1. Sea ice modeling at met.no Keguang Wang Norwegian Meteorological Institute

2. Sea ice models at met.no NorESM (Jens)  TOPAZ (Magnus)  ROMS (Keguang)  FRAMPS (Keguang)  ROMS-CICE (Jens, Keguang,...)

3. FRAMPS model description Granular ice dynamics –2-level ice category –Curved diamond yield curve [Wang, 2007, JGR] –Levy flow rule [Wang, 2006, Annals of Glaciol.], with modifications –Particle-in-cell for advection [Wang and Wang, 2009, JGR]. Comparison between RGPS-derived and modeled shear deformations (Wang and Wang, 2009)

4. Prediction domain Pan-Arctic –10 km resolution European Arctic –Fram Strait, Svalbard, Barents Sea –2 km resolution

5. Initial sea ice concentration AMSR-2 NIC ice chart CIS ice chart interpolation for North Pole

6. Initial sea ice thickness SMOS OSISAF ice type interpolation for North Pole

7. 48 hours forecast

8. forecast drift vs. CALIB_2015a (1st leg)

9. forecast drift vs. CALIB_2015c (2nd leg)

10. ROMS Nordic-4km data assimilation Operational ocean-sea ice model in met.no –TOPAZ –MIPOM –ROMS –FRAMPS Background –operational sea ice forecast system transfered from MIPOM to ROMS in 2013 –Sea ice concentration assimilated for ROMS Arctic-20km, but no yet for Nordic-4km –need for high accuracy sea ice forecast for shipping, fishing and oil and gas exploration

11. ROMS brief free-surface, terrain-following, primitive equations ocean model Very modern code, uses C-preprocessing to activate the various physical and numerical options Coupling modules for atmosphere, wave, biogeochemical, bio-optical, sediment, and sea ice applications

12. ROMS framework disgram

13. ROMS sea ice module Main developer: Budgell (2005) Two-level sea ice state EVP rheology MPDATA advection scheme One-layer ice and snow thermodynamics (Mellor and Kantha, 1989) Plume model between sea ice and water frazil ice formation (Steele et al., 1989)

14. ROMS operational system in met.no  A triply nested system: –Arctic-20km → Nordic-4km → NorKyst-800m  The global ocean model FOAM from UK Met Office gives the lateral boundary conditions to Arctic 20 km  ECMWF forcing for Arctic-20km and Nordic-4km

15. ROMS Arctic-20km

16. ROMS Arctic-20km + Nordic-4km

17. Arctic20 + Nordic4 + NorKyst800

18. Previous Nordic-4km vs. Arctic-20km

19. Nordic-4km sea ice extent vs. ice chart ice conditions on 20150909

20. available data and uncertainties

21. combined data and std

22. Combined optimal interpolation and nudging scheme (Wang et al., 2013) Optimal Interpolation: Nudging:

23. assimilation method COIN scheme –cheap computation –no abrupt shock to the model simulation –OSISAF and ice chart are daily averaged products based on multi-sensors and multi-time –observations usually has similar or even higher spatial resolution than model –weak stability, homogeneity, isotropy of the error covariance –no spatial variation of nudging coefficient make a 5-year simulation of ROMS Arctic-20km interpolate the Arctic-20km results to Nordic-4km domain for boundary and initial fields

24. assimilated sea ice extent and mean std

25. correct rate of the simulation

26. Nordic-4km 1-day forecast skill Ice edge position ice concentration ice extent

27. Nordic-4km vs TOPAZ

30. Thanks for your attention !!!