ALATNET developments in 2000 in the ALATNET centers
1. In
Toulouse (France) during the last 6 months of 2000
The ALATNET actions in Toulouse during the first 6 months of 2000 were
reported in Newsletter 18, only the actions of the last 6 months of 2000
are presented here.
1. Theoretical aspects of non-hydrostatism (NH) & 2. Case
studies aspects of NH (2c: Solving residual instability problems in
the two-time-level semi-Lagrangian advection scheme)
- See the article of Pierre Bénard
in this Newsletter.
4. Removal of the thin layer hypothesis
- Karim Yessad and Jean-Marcel Piriou
completed the analysis of the required code modifications, focusing on
physical fluxes and diagnostics (DDH). Afterwards Karim Yessad updated
consequently the ARPEGE code for Eulerian dynamics and performed preliminary
3d tests to check that conservative properties and scores don't change
at the current scales.
6. Specific coupling problems (6a: Blending of fields in data
assimilation for preserving high resolution forecast details)
- Olivier Pédoussaut and Claude
Fischer resumed the work of Karim Bergaoui and Jean-Daniel Gril, to adapt
blending for ALADIN / France. Scripts were carefully updated and corrected,
and the first new experiments don't exhibit such puzzling features as the
previous ones.
A specific paper will be devoted to blending in the next Newsletter.
6. Specific coupling problems (6b: Tendency coupling for surface
pressure and other technical variations around Davies' technique of field
coupling in a buffer zone)
- Gabor Radnoti proposed alternatives
to the current blending technique, to avoid such problems as those encountered
during the 1999's Christmas storms. Piet Termonia undertook a detailed
investigation of the related problems, using the simplified 2d model designed
by Ilian Gospodinov.
6. Specific coupling problems (6c: Coupling problems in variational
data assimilation)
- The IFS team intends to suppress the
current 3d-var code soon. It would be replaced by a derived product of
4d-var called FGAT. However the first examination of the associated coupling
problems led to the demand of a longer delay (Claude Fischer).
8. Adaptation of physics to higher resolution (8b: Test, retuning
and improvement of the various physical parameterisation in the framework
of a very high resolution)
- Philippe Nomerange analysed the behaviour
of the so-called "lift" parameterisation in the framework of ALADIN
/ France, I.e. at a resolution
of ~ 10 km. This scheme aims at representing
the transversal forces applied by the mountains on the wind flow. It has
been used for some years in large-scale global models, where it led to
the suppression of the "envelope" orography. However several attempts to
implement it in ALADIN failed, whatever the representation of the mountain
height. This study showed that, while the impact of the "form-drag", "lift"
and "envelope" parameterisations are of similar magnitude at large scales,
this is no longer true in a meso-scale model. The impact of "lift" is far
larger than that, essentially local, of "form-drag", and is itself too
weak to allow a full suppression of the "envelope" concept.
- Dagmar Merkova tested the relative
parts of changes in the physics and increases of resolution on July 1997
(flash floods over Central Europe). She compared the initial, CYCORA and
CYCORA-bis packages, within ARPEGE, ALADIN /LACE
and ALADIN / CZ
- Eric Bazile compared the Kain-Fristsch
and the Bougeault convection schemes on two situations with an ALADIN domain
equivalent to one of the HIRLAM / Spain models (with a resolution of 15.8
km). The Spanish HIRLAM team performed parallel experiments, with another
version of the Kain-Fristsch scheme compared to the reference combination
of Kuo and Sundquist schemes.
8. Adaptation of physics to higher resolution (8c: Improved representation
of boundary layer)
- Martin Bellus, Margarida Belo Pereira
and Jean-François Geleyn improved the parameterisation of vertical
diffusion. A new formulation of the modified Richardson number was proposed,
with a dependency on height. It enables to prevent the too frequent erosions
of the inversion layer encountered last winter, while preserving the ability
of the model to forecast cyclognesis (such as the 2000's Christmas storm).
