Towards the construction of a validated numerical system to study the mesoscale dynamics of the North East Atlantic (2003-2006)

UNIVERSIDAD DE LAS PALMAS DE GRAN CANARIA Facul ad de Ciencias del Ma Depa amen o de Física P oyec o Final de Más e – Tesina: Towa ds he cons uc ion o a alida ed nume ical sys em o s udy he mesoscale dynamics o he No h-Eas A lan ic (2003-2006) case s udy. Sheila Na alí Es ada-Allis Di ec o : Ángel Rod íguez San ana Co-di ec o es: Rui Caldei a And ade y Pablo Sang à Incia e Las Palmas de G an Cana ia a 3 de Feb e o de 2011 Towa ds he cons uc ion o a alida ed nume ical sys em o s udy he mesoscale dynamics o he No h Eas A lan ic (2003-2006) S.N. Es ada-Allisa,c, X. Cou ela db, R.M. Caldei ab,c,1,∗, F. Mach´ına, A. Rod ´ıguez-San anaa, P. Sang `aa aDepa amen o de F´ısica. Facul ad de Ciencias del Ma . Uni e sidad de Las Palmas de G an Cana ia (ULPGC). Las Palmas. Spain bCen e o Ma hema ical Sciences (CCM) Uni e si y o Madei a. Madei a Island. Po ugal cIn e disciplina y Cen e o Ma ine and En i onmen al Resea ch (CIIMAR). Po o. Po ugal Abs ac A de ailed alida ion s udy has con ibu ed o he cons uc ion o a NE- A lan ic (NEA), ocean ci cula ion model, o he 2003-2006 pe iod. The compa isons be ween h ee model solu ions, emo e sensing and in si u da a, ocused on he s udy o he mos dynamical p ocesses o NEA sub- egions. Model alida ions include (i) compa isons wi h Sea Su ace Tempe a u e (SST), om AVHRR and Mic owa e-OI; (ii) Eddy Kine ic Ene gy (EKE), compu ed om al ime y; (iii) Tempe a u e and Salini y p ofiles compu ed om ARGO floa s, and (i ) sub-su ace empe a u e, measu ed in h ee buoys o ‘Pue os del Es ado’. Simple s a is ical me hods we e used o quan- i y model-da a compa isons. In gene al, model egional solu ions show a ∗Co esponding au ho Email add ess: [email p o ec ed] (R.M. Caldei a) 1CIIMAR, Rua dos B agas, 289, 4050-123 Po o, Po ugal, Ph (+351) 22 340 18 00, Fax (+351) 22 339 06 08 P ep in submi ed o Ocean Modelling Augus 10, 2011 good co ela ion and small Roo Mean Squa ed E o (RMSE) wi h SST and EKE. The main wa e masses we e well depic ed by he egional model; while he egion wi h high salini y alues, o en domina ed by Medi e anean In e media e Wa e (MIW), was no accu a ely esol ed. The ini ial condi- ion and bounda y o cing o he egional model was e alua ed, pa icula ly conce ning he usage o Ocean Gene al Ci cula ion Models (OGCM’s), as an al e na i e o he classical clima ological o cing. The analysis o hei Kolmogo o ene gy spec um de e mined hei effec i e esolu ions and hei EKE le els. Me ca o global solu ion a 1/4◦, was shown o be an adequa e OGCM solu ion o s udying he 2003-2006 pe iod. The ene gy spec um analysis also showed ha new 1/12◦(downscaled), egional solu ion esol ed mo e ene ge ic scales han he o iginal OGCM, confi ming he need o use high spa ial esolu ion egional ocean ci cula ion models o esol e mesoscale and sub-mesoscale phenomena. The egional model (ROMS) was able o e- p oduce an Ibe ian Peninsula upwelling e en as well as o he p e iously documen ed NEA p ocesses, such as he wes wa d p opaga ing eddies asso- cia ed wi h he Azo es F on , wi h he Cana y and wi h Madei a A chipela- gos. The analysis also showed ha a mosphe ic o cing, is impo an o adequa ely esol e su ace dynamics. Howe e , o he mo e conse a i e as- pec s o he ocean such as he wa e mass composi ion, a e be e ep esen ed wi h clima ological o cing. Keywo ds: ROMS, No h Eas e n A lan ic, model alida ion, a mosphe ic-ocean in e ac ions, bounda y condi ions. 2 1. In oduc ion The No h Eas A lan ic ocean (NEA) is a e y dynamic oceanic e- gion wi h iden ifiable sub-sys ems; hese sub-sys ems can be conside ed as “c i ical egions”, cha ac e ized by he high alues o eddy kine ic ene gy (EKE), (Fig. 1). The main c i ical egions o he NEA, iden ified using al-5 ime y de i ed da a o hei high EKE ac i i y (Fig. 1), a e: (i).- Ibe ian Peninsula (IP); (ii).- The Medi e anean Wa e Ou flow (MWO); (iii).- The Azo es F on egion (AF); (i ).- Madei a A chipelago (MA); ( ).- Cana ies A chipelago (CA) and ( i).- The Azo es A chipelago (AA). The IP sub- egion is pa o he ou main d i en eas e n bounda y up-10 welling zones, he ea e e e o as IPUS (Ibe ian Peninsula Upwelling Sys- em), wi h a s ong seasonal a iabili y (e.g. Al a ez e al., 2009). Two wa e masses play an impo an ole: (i) he Medi e anean In e media e Wa e (MIW), since he upwelling is in he pa hway o an icyclonic mesoscale lens o wa m sal y o Medi e anean Wa e (Meddies) (e.g. A mi and S ommel,15 1983; Rel as e al., 2007); (ii) he Eas e n No h A lan ic Cen al Wa e (ENACW), ha is, in gene al, obse ed in upwelled wa e s (e.g. Polla d e al., 1996). In absence o coas al upwelling, he su ace ci cula ion off Wes e n Ibe ia is p edominan ly polewa d (e.g. Peliz e al., 2002). Ano he impo an and ecu en s uc u e in he IPUS a e he