L. Parent , N. Ferry, B. Barnier, G. Garric, C. Bricaud, C-E Testut, - PowerPoint PPT Presentation

GLOBAL Eddy-Permitting Ocean Reanalyses and Simulations of the period 1992 to Present L. Parent , N. Ferry, B. Barnier, G. Garric, C. Bricaud, C-E Testut, O. Le Galloudec, J-M Lellouche, E. Greiner, M. Drevillon, E. Rémy J-M Molines, Mercator Océan, LGGE-CNRS, Coriolis, CLS Sixth WMO Symposium on Data Assimilation NOAA Center for Weather and Climate Prediction / Univ of Maryland, College Park, USA, 7-11 October 2013

Outline 1. Introduction : The GLORYS project overview and the European context 2. GLORYS2 : an eddy permitting (1/4 ) global ocean reanalyses of the « altimetric era » System overview and Performances 3. Conclusions & Perspectives 2

1. Introduction 3

GLORYS project: National level GLORYS: GLobal Ocean ReanalYses and Simulations - French Reanalysis project, supported by GMMC (Mercator, Coriolis). PI: B. Barnier - main partners: Drakkar consortium, CORIOLIS, MERCATOR - project started at national level in 2008 + cooperation with EU funded FP7 MyOcean project MOTIVATION : The need for a realistic description of the ocean state and variability over the recent decades , at the global scale , and at the scale of the ocean basins and regional seas. OBJECTIVES : - Produce an eddy permitting global ocean/sea-ice reanalysis spanning the “altimetric + ARGO" era 1992-2009 - To iterate / produce different reanalysis along the 1992-today time period - Start to design the ERA-Interim reanalysis scenario : 1979-today - Promote the use of reanalysis products in the climate community 4

GLORYS: different Streams Number of Obs. / 7 days GLORYS1 T/S prof .: 2000~4500 Stream1 : 2002-2008 SST : 0,5 x0,5  0,1 x0,1 “ARGO” SLA : 3~4 satellites GLORYS2 T/S prof .: < 1000 (+ sea mammals) Stream2 : 1993-2010 SST :1 x1 “altimetry” SLA : 2~3 satellites SIC : CERSAT data 25km (12km avail) GLORYS3 T/S prof .: < 800 SST :1 x1 Stream3 : 1979-2012 SLA : none “ERAinterim” SIC : NSIDC data 1979 1992 2002 NOW 5

Global ocean reanalyses at EU level MyOcean project : www.myocean.eu.org MyOcean1: 2009-2012, MyOcean2: 2012-2014 Basic ingredients : - NEMO Ocean source code, tuned for reanalyses, provided by CNRS - ERA Interim forcing + some corrections - Reprocessed historical observations provided by Thematic Assembly Center - Different data assimilation methods Ocean reanalyses : Ocean simulations constrained by reprocessed obs.  CMCC, Mercator, U. Reading  CNRS Ocean free simulation: Ocean state estimation based on observations only  CLS 6

2. GLORYS2: System Overview and Performances 7

Model: DRAKKAR ORCA025 configuration NEMO OGCM + LIM Sea-Ice model : Resolution : - Global 1/4 - 75 vertical levels from 1 m at the surface to 200 m at the bottom 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 corrections : of the radiation based on GEWEX satellites fluxes products of the precipitations based on GPCPv2 observations 8

GLORYS: DATA ASSIMILATION SCHEME DATA ASSIMILATION SYSTEM : SAM2v1 used in real-time global forecasting systems SAM2V1 assimilation platform : - Reduced order extended Kalman filter family (SEEK kernel) -This approach is similar to the Ensemble optimal interpolation ( EnOI ) developed by Oke et al., (2008) which is an approximation to the EnKF that uses a stationary ensemble to define background error covariances - Innovation is calculated at the First Guess at Appropriate Time ( FGAT ) approx. - Analysis is performed at the middle of the 7-day assimilation cycle 3D-VAR Bias correction : to correct large-scale temperature and salinity biases Incremental Analysis Updates (IAU) : inserting increments over all model time steps  smooth trajectory, more costly 9

Delayed time observations for data assimilation Along track DT SLA (SSLATO/DUACS) : Jason1, Jason2, Envisat, GFO, ERS1, ERS2, Topex/Poseidon + use of an adjusted CNES-CLS09 Mean Dynamic Topography (GOCE obs) 10