The anti-fibrillation scheme was modified consistently. A detailed study
of the sensitivity of the new exchange coefficients (dependency on height
and on Richardson number) to the namelist-tunable parameters of vertical
diffusion enabled to check that sensible values were chosen for the CYCORA-bis
package.
- Siham Sbii and Jean-Marcel Piriou worked
on the representation of low-level cloudiness, focusing on the too weak
development of stratocumulus over sea in anticyclonic areas. Tuning the
current formulation of stratiform cloudiness or introducing other parameterisations
(two new formulations were tested) didn't yield to an improvement. A diagnostic
study showed that there is enough liquid water in the corresponding areas
to produce a sensible cloudiness, provided the formulation is slightly
modified. Besides a new tuning of the present formulation of convective
cloudiness was proposed, which could lead to a better spatial distribution
of clouds.
9. Design of new physical parameterisations (9b: Use of liquid
water and ice as prognostic variables, implementation of a new microphysics
parameterisation)
- Eric Bazile has nearly finished the
design of a "functional boxes" approach for the parameterisation of liquid
water and ice, in cooperation with HIRLAM.
9. Design of new physical parameterisations (9d: Improved representation
of exchanges at sea surface)
Eric Bazile tested two modifications of the parameterisation of evaporation
over sea using parallel one-month long 4d-var assimilation experiments
:
- reduction of the residual roughness
length over sea (i.e. suppressing the change of VZ0CM in the first CYCORA
package)
- the same but adding the parameterisation
of Redelsperger, which aims at increasing evaporation by introducing a
dependency on convective activity.
The global hydrological cycle and the trade winds are better described,
though an improvement of scores is noticed only for buoy observations in
the Tropics.
9. Design of new physical parameterisations (9e: Improved
representation of land surface, including the impact of vegetation and
snow)
- Olivier Latinne implemented a new global
high-resolution (~ 1km) land-cover map produced by the MAI & MC2 teams
of CNRM (for more details contact Jean-Louis.Champeaux++at++meteo.fr or
Valery.Masson++at++meteo.fr). He developed the interface to the input
files required by configuration 923 : changes in resolution (with aggregation
of fields to a resolution of 5'), derivation of the necessary fields, formatting,
and adapted the 923-script too. These new data are to be introduced together
with the FAO soil data. The differences with the previous data where carefully
examined, field by field and month by month. An assimilation experiment
along last summer was launched to evaluate the impact of the new surface
characteristics.
- Valery Spiridonov, Andrey Bogatchev
and Eric Bazile pursued the improvement of the snow-cover parameterisation,
taking into account the differences between the albedos of vegetation and
bare ground, and the time-evolution of the albedo of snow. The first tests
are quite encouraging, but the operational implementation is likely to
be postponed to next winter.
10. Use of new observations (10a: Yet unused SYNOP observations)
- The use of new SYNOP observations was
first investigated in the framework of the optimal interpolation of surface
and boundary layer fields : snow height, visibility, hourly observations.
The corresponding work is detailed in the ALADIN Newsletter.
- The other developments related to observations
were performed in the framework of ARPEGE (for more details contact Florence.Rabier++at++meteo.fr)
11. 3D-Var analysis and variational applications
Beside the long term studies of Cornel Soci and Wafaa Sadiki, detailed
in specific articles, most of the work along the last months was devoted
to :
- searching an optimal combination of
3d-var analysis and digital filtering (Vladimir Ivanovici, Claude Fischer,
Dominique Giard and Elisabeth Gérard) : a coupling strategy for
incremental initialisation was designed and a bug corrected (management
of mean wind along incremental digital filtering);
- updating the tangent-linear and adjoint
code for non-hydrostatic dynamics (Claude Fischer).
A review of the latest developments around 3d-var assimilation is also
available in this Newsletter.
12. 4D-Var assimilation (12d: Improvement of the treatment
of humidity in data assimilation)
- This problem was indirectly addressed
by Radi Ajjaji through the study of corresponding problems in ARPEGE, and
their links with physics and scale selection. His results are presented
in a longer article.
