|
Titel |
The representation of convection in high resolution (km-scale) NWP models. |
VerfasserIn |
Carol Halliwell, Humphrey Lean, Kirsty Hanley, Emilie Carter |
Konferenz |
EGU General Assembly 2013
|
Medientyp |
Artikel
|
Sprache |
Englisch
|
Digitales Dokument |
PDF |
Erschienen |
In: GRA - Volume 15 (2013) |
Datensatznummer |
250076639
|
|
|
|
Zusammenfassung |
As the resolution of NWP models increases, the representation of physical
processes in the model may change. The aim of this work is to improve the
representation of convection in the Met Office Unified Model (UM) in high
resolution (km gridlengths and below) models and at these resolutions
explicit representation of convection may replace parametrisation.
However, the exact resolution at which this transition takes place is
not clear cut leading to the problem of the so called "grey-zone".
The Met Office highest resolution operational model uses a gridlength of
1.5km and subgrid mixing is represented by a boundary layer scheme and a
Smagorinsky-type turbulence scheme, not a convection scheme. At this
resolution, convection is still under-resolved and convective cells are
often seen to be too large, too far apart and overall there is too much
heavy rain and a lack of light rain.
Improvements have been seen in the precipitation fields by increasing the
vertical resolution and results can also be sensitive to the details of the
subgrid mixing scheme. This work aims to determine the optimum configuration
of the UM at various high resolutions, in particular with a gridlength of 1.5km.
Computer power puts constraints on the resolution, both horizontal and vertical,
used in operational models but a study of convection in models with horizontal
gridlengths as small as 100m may inform decisions on the improvement of
convection in km-scale models as well as providing an understanding of the
trends of model behaviour with gridlength and an assessment of the capability
of future high resolution forecast systems.
Investigations into how well convection is represented in models with gridlengths
of a kilometre and below are being carried out using a suite of nested UM
models centred over the Chilbolton Advanced Meteorological Radar in southern
England. The models have gridlenths of 4km, 2.2km, 1.5km, 500m, 200m and 100m
with each model downscaling the next larger one. A statistical assessment has
been carried out, using data gathered on 40 days during the DYMECS (Dynamical
and Microphysical Evolution of Convective Storms) project, a collaborative
project between The University of Reading and the Met Office. Comparisons are
also made with NIMROD radar data. Diagnostics including trends in size, number
and strength of convective cells, mass fluxes and spectra of vertical velocity
as the resolution increases are studied to understands trends in the behaviour
of the models at different resolutions. Evidence of convergence of results
with increasing horizontal resolution and the optimum configuration will also
be discussed. |
|
|
|
|
|