1 / 14

Assimilation of S(T) from ARGO

Assimilation of S(T) from ARGO. Keith Haines, Arthur Vidard * , Xiaobing Zhou, Alberto Troccoli * , David Anderson * Environmental Systems Science Centre, (ESSC) Reading University * ECMWF. Surface Freshening. Surface Warming. T/S relations and air-sea fluxes. Bindoff and McDougall (1994).

soren
Télécharger la présentation

Assimilation of S(T) from ARGO

An Image/Link below is provided (as is) to download presentation Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author. Content is provided to you AS IS for your information and personal use only. Download presentation by click this link. While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server. During download, if you can't get a presentation, the file might be deleted by the publisher.

E N D

Presentation Transcript


  1. Assimilation of S(T) from ARGO Keith Haines, Arthur Vidard*, Xiaobing Zhou, Alberto Troccoli*, David Anderson* Environmental Systems Science Centre, (ESSC) Reading University *ECMWF

  2. Surface Freshening Surface Warming T/S relations and air-sea fluxes Bindoff and McDougall (1994) Changes in temperature and salinity on z levels and on isopycnals allow surface forcing signature to be determined. Assimilation induced changes in water masses in OCCAM model with T only assimilated (Fox et al 2003)

  3. Temperature profile assimilation at ECMWF • All T profiles assimilated together, including those from CTD/ARGO data (i.e. where salinity also available) • ΔT innovations spread out horizontally only using gaussian decorrelation function (level by level assimilation) K ~ exp –[(Δx/Rx)2 + (Δy/Ry)2]; Rx= 15°; Ry = 3° equator • Analysed Ta down to deepest observation depth zmax • Model background Tb displaced vertically to match Ta(zmax) to give Ta (z>zmax) • S1 Salinity increment to give Sa consistent with no change in S(T) (Troccoli and Haines; 1999)

  4. New S(T) assimilation scheme • Start with Ta ; Sa = Sb +S1from temperature assimilation. • At CTD/ARGO observation points calculate salinity increments ΔS2 = [So(To) - Sb(To)]at temperature To • ΔS2 is a now direct measure of change in S(T) • Store ΔS2 for several To in a profile. ECMWF store 1 per model level; could have more • How to use ΔS2(To) at distance Δr to influence Sa(Ta)? • Use covariance K ~ exp –[((To – Ta)/ RT)2 + (Δx/Rx)2 + (Δy/Ry)2]; Rx; Ry; RT ? What scales to choose?

  5. ECMWF Seasonal Forecasting Assimilation Aug 2002 – Aug 2003: One year of Temperature … and Salinity data

  6. Salinity increments from ARGO assimilation at ECMWF • New S(T) assimilation leads to 2 increments • Balancing increment S1associated with • T assimilation keeps S(T) unchanged • (already operational at ECMWF for past • 2 years, Troccoli et al 2002) • Salinity assimilation increment S2 • associated with observed S(T) changes • (under test, 1 year assimilation complete) First assimilation increments Aug02 (averaged over upper 300m) S1 S1 + S2 S2 Mean N. Atl. Salinity Top 300m S1 only Aug02 Aug03

  7. Salinity Black= rms (obs-back) Red= rms (obs-anal) Mean Salinity top 300m Trop Pac box Aug02 Aug03 ARGO S1 + S2 S1 only

  8. Covariance scales for salinity S K ~ exp –[((To – Ta)/ RT)2 + (Δx/ Rx)2 + (Δy/Ry)2]; • How to choose Rx= ; Ry = ; RT= ? • Consider To = Ta : then Rx and Ry are clearly correlation scales on T surface • Calculate correlations from model data sets 4 years of OCCAM high ¼ degree data every 5 days 50 years HadCM3 1.25 degree data every month • Scales must represent the right kind of S(T) variability, i.e. variability associated with climatic changes!! (model drift?)

  9. One-point correlation S(12C). Example S(12C) Shear dispersion only Noise Mesoscale One point correlation S(355m) Seasonal cycle not removed! Example S(355m) Different Scale OCCAM ¼ degree model run for 4 years

  10. HadCM3 model: 50 years data S(301m) one pt covariances S(12C) one pt covariances

  11. HadCM3 model run S(301m) one pt covariances S(12C) one pt covariances 50 yrs 4 yrs

  12. HadCM3 model: 50 years data S(301m) one pt covariances S(12C) one pt covariances x exp –[(Δr/R)2 + (ΔT)/ RT)2] S(T) covariances at one location

  13. S(T) Covariances • Covariance scales for S(T) should be larger than covariance scales for S(z)=Mesoscale • Models must be run long enough to have realistic S(T) variability which is not simply model drift • Best illustration would come from a long run of mesoscale model with stable climate! • Tune scales during assimilation based on model-data misfits (common in meteorology)? May require long time period to capture interesting S(T) variations.

  14. Further work • Tuning of S assimilation at ECMWF • Covariance scales from models or by tuning (eg. 2 or OmF stats.) • Compare scales with QC scales cf. Boeme/Send! • 40 year ocean reanalysis (EU ENACT project) • Analyse changes in T/S properties to detect climate signals as in Bindoff and McDougall or Walin • Impact of Salinity assimilation on seasonal/mesoscale forecasting (ECMWF, Met Office)

More Related