2. In Bruxelles (Belgium)
Last (and first) report in Newsletter
18.
3. In Prague (Czech Republic)
Topic 1: Theoretical aspects of non-hydrostatism
- Top and bottom boundary conditions
(C. Smith)
The study has been started using a simulation of trapped lee waves, both
in semi-Lagrangian 3TL and Eulerian schemes. Refer, please, to the report
of Chris in this Newsletter.
- Predictor-corrector scheme (J. Vivoda)
The basic predictor-corrector scheme has been developed in the framework
of a vertical plane model, both for Eulerian and semi-Lagrangian advection
schemes. At the same time a linear stability analysis tool has been created
following the Simmons-Hoskins-Burridge (SHB) approach (in a close collaboration
with P. Bénard). It helped to reveal the following very important
points: (i) the SHB instability with respect to the departure of temperature
from its semi-implicit basic value (there is a different dependency for
the fully compressible equations than for hydrostatic primitive equations
(HPE)); (ii) a discretization of the "non-hydrostatic"" part of pressure
force which was not in accordance with its semi-implicit treatment and
hence, there was a hidden source of instability; (iii) a good step forward
to stabilize the 2TL SL scheme by finding terms which extrapolation leads
to instability (confirming experiments from 1997 partly done by M. Janousek
in Toulouse). Refer, please, to the report of Jozef for more details.
- Other discretization aspects
(R. Brozkova)
A revision of the pressure force discretization respecting the angular
momentum conservation was done both for HPE equations using the "NH-type""
of the logarithmic pressure thickness approximation (NDLNPR=1) and for
the compressible equations. It was found that the so-called "happy cancellation""
of terms in the total pressure force expression is valid as well for NDLNPR=1
setting contrary to past analysis of the problem. The examination of the
"compressible"" case led to a problematic point on how to ensure the angular
momentum conservation as well for the NH pressure departure part (due to
the different vertical staggering of the hydrostatic pressure and its NH
complement). This point is in connection with the above mentioned current
choice of an unstable" discretization, which respects the angular momentum
conservation but which is not secured by the linear S.-I. correction. A
solution how to conciliate both the stability and the full angular momentum
conservation has not been really thought over yet. Another little revision
dealt with the top boundary condition for the vertical momentum equation,
where both Dirichlet and von Neumann types of conditions for the NH pressure
departure variable were formulated and tested. Finally, Dirichlet condition
was retained as more stable for the vertical momentum equation.
- 3D academic experiments with NH model
(T. Szabo)
The experiments were made for the "SCANIA"" type of idealized atmosphere
(just orographic forcing at geostrophic balance and vertical static stability
of the flow). There is a strong vertical temperature gradient with top
model temperatures below 50K. Under such conditions there is a violent
SHB instability in the compressible model. The goal of experiments was
to obtain as many instability diagnostics as possible, in order to confirm
indirectly the SHB analysis results, to see the response of purely non-linear
terms (not covered by the linear stability analysis) and to possibly indicate
what kind of dependency the instability has on the hydrostatic pressure
semi-implicit reference state. However, the experimental environment was
too complex and the results were not very conclusive.
Topic 2. High resolution runs (D. Cemas, U. Strajnar, N. Pristov,
B. Ahrens)
In parallel to the theoretical work and academic experiments, the real-case
studies using a very high horizontal resolution started. One of the first
objectives was to prepare an experimental protocol leading to the tests
at 1km grid-size, e.g. to define ALADIN domains going progressively from
ALADIN/LACE (12km) to a domain of 1km, considering also a proportional
increase of the vertical resolution. In order to obtain a good description
of the model orography at 1km grid-size, a very fine mesh input data set
of orography is needed. Therefore a domain over Slovenia was chosen in
order to profit from both the local ee923 configuration and available data-set.