upwelling filamen s20 (e.g. Ba on e al., 2001), as well as he long apped filamen s (see Meunie e al., 2010). Sa elli e da a (Haynes e al., 1993) and models esul s (Haid- ogel e al., 1991) show ha la ge filamen s a e o en closely ela ed wi h he loca ion o capes and p omon o ies. Labo a o y s udies by Mau i zen e al. (2001) hypo hesize ha he MWO25 3 Medi e anean Wa e Ou flow exe s a s ong influence in he No h Eas A lan ic dynamics. Downs eam o he Gib al a S ai , highe salini ies we e measu ed in he su ace wa e s o he Gul o Cadiz, h ough a de ainmen p ocess (i.e. due o diapycnal mixing o flux o salini y om he MWO owa ds he low-densi y Cen al Wa e ). The su ace wa e s a el wes wa d30 and no hwa d, p omo ing he salini y inc ease in he Cen al Wa e off he Wes IP (see Rel as e al., 2007). Fu he mo e, p e ious modeling s udies sugges he exis ence o a di ec link be ween he Azo es Cu en and he MWO (e.g. Rel as e al., 2007; Volko and Fu, 2010). The o ma ion o he well-defined zonally o ien ed Azo es Cu en may be he esul o wa e mass35 ans o ma ion associa ed wi h MWO in he Gul o Cadiz (Volko and Fu, 2010). The Azo es Cu en , and i s associa ed on (AF), is a quasi-pe manen NEA ea u e h oughou he yea , cen e ed be ween 33◦and 35 ◦N (e.g. Klein and Siedle , 1989). The complex mesoscale a iabili y o he AF is cha -40 ac e ized by long pe iods (∼250 days) and la ge wa eleng hs (∼600km), wi h wes wa d p opaga ion, associa ed wi h Rossby wa es (Le T aon and De Mey, 1994) and a e la gely due o ba oclinic ins abili ies (e.g. Al es and De Ve di`e e, 1999). Chel on e al. (2007), also sugges ed ha he wes - wa d ene gy p opaga ion a mid-la i udes is mo e ep esen a i e o nonlinea 45 e ically-cohe en eddies, as an al e na i e o he Rossby wa e p opaga ion heo y. Islands a e also egions wi h s ong mesoscale ac i i y in he NEA. Sang `a e al. (2009) epo ed he exis ence o se e al wes p opaga ing eddy co i- do s a he AF and leewa d o he islands. Two small eddies co ido s we e50 4 iden ified no h and sou h o he AF. These wes p opaga ing cyclonic eddies we e fi s obse ed a 32.2 ◦N om in-si u da a (Ping ee and Sinha, 2001) and p esen ed as an al e na i e hypo hesis, which had p e iously associa ed hese wes wa d p opaga ion wi h he occu ence o plane a y Rossby wa es. Ano he zonal co ido was also de ec ed a 31 ◦N, sou h o MA and sou h55 o he CA. The Cana y Eddy Co ido , ex ending om 22 ◦N o 29 ◦N, and popula ed mainly by an icyclones. Bo h Madei a and Cana y eddy co ido s, had well defined sou ce egions i.e. leewa d side o he islands. The fi s a emp s o s udy he dynamical mesoscale sub-sys ems o he NEA, using only Ocean Global Ci cula ion Models (OGCM), did no ep o-60 duce well he mesoscale s uc u es o en obse ed om in-si u and emo e sensing da a (e.g Ping ee, 2002). Mesoscale eddies a e pa ame e ized as a iscous e m playing an impo an ole in ene gy and momen um dissi- pa ion in ocean ci cula ion models, no well esol ed in OGCMs (Jochum e al., 2008). Ano he OGCM limi a ion is hei di ec impac on deep-ocean65 cu en s, which a e la gely cons ained by he poo ep esen a ion o he oceanic ba hyme y (Ba h e al., 2008). Ne e heless, OGCMs can p o ide adequa e bounda y and ini ial condi ions o high- esolu ion egional ocean models (Ba h e al., 2008; Melsom e al., 2009; Al e a-Azc´a a e e al., 2011). Cu en ly, he e a e se e al OGCM solu ions eely a ailable o he scien ific70 communi y. These include: (i) Me ca o , (ii) SODA, (iii) ECCO and (i ) HYCOM. In o de o de e mine which OGCM o use, hei accu acy and a ailabili y o hei solu ions we e conside ed he ein. Classical clima ological o cing was also conside ed, as an al e na i e con ol expe imen . P e ious s udies (Mason e al., 2011) used high esolu ion egional models75 5 o s udy some aspec s o he NEA dynamics, o cing hei model wi h clima- ological da a. To he bes o ou knowledge, no p e ious s udies e alua ed he ela i e ole o OGCMs and clima ological bounda y o cing, o s udy he NEA dynamics. Ul ima ely, ou main goal is o cons uc a alida ed ocean egional model, conside ing he ools and he da a a ailable oday, in o de 80 o adequa ely con inue o in es iga e he egional dynamics. I is expec ed ha his sys em and i s alida ion p o ocols will e ol e in o a egional ocean o ecas ing sys em. The layou o his epo is as ollows, a e he in oduc ion (sec ion 1), Sec ion 2 desc ibes he diffe en sou ces o da a used, as well as, a de ailed85 desc ip ion o he egional nume ical modeling sys em, he diffe en aspec s o he expe imen s, hei bounda y and ini ial condi ions and he s a is i- cal me ics used o compa ing model esul s wi h da a. Sec ion 3 discusses he main esul s including: (i) The ep esen a ion o sub-mesoscale NEA p ocesses; (ii) The ep esen a ion o he su ace dynamics; and (iii) he ca-90 pabili y o ou egional model o ep oduce p e iously documen ed dynamic p ocesses, such as he wes wa