GLORYS 2 A Global ocean reanalysis of the

GLORYS 2: A Global ocean reanalysis of the period 1992 -present N. Ferry, B. Barnier, L. Parent, G. Garric, C. Bricaud, C-E Testut, O. Le Galloudec, J-M Lellouche, E. Greiner, M. Drevillon, E. Rémy J-M Molines, C. Cabanes LEGI, Mercator Océan, Coriolis

Outline 1. Introduction: - GLORYS project overview - European context 2. GLORYS 2 : an eddy permitting (1/4°) global ocean reanalyses of the « altimetric era » - reanalysis system overview - results 3. GLORYS / My. Ocean reanalysis meeting (June 2011): outcome 4. Conclusions and perspectives 2

1. Introduction 3

GLORYS project: National level GLORYS: GLobal Ocean Reanal. Yses and Simulations • French Reanalysis project, supported by GMMC (Mercator, Coriolis). PI: B. Barnier • main partners: LEGI, LPO, LOCEAN, CNRM, CORIOLIS, MERCATOR • project started at national level in 2008 • ARGO era reanalysis (2002 -2008) produced in 2009 : GLORYS 1 V 1 Objectif en 2011: • Achever la production de GLORYS 2 V 1 (1993 -2009) • Valider GLORYS 2 V 1 et le run de reference MJM 95 selon le protocole de validation commun My. Ocean • Distribuer GLORYS 2 V 1 et MJM 95 • Quels défis pour les réanalyses ? organisation d’un meeting réunissant les utilisateurs et les producteurs de réanalyses océaniques. 4

GLORYS: different Streams Number of Obs. / 7 days GLORYS-1 Stream 1: 2002 -2008 “ARGO” GLORYS-2 Stream 2: 1993 -2007 “altimetry” T/S prof. : 2000~4500 SST: 0, 5°x 0, 5° 0, 1°x 0, 1° SLA: 3~4 satellites T/S prof. : ~1000 SST: 1°x 1° SLA: 2~3 satellites 1992 2002 NOW GLORYS 1 V 1: produced and delivered to users early 2009 GLORYS 2 V 1: produced and delivered to users early 2011 6

Global ocean reanalyses at EU level My. Ocean project: www. myocean. eu. org WP 4: Global ocean reanalysis production & assessment Target: eddy permitting (1/4) global ocean reanalyses Basic ingredients: • NEMO 3 Ocean code, tuned for reanalyses, provided by CNRS • ERA Interim forcing • Reprocessed historical observations provided by TACs • Different data assimilation methods • Ocean reanalyses: Ocean simulations constrained by reprocessed obs. CMCC, Mercator, U. Reading • Ocean free simulation: CNRS • Ocean state estimation based GLORYS reanalyses 7 on observations only: CLS

Global ocean reanalyses at EU level My. Ocean project: www. myocean. eu. org Within WP 4: Global ocean reanalysis production & assessment Two reanalyses versions are planed: V 1 reanalysis: “June 2011” V 2 reanalysis: “beginning 2012” (improved V 1) delivered products : monthly fields, T, S, U, V, SSH, Sea Ice (conc. , thickness, velocity) Documentation describing the reanalysis (strategy, implementation) Common CA/VAL protocol for reanalyses evaluation and assessment. 8

2. GLORYS 2 V 1: reanalysis system overview 9

GLORYS 2: reanalysis system overview Requirements for a global ocean/sea ice reanalysis • Global ocean/sea-ice « eddy-permitting » circulation model: ORCA 025. L 75 - NEMO OGCM (Ocean: OPA, sea-ice: LIM 2_EVP) - Configuration set up by DRAKKAR consortium (Barnier et al, 2006, Drakkar group, 2007). • Data assimilation system - Adapted from operational SAM systems used by MERCATOR-Ocean for operational forecasts - SEEK Filter • A delayed-time quality-controlled observations for assimilation - In-situ: CORA data base (Coriolis data center) - Altimetry: AVISO data base (CLS) - SST maps 10