Unfortunately, the ee923 configuration did not work any more in Ljubljana
(due to the change of cycle and machine) and not yet in Prague (port of
this configuration has not yet been completed), hence all the experiments
were done for the time being on the domain "GORI"" with 2.5 km horizontal
resolution. The testing situation was one of MAP IOPs wet cases. The experiments
were done always with HPE and NH versions of ALADIN. Concerning the nesting
of domains, GORI domain simulations were coupled either directly with LACE
model or with a nested intermediate 5km grid-size model. As short and quite
preliminary conclusions, problems were noticed with the parameterization
of deep convection at 2.5 km mesh, where the convective rain pattern became
quite noisy and unrealistic. Perhaps it would have been better to fully
switch off the deep convection scheme at these scales. In addition, we
could notice worse results when GORI domain was directly coupled with LACE
domain (probably due to both spatial and temporal resolutions of the lateral
boundary conditions: the smaller the domain was, the more frequently refreshment
of the lateral boundary was needed).
Bodo Ahrens from the University of Vienna proceeded along similar lines
and did fine-mesh simulations over the Ticino-Verzaska-Maggia region with
ALADIN NH at 9.6 km and 4 km horizontal resolutions. After a short stay
in Prague to get familiar with ALADIN NH version he continued his work
in Vienna, concentrating on the quantitative precipitation forecasts. Please,
refer to his report in ALADIN Newsletter for details.
Topic 6. Special coupling issues
- Evaluation of blending (D. Klaric,
K. Stadlbacher, H. Seidl)
ALADIN/LACE forecasts starting from blended initial states were evaluated
by both objective methods and forecaster's experience. Regarding the objective
verification, standard scores were computed for both 00 and 12 UTC network
times (D. Klaric). There was a nice reduction of bias at the initial time,
especially in humidity and wind fields. Otherwise the scores were rather
neutral with a few weak improvements. On the other hand there was a yet
unexplained worse bias in surface pressure for the network time of 12 UTC,
while better bias results could be noticed for 00, 06 and 18 UTC after
a more profound investigation. The hypothesis to be still verified is that
there was a tidal wave pattern at work.
The standard scores were completed by an objective evaluation of the
spin-up measured by the physical fluxes of water vapor (balance of precipitation
and evaporation) and radiation (K. Stadlbacher, D. Klaric). While the balance
of radiation flux was rather indifferent, the water vapor flux had clearly
a reduced spin-up both for stratiform and convective precipitation. Even
the convective precipitation flux was a bit too active for 12 UTC network
time, a fact deserving as well some deeper investigations.
Finally, a forecaster's assessment of the blending" benefices was done
last October (H. Seidl) for a few recent cases of the late summer convection.
The examined forecast range was from 0 to 6 hours. The conclusion was that
the blending forecast built up more realistic (organised) structures in
some cases but not noticeably in all cases. In this exercise it was quite
difficult to find some purely objective method of evaluation: due to coarse
conventional observations a verifying analysis (a diag-pack type with a
lot of fit to the observations) provided just very smooth fields with no
real meso-scale structures in. Hence, satellite pictures were compared
subjectively to the cloudiness and potential vorticity fields of the model.
The overall result of these evaluations is that the blending has beneficial
effects on the forecast namely in reducing the spin-up. Therefore it is
considered to be employed operationally in ALADIN/LACE after some basic
cleaning of the scripts to respond to the operational standards.
- 3DVAR strategy in ALADIN (M. Siroka,
G. Boloni)
Just before Christmas first comprehensive tests of variously built 3DVAR
algorithms were run on ALADIN/LACE domain. The algorithmic tree offers
quite many possible solutions to choose a proper first guess (ALADIN forecasts,
blended initial states, ARPEGE analysis) with the Jb model (classical or
lagged forecast error statistics), yet all that possibly combined with
the DFI incremental approach to get-rid of the initialization wiping-out
of small scales" in the model forecast. There are plenty of obviously looking
but still unanswered questions, for example whether to make first the blending
and second the 3D VAR analysis or the contrary, how to combine all that
with the surface analysis, etc. All these possible ingredients would please
even the best royal alchemists at the famous court of the emperor Rudolf
II. This first attempt to assess a potential of the ALADIN 3D-VAR data
assimilation might give us an indication which algorithms were the most
prospective ones, though there were still serious limitations in this experiment:
(i) an incremental DFI approach was still missing; (ii) we used a hypothesis
that finer structure functions in ALADIN might hopefully play a positive
role even without a higher density of observations (in this experiment
the set of observations but TOVS was the same as in the global model assimilation;
the missing TOVS were hopefully not much penalising over the rather continental
domain of LACE).