d p opaga ion eddies, he ep esen a ion o main wa e masses and an IPUS episode. Sec ion 4 sums up he esul s and p oposes u u e di ec ions. 2. Da a sou ces and Me hods95 2.1. The egional ocean ci cula ion modeling sys em The egional model used in his s udy was he Regional Oceanic Mod- eling Sys em (ROMS). Fo a comple e desc ip ion o he model e e ed o (Shchepe kin and McWilliams, 2003, 2005). ROMS is a spli -explici , ee- 6 su ace and e ain- ollowing e ical coo dina e oceanic model, whe e sho 100 ime s eps a e used o ad ance he su ace ele a ion and ba o opic momen- um equa ion, and a la ge ime s ep is used o empe a u e, salini y, and ba oclinic momen um. ROMS employs a wo-way ime-a e aging p ocedu e o he ba o opic mode which sa isfies he 3D con inui y equa ion. The spe- cially designed p edic o -co ec o ime-s ep algo i hm allows a subs an ial105 inc ease in he pe missible ime-s ep size. The hi d-o de , ups eam-biased, dissipa i e ad ec ion scheme o momen um allows he gene a ion o s eep g adien s, enhancing he effec i e esolu ion o he solu ion o a gi en g id size Shchepe kin and Mcwilliams (1998). Fo ace s, he RSUP3 scheme whe e diffusion is spli om ad ec ion and is ep esen ed by a o a ed bi-110 ha monic diffusion scheme wi h flow-dependen hype -diffusi i y, is used in o de o a oid excessi e spu ious diapycnal mixing associa ed wi h sigma coo dina es Ma chesiello e al. (2009). Explici la e al iscosi y is null e e y- whe e in he model, excep in sponge laye s nea he open bounda ies whe e i inc eases smoo hly on se e al g id poin s. A K-p ofile pa ame e iza ion115 (KPP) bounda y laye scheme La ge e al. (1994) pa ame e izes he sub-g id e ical mixing p ocesses. In o de o encompass he mos ele an dynamic ea u es o he NEA ci cula ion, and conside ing he oceanog aphic da ase a ailable o model alida ion, we ha e designed an ex ended ec angula g id om 25 ◦N o 45120 ◦N in la i ude and om 35 ◦W o 5 ◦W in longi ude. The model g id, o cing ini ial and bounda y condi ions a e buil using an adap ed e sion o he ROMSTOOLS package (Pen en, 2003). The bo om opog aphy is de i ed om a 30 a c-second esolu ion da abase GEBCO 08 7 (www.gebco.ne ). Al hough a new p essu e g adien scheme associa ed o a125 modified equa ion o s a e limi s compu a ional e o s o he p essu e g a- dien is cu en ly implemen ed in ROMS (Shchepe kin and McWilliams, 2003), he ba hyme y s ill needs o be smoo hed, so ha he “slope pa- ame e ” =∆h/h (Beckmann and Haid ogel, 1993) emains unde 0.2. To p ese e a sufficien esolu ion in he uppe ocean, we use 50 e ical le els130 wi h s e ched s-coo dina es, using su ace and bo om s e ching pa ame e s θs=6, θb=0 (Song and Haid ogel, 1994). Th ee ROMS expe imen s we e buil o his s udy, hey diffe ed on he sou ce and na u e o hei oceanic bounda y condi ions and on he sou ce and na u e o hei a mosphe ic o cing condi ions. These include: (1) R M,135 ROMS o ced wi h Me ca o a he oceanic bounda ies. Hal deg ee, daily mean wind s ess was ex ac ed om he QuikSCAT sa elli e sca e ome e da a, p o ided by CERSAT (www.i eme . /ce sa /en/index.h m); hea and esh wa e fluxes we e ex ac ed om NCEP2, using he bulk o mula Fai all e al. (1996, 2003); (2) R NQ, ROMS was o ced wi h WOA05 clima ology140 (Loca nini e al., 2006; An ono e al., 2006) a he ocean bounda y, momen- um fluxes we e ex ac ed om QuiKSCAT and hea fluxes om NCEP2; (3) R C, ROMS o ced wi h WOA05 clima ology a he oceanic bounda ies, and COADS clima ology (da Sil a e al., 1994) (a mosphe ic hea and momen- um).145 The la e al bounda ies acing he open ocean, a mixed passi e-ac i e, im- plici , adia ion condi ion connec s he model solu ion o he su oundings (Ma chesiello e al., 2001). Rega ding he Me ca o inflow condi ions, he so- lu ion a he bounda y is nudged owa d daily ime-a e aged ou pu s, which 8 (sub)mesoscale a iabili y, he e is a need o use high- esolu ion model g ids. Compa ing he OGCM solu ions (figu e 4), Me ca o (1/4◦) and HYCOM (no shown) esol e well up o he 100 km ange, bu i needs o be coupled wi h ROMS, in o de o ully esol e he sub-mesoscale a iabili y (10-100 km). On he o he hand, ECCO and SODA a e only ep esen ing la ge 295 scales (100-1000 km). Al hough his analysis is done o he same egion, and o a pa icula ( andom) momen in ime, i is no expec ed ha i will a y much o e space and ime. The e o e, he abili y o nume ical ocean models o adequa ely esol e he ene gy spec um o he diffe en scales is e y much a unc ion i s g id esolu ion. The ens ophy ans e which is300 diagnosed by k−3, is only esol ed by he highe esolu ion egional models i.e. ROMS. The e a e also no iceable diffe ences be ween he spec al densi y esol ed by clima ological o ced egional model (R C) and he o he ROMS solu-305 ions (e.g. R M;R NQ). Clima ological o ced ROMS esol e p ocesses wi h lowe spec al densi y signa u es (10−2 e sus10−1 e ol ed by he o he ROMS