1. MSG cloud mask initialisation in hydrostatic and non-hydrostatic NWP models Sibbo van der Veen KNMI De Bilt, The Netherlands EMS conference, 10 -14 September 2012, Lodz, Poland

2. Changing initial clouds in Hirlam / Harmonie: Use:  cloud mask nowcasting SAF  MSG cloud top temperatures  Synoptic cloud base heights

3. Relation between cloud amount and specific humidity: Preserve buoyancy when changing humidity (keep virtual T constant) especially important in Harmonie! Correction: N: 3-D cloud cover (N>0)(N=0)

4. Apply to: 1) Hirlam (hydrostatic) * 4 seasons verification See Van der Veen, S. H., 2012: Improving NWP model cloud forecasts using Meteosat Second Generation imagery. Accepted for publication in Mon. Wea. Rev. * RUC (in cooperation with Siebren de Haan, KNMI) 2) Harmonie (non-hydrostatic)

5. Spin up / spin down rainfall / cloud cover Limit initial humidity changes to 10% (optimisation of precipitation forecasts) results based on ~ 140 runs

6. Example: cloud creation and cloud removal

7. 1. Hirlam Rapid Update Cycle Semi-operational (version 7.3), 11 km 3d VAR Observations: Mode S, AMDAR, synop, GPS Analysis every hour Forecast length: 6 h Period: 5 May – 23 December 2011 Control run and ‘MSG’ run (Kain-Fritsch)

15. Observations used for verification: AMDAR: upper air temperatures Synop: cloud amounts Synop: pressure Synop: 2-m temperature

16. Verification of forecast cloud amounts:

17. Standard deviation / bias cloudiness averaged over whole period

18. Verification of precipitation forecasts

19. Verification of upper air temperatures:

20. Verification of 2-m temperatures: Standard deviation: slightly better Bias: worse! (radiation module?)

21. Verification of forecast surface pressure:

22. 2a) Convection in Harmonie Case study: 10 July 2010 Analysis times: 12 / 18 UTC Severe thunderstorms over the Netherlands

23. CNTR 18+(3-2) MSG

24. CNTR 12+(9-8) MSG

25. 2b) Fog in Harmonie Two experiments: with and without cloud mask / synop initialisation Cycling frequency: once every 6 hours Period: 15 – 30 March 2012 Area: The Netherlands / North Sea Example showing impact: 22 March, 06 UTC + 24h forecast

26. CNTR +1 MSG

27. CNTR +2 MSG

28. CNTR +3 MSG

29. CNTR +4 MSG

30. CNTR +5 MSG

31. CNTR +6 MSG

32. CNTR +7 MSG

33. CNTR +8 MSG

34. CNTR +9 MSG

35. CNTR +10 MSG

36. CNTR +11 MSG

37. CNTR +12 MSG

38. CNTR +13 MSG

39. CNTR +14 MSG

40. CNTR +15 MSG

41. CNTR +16 MSG

42. CNTR +17 MSG

43. CNTR +18 MSG

44. CNTR +19 MSG

45. CNTR +20 MSG

46. CNTR +21 MSG

47. CNTR +22 MSG

48. CNTR +23 MSG

49. CNTR +24 MSG

50. Summary (1): > Initialisation procedure Hirlam / Harmonie: * change spec. humidity after analysis (MSG, synoptic cloud base) > Verification results for Hirlam: * better cloud cover, precipitation, upper air temperatures, surface pressure for forecasts * worse bias of 2-m temperatures > Conclusion: * (variational) data assimilation not necessary for humidity in Hirlam…

51. > Convection in Harmonie: Initial MSG cloud mask in Harmonie increased thunderstorm precipitation considerably in +3h and +9h forecasts > Fog in Harmonie: MSG cloud mask can remove erroneous fog fields in forecasts (up to +24 h) > ‘buoyancy correction’ in Harmonie: significant impact (more rain if not applied) Summary (2):

52. Outlook: -Improve initial cloud base over ocean -MSG in Harmonie semi-operational -More verification of Harmonie