Low (web) Quality - BALTEX
Low (web) Quality - BALTEX
Low (web) Quality - BALTEX
Create successful ePaper yourself
Turn your PDF publications into a flip-book with our unique Google optimized e-Paper software.
116<br />
Overview of recent developments of COSMO-CLM numerics and physical<br />
parameterizations for high resolution simulations<br />
Andreas Will (1) and Ulrich Schättler (2)<br />
(1) Chair of Environmental Meteorology, Brandenburg University of Technology (BTU), P.O. Box 10 13 44, D-03013<br />
Cottbus, Germany, e-mail: will@tu-cottbus.de<br />
(2) Research and Development (FE13), Deutscher Wetterdienst, Offenbach, Germany<br />
1. Motivation<br />
Several RCMs are based on NWP models. They use the<br />
NWP dynamical core and extend the model physics relevant<br />
on long time scales. This is also the case with COSMO-<br />
CLM.<br />
One basic idea of the COSMO-CLM development is that a<br />
model improvement should reduce the inaccuracy of<br />
operational numerical weather prediction and of regional<br />
climate simulations with perfect boundary conditions.<br />
From version 4 on the COSMO-CLM is a unified model for<br />
NWP and RCM designed for high resolution applications of<br />
up to 100m horizontal resolution. In NWP mode the model<br />
is developed and tested at the 2 to 7 km space and the synop<br />
tic time scale. The corresponding RCM mode is used and<br />
evaluated at space scales between 2 and 50 km and<br />
climatological time scales. Both "views"<br />
- the time development of single weather situations (case<br />
studies) investigated in NWP mode and<br />
- the statistics of climate investigated in RCM mode<br />
are regarded as complementary pictures and contribute to<br />
the further improvement of the model system.<br />
The main issue of the presentation is to give an overview of<br />
the COSMO-CLM model system, to present the recent<br />
developments of numerics and of physical parameterizations<br />
and to discuss their relevance for regional climate<br />
simulations.<br />
2. Developments of Model Physics and Numerics<br />
for High Resolution Simulations<br />
The basic non-hydrostatatic model version LM_3 / CLM_3<br />
of the COSMO-CLM uses the leapfrog-dynamical method<br />
as proposed by Klemp-Wilhelmson, moist physics for rain<br />
and snow, 1d radiation scheme of Ritter-Geleyn and a multilayer<br />
deep soil model (see Steppeler et al. (2003) for<br />
details). This model dynamics has the advantage of<br />
efficiency and exhibits an acceptable local accuracy in<br />
NWP (LM) and regional climate mode (CLM) (see Hollweg<br />
et al (2008)).<br />
The model physics of the LM_3 was designed for<br />
resolutions between 10 and 50km and had to be further<br />
developed for applications at much higher resolutions. The<br />
strong relation between the physical parameterizations of<br />
unresolved processes and the simulation of grid scale<br />
phenomena led to the development of a new dyanamical<br />
core in order to achieve higher local accuracy.<br />
schemes for passive tracer fields (water constituents) have<br />
been introduced and the so called "shallow atmosphere<br />
approximation" was removed. Now all advection compo<br />
nents and all coriolis terms are implemented.<br />
At meso-scale resolving simulations also more and more<br />
physical processes are explicitely resolved and/or become<br />
important. In COSMO-CLM the horizontal transport,<br />
evaporation of falling water constituents and a further type<br />
of precipitation (graupel) have been introduced, the cloud<br />
microphysics and the convection scheme have been<br />
modified and a shallow convection parameterization has<br />
been introduced.<br />
The improved numerics (RK dynamics) and higher spatial<br />
resolutions modified the unresolved dynamics, which has<br />
to be parameterized. In COSMO-CLM an explicit parame<br />
terization of the effect of the unresolved orography on the<br />
dynamics has been introduced (Lott and Miller (1997))<br />
and a prognostic turbulent kinetic energy equation is<br />
solved.<br />
At higher resolutions and for climate applications the<br />
heterogenity of the land surface becomes more important.<br />
A lake model has been introduced to calculate the surface<br />
temperature of the grid scale lakes and the albedo of snow<br />
covered evergreen forests has been introduced. The last<br />
improves the model behavior especially in late winter.<br />
The developments have not been evaluated independently<br />
on climatological time scales and/or for different regions<br />
on the globe. Therefore, results will be presented<br />
illustrating the influence of each of the developments for<br />
idealized test cases (numerics), extreme weather<br />
conditions and/or on seasonal to inter-annual time scales.<br />
References<br />
Hollweg, H.-D., U. Boehm, I. Fast, B. Hennemuth, K.<br />
Keuler, E. Keup-Thiel, M. Lautenschlager, S. Legutke,<br />
K.Radtke, B. Rockel, M. Schubert, A. Will, M. Woldt,<br />
C. Wunram, Ensemble simulations over Europe with<br />
the regional climate model CLM forced with IPCC<br />
AR4 global scenarios. M&D Technical Report, No. 3,<br />
145 pp. Max-Planck-Institute for Meteorology, 2008,<br />
www.mad.zmaw.de/projects-at-md/sg-adaptation/<br />
J.Steppeler, G.Doms, H.-W.Bitzer, A.Gassmann,<br />
U.Damrath and G.Gregoric, Meso-gamma scale<br />
forecasts using the nonhydrostatic model LM,<br />
Meteorology and Atmospheric Physics (2003), 82, 75-<br />
96<br />
A new dynamical core based on a non-dissipative 3 rd order<br />
Runge-Kutta time integration method has been introduced<br />
together with 3 rd and 5 th order upwind advection schemes in<br />
order to reduce the phase error. Furthermore, different