Preparation of the SWOT Mission M.Benkiran, E. Greiner, E. Rémy, P.Y. Le Traon and the Mercator Ocean team. Study done in the framework of a CNES/Mercator Ocean convention, in collaboration with CLS. GODAE OceanView; Toulouse, 10-11 December 2014-1 -
Questions Altimeter data along the tracks (T/P, Jason.) Futurs missions along the swaths ( SWOT) Questions : How can we manage the large number of data assimilation systems? How Swot High resolution observations can impact Ocean Analysis and Forecast? What is the impact of repetitivety in the analysis and in the forecast skill, Especially for the meso-scale structures Develop an ocean OSSE (Observing System Simulation Experiments) system capable of providing valid impact assessments of Swot measurement. GODAE OceanView; Toulouse, 10-11 December 2014-2 -
Outline Ocean model and assimilation system Ocean observation simulation, white-noise level Experimental setup : Observing System Simulation Experiment (OSSE) Preliminary results Conclusions / Perspectives SWOT KaRIN instrument draft (http://swot.jpl.nasa.gov/science/technology/). Study done in the framework of a CNES/Mercator Ocean convention, in collaboration with CLS. GODAE OceanView; Toulouse, 10-11 December 2014-3 -
OSSE With IBI System (regional system on Iberia-Biscay-Ireland area) System configuration: SSH IBI12 : 12/03/2009 Model : Nemo (1/12 ~ 8Km; 75 levels) Tides (including potential) : M2, S2, K2, N2, K1, O1, P1, Q1, M4,Mf, Mm Atmospheric pressure forcing Atmospheric forcing: 3h ERA-interim reanalysis products; CORE bulk formulae Open Boundaries Conditions : GLORYS (Mercator global reanalysis) System Assimilation: SAM2V1 (SEEK) : Operationel real time system Adaptation of the scheme to assimilate a bigger amount of observations, higher resolution signal. GODAE OceanView; Toulouse, 10-11 December 2014-4 -
Observations : Nature Run (NR: Realistic simulation) Model : Nemo (1/36 ~3km, 50 levels) SSH IBI36 : 12/03/2009 Tides (including potential) : M2, S2, K2, N2, K1, O1, P1, Q1, M4,Mf, Mm Atmospheric pressure forcing Atmospheric forcing: 3h ECMWF operational analysis (CORE bulk formulae) Open Boundaries Conditions : PSY2 (Operational Mercator-Ocean system) we use a validated simulation and reproduces the physical phenomena in the region GODAE OceanView; Toulouse, 10-11 December 2014-5 -
Observations Simulated Observations from IBI36 (Free Model, With 1/36 ~3km, 5 Day-Assimilation 2009) : Window we have good coverage under Swath (Eddies, Insitu : Temperature and salinity profiles (CORA Data gradient positions)..) but we have some of the area SST : Daily Mean with 25 Km for horizontal resolution not covered SSH : (25 hours mean ; Inverse Barometer and tide removed) Altimeters : J2, J1n, En Swot ( 7Km) SSH Swot IBI36 : Cycle : 12/03/2009 is 20.9 days; 10.9-day sub-cycle J2; J1n; of En the baseline orbit Swot SSH From NR(IBI36) : 09-14/03/2009 (5day-Assimilation window) OSSE : Add. White Noise to each type of data Not Add. Roll Observation Error (Swot) GODAE OceanView; Toulouse, 10-11 December 2014-6 -
OSSE system : Evaluation Experiments : FreeSim : Free model (All data are used in verification mode) OSSE1 : Assimilated : T & S profiles; SST and Conventional Altimetry Observations (J2, J1n, En) OSSE2 : Assimilated : T & S profiles; SST and Swot Observations (7Km x 7Km) All experiments are initialized from the same state (free model) on 01/01/2009; Objective: 1-year simulations in 2009 test the impact of data assimilation along swaths instead of the data along the tracks. GODAE OceanView; Toulouse, 10-11 December 2014-7 -