expe imen s, R M;R NQ). Fu he mo e, he a mosphe ically o ced ROMS show a g ea e u bulen a iabili y, when compa ed wi h he clima- ologically o ced expe imen .310 3.2. Regional ocean ci cula ion model alida ion The alida ion o ou egional ocean ci cula ion model was achie ed s udy- ing some o he p e iously documen ed dynamics o he NEA c i ical sub- egions. Fi s , he su ace dynamics we e s udied by compa ing model so- lu ions wi h EKE, SST and buoys da a. Secondly, he wes p opaga ion o 315 15 mesoscale eddies we e s udied compa ing model o AVISO da a. Thi dly, a en ion was gi en o he model ep oducibili y o he main NEA wa e masses. Finally, a case s udy o an Ibe ian Peninsula Upwelling episode, which occu ed be ween he 5 h and he 20 h Augus o 2005, helped demon- s a e he accu acy o he egional model solu ion o ep esen a specific320 oceanic e en . 3.2.1. Rep esen a ion o su ace dynamics Ba h e al. (e.g. 2008), showed ha OGCM bounda y condi ions im- p o ed he egional model solu ion, pa icula ly on he shel dynamics, e en when he open bounda y is loca ed a , in he open ocean. Clima ological325 o cing p oduced low-densi y g adien s and less ene ge ic egional solu ions (Ba h e al., 2008). Ou esul s show ha o adequa ely ep oduce he in e - annual SST a iabili y (see figu es 5) ano he impo an ac o o conside is he use o app op ia e a mosphe ic o cing, such as be e ep esen a i e winds and hea fluxes. Likewise, in ou s udy be e SST compa isons we e330 achie ed using OGCM o ced ROMS han clima ological o ced 17(Table 4). Expe imen s R M and R NQ show be e compa isons wi h AVHRR de i ed SST da a in ela ion o he pu ely clima ologically o ced expe imen (R C), (Fig. 5). The bias alues a e close o 0 n R M and R NQ expe imen s, while he R C expe imen shows a g ea e a iabili y. The RMSE alues a y be-335 ween 0 and 1 in R M and R NQ whe eas in R C i eaches 1.9. The analysis o he 2coefficien shows lowes co ela ion in win e mon hs, p obably due o limi ed da a a ailabili y i.e. highe cloud co e age (e.g. Reynolds e al., 2007). Non-spi e he ac ha he sa elli e SST da a uses OI o fill he miss- ing alues, i is expec ed he use o mo e SST alid da a poin s du ing cloud340 16 ee pe iods, hus in e pola ed p oduc s a e expec ed o ha e mo e accu a e ep esen a ion. The compa isons be ween AVHRR de i ed SST and R C wo sens wi h ime, 2003 RMSE and Bias a e highe o 2006. Co ela ions a e also weake o 2006 han o he 2003-2005 pe iod. Recen alues o he DJFM NAO345 index, shows a p og essi e change o he same pe iod (2003-2006). NAO indexes a e -0.20 in he win e mon hs o 2003, -0.11 in he win e o 2004, -0.82 in he win e o 2005, becoming 1.83 (posi i e) in he win e o 2006 (h p://www.c u.uea.ac.uk/ imo/da apages/naoi.h m). In o de o accoun o he in e -annual a mosphe ic a iabili y, cha ac e is ic o pe iods wi h350 posi i e and nega i e NAO-yea s, i is impo an o conside non-clima ological a mosphe ic o cing condi ions. The leading mode o sea su ace empe a- u e (SST) a iabili y o e he No h A lan ic du ing he posi i e NAO win e consis s o a ipole pa e n: (i) wi h a cold SST anomaly in he sub-pola egion, (ii) a wa m anomaly in he middle la i udes cen e ed off Cape Ha -355 e as (NEA case); (iii) a cold sub opical anomaly be ween he equa o and 30 ◦N(Dese and Blackmon, 1993; Kushni , 1994). The empo al a e aged SST shows he bes model-da a compa isons in he open-ocean (Fig. 6), hus he e is a need o complemen his analysis compa ing model esul s wi h nea -su ace empe a u e collec ed using in si u360 buoys (see e.g. Ba h e al., 2008; I ano e al., 2009). Va iabili y o coas al (nea -sho e) p ocesses a e no well ep esen ed in me ged/in e pola ed sa el- li e p oduc s. The sub-su ace (3m) empe a u e ime-se ies o model so- lu ions a e in good ag eemen wi h he obse a ions collec ed by he VS (Villano Sisa gas), wi h he CS (Cabo Sillei o) and EB (Es aca de Ba es)365 17 buoys as shown in empo al se ies o empe a u e in figu e 8. The ROMS expe imen s using high esolu ion a mosphe ic o cing (R M and R NQ) a e able o ep oduce he in e -annual a iabili y in he buoys, whe eas he cli- ma ologically o ced expe imen (R C), showed always weake co ela ions. Wi h espec o EKE de i ed om al ime y da a, he bes model ep-370 esen a ion is ob ained in offsho e egions and using clima ological o cing condi ions (figu e 7). I is hypo hesized ha hese diffe ences a e due o (i) he limi ed spa ial and empo al esolu ion o AVISO da a (1/3◦), and (ii) due o he ac ha he models ha e highe esolu ion compa ed o AVISO da a, and hus hey a e mo e ene ge ic. Fu he mo e, he lack o AVISO da a375 nea he coas a o s he compa isons wi h offsho e egions, since AVISO da a accu acy is limi ed o acqui e da a up o 45 km om he coas (Du and e al., 2008). 