|
Titel |
Mesoscale impacts of explicit numerical diffusion in a convection-permitting model |
VerfasserIn |
Wolfgang Langhans, Jürg Schmidli, Christoph Schär |
Konferenz |
EGU General Assembly 2010
|
Medientyp |
Artikel
|
Sprache |
Englisch
|
Digitales Dokument |
PDF |
Erschienen |
In: GRA - Volume 12 (2010) |
Datensatznummer |
250038810
|
|
|
|
Zusammenfassung |
Previous studies have highlighted the effects of both numerical and sub-grid turbulent
filtering in convection-permitting simulations (Îx - O(1Â km)), since both significantly
modify cloud dynamics by enhancing the dissipation within moist convective plumes. In this
study we are interested in impacts of the model’s numerics and physics upon intermediate
scales, at which energy is not primarily attenuated by the explicit numerical filter, e.g.,
the scale of the European Alps. Furthermore, we study the influence of numerical
small-scale filtering of specific prognostic variables on the up-scaling of energy to larger
scales.
Numerical simulations of a period characterized by summertime convection over Alpine
terrain are performed using the COSMO-CLM mesoscale model at a grid-spacing of 2.2Â km.
The feedback to larger scales is explored using spectral analysis and by computing bulk
Alpine heat and moisture budgets. The latter are evaluated by a recently implemented budget
diagnosis tool, which extracts all physical and dynamical contributions to the net temperature
and moisture scalar tendencies.
Both the peak of the mean diurnal cycle and the total amount of precipitation in a large
Alpine region is reduced by up to 37 % in the case of strong numerical dissipation
applied to the prognostic variables. Besides a direct impact on cloud structures, the
spectral analysis and the heat budgets reveal a substantially reduced feedback of
small-scale energy to the mesoscale. Less energy is contained at the Alpine scale and the
bulk Alpine net heat budget is modified such that less heating occurs within the
PBL during daytime and the upper tropospheric heating is reduced by ~ 30Â %
in the late evening. The former is mainly a consequence of reduced vertical heat
advection within the PBL, while the latter is primarily a result of minimized latent heat
release.
In order to determine the origin of the amplifying grid-scale energy, which is either of
physical origin or the result of computational noise, we compare different moisture advection
schemes. On the one hand a Semi-Lagrangian (SL) scheme, which is characterized by
spurious small-scale oscillations, and on the other hand different monotonic schemes. In
contrast to the monotonic schemes, energy at the shortest wavelengths is increased with the
SL scheme and the up-scaling is therefore significantly influenced by explicit filtering of
moisture scalars.
The results agree surprisingly well with the linear theory of convective growth,
which is also used to better understand the initial period of convective growth of the
numerically generated perturbations. Especially the filtering of horizontal momentum and
buoyancy determines the characteristic amplification times of small-scale instabilities. |
|
|
|
|
|