Determining impact of swot assimilation (Innovation = Data Model Forecast; 5-Day assimilation Window) SLA Score Along Jason2 tracks : FreeSim OSSE1 OSSE2 SLA Score Along Swot swaths: FreeSim OSSE1 OSSE2 5930 67677 The assimilation system is able to assimilate a large amount of data OSSE1 : significant reduction Rms of Innovation (data Model Forecast) with assimilation of altimetry data: 6.33 to 4.78 Cm in mean OSSE2 : With the assimilation of Swot data, we reduce this Rms along the tracks Jason2 (~0.4Cm) A significant reduction (fall) after two analysis cycles (10 days), once the basin is covered by Swot data Impact Assessments: We have a large data number (Swot) per an analysis cycle (x11) Rms Error decreases in OSSE2 Here we have the answer to our first question: How can we manage the large number of data assimilation systems? GODAE OceanView; Toulouse, 10-11 December 2014-8 -
Ssh Correlation (2009) : NR(Data) vs FreeSim vs OSSE1 vs OSSE2 : «True» Ocean (IBI36) vs daily model forecast OSSE1 : Some improvement compare to the FreeSim Some time discontinuity exist (exp: May ) OSSE2 : Good agreement between the model forecast and true Ocean (IBI36) Not time discontinuity in the model forecast when Swot Assimilated. NR (IBI36, True Ocean ) FreeSim; OSSE1; OSSE2 NR/FreeSim NR/OSSE1 NR/OSSE2 0.0 0.5 1.0 Mean : 59% 0.0 0.5 1.0 Mean : 72% 0.0 0.5 1.0 Mean : 80% GODAE OceanView; Toulouse, 10-11 December 2014-9 -
Vorticity (28/12/2009) : NR(IBI36) vs FreeSim vs OSSE1 vs OSSE2 Impact in surface velocity with Swot Assimilation NR(IBI36) OSSE1 FreeSim OSSE2 GODAE OceanView; Toulouse, 10-11 December 2014-10 -
Vorticity (Forecast,08/12/2009) : NR(IBI36) vs FreeSim vs OSSE1 vs OSSE2 NR(IBI36) «True» Ocean FreeSim OSSE1 OSSE2 GODAE OceanView; Toulouse, 10-11 December 2014-11 -
Conclusion & Perspectives The assimilation system is ready to handle the SWOT data set: Deal with a large number of data ( x11 / conventional SLA). Improvement due to SWOT data assimilation: Better reconstruction of the SSH fields, Analysed fields «smoother» than with the conventional tracks: control of the 2D gradients in SSH?, Structure reconstruction, comparison with the true (IBI36) ocean, velocity fields Positive impact on the in situ T/S profiles: correct vertical projection. Toward a more realistic configuration: Improvement/test of observation error variance, up to now uncorrelated observation errors Rolling error simulation. GODAE OceanView; Toulouse, 10-11 December 2014-12 -
Lyapunov exponent FreeSim OSSE1 OSSE2 IBI36(NR) GODAE OceanView; Toulouse, 10-11 December 2014-13 -
SSH Spectrum, in the native model grids 44 N SSH: FreeSim 10/09/2009 SSH: OSSE1 10/09/2009 SSH: OSSE2 10/09/2009 SSH: IBI36( NR) 10/09/2009 1000.0 100.0 50.0 10.0Km 34 N 19 W 11 W IBI36 Obs Forecast OSSE2 Forecast OSSE1 Forecast FreeSIm FreeSim : Free Run OSSE1: Assim ALt OSSE2 : Assim Swot IBI36 : Data SSH Mesoscale (10/09/2009) GODAE OceanView; Toulouse, 10-11 December 2014-14 -
Surface Current : IBI36 vs OSSE1 vs OSSE2 vs FreeSim 09/09/2009 (5day-Forecast) Current IBI12 (FreeSim) Current IBI12 (OSSE1) Current IBI12 (OSSE2) Current IBI36 GODAE OceanView; Toulouse, 10-11 December 2014-15 -
Observations (3/3) : White-noise levels White-Noise : Mean Error = 1.7 Cm for 1Km 2 Form : in accordance with satellite design requirement (Durant et al. 2010; E. Rodriguez 2011) Height Height Noise Noise SSH+ Noise : 08/01/09 OSSE : Not Add. SSH : 08/01/09 Roll Observation Noise Error : 08/01/09 GODAE OceanView; Toulouse, 10-11 December 2014-16 - -30. -20. -10. 0. -30. -20. -10. 0. Cm -3. -2. -1. 0. 1. 2. 3. Cm