3.2.2. Wes p opaga ion o eddies Figu e 9, shows he wes wa d p opaga ion o mesoscale eddy s uc u es380 (see whi e a ow in figu e 9a and b), (34.0423 ◦N be ween 19 ◦ o 35 ◦W). Sang `a e al. (2009) obse ed wo small co ido s o wes wa d p opaga ing ed- dies, no h and sou h o Azo es F on . These co obo a e he esul s showed in figu es 9a and b. Also appa en , in ou esul s, is he ac ha hese s uc u es li e less han 28 weeks (∼200 days). Ping ee and Sinha (2001),385 a e analyzing in a ed, al ime e and in-si u measu emen s, sugges ed ha he wes wa d mo emen o hese la ge s uc u es a e due o cold (cyclonic) s uc u es called STORMS, p opaga ing wes wa d a ound 32.2 ◦N. The e- a e , Ping ee (2002) showed wes wa d p opaga ing anomalies (in hei fig- u e 3), occu ing in he same a ea ep oduced by ou ROMS expe imen s390 18 (figu e 9c and d). Fu he mo e, he e is good quali a i e ag eemen be ween AVISO EKE da a and he simula ed EKE ex ac ed om he R M expe i- men . Mo eo e , he mesoscale s uc u es obse ed in he figu e 9c and d, co espond o he seasonal a iabili y o EKE showed in Sang `a e al. (2009). The highes alues o EKE a e ound sou h o he CA a chipelago in sp ing,395 summe and au umn. 3.2.3. Wa e masses ep oducibili y The wa e s masses desc ibed o he NEA a e be e depic ed by he cli- ma ological o ced egional model (R C), and ep esen ed o a lesse deg ee o ealism in he expe imen s using high esolu ion a mosphe ic o cing. I 400 is expec ed ha clima ological o ced ROMS bes ep oduces wa e masses composi ion due o i s conse a i e na u e. Deep-wa e masses wi h slow o e u ning a es, o en akes hund eds o yea s o a comple e e-ci cula ion. On he o he hand, his also challenges he classical iew ha a mosphe ic o cing only affec s he ocean su ace dynamics. Somehow, h ough e ical405 mixing (ad ec ion / diffusion), a mosphe ic o cing seems o also play an im- po an ole on wa e mass composi ion. In ac , he e a e ecen examples in he li e a u e sugges ing ha mesoscale su ace-induced ea u es, ha e a sig- nifican influence in ‘ en ing’ deep-wa e masses, pe u bing i s conse a i e flow egime.410 The cen al wa e , No h A lan ic Cen al Wa e (NACW; Fig. 10), cha ac e ized by densi y alues o 27.38 (kg/m3), wi h po en ial empe a- u e a ying (θ) be ween 11-18 ◦C and salini ies a ying be ween 35.5-36.5 (Mach´ın e al., 2006), showed good ep oducibili y by he model. Howe e , he less ep oducible wa e mass was indeed, he wa me and sal ie Medi e -415 19 anean In e media e Wa e (MIW; Fig.10). MIW is cha ac e ized by θ= 10 ◦C, wi h S highe han 35.6, and densi ies a ying be ween 27.38-27.922 (kg/m3) (Mach´ın e al., 2006). The An a c ic In e media e Wa e was also well ep esen ed in mos egional models (AAIW; θ= [7-8] ◦C; S = [<35.4] and a γ= [27.38-27.92] kg/m3).420 In e ms o he diffe en NEA sub- egions, he bes co ela ion was ound o he AF sub- egion (Fig. 10a), wi h 2∼0.99, o empe a u e and 2 ∼0.97, o salini y. As expec ed he lowes co ela ion is ound in MWO sub- egion (Fig. 10b) whe e he 2a e 0.94 and 0.86 o empe a u e and salini y, espec i ely. Despi e he ac ha CA and MA sub- egions a e be e 425 co ela ed wi h ARGO p ofiles, han he IP sub- egion, i is impo an o no e ha he CA and MA sub- egions had less p ofiles han he IP egion and hus measu ed less in insic a iabili y. An o e all s a is ical analysis was also pe o med o indi idual floa a- jec o ies in he whole NEA egion. An example ep esen a i e o hese com-430 pa isons is shown in figu es 11 a 16 , whe e he la ges Bias be ween ROMS (R C) solu ion and ARGO was o en ound in he sub- egions domina ed by he MIW (be ween 1000 and 1700 m o dep h) such as, AF (Fig. 11), MWO (Fig. 12) and IP (Fig. 13). Howe e , be e compa isons we e ound in egions away om he Gib al a S ai namely: MA (Fig. 14), CA (Fig.435 15) and AA (Fig. 16). Gene ally, in he h ee expe imen s simula ions he alues o Bias a e in he ange o -2 and 2 ◦C o empe a u e and -1 and 1 o salini y o each ARGO floa compa ison. These esul s, sugges ha he poo es ep esen a ion o he high salini y In e media e Medi e - anean Wa e in he model, i.e. he salini y was unde es ima ed in ROMS.440 20 The eason o hese la ge diffe ences a e no in he numbe o sigma le els, since simula ions wi h diffe en numbe s o sigma-le els we e also conside ed, wi h no significan ly diffe en esul s. Ne e heless, he e is a need o con- inue hese expe imen s wi h coupled a mosphe ic-ocean models, conside ing a mo e ealis ic in e change be ween he a mosphe e and he ocean. Haid-445 ogel e al. (2000) sugges s ha wi hou a con inuous sou ce o wa e wi h S>36.5, T>11.8 ◦C and ◦=27.9 kg/m3, a significan eshening akes place du ing he model configu a ion, so he appa en ly be e ep esen a ion o MIW. Ne e heless, hese co ec ions a e o en applied in longe model uns (+10 yea s), and hus appa en ly no so ele an o ou 4 yea NEA s udy.450 Peliz e al. (2007) and Mason e al. (2011), also p esc ibe he MIW ou flow in he Gul o Cadiz egion, al hough his migh be an accep able band-aid pa ame e iza ion, when using