Large-eddy simulation of an observed evening transition boundary layer Here is a nice picture of a stable bl in California. Dr Bob Beare John M Edwards and Alan Lapworth © Crown copyright 2004
Aims Establish if LES of the evening transition is possible. Nieuwstadt and Brost 86 and Sorbjan 97 look at decay of convective turbulence but not full transtion. Domain size for CBL (>2-4km in horizontal) and grid length requirements for SBL (1-10 m). Comparison with surface and tethered balloon measurements. Three types of run: full transition (CBL-SBL), nocturnal development (just SBL, smaller domain), degraded resolution (just SBL). See Beare et al 2005, to appear in QJRMS. © Crown copyright 2004
Cardington observations Reasonably flat grassy site, z0m = z0h = 5 cm near Bedford UK. Times series of 1.2 m temp., 4 m sensible heat flux and friction velocity. Tethered balloon observations for transitions 7-8 April and 23-24 September 2003 (clear skies, moderate gradient winds ~ 7m/s, moderately stable: h/L~2-4), every hour after sunset. Sept case had significant nocturnal jet and inertial oscillation, thus focus of study. See Lapworth and Mason (1988) and Lapworth (2003) for more info about site and obs method. © Crown copyright 2004
Large-eddy simulation (LES) model Full transition run: 12Z – 0Z; 2km x 2km x 3km domain; 10 m grid length. Mainly temperature boundary conditions (some flux). Smagorinsky diffusion (blue) ; Stochastic backscatter (red), Brown et al. (1994). Zero moisture; radiation code off (approximated by constant interior cooling forcing). Temperature bottom boundary condition for other runs; grid length typically 5m; 500mx500mx1km domain. LES equations: momentum, heat ,continuity dry, advection, coriolis, p-grad, buoyancy, SUB-GRID model… © Crown copyright 2004
Full transition: surface parameters Area averaged u* and SHF (hence smoother) vs surface obs. Sunset ~ 18Z Similar mag in CBL, but delayed decay. Over prediction of u*, and SHF in SBL phase (a bit like large scale models). Slightly better timing with flux bcs. © Crown copyright 2004
Full transition: flux profiles, every hour from 19Z Full transition: flux profiles, every hour from 19Z. Buoyancy and TKE fluxes. Normalise buoyancy and TKE flux after transition. Curvature of wb going from curved to linear. cf obs of Grant 97. Collapse of TKE flux; quite suddenly collapse. © Crown copyright 2004
Mean profiles. Observations thick dashed lines. Analytical initialiastion Full transition Full transition runin early eve. has too shallow a SBL and too weak ageostrophic wind… A problem (uncertainty in intial profile (from mes) or poor simulation of the collapse of the bl) So to make progress with comparing LES with obs for the SBL, we had to reinitialise with an analytic fit to obs. © Crown copyright 2004
Comparison of LES and tethered balloon observations for Cardington, Bedford; nocturnal development LES area averaged. Captures positive curvature and jet. Differences low down (decoupling?) © Crown copyright 2004
Sensitivity to initial wind profile; nocturnal development Sensitivity to initial conditions. Jet all but disappears with geostrophic initialisation. © Crown copyright 2004
Nocturnal development: sensitivity to forcings and model Sensitivity to forcings bigger than that for the model. Although double resolution pushes solution further away from obs.can’t say with conviction that this is a worse simulation as forcings are uncertain. © Crown copyright 2004
April 2003 case Very good match in theta; no low level jet in OBS. © Crown copyright 2004
Grid length: 1000m in horizontal, 100m in vertical Cardington case: sensitivity to stability function; degraded resolution runs Grid length: 1000m in horizontal, 100m in vertical SGS model providing all the fluxes; SHARP is control ; long tails like large-scale models © Crown copyright 2004
Vertical velocity cross-section and spectra Distinct peak Red positive, blue down © Crown copyright 2004
Momentum budget (Sept 2003 case, 0230Z) Momentum budget indicating significant inertial oscillation above SBL. © Crown copyright 2004
Summary LES of evening transition is possible (although robustness not fully explored due to computational burden) and qualitatively correct. Significant differences between model and obs in details: ageostrophic wind and collapse of the CBL stress (cf Mahrt 1981, for Wangara). Good comparison for nocturnal development runs, initialised with obs in early evening. Uncertainties in the geostrophic, surface and radiative forcings can make detailed comparison difficult and hard to say which model configuration is better. Nesting may be helpful for full transition at higher resolution. © Crown copyright 2004