into applyingOSW data from the HSCAT-B an
28、d the Chinese-FrenchOceanography Satellite in the future.The remainder of this paper is organized as follows.Sec-tion 2 discusses the assimilation system and experimentaldesign.Section 3 presents the assimilation data and theresults.Section 4 provides a discussion and conclusion.2.Assimilation syste
29、m,HSCAT-B oceansurface wind,and experimental design2.1.GRAPES global four-dimensional variational dataassimilation systemThe GRAPES global four-dimensional variational dataassimilation system (GRAPES_GFS 4DVAR)adopts theincremental analysis update scheme (Zhang et al.,2019).Compared with the three-d
30、imensional variational data assimi-lation system(3DVAR),the amount of observational datathat can be effectively ingested in the 4DVAR system isincreased by 50%;consequently,the analysis and forecasterror amplitude significantly decreases(Zhang et al.,2019).The assimilation window of GRAPES_GFS 4DVAR
31、 is sixhours,and the observational time slot is 30 minutes,whichmeans the GRAPES_GFS 4DVAR can assimilate the observa-tional data with high temporal resolution.The assimilatedobservational data in this study include the conventionalobservational data and the satellite observational data,as sum-mariz
32、ed in Table 1.The forecast model used in this study isGRAPES_GFS V3.0,which is currently operated at theChina Meteorological Administration with a horizontal reso-lution of 0.25 0.25 with 87 vertical layers.The mainphysical parameterization schemes in GRAPES_GFS V3.0are shown in Table 2.The long-and
33、 short-wave radiationscheme is generated by the Rapid Radiative Transfer Model(RRTMG)(Morcrette et al.,2008).The land surface schemeused is the Common Land Model(CoLM)(Dai et al.,2003).JANUARY 2023WANG ET AL.45The planetary boundary layer scheme used is the Medium-Range Forecast(MRF)(Hong and Pan,19
34、96).The deep andshallow cumulus convection parameterization scheme isgiven by the New Simplified ArakawaSchubert(NSAS)sub-routine(Arakawa and Schubert,1974;Pan and Wu,1995;Liu et al.,2015).The cloud physics schemes include a explic-itly prognostic cloud cover scheme(Ma et al.,2018)and theCMA double
35、moment microphysics scheme,the macro-physics cloud condensation scheme,the impacts of detrain-ment of deep/shallow convection on grid-scale clouds to rep-resent the processes of formation and extinction for allhydrometeros.(Tan et al.,2013;Jiang et al.,2015;Chen etal.,2021).In the incremental 4DVAR,
36、the numbers of verticallayers of inner and outer loops and the horizontal resolutionof the outer loop are the same as those in the GRAPES opera-tional model.The horizontal resolution of the inner loop is1.0 1.0.In GRAPES_GFS 4DVAR,the OSW data isdirectly assimilated as the 10 m neutral wind over the
37、 oceansurface.The deblurred zonal wind component(u)and merid-ional wind component(v)products of HY-2B are also assimi-lated.2.2.HSCAT-B ocean surface windThe HSCAT-B OSW data retrieved by the National Satel-lite Ocean Application Service(NSOAS)is assimilated inthis study,with a horizontal resolution
38、 of 25 km.The NWP-based ocean calibration (NOC)algorithm is used for theHSCAT-B OSW retrieval,which considers the influence ofsea surface temperature and exerts strict quality control(Ver-speek et al.,2012;Wang et al.,2015,2017;Lin et al.,2017a).The results show that the root mean square errors(RMSE
39、s)of wind speed and wind direction of the HSCAT-B OSW data after strict quality control are 2.26 m s1 and16.6,respectively,evaluated against the buoy wind data.Table 1.Observations and variables assimilated in the control experiment(CTRL).Observation typeInstrumentPlatformAssimilated observation ele
40、mentConventional observationTEMPWind,temperature,relative humiditySYNOPAir pressureSHIPAir pressureBUOYWindAIREPWind,temperatureSatellite observationAMSUANOAA-15,-18,-19,Metop-A,-BRadianceAMSUBNOAA-15,-18,-19,Metop-A,-BRadianceMWTS-2FY-3DRadianceATMSSoumi-NPPRadianceMWHS-2FY-3DRadianceMWRIFY-3DRadia
41、nceHIRASFY-3DRadianceIASIMetop-A,-BRadianceAIRSAQUA(EOS-2)RadianceGIIRSFY-4ARadianceAGRIFY-4ARadianceS-VISSRFY-2HRadianceGNSS ROCOSMIC,Metop-A/B/C GRAS,GRACE-A,TerraSAR-X,FY-3D GNOSRefractivityGPS-PWAtmospheric column water vapor contentAMVsFY-2E,GOES-13,-15,METEOSAT-10,Himawaii-8Wind(u,v)Table 2.Pa
42、rameterization schemes for physical processes used in GRAPES_GFS V3.0.Physical processParameterization SchemeReferenceLong-wave radiationRRTMG_SWMorcrette et al.,2008Short-wave radiationRRTMG_LWMorcrette et al.,2008Land surface processCoLMDai et al.,2003Planetary boundary layer processMRFHong and Pa
43、n,1996Cumulus convection processNSASArakawa and Schubert,1974;Liu et al.,2015Cloud coverCMA cloud cover prognostic schemeMa et al.,2018Cloud microphysical processesCMA cloud scheme including the large-scalecondensation,double-moment microphysics,sub-grid scale convection detrainment pro-cessesTan et
44、 al.,2013;Jiang et al.,2015;Chen et al.,202146HY-2B OSW DATA ASSIMILATION IN GRAPES_GFS 4DVARVOLUME 40The accuracy of the HSCAT-B OSW data is equivalent tothat of the ASCAT-B OSW data(Wang et al.,2020).2.3.Observational errorWith the wind data observed by oceanographic buoysas a reference,the RMSEs
45、of u and v of the HSCAT-BOSW data are 1.70 m s1 and 1.76 m s1,and the biases are0.03 m s1 and 0.08 m s1,respectively(Wang et al.,2020).Although the bias of HSCAT-B is relatively small,previousresearch showed that the bias is quite sensitive to the actualwind speed.The bias of the HSCAT-B wind speed
46、is particu-larly obvious when the wind speed becomes larger than20 m s1(Wang et al.,2020).However,there is no bias correc-tion procedure especially for the OSW data inGRAPES_GFS.To mitigate the potential influence of thebias,the observational error should be appropriatelyenlarged.In addition,the spa
47、tial resolution of the u and v com-ponents of the OSW retrieval is set at 25 km,which isslightly finer than that of GRAPES_GFS.In general,the cor-relation between the observational errors of adjacent gridswithin the current resolution of 25 km should be considered.Previous studies also demonstrated
48、that the spatial correlationof ASCAT-B observational errors approaches zero when thegrid interval is beyond 50 km (Valkonen et al.,2017).Nonetheless,in GRAPES_GFS 4DVAR assimilation sys-tem,the correlation between observational errors of adjacentgrids is not considered in the observational error cov
49、ariancematrix due to computational cost.Therefore,we process theHSCAT-B OSW observations with the thinning step andenlarge the observational errors in the assimilation systemto 2 m s1 for the u and v wind components.2.4.Quality controlThere are strict quality controls in the retrieval processof the
50、HSCAT-B OSW (Lin et al.,2017a;Wang et al.,2020).The quality control algorithm for the HSCAT-BOSW L2b products is used to eliminate the data over theareas covered with sea ice and those polluted by precipitationbefore the GRAPES_GFS 4DVAR assimilation.The qualityof observations is controlled by compa