1. Stable Boundary Layers over Land Bert Holtslag Wageningen University, NL Towards a better representation of the Atmospheric Boundary Layer in Weather and Climate models Presentation mostly based on achievements within GEWEX Atmospheric Boundary Layer Study (GABLS)

2. Climate: Many factors

3. Why is the stable boundary layer (SBL) important? Surface temperature forecasting at night Fog forecasting Polar climate Land Climate (night and in winter) Dispersion studies Built up of high CO2 (and other scalar) concentrations at night over land...

4. Example: Mean model bias for the 2 meter temperature in present winter climate (30 years) Courtesy, Geert Lenderink, KNMI Also impact on diurnal cycle

5. Mean model difference in 2 meter temperature for January 1996 using two different stabilty functions in ECMWF model (Courtesy A. Beljaars)

6. Stable boundary layer mixing Diffusion coefficients by updated ‘Monin- Obukhov (MO)’ versus alternatives (LTG) FmFm FhFh LTG Revised LTGMO Revised LTG LTG MO MO based on Cabauw data (Beljaars and Holtslag, 1991) LTG ‘s used in ECMWF model (Louis et al; Beljaars et al)

7. Sensitivity to SBL parameterization in Hadley Centre Climate Model, over Antarctic Difference between new (2nd order closure ) and current scheme (1st order closure ) for 1.5m Temperature (K), JJA season, 5 year mean (King et al. 2001, QJRMS)

8. State of the Art Large Bias and Sensitivity to Stable ABL formulation (at least over Land and Ice!) Operational models typically like enhanced mixing in stable cases What can we learn from fine-scale modeling (LES)? How do operational models compare?

9. GABLS first inter comparison case Simple shear driven case (after Kosovic and Curry, 2000) Prescribed surface cooling 0.25 K/h (over ice) for 9 hours to quasi- equilibrium; no surface and radiation scheme Geostrophic wind 8 m/s, latitude 73N

10. An intercomparison of large-eddy simulations of the stable boundary layer Coordinated by Bob Beare, Malcolm MacVean, Anne McCabe Met Office, UK See: http://www. gabls.org for details Domain 400m x 400m x 400m Resolutions: 12.5m, 6.25m, 3.125m, 2m, 1m 10 results sets, 17 investigators Beare et al, 2005 (Accepted for BLM)

11. Mean potential temperature of LES models after nine hours of cooling Significant spread in results, but convergence at high resolution (Sensitivity to sub-grid model) Beare et al, 2005

12. Mean heat fluxes cf linear heat flux profile derived by Nieuwstadt (1984).

13. Mean wind

14. Mean stress

15. Normalized fluxes Crosses are based on Cabauw observations (Nieuwstadt 1984), with the standard deviation of the means shown by the shaded regions.

16. Eddy-Diffusivities In LES: Pr = Km / Kh < 1, but observations indicate Pr ~ 1 or > 1

17. Summary LES results Significant spread in results, but convergence at high resolution Sensitivity to sub-grid model Overall agreement with observations is fair! Effective stability functions in agreement with observations and sharper than those typically used in Operational Models!

18. Intercomparison of Single-Column Models Coordinated by Joan Cuxart i Rodamilans, Maria Jiménez, Laura Conangla Universitat de les Illes Balears (Mallorca, SP) Results of more than 20 models (Operational and Research): Various SBL parameterizations and resolutions Next slides focus on operational models in comparison with LES Cuxart et al, 2005 (BLM, accepted) http://turbulencia.uib.es/gabls/

19. GABLS 1D Intercomparison Resolution (most) operational models is set to 6.25 m! ‘Operational’ Single-Column Models versus LES (Cuxart et al, 2005)

20. Heat Fluxes Momentum Fluxes

21. Apparent Diffusivities for Momentum and Heat

22. ’Ekman spirals’ (Svensson+Holtslag, 2005) (u ms -1 ) LES Operational Research- Meso Research 

23. Boundary-layer height and surface angle LES consistent with Cabauw data!

24. Steeneveld, Van de Wiel, Holtslag, 2005 (BLM, accepted) Models can represent main LES results after adjusting, e.g. length scale:

25. Why do models need Enhanced Mixing? To compensate for model errors and to prevent ‘runaway’ surface cooling To have sufficient ‘Ekman pumping’ Future work: Relate mixing to heterogeneity land surface… Study impact of soil characteristics and resolution

26. Alternative runs with coupling to surface Steeneveld, Van de Wiel, Holtslag, 2005 (BLM, accepted)

27. GABLS referenceAlternative with coupling to surface Steeneveld, Van de Wiel, Holtslag, 2005 (BLM, accepted)

28. Impact of resolution in atmosphere and ‘soil’ 

29. Summary Large variation among 1D models, but all operational models show too strong mixing! Apparently the turbulence schemes are used to increase operational model performance but this decreases representation of ABL! Details of Atmosphere-land interaction do matter! 8 papers are currently accepted for special GABLS issue in Boundary Layer Meteorology, to appear in 2005/6

30. What next in GABLS? Intercomparison for a couple of day and night cycles in CASES99 (based on Steeneveld, Van der Wiel, Holtslag, 2005, JAS cond.accept.) Preparation of case by G. Svensson et al for 1D models and T.Lund for LES focussing on PBL (prescribed surface temperature or flux): Case released in May 2005! See www.met.wau.nl/gabls.htmlwww.met.wau.nl/gabls.html

31. Coupling of new GABLS case study to Land Surface within LOCO (Local Coupled action of GEWEX Land Study) Results to be discussed at GABLS/LOCO workshop at KNMI, NL, 19-22 september, 2005 Thanks to all the participating scientists within GABLS (about 50) and many colleagues for providing inputs, model runs, advice, et cetera

32. Modeling the evolution of three contrasting nights in CASES99 Gert-Jan Steeneveld, Bas van de Wiel, Bert Holtslag

33. Example from MM5: Predicted vegetation temperature Large scale model shows small predictive skill in the SBL The Problem Can we improve this?

34. GABLS case study with coupling with the surface Surface temperature and sensible heat flux H, are interdependent Detailed observations show relatively large heat flux from the soil compared to H and net radiation

35. Int. turb Fully turb Non- turb Results with 1D model Duynkerke (1991) Extended with heat diffusion in the soil Steeneveld et al. (2005)

36. Int. turb Fully turb Non- turb

37. Int. turb Fully turb Non- turb

38. Int. turb Fully turb Non- turb

39. Wind and temperature profiles

40. Conclusion A series of three diurnal cycles for which the nights differ strongly in their dynamics is available from CASES-99 Column-Model with high resolution and prescribed dynamics is performing well in comparison with the observations!

41. Momentum stability functions LES is in equilibrium, flat terrain, ‘Sharp tail’ corresponds to observations!

42. LES 1D Striking differences overall Even for this simple case!

44. GABLS 1D Intercomparison Test with one model (UIB-UPC) changing the mixing length