KILOMETER-SCALE SIMULATIONS OF ALPINE SUMMERTIME
Transcription
KILOMETER-SCALE SIMULATIONS OF ALPINE SUMMERTIME
K ILOMETER - SCALE C2SM Center for Climate Systems Modeling SIMULATIONS OF CONVECTION A LPINE SUMMERTIME Wolfgang Langhans, Jürg Schmidli, and Christoph Schär Institute for Atmospheric and Climate Science, ETH Zurich Contact: wolfgang.langhans@env.ethz.ch Introduction and Motivation Mean diurnal cycle of the heat budget Convection-permitting simulations using grid-resolutions of ∼ 1 km outperform coarse-grid simulations using parameterizations for convection (e.g., Hohenegger et al. 2008). However, horizontal resolutions in the range from ∼10 km to ∼100 m are often referred to as “Terra Incognita” (Wyngaard 2004), since parameterizations are designed for larger scales and small-scale turbulent processes still remain unresolved. Previous studies highlighted the importance of sufficiently small grid-scacings in order to resolve moist convective features (Bryan et al. 2003; Craig and Dörnbrack 2008). In this study kilometer-scale simulations are performed for a 9-day period in July 2006, which was characterized by a quasi-stationary synoptic high-pressure situation (see Fig. 1) and vigorous moist convection over the Alps during daytime. We analyze the convergence of simulations with horizontal grid-spacings reaching from ∆x = 4.4 km to ∆x = 1.1 km by studying bulk vertical energy and moisture transport over the Alps. ∂θ θ Lv l,f θ 1 θ ~ = −~v · ∇θ + S − ∇ · H + Qr ∂t T cpd T ρcpd T TOT ADV MICPHY TURB RAD TOT ADV MICPHY Main goals • How do single numerical components influence predicted convective precipitation? • How is mean vertical heat and moisture transport related to horizontal grid-spacing? TURB CCLM/COSMO 4.3 setup • Split-explicit 3rd-order Runge-Kutta scheme • Prognostic TKE-based scheme (closure on level 2.5) for sub-grid vertical turbulence • Topographic corrections of radiative fluxes (Buzzi 2008) • TERRA multi-layer soil model • 1-moment graupel scheme • NO CUMULUS PARAMETERIZATION • IC/LBC provided by ECMWF analysis (except soil moisture/temperature) • Pressure-based hybrid vertical coordinate (45 levels) • Smooth 4.4 km topography for all simulations (see Fig. 1) • Simulation period: 11 July - 20 July 2006 (see Fig. 1) Figure 3: Mean diurnal cycle of the bulk heat budget. Potential temperature tendencies (K s−1 ) have been spatially averaged over an Alpine subdomain (see Fig. 1): (1. column) Potential temperature tendencies from ∆x = 4.4 run, (2. column) Difference between ∆x = 4.4 run using a smaller timestep and ∆x = 4.4 run, (3. column) Difference between ∆x = 2.2 run and ∆x = 4.4 run, (4. column) Difference between ∆x = 1.1 run and ∆x = 4.4 run. Bluish colors in columns 2 to 4 indicate a decrease of heat tendencies (cooling) compared to the reference ∆x = 4.4 run. Mean vertical fluxes of heat and moisture at 4 km above sea level Figure 4: Simulated mean diurnal cycle of vertical (top) heat and (bottom) moisture transport (W m−2 ) over the Alps through a horizontal plane in 4 km above mean sea level: (Left) Mean fluxes, (middle) grid-scale fluxes, and (right) sub-grid scale fluxes for several simulations. Figure 1: (Left) Mean geopotential height (gpdm, contour lines) and standard deviation (gpdm, color shaded) at 500 hPa calculated from 6-hourly ECMWF analysis between 06 UTC 11 July 2006 and 06 UTC 20 July 2006. The black box indicates the position of the model domain. (Right) Topography in the model domain, the black box indicates the subdomain used for computation of spatial averages. Vertical velocity at 17 UTC 17 July 2006 Diurnal cycle of precipitation First simulations revealed a strong reduction (up to ∼ 35 %) of convective precipitation with increasing quasi-horizontal diffusion of temperature (T’) and/or horizontal wind components U,V. Furthermore, decreasing the timestep showed a decrease of precipitation, if non-dimensional diffusion coefficients α were kept constant. Therefore, we choose a setup of minimal diffusion for our simulations and adapt the non-dimensional hyperviscosities α to modifications of the timestep, such that diffusion per time is constant. Figure 5: Vertical velocity (m s−1 ) on model level 25 (∼ 2 km above ground) at 17 UTC 17 July 2006 for simulations with different horizontal grid-spacing. The subsection is located in the southwestern part of the model domain: (Left) ∆x = 4.4 km, (middle) ∆x = 2.2 km, and (right) ∆x = 1.1 km. The lengths of the black lines in the southwestern parts of the figures correspond to 10∆x. Conclusions • The strength of explicit horizontal diffusion strongly influences precipitating convective updrafts • Astonishingly, decreasing the time step highly dampens convection • The increase of the model’s horizontal resolution (and time resolution) results in: ? Reduction of mean convective precipitation over the Alps ? Less total heating between 12 and 18 UTC, primarily due to reduced latent heat release ? Smaller vertical heat and moisture exchange between planetary boundary layer and free troposphere Outlook Figure 2: Mean diurnal cycle of precipitation rates (mm−1 h) during the simulation period and averaged over an Alpine subdomain (see Fig. 1). References • • • • Investigate mechanisms responsible for decreasing upward directed fluxes at higher resolutions (smaller timesteps) Investigate numerical simulation at 0.55 km grid-spacing Repeat study with adapted resolution of topography Impact of thermally driven mountain circulations on the diurnal cycle of moist convection Bryan, G.H., J. Wyngaard, and J. Fritsch, 2003: Resolution requirements for simulations of deep moist convection. Mon. Wea. Rev., 131, 2394–2416. Buzzi, M., 2008: Challenges in operational numerical weather prediction at high resolution in complex terrain. Ph. D. thesis, ETH Zurich. Nr. 17714. Craig, G.C., and A. Dörnbrack, 2008: Entrainment in cumulus clouds: What resolution is cloudresolving? J. Atmos. Sci., 65, 3978–3988. Hohenegger, C., P. Brockhaus, and C. Schär, 2008: Towards climate simulations at cloud-resolving scales. Meteor. Z., 17, 383–394. Wyngaard, J., 2004: Toward numerical modeling in the "Terra Incognita". J. Atmos. Sci., 61, 1816– 1826. Acknowledgements We would like to thank MeteoSwiss (especially Oliver Fuhrer and Tanja Weusthoff) for their support during this ongoing PhD project. The Swiss National Supercomputing Centre is acknowledged for providing access to its high-performance compute clusters.