PROVISIONAL ONLY ....Summary of Microscale Modelling Workshop held on
16th December 2003, 9:00 - 16:00, at the
University of Leeds .
Presentations
1. Introduction: Stephen Mobbs.
(link to presentation).
A review of the background, timetable and comments on the 1st workshop
was presented. Questions particularly relating to
particularly, grid structure and adaptation need to be addressed, in
addition to those in the list provided in the
presentation. A brief discussion of the results of the UWERN management
committee 27/10/03 discussed the need to provide interim modifications
to the BLASIUS code.
2. Brief Review: Alan Gadian.
(link to presentation).
A brief review of progress made was presented. An inter-comparison
between CLARK, Smolarkiewicz, BLASIUS and the UM demonstrated many similarities.
Future progress with microphysics in BLASIUS, and pressure solver issues
were discussed. There were detailed comments relating to use of pressure
solvers, whether the modification of the BLASIUS code was
a distraction, and the merits, or otherwise of the use of different types
of grids.
3. Building / Urban / Surface Scale modelling initiative: Ian Castro.
(link to presentation). Developments in the urban scale model
proposal
(UWERN management meeting) and discussions with cfd vendors were
described. There was much discussion as whether a Microscale model
could ever be applied on the urban scale, with buildings. It was agreed
that, as in workshop1, current Microscale models were not compatible, but
if there were developments with unstructured grids and or terrain
interstecing grids there might be some applicability. The developments
and collaborations with cfd vendors were seen as positive and parallel
developments to the Microscale model.
4. UWERN meso-scale model developments : Bob Plant
(link to presentation).
Progress in the area of meso-scale science areas were presented. The
importance of being able to model slant-wise convection was emphasised. A
workshop on 13th February was going aimed at bringing the community together
to discuss these issues. There was seen to be overlap between Microscale
and meso scale activities in UWERN and these should be pursued.
5. Issues for fine scale algorithms equations: Mike Cullen
(link to presentation).
A discussion of the important elements relating to a Microscale
model were presented. The advantages of a unstructured approach, for
steep terrain flow were discussed. Mesh structures, semi Lagrangian
and use of the GCR + pre conditioner all have advantages. The suitability
of using the UM, and the advantages of the equation structure and grids
was illuminated, suggesting this is a viable platform for a Microscale
model.
6. Different vertical grids and choices of Thermodynamic Variable
: John Thuburn
(link to presentation) and
(link to figures). A presentation of the optimal choice of grid structure
showed that the C grid structure / Charney Phillips approach had advantages
for representation of normal modes, although C grids may have better
conservation properties. The recommendation, from a purely dynamical
viewpoint, is that a derivation of the C / Charney Phillips grid is optimum.
7. Efficient semi-implicit discretizations on terrain intersecting
grids for high resolution atmospheric models : Luca Bonaventura
(link to presentation)
This talk presented the use of terrain intersecting grid structures for
a atmospheric Microscale model. Advantages of the C grid, semi Lagrangian,
approach were demonstrated with model results with rapid convergence to a solution. Development of the surface "skin" layer lower boundary conditions
are improving the near surface solution. The approach of using a terrain
intersecting Cartesian grid was demonstrated to be efficient and effective.
This method is likely to be used as an alternative dynamical core in
the LOCALE / DWD model, and was demonstrated as a dynamical structure
for a Microscale model.
8. Results from a terrain interesting local area model: David Woodhead
(link to presentation).
This presentation showed some initial 2-d and 3-d results from work
completed as a research student. Terrain intersecting, variable grids
were displayed, first in two dimensions, and then in 3 dimensions. The
equation set used was the "RAMS" formulation, with a split time
scale technique for sound waves. Potential temperature is
currently being included as a variable, and other issues include the
lower boundary condition representation with a law of the wall formulation.
9. Development of a finite element ocean model at IC with adaptive
meshing: David Marshall
(link to presentation).
This talk presented the progress made with an ocean model, using a variable
mesh domain. The need to deal with steep topography and geometrically
difficult boundary conditions largely influences the unstructured
approach required. Good computationally efficient results were obtained and
different Reynolds number solutions were produced. The application of this
unstructured Finite Element Approach is well suited for this ocean modelling.
10. Issues about a fine scale model Simon Vosper (given by Alan Gadian):
(link to presentation).