The previous research results on Jb statistics are explained in details
in a RC LACE Technical Note (available on request) and summed up in an
article soon to be submitted for publication.
- Coupling of surface pressure tendency
(S. Ivatek-Sahdan, T. Szabo)
A development has been started to test a new way of coupling surface pressure.
The motivation (after Gustafsson, personal communication) is the following:
Within the spatial interpolation procedure from the coarser to the finer
mesh, a weak discordance is introduced between the fields of orography
and surface pressure. When these too fields are not well adjusted to each
other, a gravity wave noise is generated. After some time the model finds
a balanced state (within a DFI session for example) except at the relaxation/coupling
lateral belt of the domain, where the large scale" solution, containing
the above mentioned discordance due to the interpolation, is partly or
fully retained. A way to repair this weakness is to couple the tendency
of surface pressure instead of its absolute value. In this case it is hoped
that the orography-pressure discordance is not induced at every coupling
event (every time-step) to the model solution. The modified coupling scheme
has been developed and is currently under testing, first in a simplified
framework of a 2D vertical plane model.
4. In Budapest (Hungary)
INTRODUCTION
The activities around ALATNET started very effectively in 2000, which
gives a good hope for the overall execution of the scientific objectives
of the project. Regarding the topics where the Hungarian Meteorological
Service is concerned most of them started in 2000. Hereafter these topics
are described and then detailed status report will be given on the topics
which are executed in Budapest. It is mentioned that unfortunately we couldn't
manage to find a post-doc student working on the proposed 3DVAR-related
topic in Budapest. New call for candidates are being issued at the very
beginning of the year 20001.
ACTIVITIES ON ALATNET TOPICS
Based on the work plan of the ALATNET project the topics (where HMS
is concerned) where work had been started will be listed (there will not
be detailed descriptions due to the fact that those are available at the
given center's report):
Topic 1. Theoretical aspects of the non-hydrostatism
- 1.c. Improvement of the lower boundary
condition (analysis, test in 2d then 3d models) : the work started in Toulouse
and Prague respectively.
Topic 2. Case studies aspects of NH
- 2.d. Validation of the new developments
in dynamics (upper and lower boundary conditions, noise control) or coupling
: Some stability problems of the non-hydrostatic dynamics were studied
in Toulouse.
Topic 5. Coupling in high resolution modes
- The theory of the phase+amplitude coupling
was discussed and written down and some preliminary simple tests were carried
out.
Topic 6. Specific coupling problems
- 6.b. Tendency coupling for surface
pressure and other technical variations around Davies' technique of field
coupling in a buffer zone : The study (in Prague) of the surface pressure
tendency coupling started, with the theoretical clarifications and the
consideration of the practical difficulties.
Topic 9. Design of new physical parametrisations
- 9.c. New parametrisation of exchanges
at lake surface : A 2-month work was carried out in Toulouse and detailed
report can be found in ALATNET Newsletter 1.
Topic 11. 3DVAR analysis and variational applications
- 11.a. Definition and calculation of
new background error statistics, impact of domain resolution and extension:
The work started in Budapest and is detailed in the next chapter.
ALATNET ACTIVITIES IN BUDAPEST
In this chapter more details will be given in the topics being carried
out in Budapest, mainly the 3DVAR-related work will be mentioned.
At the end of May and the beginning of June a 3DVAR workshop was held
in Budapest with the participation of Claude Fischer (France) and Maria
Siroka (Slovakia). As a result of this workshop a prototype 3DVAR assimilation
suite was implemented. See the ALATNET Newsletter 1 for more details.