OGCM bounda y condi ions his migh in ac comp omise he ini ial and bounda y condi ions and hus in oducing unex- pec ed a iabili y. To accoun o he in e -decadal a iabili y o he MIW455 ou flow, pe haps a be e solu ion would be o ep oduce such a iabili y a he OGCM le el, which would hen p opaga e his on o he egional model solu ions. 3.2.4. Case s udy o an IPUS episode (Aug 2005) An Ibe ian upwelling e en was de ec ed wi h an 8-day MODIS-Aqua460 composi e o su ace de i ed chlo ophyll-a (Fig. 17a), o he 20 o Augus 2005. The same episode was also cap u ed in he AVHRR de i ed SST map, as shown in he figu e 17b. Fo he same pe iod, ROMS calcula ed SST (R NQ) (Fig. 17c, d and e) showed simila low- empe a u es sugges ing he de elopmen o an upwelling e en , accompanied by he o ma ion o an465 21 upwelling filamen , in he Ex emadu a p omon o y (be ween 38.5 and 39.5 ◦N, 17c). The same e en was p e iously s udied by Meunie e al. (2010) using SST and chlo ophyll maps. Peliz e al. (2003) obse ed ha in such e en s eddy shedding migh occu as a esul o he in e ac ion be ween he flow wi h he opog aphy discon inui ies, like he Ex emadu a p omon o y,470 cohe en wi h he loca ion o some SST on s. Recen ly, Ba een e al. (2007) iden ified h ee main easons o explain he gene a ion o filamen s which include: (i) he ba oclinic ins abili y o he upwelling on ; (ii) he effec o capes and p omon o ies; (iii) plane a y be a effec and bo om opog aphy. In ac , all ou ROMS simula ions can ep oduce lowes empe a u es475 nea he coas , sugges ing he de elopmen o an IPUS e en , howe e , R M and R NQ (Fig. 17c and d), compu ed he bes esul s. The black box in he op panel o figu e 8 highligh s he s ong upwelling e en (see clo ophyll-a map in figu e 17a) o he VS buoy. The second black box ep esen s he da a collec ed by he CS buoy, also deno ing ano he IPUS e en (Augus , 2006).480 These esul s show ha he expe imen s including adequa e a mosphe ic o cing, a e capable o ep oducing he in e -annual a iabili y measu ed by he buoys. The same can no be said o he clima ological expe imen R C which main ains a cons an empe a u e, du ing bo h IPUS e en s. Quan- i a i ely, he lowes spa ial RMSE was ound using R M, wi h an a e age485 alue o 0.73 ◦C o he whole mon h o Augus , ins ead o 1.06 ◦C and 1.27 ◦C o R NQ and R C, espec i ely. As is expec ed, Bias alues a e also lowe o he R M (0.42 ◦C) and o he R NQ (0.65 ◦C), compa ed o he R C expe imen (0.91 ◦C). 22 4. Conclusions and sugges ions o u u e wo ks490 1 - The esul s om his wo k show he impo ance o a mosphe ic fluxes o gene a e a ealis ic egional ocean model solu ion o he NEA. The e a e also some ad an ages on downscaling an OGCM solu ion in o a egional model o s udy he dynamics o he mesoscale p ocesses and no elying only on clima ological o cing. Cu en OGCM solu ions o eddy- esol ing495 (1/12◦), a e only a ailable om 2009 o he p esen . The kine ic ene gy spec a (Fig. 4) shows ha ROMS seems o econs i u e he heo e ical spec um o oceanic mesoscale, ep esen ing he slopes k−5/3and k−3, ac- co ding o Kolmogo o u bulen cascade ene gy. The effec i e esolu ion o OGCM is a scales o ∼500km o SODA, ∼600km o ECCO and ∼100km500 o Me ca o , whe e he ene gy le els d ops down he slope o k−3. 2 - The s a is ical analysis show a good da a-model compa isons, wi h SST and EKE de i ed om sa elli e da a, excep in coas al a eas whe e he sa elli e de ec ion capabili y is limi ed. The be e co ela ions a e ound in offsho e egions. Is equally impo an o no e ha model-da a compa isons505 wi h AVHRR de i ed SST, showed be e s a is ical ag eemen , han when compa ing model wi h MW SST da a. Model compa isons wi h in-si u buoy da a, R M and R NQ ep oduced well he daily sub-su ace a iabili y (Fig. 8). 3 - ROMS nume ical expe imen s also eplica e, a he well, he de ec-510 ion he eddy co ido s documen ed in Sang `a e al. (2009), using al ime y da a ex ac ed om AVISO. This eddies p opaga e wes wa d as o iginally p oposed by Ping ee (2002) and ecen ly confi med by Sang `a e al. (2009). 4 - The wa e column composi ion is bes ep esen ed using ROMS o ced 23 wi h clima ological da a. The cen al and deep wa e s, ep esen ed by R C515 a e in good ag eemen wi h he ARGO p ofiles, whe eas he dep h o occu - ence o he Medi e anean In e media e Wa e (MIW) is no as accu a ely ep esen ed, since ROMS unde es ima es he highes salini y alues. 