Geoscience Reference
In-Depth Information
Precipitation Processes
The formation of precipitation from cloud condensate,
q
c
, is parameterized accord-
ing to
Sundqvist et al.
(
1989
), but the Bergeron-Findeisen mechanism and collection
processes are currently disregarded. Precipitation formed from cloud liquid water
at temperatures below the freezing point is assumed to freeze instantly, which
corresponds to term
F
rain
in (
11.33
). On the other hand, precipitation evaporation
is estimated from the overlap of precipitation with the uniformly distributed subgrid
fluctuations of humidity inside the clear-sky fraction of the grid box.
Regularization
Perturbations of
C
strat
were found to cause spurious instabilities in TL and AD
integrations and are therefore artificially reduced according to the value of
C
strat
in the trajectory. A reduction of perturbations in the autoconversion of cloud
condensate to precipitation is also needed.
11.5.2
A Few Remarks
The set of physical parametrization schemes developed for the ECMWF linearized
model was described in Sect.
11.5.1
. Although there are some simplifications and
regularizations applied in the different parametrization schemes, the whole package
is comprehensive and its non-linear form is able to provide up to 3 days forecasts
that show a degree of realism which does not depart too much from that of the non-
linear physics. Different levels of simplification of the schemes have been driven
either by the requirement to decrease computational cost for operational applications
or the necessity to avoid unrealistic perturbations in the linearized version of the
scheme. The applied regularizations and simplifications allow global integrations of
the linearized model with elaborated physical parametrization schemes even up to
48 h without producing spurious noise.
Overall, the presented package is a result of compromise between realism,
linearity and computational cost while at the same time the level of complexity
for the parametrization schemes is also influenced by the required applications. It is
a constant challenge to maintain the best tradeoff between all those requirements.
11.5.3
Benefits of Regularization
The validity of the tangent-linear approximation can be highly degraded due to the
non-linear and discontinuous nature of physical processes (see Sect.
11.3.2
). If the
Search WWH ::
Custom Search