Now details will be given on the work devoted to the computation of
background error statistics for the ALADIN/HU model. The ALADIN/HU model
is the operational version of ALADIN exploited in Budapest, with 8km horizontal
and 31 levels vertical resolutions (the domain size is around 1600 km *
1150 km). The first objective of our work was to compute background error
statistics based on the archived data available from the operational suite
and compare the results with those obtained for different ALADIN domains
and resolution.
The main tasks to be executed for this first goal were as follows:
- Adaptation of the "festat" tool (which
is a software computing background error statistics based on the NMC method;
Parrish and Derber, 1992) originally developed by M. Monteiro (Monteiro,1998)
and modified by L. Berre (Berre, 1999) to the SGI Origin 2000 machine available
at the Hungarian Meteorological Service. It is mentioned that more other
utilities were adapted for providing valuable diagnostics on the background
errors.
- Adaptation of the visualisation softwares
developed partly by M. Monteiro, L. Berre and M. Siroka (Siroka and Bubnova,
2000) for the visualisation of all the diagnostics related to the background
error statistics.
- Execution of the festat tool (NMC method)
on the cca. 3 months archive of the ALADIN/HU model's products for computing
the background error statistics.
- Evaluation of the results and comparison
to the already obtained results for different ALADIN domains (France, LACE
- Siroka, 2000, and Maroc - Sadiki et al., 2000).
- Simple sensitivity tests using different
integration lengths and intervals for the NMC method.
- Test on the applicability of the background
error statistics for 3DVAR having single-obs experiments (see ALATNET Newsletter
1: 3DVAR workshop in Budapest).
The main conclusions of this work can be summarised hereafter:
- The festat tool and the corresponding
other utilities were adapted properly (some slight modifications were also
carried out).
- The background error statistics for
the ALADIN/HU domain were proven to be of good quality as far as the diagnostics
(dynamical relevance) and the applicability for 3DVAR analyses (single-obs
experiments) are concerned.
- Small sensitivity was found with respect
to the forecast length and time interval of the NMC method, basically the
absolute values of the different covariances changed, but not their structure.
- While comparing the results with those
available for other ALADIN domains it was found that basically they are
very similar and only just some slight differences were found. The small
differences are coming from the different domain and resolution of the
models.
For the second part of the year the work had been continued along the ideas
of Claude Fischer regarding the sensitivity of the NMC method with respect
to the applied forecast lengths and time intervals for the lagged-type
version (Siroka, 2000) of the NMC method. The main idea having lagged-type
of statistics instead of using the classical NMC method is to rule out
the effect of the large scale boundary conditions and take into account
only those scales which are really important for such mesoscale analyses
applications. The lagged-type notion means that the shorter forecasts for
the NMC method is recomputed using the same boundary conditions as it is
for the longer forecasts. The sensitivity tests consist in varying the
forecast lengths and the time interval (the difference between the two
forecast lengths). From this study it is expected that we will find the
optimal length and time intervals to be used for the computation of background
error statistics and consequently the optimal background error weights
in 3DVAR. As it was mentioned the work had been started and the necessary
new model runs are under execution. Already 3 months of data is accumulated
allowing the start of the evaluation procedure. The evaluation is to be
started at the beginning of 2001.
CONCLUSIONS
As a summary it can be said that good start can be noticed in the activities
where HMS is concerned. It is valid also for the work carried out in Budapest.
It is unfortunate that the post-doc position was not filled in Budapest
during 2000, however it is strongly hoped that the work can be extended
in 2001 with the help of a post-doc student. The plan for the last part
of 2001 is to start the work on the variational type applications (sensitivity
studies) with the help of a Ph. D. student.
REFERENCES
L. Berre, 1999: Estimation of synoptic and meso scale forecast error
covariaces in a limited area model. Accepted to Monthly Weather Review.
M. Monteiro, 1998: Balanced Statistical Structures of Background Errors
for ALADIN/LACE 3DVAR. ALADIN Internal Report. Available at Meteo France.
D. Parrish and J. Derber, 1992: The National Meteorological Center's
spectral statistical interpolation analysis system. Monthly Weather Review,
120, 1747-1763.
