The 9th June case.
Damian Wilson
This is a convective case, with widespread showers formed from weak convection.
The radar returns clearly have a melting layer, which suggests graupel
formation and strong updraughts do not exist in this case.
Observational Data
Radar data,
from the Chilbolton radar, for this event can be found via the
University of Reading
Radar Group website.
Aircraft data should follow soon.
Model simulation
- The model appears to struggle somewhat more in this case. This may
be more due to a problem in comparing model and observations in a
convective situation, where a model cross section may miss a
convective cell altogether. Also, we have been trying to compare the
large-scale schemes, whereas a lot of the condensate will be diagnosed
from the convection scheme. However, the model does get the convective
region in the correct place. If we make the assumption that the
convective condensate can be considered as ice above the melting layer
and has the same microphysical characteristics as the large-scale
water contents (probably quite reasonable in this case) then we can
calculate a combined reflectivity
. This gives reflectivities which are a few dB too large, at over 35
dBZ in ice cloud. This suggests there is too much condensate diagnosed
by the convection scheme in this case. This can be reduced a little if
some of the condensate is assumed to be liquid instead of ice.
- The 3C scheme appears to give better results than the 3B scheme over
the Borders region when compared to the Nimrod analysis, but we are not
sure why.
- Running with the precipitation part of the convection scheme switched
off produces siginificant degredation of the forecast in terms of the
positioning of the precipitation. The large-scale water
contents are, as expected, much increased as a result, and compare somewhat
to the values inferred from the Chilbolton radar.
- If we searching for some convective regions in the model then we can
reproduce significant evaporative regions underneath ice cloud, which were
observed by the radar. We don't get much supercooled water in the 3B scheme,
but get more in the 3C scheme and in reality, which shows supercooled water
to the top of the convective plumes.
- Overall, the model needs to get the convection correct in order for
the large-scale to be correct. There is more of a problem in comparing the
radar and the model in this case due to the convection. Also note that the
subgrid scale part of the convective microphysics scheme is vitally important
in producing a good forecast (the large-scale microphysics would be
quite reasonable in the convective regions in this case). The 3C scheme is
probably slightly better than 3B.
The model requires a good prediction of convection as well as large-scale
microphysics. It is difficult to assess the large-scale parameterizations
with convective cases: There is a representivity problem and much of the
condensate is only around within the convective regions.