Model: DRAKKAR ORCA 025 configuration NEMO 3. 0/3. 2 OGCM + LIM 2 EVP Sea-Ice model: Resolution: - Global 1/4° ORCA-type grid (1442 x 1021 grid points) - 75 vertical levels from 1 m at the surface to 200 m at the bottom for GLORYS 2 Initialization: December 1991 - Levitus 1998 climatology + Sea-Ice Concentration from NSDIC Bootstart products Parameterizations: Filtered free surface, Partial step, Energy and Enstrophy conserving advection scheme, Isopycnal diffusion for tracers, Biharmonic for momentum, TKE turbulence scheme Atmospheric forcing: - Bulk CORE Formulation (Large&Yeager, 2004) - ERA-Interim reanalysis products: 3 hourly for turbulent fluxes Daily for radiation (analytical diurnal cycle for solar) In house correction of the radiation based on GEWEX satellites fluxes products. 11

correction of ERA-Interim radiative fluxes SST spatial errors structures Downward SW Flux 2002 ERAinterim - GEWEX SST (2002) model – RTG method to correct SW, LW Large Scale ERAinterim Fluxes with GEWEX 12

Method of correction ERAInterim : (monthly running mean) daily mean Climatology Satellite Obs : (monthly running mean) daily mean Climatology Daily Climatology of corrective factor x Daily ERAInterim field: large scale Daily Corrected ERAInterim: large scale + Daily ERAInterim field: small scale Daily Corrected ERAInterim field The correction is local, (i, j) grid point. Large scales correction only. Gewex climatology : 1999 -2006; Indian Gap before. Gewex [40 S-40 N] : -3% SW and -6% LW (as Large & Yeager 2004 for ISCCP) GPCPV 2. 1 Climatology: 1997 -2008. No correction northward 65 N. No correction for ERAinterim climatology values less than 1 unit (W. M-2 or mm. day-1). • No change of interannual signal. • No change of synoptic patterns (cyclones). • The method can be applied outside the satellite period. 13

Correction of SW & LW ERAI/GEWEX Mean SST bias with SST RTG ERAI/SW&LW corrected Large improvement with Gewex data Same solution with corrected ERAinterim SW & LW fluxes Errors remains in upwellings & large mesoscale activity areas. FLUX ERA-Interim corrigés: utilisés à Mercator, LEGI, CNRM, IC 3. CAT 14

GLORYS: DATA ASSIMILATION SCHEME DATA ASSIMILATION SYSTEM: SAM 2 v 1, SEEK formulation SEEK Filter : - Innovation is calculated at the First Guess at Appropriate Time (FGAT) approximation - Control vector comprises the barotropic height, T, S, U and V - Background error covariance calculated from an ensemble of 3 D anomalies from a reference simulation -Adaptive error variance is consistent with innovation vector (a posteriori diagnostic) -The SEEK filter is weakly sensitive to the number of obs. to assimilate 3 D-VAR Bias correction: for T and S Incremental Analysis Updates (IAU): inserting increments over all model time steps smooth trajectory 15

Delayed time observations for data assimilation • Assimilated data : • Along track DT SLA (SSLATO/DUACS) : Jason-1, Jason-2, Envisat, GFO, ERS 1, ERS 2, Topex/Poseidon 17

Delayed time observations for data assimilation • Assimilated data : • Along track DT SLA (SSLATO/DUACS) : Jason-1, Jason-2, Envisat, GFO, ERS 1, ERS 2, Topex/Poseidon • Reynolds AVHRR-only 0. 25° SST, 18

Delayed time observations for data assimilation • Assimilated data : • Along track DT SLA (SSLATO/DUACS) : Jason-1, Jason-2, Envisat, GFO, ERS 1, ERS 2, Topex/Poseidon • Reynolds AVHRR-only 0. 25° SST, • in situ temperature & salinity profiles (Argo network + Xbts, CTDs, etc. . ) : CORA 2. 3 data base 19

