1 / 14

Atmosphere at rest experiments with the latest COSMO version and comparison with EULAG

Atmosphere at rest experiments with the latest COSMO version and comparison with EULAG. Oliver Fuhrer, MeteoSwiss. Introduction. Why again? experiments performed with EULAG, but different setup latest COSMO model version (4.14) more sensitivity studies What is tested?

Download Presentation

Atmosphere at rest experiments with the latest COSMO version and comparison with EULAG

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. Atmosphere at rest experiments with the latest COSMO version and comparison with EULAG Oliver Fuhrer, MeteoSwiss

  2. Introduction Why again? experiments performed with EULAG, but different setup latest COSMO model version (4.14) more sensitivity studies What is tested? terrain following coordinate transformation introduces additional truncation error term for flows which are nearly hydrostatic how large is this error? Basic setup topography u = v = w = 0 hydrostatic equilibrium

  3. Why does not everything cancel? 3 2 1 1 and BC determine p’ completely. 2 and 3 only cancel out to precision of discretization.

  4. Ideal test case I • 2-dimensional • Schaer et al. MWR 2002 topography • Gal-Chen coordinates • ∆x = 1 km, Lx = 320 km • ∆z = 400 m, Lz = 20 km • ∆t = 10 s • Reference atmosphereN = 0.01 s-1 • Initial stateT0 = 288.15 K, p0 = 105 Pa, dT/dlogp = 42

  5. Sensitivity: Topography height h = 0 m h = 1 m h = 10 m h = 100 m h = 300 m h = 500 m h = 1000 m h = 2000 m h = 4000 m Crash!!!

  6. Sensitivity: Mountain Height

  7. Sensitivity: Mountain Width

  8. Sensitivity: Timestep

  9. Sensitivity: Summary • Mountain height / steepness play key role • Explicit vertical advection (EVA) helps • Timestep has small influence • θ or θ’ dynamics worsens situation • Independent of lower BC • Explicit hyper-diffusion on model levels helps • Time weighting (β) in fast-modes has no influence • Order of horizontal advection has negligible influence

  10. Ideal test case II • 2-dimensional • Schaer et al. MWR 2002 topography • Gal-Chen coordinates • ∆x = 1 km, Lx = 320 km • ∆z = 400 m, Lz = 20 km • ∆t = 10 s • Reference atmosphereN = 0.01 s-1 • Initial stateT0 = 288.15 K, p0 = 105 Pa, dT/dlogp = 42 • Rayleight sponge (> 13 km)

  11. Sensitivity: Topography height h = 0 m h = 1 m h = 10 m h = 100 m h = 300 m h = 500 m h = 1000 m h = 2000 m h = 4000 m Crash!!! Crash!!!

  12. Comparison COSMO vs. EULAG 2.0 10-12 1.8 10-5 1.0 10-4 2.2 10-4 4.5 10-4 6.0 10-3 (crash) 2.0 10-12 6.4 10-5 2.5 10-4 6.4 10-4 9.3 10-3 (crash) (crash) 7.1 10-13 2.3 10-2 8.5 10-2 1.4 10-1 (crash) (crash) – 7.1 10-13 4.9 10-2 9.1 10-2 1.6 10-1 (crash) (crash) –

  13. Comparison EVA vs. IVA 2.0 10-12 1.8 10-5 1.0 10-4 2.2 10-4 4.5 10-4 6.0 10-3 (crash) 2.0 10-12 6.4 10-5 2.5 10-4 6.4 10-4 9.3 10-3 (crash) (crash) 7.1 10-13 2.3 10-2 8.5 10-2 1.4 10-1 (crash) (crash) – 7.1 10-13 4.9 10-2 9.1 10-2 1.6 10-1 (crash) (crash) – 1.8 10-12 2.0 10-5 1.1 10-4 2.4 10-4 4.7 10-4 6.3 10-3 2.9 10-1 1.8 10-12 7.1 10-5 2.7 10-4 6.6 10-4 1.1 10-2 3.1 10+1 (crash) 6.8 10-13 8.3 10-3 1.4 10-2 1.7 10-1 4.9 10-2 6.4 10-2 5.8 10-1 6.8 10-13 1.3 10-2 2.4 10-2 3.6 10-2 1.6 10-1 2.3 10+1 –

  14. Conclusion Results with model version 4.14 are better than with model version 4.7 Results for stable experiments compare well to EULAG and are always within one order of magnitude Model still crashes for too steep and high topography Explicit vertical advection (EVA) and some explicit hyper-diffusion go some way in stabilizing model, but do not solve problem Other factors have little or not influence

More Related