1 Large Eddy Simulation of Stable Boundary Layers with a prognostic subgrid TKE equation 8 th Annual Meeting of the EMS, Amsterdam, 2008 Stephan R. de Roode and Vincent Perrin Clouds, Climate and Air Quality, Dept. of Applied Sciences, Delft University of Technology, Delft, Netherlands
2 Contents Problem/question - Dutch LES model: Stable boundary layer simulation dominated by subgrid contributions Strategy - Analysis of subgrid prognostic TKE model LES results - subgrid vs resolved - similarity relations - high resolution results Conclusions 8 th Annual Meeting of the EMS, Amsterdam, 2008
3 Prognostic subgrid TKE equation (Deardorff 1980) subgrid fluxes, eddy diffusivity length scale subgrid TKE 8 th Annual Meeting of the EMS, Amsterdam, 2008
4 GABLS SBL intercomparison case Neutral layer becomes stable due to a prescribed surface cooling (-0.25 K/h) Original set up according to Beare et al. (2003): x= y= z=6.25 m Length scale correction turned off: = =( x y z) 1/3 c h =c m (c h,1 +c h,2, ) = c m (c h,1 +c h,2 ) c m =0.12, c h,1 =1, c h,2 =2 8 th Annual Meeting of the EMS, Amsterdam, 2008
LES results: Examples taken from the 5th hour Turbulent fluxes dominated by subgrid contribution
Solution close to Smagorinsky model's solution Smagorinsky subgrid TKE solution: LES subgrid constants: c f =2.5 c m =0.12, c =0.76 corresponding Smagorinsky constant: c s =0.22
Changing the filter constant c f =2.5 2 Less filtering more resolved motions
8 Subgrid constants c m and c h c m more mixing of hor. winds c h more mixing of pot. temp.
Similarity relations Solution if solution is 100% subgrid (Baas et al., 2008)
Similarity relations: c f =2 (c m =0.096) DNS Van der Wiel et al. (2008)
High resolution: x= y= z=1.5626m
12 Conclusions 1. =6.25 m resolution not enough - Solution dictated by Smagorinsky subgrid TKE solution - too much dependency on subgrid constants: bad simulation - recommendation: refine grid resolution (smaller ) 2. High resolution simulation - smaller gradient for m and h compared to observations and DNS