2. GLORYS 2 V 1 reanalysis : results 20

Reanalysis validation strategy Validation protocol for reanalyses : MERSEA-GODAE metrics, CLIVAR/GSOP diagnostics, My. Ocean quality control for operational systems adapted for reanalyses 6 types of diagnostics: CLASS 1: computations are done on the model native grid. CLASS 2: comparison of reanalysis products moorings, tide gauges and indexes (e. g. Nino boxes) comparisons CLASS 3: Mean and variability of integrated quantities like transports (volume, heat, freshwater), sea ice extent, global/regional averages of various parameters (SST, SLA, sea ice concentration, …) CLASS 4 : collocation of observation with model DATA ASSIMILATION DIAGNOSTICS: innovation statistics. EOFs analysis 21

Results: GLORYS 2 V 1 reanalysis : 1992 -2009 ARGO is getting in Data assimilation diagnostics for T Global misfit average No IAS B Global misfit RMS Si M s e abl st R 22

GLORYS 2 V 1 reanalysis results Data assimilation monitoring : SLA RMS ~7 cm No BIAS 1992 2009 23

Results: GLORYS 2 V 1 reanalysis : 1992 -2009 CLASS 1 metric - difference with climatology GLORYS 2 REFERENCE: MJM 95 SST (0. 5 m depth): difference with Levitus climatology 24

Results: GLORYS 2 V 1 reanalysis : 1992 -2009 CLASS 1 metric - difference with climatology GLORYS 2 REFERENCE: MJM 95 Temperature (at 100 m depth): difference with Levitus climatology 25

Results: GLORYS 2 V 1 reanalysis : 1992 -2009 CLASS 1 metric - difference with climatology GLORYS 2 REFERENCE: MJM 95 Salinity (at 0. 5 m depth): difference with Levitus climatology 26

Results: GLORYS 2 V 1 reanalysis : 1992 -2009 CLASS 1 metric - difference with climatology GLORYS 2 REFERENCE: MJM 95 Salinity (at 100 m depth): difference with Levitus climatology 27

Results: GLORYS 2 V 1 reanalysis : 1992 -2009 1022 J Ocean Heat Content 0 -700 m Levitus et al. , GRL 2005 GLORYS 2 V 1 28

Results: GLORYS 2 V 1 reanalysis : 1992 -2009 Ocean Heat Content 0 -700 m GLORYS 2 V 1 reanalysis Observations only GLORYS 2 V 1 Lyman et al. 2010 1992 2009 29

Results: GLORYS 2 V 1 reanalysis : 1992 -2009 Ocean Heat Content 0 -700 m MJM 95 reference simulation Observations only Lyman et al. 2010 1992 2009 30

Results: GLORYS 2 V 1 reanalysis : 1992 -2009 Ocean global heat budget [0 -2000 m] global mean temperature GLORYS 2 V 1 Levitus 1998 Climatology Mean temperature trend = + 0. 5 W/m 2 1992 2009 31

Results: GLORYS 2 V 1 reanalysis : 1992 -2009 SST trends ___ GLORYS: 0. 10°C/decade - - - OBS. : 0. 17°C/decade 32

Results: GLORYS 2 V 1 reanalysis : 1992 -2009 Sea level rise Obs. (Altimetry) Mean sea level rise 2. 8 mm/year 33

Results: GLORYS 2 V 1 reanalysis : 1992 -2009 Arctic Sea Ice: Ice extent SEA ICE EXTENT INTERANNUAL ANOMALIES Correlation = 0. 9 34

Results: GLORYS 2 V 1 reanalysis : 1992 -2009 Arctic Sea Ice Extent: September 1996 maximum 35

CLASS 1 metric - difference with climatology Arctic Sea Ice Concentration 36

Results: GLORYS 2 V 1 reanalysis : 1992 -2009 Arctic Sea Ice Extent: September 2007 minimum 37

Results: GLORYS 2 V 1 reanalysis : 1992 -2009 CLASS 1 metric - difference with climatology ed bias are ties 1 ci velo L 201 er drift t al. GR SVP ky e s 15 m zonal velocity Grod Under estimation of surface velocities Still major flaws in the representation of equatorial undercurrents (the FREE RUN is far better) 39

Results: GLORYS 2 V 1 reanalysis : 1992 -2009 CLASS 2 metric – local comparison with independent data GLORYS 2 Comparison with Tide Gauges: GLORYS 1 40

