Providing an improved wind stress parameterization and QuikSCAT wind stress constraints for ECCO2 - PowerPoint PPT Presentation

providing an improved wind stress parameterization and quikscat wind stress constraints for ecco2 n.
Skip this Video
Loading SlideShow in 5 Seconds..
Providing an improved wind stress parameterization and QuikSCAT wind stress constraints for ECCO2 PowerPoint Presentation
Download Presentation
Providing an improved wind stress parameterization and QuikSCAT wind stress constraints for ECCO2

play fullscreen
1 / 20
Download Presentation
Download Presentation

Providing an improved wind stress parameterization and QuikSCAT wind stress constraints for ECCO2

- - - - - - - - - - - - - - - - - - - - - - - - - - - E N D - - - - - - - - - - - - - - - - - - - - - - - - - - -
Presentation Transcript

  1. Providing an improved wind stress parameterization and QuikSCAT wind stress constraints for ECCO2 David Moroni1 Holger Brix2 Dimitris Menemenlis1 Hong Zhang2 Jet Propulsion Laboratory California Institute of Technology Pasadena, California 1. Jet Propulsion Laboratory / California Institute of Technology 2. University of California, Los Angeles

  2. Motivation • More than 10 years of QuikSCAT data are available at 90% global daily coverage provides exciting opportunity for improving the ECCO2 MITgcm. • An improved, ‘prototype’ wind stress analysis product would have enormous impacts for both operations and research. • Potential for optimizing of high resolution simulations. • Potential contributions to the fundamental understanding of the physics and dynamics of oceanographic processes.

  3. Objectives • Test an improved wind stress parameterization in MITgcm. • Run surface-relative and absolute wind integrations. • Compare with Large and Pond (1981) parameterization. • Compare all runs with QuikSCAT retrievals. • Incorporate improved parameterization into MITgcm. • Quantify model vs. data errors through comparisons of QuikSCAT-ECCO2 wind stress. • Requirement for assimilation. • Repeat 1/8° simulation (Hill et al. 2007) using both: • QuikSCAT wind stress constraints and • improved, sea-state-dependent (hypothetically ideal) surface wind stress parameterization.

  4. Power Spectrum: MITgcm vs. TOPEX Credit: Hill et al. (2007)

  5. Relative vs. Absolute Wind Mean Surface Velocities: 1992-2002 Zonal: Absolute – Relative Meridional: Absolute – Relative

  6. Latent Heat Flux: Global Probability Density Momentum Flux: Global Probability Density

  7. Comparisons to In Situ Stress • Recent developments in sea-state dependent stress parameterizations have been adjusted to provide a better fit to in situ stress observations (Bourassa 2004, 2006): • Created a ‘prototype’ QuikSCAT Level-2 wind stress retrieval product based on the Bourassa (2006) parameterization. Bourassa (2006) Large and Pond (1981)

  8. QuikSCAT-ECCO2 Inconsistencies Zonal Stress RMS Difference (N m-2) > 95% Data Errors Below 0.025 N m-2

  9. Concluding Remarks • An improved 1/8° simulation would raise the bar for higher resolution simulations. • Global, gridded stress fields from ECCO2 could be provided as an open-source analysis product. • Potential for climate studies and short-term forecasting applications. • Potential for cal/val with other OGCMs. • Potential for cal/val with current and future scatterometer stress GMF development. • Re-estimation of wind work upon the ocean circulation, continuing in the direction of Wunsch (1998) and von Storch et al. (2007).

  10. Thank you!

  11. Appendix Slides

  12. Technical Approach • How do we proceed? • Perform long-term (1999-present) statistical comparisons between QuikSCAT and ECCO2 wind stress; analysis will be updated as new and improved QuikSCAT stress products become available. • Quantify the differences between data and model errors to show the “representation” error. • Error partitioning will be used to assimilate QuikSCAT into ECCO2. • Include new and improved stress paramaterizations into the ECCO2 MITgcm, which depend upon sea-state, surface-relative wind, and atmospheric stability. • Perform series of forward model sensitivity experiments following a Green’s function approach. • Determine which stress parameterization within ECCO2 provides the best consistency with QuikSCAT-derived stress. • Using the adjoint method, include the QuikSCAT-derived wind stress constraints in ECCO2 (i.e., data assimilation). • Re-iterate as new QuikSCAT stress products become available. • Re-compute cost-function to ensure optimization. • Address the wind work question once the ECCO MITgcm has been optimized.

  13. Science Questions • Energy needed to maintain general circulation estimated to be 2.1 TW (Munk and Wunsch 1998); approx. ½ this amount is thought to be provided by wind work on ocean surface • Problem: previous estimates of the total wind work on ocean circulation vary considerably. • Early estimates suggest 1 TW (Faller 1966). • Wunsch (1998) derived the same estimate but assumed only geostrophic flow. • von Storch et al. (2007) estimated 2.7 TW on ageostrophic flow and 1.1 TW on geostrophic flow. • Modeling and observational studies agree that surface currents provide a significant sink (~0.2 TW) for the wind work driving the geostrophic flow (Duhaut and Straub 2006; Dawe and Thompson 2006; Hughes and Wilson 2008; Risien and Chelton 2008; Xu and Scott 2008). • Can we improve quantitative estimates of wind work upon both the surface geostrophic and ageostrophic circulation? • Can these estimates be made consistent with satellite observations?

  14. Scientific Motivation • Wind work is largely a function of wind stress: • Wind stress is most commonly estimated using a drag coefficient parameterization. • ECCO2 utilizes a formulation derived from Large and Pond (1981): • Assumes neutral atmospheric stability, non-variable sea-state, and only calibrated for wind speeds < 26 m s-1

  15. Scientific Motivation - Continued • The MITgcm within ECCO2 is currently optimized to provide wind stress as a function of earth-relative wind. • Recent ECCO2 simulations with a relative-wind integration reveal: • Regions containing 50% reductions in 10yr mean zonal surface velocity. • A 30% reduction in Global EKE production. • Larger overall model-data differences which increase the cost function.

  16. Scientific Motivation - Continued

  17. Scientific Motivation - Continued • Contrary to what is already done for SST and SSH, the MITgcm in ECCO2 is currently not constrained by stress observations. • Time mean and RMS differences between QuikSCAT-derived wind stress and ECCO2-derived wind stress reveal startling contrasts and inconsistencies. • An out-dated and inadequate stress parameterization within ECCO2 is the likely culprit.

  18. Scientific Motivation - Continued QuikSCAT-ECCO2 Time Mean Difference (2004-2005) QuikSCAT-ECCO2 RMS Difference (2004-2005)

  19. Global Near-Surface Stability Distribution

  20. Effect of Stability on Transfer Coefficients • This particular example is using the stress parameterization of Smith (1988). • Lines are isopleths of Tair-Tskin Heat Transfer Coeff. vs. Wind Speed Drag Coefficient vs. Wind Speed Unstable Unstable CD (x10-4) CH (x10-4) Stable Stable U10(m/s) U10(m/s)