Delayed time observations for data assimilation Reynolds AVHRR-only 0.25° SST ftp://eclipse.ncdc.noaa.gov/pub/OI-daily-v2/NetCDF/ assimilated once at the date of the analysis (4 th day, 0h) 11

Delayed time observations for data assimilation - in situ temperature & salinity profiles : CORA3.3 data base Argo network + Xbts,CTDs , etc… + sea mammal (elephant seals) database (Roquet et al., 2011) 12 Courtesy from the Sea Mammal Research Unit (SMRU)

Delayed time observations for data assimilation - Sea Ice concentration : IFREMER/CERSAT products (Ezraty et al., 2007). Assimilated once at the day of the analysis (4 th day, 0h), same as SST March 17, 2003 from QuikSCAT sensor January 5, 2009 from ASCAT sensor 13

Quality Control : Quality Control on in situ data performed in GLORYS reanalysis : in order to minimise the risk of erroneous observed profiles being assimilated in the model Obs. CLIM FCST Distrib INNOV INNOV “suspicious” Temperature profiles in 2009 14

Some results: assimilated data Altimetry, In-situ Misfit average Rms misfit Altimetry ARGO is getting in Weak warm bias In-situ Temp 15

Some results: assimilated data SLA Domain averaged sea level Imbalance E-P Obs, gridded data GLORYS2V3 Reference sim - Good agreement with SLA CCI gridded data - Ref sim: imbalance in the E-P forcing 16

Some results: Tide gauges 1993-2011 time period Correlation Good agreement, except along some coasts (no tidal model) 17

Some results: surface velocity, 1993-2011 mean Zonal surface currents GLORYS2V3 NOAA AOML Good agreement but there is a general underestimation: artefact of AOML surface current ? Grodsky et al. 2011, one explanation : undrogued drifters 18

Some results: Mean Kinetic Energy,1000m 2002-2009 mean reference simulation ANDRO Argo drift data base GLORYS2V3 - Good agreement with Argo all along the boundary currents and the ACC - Deep ocean currents too strong in the equatorial band (+/-10 ) 19

Some results: Meridional Overturning Circulation 2004-2010 time period Rapid data GLORYS2V3 Reference sim - The Atlantic MOC is stable throughout the reanalysis period - 0.8 correlation, underestimation of 3 Sv 20

Some results: Meridional Heat Transport 1993-2011 time period Atlanti c estimates GLORYS2V3 Reference sim Consistent with Ganachaud and Wunsch (2000), Trenberth and Caron (2001) estimates 21

Some results: Sea Ice Sea Ice concentration is assimilated CERSAT data GLORYS2V3 1993-2011 mean Antarctic March In the Antarctic, Data Assimilation has a positive impact during summer and winter Sept 22

Some results: Sea Ice Sea Ice concentration is assimilated GLORYS2V3 CERSAT data Extreme Events Sept 1996 Good behaviour of GLORYS2V3 during extreme events Sept 2007 23

Some results: Sea Ice Sea Ice concentration is assimilated Arctic Antarctic Sea Ice extent anomaly CERSAT data GLORYS2V3 Reference sim Sea Ice volume anomaly not realistic 24

3. Conclusions & Perspectives 25

Conclusions: GLORYS2 (1992-2011) - Ability to produce global meso-scale reanalysis simulations: Unique collaboration between operational centers (MERCATOR, CORIOLIS) and research Labs (LGGE, …) - Stream after stream, there is an improving quality - Regional reanalysis are underway: Mediterranean sea, European coasts - Difficulties to control deep ocean, unrealistic trends T/S - Difficulties to control Sea Ice Volume in the Arctic - 40 users, various applications 26

Perspectives: GLORYS2-3 . GLORYS2VX (1992-2012) : - Ongoing effort to improve products and services - MyOcean2 project: production of reanalyses is still a priority . GLORYS3 (1979-2012) : - ERAInterim years - First stream in 2014 . Data Assimilation : - Data assimilation of surface currents - Gaussian Anamorphosis transformation: to improve DA of sea ice conc. - Multiple scale analysis (see poster H-p38, Testut) - Use of 4D (+ time) error modes  smoother approach - Ensemble approach: open …. 27