Welcome, Guest
Username: Password: Remember me
Forum header

TOPIC: HIRLAM 7.4 notes

HIRLAM 7.4 notes 7 years 10 months ago #651

  • Laura Rontu
  • Laura Rontu's Avatar
  • OFFLINE
  • Administrator
  • Finnish Meteorological Institute
  • Posts: 150
  • Thank you received: 8
In HIRLAM 7.4, sloping surface radiation (Senkova et al., 2007) and
Freshwater Lake (Mironov et al., 2010) parametrisations are switched
on by default. Usage of FLake in HIRLAM is discussed in a recent
report prepared for the ESA North Hydrology
(env-ic3-vw2k8.uwaterloo.ca:8080/) project, see a copy at
hirlam.fmi.fi/tmp/HIRLAM_lake_v1.pdf . Two recent journal
papers by Eerola et al., 2010, (Bor. Env. Res) and Rontu et al., 2012
(accepted to Tellus A) are also related to the HIRLAM implementation
of FLake.

Both of these parametrisations bring new physiography fields into the
climate files (also copied to the boundary files). In addition, FLake
introduces 11 new prognostic lake variables, with initial values in
the first daily climate file. Several new diagnostic input/output lake
variables are written to the history files. Note that the orographic
radiation physiography fields are not written to the the history
files. This is done in order to save some space, as these parameters
do not change in time. Three new accumulated variables show the fields
of time-dependent coefficients used for orographic radiation
calculation. All new variables can be found in HIRLAM 7.4 output list
(hirlam.org/trac/wiki/HirlamSystemDocumen...orecast/Outputlist74).

FLake implementation in HIRLAM requires the following elements:

* FLake code and interfaces in the forecast model. Most of the code is
located in the library phys, but the lake variables travel accross
the whole code for input/output reasons (however, not into
prpo/postpp.F)

* Description of lake depth and fraction in the physiography (climate
files) and lake climatology for the initialization of the prognostic
variables in cold start. Scripts Climate and Climate_month and
subroutines in prpo (previ.F, startlake.F90, puth3c.F) control
handling of lakes in climate generation. The source data base files
are available in a directory $LAKEDIR defined in Env_input, by
default ec:/hirlam/dat/lake . For local climate generation, please
copy these files to an appropriate location. Please note that
reading of the lake cold start data requires netCDF libraries.

Evolution of the code related to these elements was described in a
blog and following to it comments at
hirlam.org/trac/blog/author/rontu . Please see there for
details of (local) climate generation including lakes, especially
hirlam.org/trac/blog/orsulanews#comment-2. Climate generation in ECMWF is recommended when possible.

To create suitable climate files, it is necessary to switch on both
parametrisations in Env_expdesc. Climate generation will take somewhat
longer time than earlier for large domains, please be patient and
allow several hours for this process. When the climate files have been
crated, they can be used with and without these parametrisations. In
other words, the extra fields in climate files will do no harm in
reduced experiments.

WARNING. Cold start (from climatological fields) of HIRLAM with FLake
is presently possible only in late summer and autumn conditions! In
the cold start, FLake takes the initial values of the lake prognostic
variables from a lake climate data base (Kourzeneva et. al., 2012). In
its first version, this data base contains significant inaccuracies at
the Northern latitudes during spring. Its usage for cold starts during
the winter - spring period is not recommended. There are two possible
ways to avoid spring-time cold start if an experiment or operational
HIRLAM needs to start this time. Either, start preliminary experiments
over needed domain from the previous autumn, before the end of October
(this has been done for the HIRLAM 7.4 RCR at FMI), or switch FLake
off now and do a new cold start with FLake from the next autumn. For
domains located inside the FMI RCR domain, interpolation of a first
guess forecast with FLake prognostic variables to the needed grid
would offer a third possibility.

In the present form the FLake parametrisation and lake surface data
assimilation live in a peaceful coexistence. This means that the
prognostic variables of FLake are not influenced by the analysis, but
provide a diagnostic lake surface temperature for the background
(first guess) of the analysis. At each time step, the atmospheric
model is coupled with FLake, which provides the lake surface
state. Thus, also the atmospheric model does not utilise the analysis
when FLake is switched on. Usage of FLake replaces all usage of
climatological lake data (Finlake or ECMWF climatological pseudo
observations and relaxation of the first guess towards an
ocean-derived climatology).

Presently, the lake temperature and ice cover analysis can be
considered an independent output of the model, possibly useful for the
users. Please note that the
analysed values can be found in the analysis (an+000) files, while the
initialised analysis (fc+000) contains the FLake first guess.
In the future, a proper assimilation of the prognostic
variables should be developed. This will require usage of satellite
and in-situ observations on lake surface state.

A possibility to use lake observations, in the format provided by the
Finnish Environment Institute (SYKE), was added to the scripts Span
and Finlake, source codes span/finlake.F, span/finlakeob.F,
(span/finlakecl.F). This possibility is used in the FMI operational
HIRLAM RCR, with the results and data made available in the RCR
archive. These observations can be used with our without FLake
parametrisation. Only in the latter case they will influence the
atmospheric forecast model. Note, however, that SYKE observations only
cover Finland, while FLake is expected to provide relatively
reasonable results over any lake. Thus, usage of FLake is recommended
in any case. FLake is also believed to provide an optimal background
for the analysis.
Time to create page: 0.072 seconds