Results: GLORYS 2 V 1 reanalysis : 1992 -2009 Maximum Atlantic MOC at 26. 5°N RAPID-MOCHA measurements GLORYS reanalysis 20 10 = in situ observations (after, Bryden et al. , 2005) 20 10 41

Results: GLORYS 2 V 1 reanalysis : 1992 -2009 Quality Control on in situ data performed in GLORYS reanalysis: Help to improve delayed time in situ data bases “bad” Temp. profiles in 2009 “bad” Salinity profiles in 2009 43

Results: GLORYS 2 V 1 reanalysis : 1992 -2009 Quality Control on in situ data performed in GLORYS reanalysis: Help to improve delayed time in situ data bases Obs. CLIM FCST innovation T profile Obs. CLIM FCST innovation S profile 44

GLORYS users > 40 users, data volume provided > 16 TB several research areas and application fields: • Biogeochemical modelling (LSCE, Mercator) • Ecosystem modeling (CLS/MEMMS) • Sea Ice (LPO, Mercator) • TIWs (LEGOS) • p. CO 2 (NIES Japan, Bjerknes CCR) • Ocean circulation estimation and validation (CLS, SHOM) • Ocean Thermal Energy Conversion (EDF, Mercator) • Boundary / Initial conditions for regional modeling (CNRM, Infocean) • Mean Dynamic Topography inter comparisons (CLS, China) • MOC, MHT (RSMAS, UFRPE, CLS, Mercator) • South Atlantic circulation (NOAA) • Climate indexes, trends (IMEDEA, Spain) • Global ocean Mass budget (EOST/IPGS, NASA/GSFC) • object / animal drift (CLS, LPO, U. Texas, U. Hawai) • Hurricane (LEGI, CERSAT) 46 • …

3. GLORYS / My. Ocean reanalysis meeting in June 2011 47

GLORYS/MYOCEAN Workshop en Juin 2011 : Objective: The objective of this workshop is to gather the users and the producers of these reanalysis products, as well as other reanalysis products (e. g. My. Ocean. WP 4, etc) in order : to assess the scientific value of the reanalyses, to identify and prioritize ways of improvements of those products, to prepare the next generation of reanalyses • 2 -day meeting • ~ 30 participants • 17 oral presentation • discussions 48

Conclusions from GLORYS/MYOCEAN Workshop: - Extend the reanalysis in the past (1979 present) - Assimilation of Sea-Ice data (concentration and velocity). - Improving of ORCA 025 model (including tides). - Improving forcing (inter-annually varying runoffs, flux correction through data assimilation). - search for additional assimilation data (Tide gauges, SSS from SMOS, Aquarius, GRACE, etc. ). - Links with the Green Ocean. - Improving the computing efficiency of the reanalysis system. - Partnership with decadal prediction community. - Feasibility of a global reanalysis at 1/12° with the ORCA 12 model 49

4. Conclusions & Perspectives 50

GLORYS 2: Conclusions GLORYS 2 V 1 (1992 -2009): résultats 1. capacité de production de réanalyse globales à méso-echelle: collaboration unique entre l’opérationnel (MERCATOR, CORIOLIS) et la recherche (LEGI, …) 2. production de la première réanalyse au niveau européen multidonnées (SLA, SST, in situ) eddy resolving sur la période altimétrique. 3. résultats de très bonne qualité, GLORYS 2 V 1 > GLORYS 2 V 1. validation extensive dans un cadre international (My. Ocean, Godae) 4. reconnaissance au niveau international, >40 utilisateurs, applications très variées. 51

GLORYS 2: Perspectives GLORYS 2 V 2 (1992 -2009): reanalysis is close to be finished Improvements: - ERA-Interim corrected precipitation with GPCPv 2 observations - Iceberg melt parameterization in the Antarctic region - New MDT + new MDT errors from an objective analysis of 3 MDT products - On line Quality Control on of situ data temperature and salinity profiles - Assimilation of In situ profiles from sea mammals (T, S profiles) in the Antarctic (2004 -2009) → Available beginning 2012. 52