W. Sadiki, C. Fischer and J.-F Geleyn, 2000: Mesoscale background error
covariaces: recent results obtained with the limited-area model ALADIN.
Accepted to Monthly Weather Review.
M. Siroka and R. Bubnova 2000: Background error statistics for LACE
domain - a study. Technical Report of RC LACE, CHMI, Prague.
M. Siroka, 2000: Report on experiments with ALADIN/LACE 3DVAR. RC LACE
Internal Report, CHMI, Prague.
5. In Ljubljana (Slovenia)
The main focus of the ALATNET centre in Ljubljana in the first period
is "Case studies aspects of high resolution models with the special interest
in non-hydrostatic models".The work
started with the arrival of the PhD candidate, Mr. Klaus STADLBACHER, in
Ljubljana. The subject of his stay is: "Systematic qualitative evaluation
of high-resolution non-hydrostatic model".
After the first initial time dealing mostly with arrangement of working environment
(working place, computers, etc.) the work effort has especially been devoted
to the first field in the working plan (topic 2.a), namely "Definition
of the framework of experiments (domains, resolutions), choice of a set
of reference situations".
First of all, the decision has been taken to concentrate especially in the IOP
(Intensive Observation Periods) cases from MAP (Mesoscale Alpine Project)
due to availability of additional observational data with high spatial
and time resolution. There are also some non-conventional additional measurements
available for that period (like wind profiler data, aircraft measurements,
etc.) useful in this frame of our work. The main goal of the research in
Ljubljana is to systematically evaluate the behaviour of high-resolution
models. That is why the decision has been taken to find the cases where
already the reference coupling model (in our case ALADIN/LACE) is close
enough to reality (observed state).The
main question we have to answer is what additional quality can models at
higher resolution bring and not why the coupling model was wrong for some
weather situation. The task of determination the reference dates is going
on right now.
Due to high computer usage demands of non-hydrostatic model experiments, there
was quite a lot of work done in order to make the preparation of climate
files at high-resolution (so called configuration 923) easier. Although
this was a more technical work we expect that with fulfilling this task,
the creation of climatological files with arbitrary horizontal and vertical
resolution will be much faster. The main advantage will be a freedom in
moving the integration domain to the area of interest, where additional
data are available, so that the computational domain could be much smaller,
only covering the region of interest for the particular situation.
There were some initial efforts done also with the second item of the working
programme (topic 2.b): "Validation of the current physics and non-hydrostatic
dynamics: comparison to hydrostatic dynamics, to observations, identifying
problems".
Various experiments were performed to obtain the information about the model behaviour.
There were experiments repeated with the hydrostatic and non-hydrostatic
versions of the model.The output
for some meteorological fields were compared and some knowledge about the
typical difference between the two types (hydrostatic/non-hydrostatic)
of models was gathered. The fact that unrealistic rain bands appeared in
the precipitation fields called for some additional experiments aiming
to isolate reasons for that behaviour. The first conclusions after taking
a look at the physical parameterisations in ARPEGE/ALADIN might be that
the field of moisture convergence could be the reason of unrealistic rain
bands, at least in situations where the convection scheme is active due
to unstable layers. Some further studies will be needed in order to proof
this idea and possibly find a solution for that problem. However, we are
quite aware that in resolutions of about 5 km we may already be deep in
the twilight region of numerical modelling, where parameterised and resolved
phenomena appear at the same time and interact with a magnitude, uncontrollable
with the present modelling technique.
A study of the influence of the ratio in horizontal resolution between coupling/coupled
model is also on the way. First results show a significant sensitivity
(there seems to be more noise in the meteorological fields near the border
of integration domain when this ratio is bigger).
Over the next period or within about nine months (this is how much time Klaus
Stadlbacher will still work in Ljubljana for his first visit) we expect
to produce useful leads or perhaps even few instructive conclusions about
how and when does the non-hydrostatic dynamics in the model really improve
the model results and thus justify the increased computational costs.