1 / 17

Objective Obtain an eddying global ocean and sea ice state estimate using the adjoint method.

Progress towards a 16-month CS510 adjoint state estimate Hong Zhang, Dimitris Menemenlis JPL/Caltech Gael Forget, Patrick Heimbach, Chris Hill MIT/EAPS ECCO2 meeting, Pasadena, 2009. Objective Obtain an eddying global ocean and sea ice state estimate using the adjoint method. Outline

ellery
Télécharger la présentation

Objective Obtain an eddying global ocean and sea ice state estimate using the adjoint method.

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. Progress towards a 16-month CS510 adjoint state estimateHong Zhang, Dimitris MenemenlisJPL/CaltechGael Forget, Patrick Heimbach, Chris HillMIT/EAPSECCO2 meeting, Pasadena, 2009

  2. Objective Obtain an eddying global ocean and sea ice state estimate using the adjoint method. Outline 1. Introduction 2. Results a. overall cost reduction b. IC, BC adjustments c. RMS model-data difference 3. Concluding Remarks

  3. One year ago, from Gael’s presentation at the MIT ECCO2 meeting

  4. One year ago, from Patrick’s presentation at the MIT ECCO2 meeting

  5. Configuration details • CS510 cube sphere with ~18 km horizontal grid spacing and 50 vertical levels • First guess initial conditions from a 2-yr spin up from OCCA climatology • First guess atmospheric state from ECMWF • Constraints: OCCA climatology, ARGO, XBT, Jason, Envisat, and AMSR-E • Currently optimizing 16 months starting January 2004, i.e., the beginning of the ARGO period • Using 900 cpus on Columbia or 3600 cpus on Pleiades (for extra memory to reduce adjoint re-computations)

  6. Overall cost function reduction

  7. ARGO (T/S) cost function reduction

  8. At 154m At 634m Adjustments of initial temperature

  9. At 154m At 634m Adjustments of initial salinity

  10. mean std Adjustments of atmospheric zonal wind

  11. mean std Adjustments of atmospheric meridional wind

  12. At 154m At 634m rms(iter14 – Argo T) – rms(iter0 – Argo T)

  13. At 154m At 634m rms(iter14 – Argo S) – rms(iter0 – Argo S)

  14. At 154m At 634m rms(iter14 – OCCA T) – rms(iter0 – OCCA T) T

  15. At 154m At 634m rms(iter14 – OCCA S) – rms(iter0 – OCCA S)

  16. rms(iter14 – AMSRE SST) – rms(iter0 – AMSRE SST) rms(iter14 – EVISAT SSH) – rms(iter0 – EVISAT SSH)

  17. Concluding Remarks • A global, eddying, full-depth-ocean and sea-ice configuration of the MITgcm is being constrained with a variety of satellite and in-situ data products using the adjoint method • After 14 forward-adjoint iterations there is a 35% overall cost function reduction • TO DO: • additional data constraints (mean dynamic topography, QuikSCAT wind stress, sea-ice observations) • additional controls (e.g., vertical diffusivity) • improved minimization algorithms (smoothed gradients, improved line search, etc.) • longer optimization period • first science applications …

More Related