5 - The analysis o he esul s also show ha ROMS o ced wi h high empo al esolu ion a mosphe ic fluxes (expe imen s R NQ and R M), e-520 p oduces well an IPUS e en , (20 o Augus o 2005). Realis ic upwelling filamen s a e also well simula ed, ep esen ing he in e ac ion o he flow and he local opog aphy, along he Ibe ian Peninsula coas . Fu u e wo ks include, bu a e no limi ed o, he use o coupled a ocean- a mosphe ic mode, in o de o s udy he su ace dynamics wi hou comp o-525 mising he wa e masses ep esen a i i y. I is also expec ed ha he inclu- sion o ides and i e ou flows, will con ibu e o a be e ep esen a ion o he salini y alues, in pa icula he adequa e loca ion o MIW. 5. Acknowledgmen s The au ho s wish o acknowledge unds om p ojec s: RAIA (0313 RAIA 1E)530 and PROMECA (CTM2009-06993-E/MAR). Nume ical model solu ions we e calcula ed a CIIMAR HPC uni , cons uc ed using unds he FCT-Po uguese Na ional Science Founda ion plu iannual, and u he imp o ed using unds om RAIA.co p ojec , co- unded by INTERREG-IV and by FEDER (‘Fundo Eu opeu de Desen ol imen o Regional, 2007-2013’), h ough he POCTEP535 egional ini ia i e. The al ime e p oduc s we e p oduced by SSALTO/DUACS and dis ibu ed by AVISO wi h suppo om CNES. MODIS da a was ex- ac ed using he online sys em, de eloped and main ained by he NASA. 24 Ping ee, R., Sinha, B., 2001. Wes wa d mo ing wa es o eddies (s o ms) on he sub opical/azo es on nea 32.5n? in e p e a ion o he eule ian cu en s and empe a u e eco ds a moo ings 155 (35.5w) and 156 (34.4w). Jou nal o Ma ine Sys ems 29 (1-4), 239–276.675 Polla d, R., G iffi hs, M. J., Cunningham, S. A., Read, J. F., P ez, F. F., Ros, A. F., 1996. Vi aldi 1991 - a s udy o he o ma ion, ci cula ion and en ila ion o eas e n no h a lan ic cen al wa e . P og ess in Oceanog a- phy 37 (2), 167–192. Rel as, P., Ba on, E. D., Dube , J., Oli ei a, P. B., Peliz, A., da Sil a,680 J. C. B., San os, A. M. P., 2007. Physical oceanog aphy o he wes e n ibe ia ecosys em: La es iews and challenges. P og ess in Oceanog aphy 74 (2-3), 149–173. Reynolds, R. W., Smi h, T. M., 1994. Imp o ed global sea su ace em- pe a u e analyses using op imum in e pola ion. Jou nal o Clima e 7 (6),685 929–948. Reynolds, R. W., Smi h, T. M., Liu, C., Chel on, D. B., Casey, K. S., Schlax, M. G., 2007. Daily high- esolu ion-blended analyses o sea su ace em- pe a u e. Jou nal o Clima e 20 (22), 5473–5496. Sang `a, P., Pascual, A., Rod ´ıguez-San ana, A., Mach´ın, F., Mason, E.,690 McWilliams, J. C., Peleg ´ı, J. L., Dong, C., Rubio, A., A ´ıs egui, J., Ma e o-D´ıaz, A., He n´andez-Gue a, A., Ma ´ınez-Ma e o, A., Auladell, M., 2009. The cana y eddy co ido : A majo pa hway o long-li ed eddies 31 in he sub opical no h a lan ic. Deep-Sea Resea ch Pa I: Oceanog aphic Resea ch Pape s 56 (12), 2100–2114.695 Shchepe kin, A. F., Mcwilliams, J. C., 1998. Quasi-mono one ad ec ion schemes based on explici locally adap i e dissipa ion. Mon hly Wea he Re iew 126 (6), 1541–1580. Shchepe kin, A. F., McWilliams, J. C., 2003. A me hod o compu ing ho i- zon al p essu e-g adien o ce in an oceanic model wi h a nonaligned e i-700 cal coo dina e. Jou nal o Geophysical Resea ch C: Oceans 108 (3), 35–1. Shchepe kin, A. F., McWilliams, J. C., 2005. The egional oceanic mod- eling sys em ( oms): A spli -explici , ee-su ace, opog aphy- ollowing- coo dina e oceanic model. Ocean Modelling 9 (4), 347–404. Song, Y., Haid ogel, D., 1994. Nume ical simula ions o he ccs unde he705 join effec s o coas al geome y and su ace o cing. pp. 216–234. S auffe , D. R., Seaman, N. L., 1990. Use o ou -dimensional da a assimila- ion in a limi ed-a ea mesoscale model. pa i: expe imen s wi h synop ic- scale da a. Mon hly Wea he Re iew 118 (6), 1250–1277. Vallis, G. K., 2006. A mosphe ic and Oceanic Fluid Dynamics. Camb idge710 Uni e si y P ess. Volko , D. L., Fu, L. L., 2010. On he easons o he o ma ion and a ibiali y o he azo es cu en . Jou nal Geophysical Oceanog aphy 40, 2197–2220. 32 Figu e 1: Eddy Kine ic Ene gy (EKE) compu ed om AVISO da a (a e age 2003-2006 yea s) o NEA domain. The black boxes deno e he c i ical sub- egions used o he solu ion alida ion, whe e: AF-Azo es F on ; MWO-Medi e anean Wa e Ou flow; IP- Ibe ian Peninsula; MA- Madei a A chipelago; CA-Cana y A chipelago; and AA-Azo es A chipelago. Dashed ed lines show he sec ions used o plo o he H¨o moelle diag ams. 33 500 450 400 350 300 250 200 150 100 50 Figu e 2: Tempo al se ies o Sea Su ace Tempe a u e (◦C) o each OGCM (black line) compa ed wi h SST om AVHRR (blue line) and MW SST ( ed line) o he pe iod o ime since 2004 o 2006 whe e (a) is ECCO model; (b) SODA model; (c) HYCOM model and (d) Me ca o model. One can no e he diffe ences in hei empo al esolu ions. 34 Figu e 3: (a) Dep h a e age MAE o empe a u e p ofiles be ween ARGO and ROMS p ofiles (0 o 2000 m a e aged) o he 2003 o 2006 pe iod o s udy. (b) MAE o salini y p ofiles. 35 Figu e 4: Kine ic ene gy spec a (m3/s−2) o h ee OGCM models: Me ca o (1/4 deg ee; g ay dashed line); SODA (1/2 deg ee; g ay do line); ECCO (1 deg ee; g ay dashed- do ed line); and he h ee ROMS simula ion expe imen s: R M un (1/12 deg ee; blue line); R NQ (1/12 deg ee; iole line); and R C un (1/12 deg ee; ed line) o all domain a 35◦N La i ude and o one mon h o simula ion in Jan-2003. The spec a show how ROMS seems o econs i u e he heo e ical spec um o oceanic mesoscale ep esen ed by he slopes -5/3 (solid black line) and -3 (dashed black line). 