Results from the MO, anelastic, Boundary Layer, dry BLASIUS code were presented.The use of this code with Gal-Chen co-ordinates was shown to give good results,
but with slope limitations of ~ 45o. A staggered C grid formulation
proved advantageous, but the pressure solver required between ~ 40 and 68%
of the total CPU, and was seen as another limitation as a Microscale model.
CFL constraints present an issue as it uses explicit time integration. The
stretched grid system has not been widely used.
Discussion
Questions were raised during each talk, and for a prolonged session
at the end of the afternoon. Topics raised
at the end of Workshop1 were also discussed, but not
described below, where duplication occurs. Principally,
"Equations Sets"; "What is the model for";
"Use of the UM as a Microscale model" were
major repeated discussion items.
It was questioned whether
allocating effort into the BLASIUS code for pressure solver and
microphysics improvements was worthwhile. I was noted, however, that these
modules would be required in a Microscale code.
Specific discussion areas in the workshop
Equation set
At workshop 1, an equation set consistent with the incompressible equation
set was defined as being most suitable, as well as being
consistent with
the UM equation set. There was further discussion, but although
comments were expressed that at this scale, other sets could equally be
used, there were no other comments. It was questioned whether
allocating effort into the BLASIUS code for pressure solver and
microphysics improvements were worthwhile. I was
Building Scale developments.
These developments (talk 3) were well received. Although they were
seen as parallel and complementary programmes, it was also felt
that the Microscale model could apply on the building scale, if a suitable
grid system (non-terrain following) were implemented. It was hoped that
the Microscale and the building scale developments could supplement each other.
Meso-scale developments.
After the talk on meso-scale developments, with a core interest in slant-wise
convection, it was again felt that links could be maintained, but there
was some discussion as to whether the scales were larger than the
original scope for the Microscale framework.
UM development and use for a Microscale model.
This was discussed at some length, particularly at the end of
fifth and sixth talks. Concerns were expressed about turbulence closure
issues, whether the UM would continue in its present form, and how well
the Gal Chen system can really cope with steep terrain, and how well it
it could be used for steep slopes. Comments from two MO persons who
couldn't be present, argued that it might be useful to try and alternative
(e.g. terrain intersecting) grid structures as a useful experiment.
However, the alternative view was also strongly argued that the UM
was now a viable option, new micro physics was being inserted, the model
was well supported and there were changes being made, including the inclusion
of variable horizontal grids in addition to the NWP plans described in the
June 2003 letter from NWP at the MO.
The inclusion of turbulence closure was being
discussed and implemented, and this would be a possibility in future years.
Specific areas for enhancement include
1. Option to remove implicit gravity wave treatment, and explore improved
implementation of semi-Lagrangian approach that then becomes possible.
2. More accurate coordinate transformation (but clearly still limited to 45
degree slopes).
3. Better (3d) turbulence model.
4. More complete microphysics.
There was no conclusion drawn from these discussions.
Grid systems.
Limitations were again seen in current grid systems. This topic was
very much tied in with the questions raised above. A strong case
was put forwards to implement a C / Charney Phillips type grid, from
the dynamical viewpoint. However, there were no comments on this type
of grid structure for use with radiation and microphysics codes, where
other grids may be more suitable.
There were many comments in support of development of unstructured and adaptive
grids. The application to ocean models was very clear and advantageous.
However, many commented that this would present a significant effort
It was felt that this would require more man resources than were
available, and therefore unless more resource was apparent, it would
be difficult to achieve these ends.
There was also much discussion about the implementation of a terrain
intersecting grid approach. The advantages for steep terrains implementation,
for simplicity of solution of the pressure equation, and the
advantages of a Cartesian grid were stressed. A further opportunity
to work with European colleagues in this area, would also assist
with limited resources available.
There were no conclusions drawn in this discussion, and further discussions
would occur.
. ---------------------------------------------------
Actions following the workshop
An interim report has already been discussed at the UWERN management group, for
short term developments required for UWERN. In the short term, modules
for microphysics and modification of the pressure solver for BLASIUS, will
be required for the middle of the year. A end of year report will be
produced proposing a way forward, where all the above discussions will be
included and action proposed. It is planned to implement these activities
after the BLASIUS modifications have been completed
Attendees:
Juma Al-Maskari, Luca Boneventura, Alan Burns , Ralph Burton, Ian Castro, Alison Coals , Mike Cullen, Francois Faure, Alan Gadian, Huw Lewis, David Marshall, Stephen Mobbs, Bob Plant, Ian Roulstone, Alison Tomlin, John Thuburn, Dr Wen and David WoodheadFurther details email Alan Gadian