36 Specl al densily 10' ~-'-'~--------~~'-------'----'---'---'--'-'-'-'----------- -'-------'----'-- - == - ~ - M~E~R~ 0.001 0.01 Wa enumbe km- 1 """' SODA _.- ECCO - R_M - R_C -R_NQ Figu e 5: Tempo al se ies o Roo Mean Squa ed E o ( op panel), Bias (middle panel) and co ela ion coefficien s (bo om panel) o he whole a ea o NEA and du ing he ou simula ed yea s (2003-2006), be ween he h ee expe imen s made o his s udy and SST om AVHRR, whe e blue (o) a e R M, iole (x) a e R NQ and ed (+) a e R C. 37 o R_M x R_NQ + R_e 1.9 w (/) :; o: Mac Juo Sep 2003 M ac Juo Sep 2004 Mae Jun Sep 2005 Mae Jun Sep 2006 Mae Juo Sep 2003 Mae 0.95 ';,: 0.9 0.85 0.8 Ma Jun Sep 2003 Ma, Jun Sep 2004 Ma Jun Sep 2005 Ma Jun 2006 Time (daily) 30oW 24oW 18oW 12oW 6oW 28oN 32oN 36oN 40oN 44oN 30oW 24oW 18oW 12oW 6oW 28oN 32oN 36oN 40oN 44oN 30oW 24oW 18oW 12oW 6oW 28oN 32oN 36oN 40oN 44oN 30oW 24oW 18oW 12oW 6oW 28oN 32oN 36oN 40oN 44oN 30oW 24oW 18oW 12oW 6oW 28oN 32oN 36oN 40oN 44oN 30oW 24oW 18oW 12oW 6oW 28oN 32oN 36oN 40oN 44oN 30oW 24oW 18oW 12oW 6oW 28oN 32oN 36oN 40oN 44oN 2 SST (ROMS−AVHRR) 0.75 0.8 0.85 0.9 0.95 1 Figu e 6: (a). Tempo al a e ages o co ela ion coefficien ( 2), o ou yea s since 2003, compa ing SST om R M expe imen esul s wi h SST om AVHRR sa elli e da a, com- pu ing o each sub- egion. 38 , " . . 30oW 24oW 18oW 12oW 6oW 28oN 32oN 36oN 40oN 44oN 30oW 24oW 18oW 12oW 6oW 28oN 32oN 36oN 40oN 44oN 30oW 24oW 18oW 12oW 6oW 28oN 32oN 36oN 40oN 44oN 30oW 24oW 18oW 12oW 6oW 28oN 32oN 36oN 40oN 44oN 30oW 24oW 18oW 12oW 6oW 28oN 32oN 36oN 40oN 44oN 30oW 24oW 18oW 12oW 6oW 28oN 32oN 36oN 40oN 44oN 30oW 24oW 18oW 12oW 6oW 28oN 32oN 36oN 40oN 44oN 30oW 24oW 18oW 12oW 6oW 28oN 32oN 36oN 40oN 44oN RMSE EKE (ROMS−AVISO) 100 200 300 400 500 600 700 Figu e 7: Tempo al a e ages o Roo Mean Squa ed E o (RMSE) o ou yea s since 2003, be ween EKE om R M expe imen and AVISO al ime e da a, compu ed o each sub- egion. Uni s a e in cm2/s2 39 Figu e 8: Tempo al se ies du ing 2003-2006 o empe a u e a 3 m dep h om Pue os del Es ado buoys (black dashed lines): Villano Sisa gas (VS), Cabo Sillei o (CS) and Es aca de Ba es (EB) compa ed a he same dep h wi h R M (blue line), R NQ ( iole line) and R C ( ed line). Mean alues o co ela ion coefficien s a e also showed in he op o hese figu es, highe alues we e a chi ed o R M and R NQ han o he R C un. Black boxes co espond o an IPUS e en o Augus 2005 o VS buoy and o he a Augus 2006 o CS buoy. No e ha he empo al axis a e no always he same due he miss da a in some days o he buoy. The loca ion o he h ee Pue os del Es ado buoys a e also shown in he bo om panel wi h he ba hyme y o he a ea. 40 Figu e 15: ARGO buoy numbe 6900506 in he CA sub- egion. 47 -500 -1000 -1500 -2000 09/24/06 -500 -1000 -1500 -2000 09/24/06 Ve ical sec ion lo bouy 6900506 -Bias 01 Tempe a u e (10 p oliles) 10/04/06 10/04/06 10/14/06 10/14/06 26°N 10/24/06 11/03/06 11/13/06 Bias 01 Salini y 10/24/06 11/03/06 11/13/06 1-- ARGO -- ROMS 1 " , 04 ." .• 25°N '-==--~=~- 18°W 16°W 11/23/06 11/23/06 12/03/06 12/13/06 0.5 -0.5 -1 12/03/06 12/13/06 Figu e 16: ARGO buoy numbe 6900166 in he AA sub- egion. 48 Ve ical sec ion la bouy 6900166 -Bias 01 Tempe a u e (61 p oliles) -200 - .., ""'"" --- ~- ""!!!I • • ~ -400 -600 -800 -1600 -1800 , ! 12/07/03 12/17/03 12/27/03 01/06/04 01/16/04 01/26/04 Bias 01 Salini y -200 -400 0.5 -600 -800 -1000 -1200 -1400 -05 -1600 -1800 -1 12/07/03 12/17/03 12/27/03 01/06104 01/16/04 01/26/04 1-- ARGO -- ROMS 1 400N ¡==~~=====~~ ... • ..~ . Figu e 17: (a) Map o chlo ophyll-a concen a ion (log10*100 mg/m3) om MODIS o 8 days a e ages (13 o 20 Augus 2005) indica ing an IPUS e en in he Ibe ian Peninsula coas . (b) SST ◦C om AVHRR a 20 Aug 2005 show he same upwelling e en . (c), (d) and (e) maps o SST ◦C om he expe imen R NQ, R M and R C, espec i ely, o he same IPUS e en o 20 Aug 2005. 49 Chl-a (MODIS) SST (AVHRR) 27 440N 44 'N b 26 25 42 0N 42'N 24 2.5 23 40 0N 40 'N 22 21 3S0N 38 'N 20 1.5 19 36 0N 36 'N 18 1S0W 1S0W 12"w gOW SOW 17 SST (R_NO) SST(R_M) 27 27 44 0N 2. 44 0N 2. 25 25 42 0N 2. 420N 23 23 400N 22 400N 21 2 1 3S0N 20 380N 20 l. 1. 36 0N l. 360N " 17 17 SST(R_C) 27 44 0N 26 25 42 0N 2' 23 400N 22 21 3S0N 20 l. 36 0N l. 17 Table 1: Sou ces o da a used in his s udy o he alida ions and compa ison wi h he diffe en ocean models 50 Table 2: Mainly cha ac e is ic o diffe en OGCMs used in his s udy 51 Table 3: Tempo al means o s a is ic pa ame e s: MAE (Mean Absolu e E o ); RMSE (Roo Mean Squa ed E o ) and Bias o EKE o each OGCM model s EKE om AVISO da a. G ey shaded show he OGCM wi h be e cha ac e is ic acco ding o alues o me ics alida ions. 52 Table 4: Valida ions made o e empo als means o he whole a ea o NEA be ween he h ee ROMS expe imen s and SST sa elli e da a om AVHRR and MW SST. 53