Version 24/03/03_2

Microscale Modelling Workshop. 27^th March 2003, room 401B, UK Met Office , Bracknell, 10:30 for 11:00 start. Lunch ~ 12:30-13:15, and 16:00 finish.

Background:

The main goal of the microscale modelling project, which was set up by UWERN and started on 1^st January 2003 is to develop a dynamical atmospheric code suitable for the scientific community. The aim is to resolve physical and dynamical scales down to a few metres, and enable resolution of flow over steep terrain and other atmospheric high resolution processes (e.g. microphysics) in the atmosphere. It is envisaged that this is an eight? year project, with a useful lifetime of not less than fifteen years. An objective is to promote input from a wide cross section of the UK community particularly the MO and research groups funded by UWERN and other research institutes.

The proposal for the first year of the project is to review current requirements and make initial tests of likely schemes, with a design proposal to be considered by the UWERN steering / management committee in Spring 2004. Scientific issues to be resolved include the equation set - e.g. compressible / anelastic etc.-, grid structure, co-ordinate system - variable or nested, data assimilation from existing larger scale models - e.g. MO -, representation of orography and vertical terrain following co-ordinates, numerical schemes for advection and time-stepping and turbulence schemes. It is currently envisaged to use Fortran 90 with MPI and Net-CDF data output, which will be compatible with standard graphical packages such as IDL and with UWERN diagnostics such as JPLOT.

A series of workshops are proposed. This is the first of these and will occur on March 27^th. The Co-Pis and those who have expressed interest in the project have been invited, but the meeting will be open, given accommodation restrictions. A limited amount of funds are available to assist with travel costs if participants do not have other sources of funding. Talks on specific topics have been scheduled to focus the discussion, but specific time has been allocated for discussions, and it is hoped that participants will express their views.

Workshop Outline:

The (outline) objectives of the microscale model project are:
1. To carry out a thorough review of current capability to model atmospheric flows and processes down to the building scale, including the modelling of building-scale flows themselves.
2. To adapt or build from scratch a core dry dynamics and thermodynamics code capable of addressing the research objectives of UWERN within the orographic flow and boundary layer areas (including urban flows). This will inevitably involve addressing the problem of very steep orography (i.e. cliffs).
3. In due course, to add to (2) physical and chemical models for a wide range of processes.
4. Provide support to UWERN for the modelling system.
5. Run workshops on the development and usage of the system.

The first workshop will be principally concerned with the first two objectives. The outcomes of the workshop should be:
1. A plan for which models, which test problems and which contributors will be involved in the review.
2. A plan for who will contribute, and on what timescale, to answering the following three questions:
a. Which equation set(s) should be used.
b. What type of grid system should be used (e.g. terrain following, boundary fitted, terrain intersecting...)
c. What solution method should be adopted (e.g. finite volume, etc)

Draft Programme: 1^st Microscale Modelling workshop: A review

Each talk should include 5 minutes for comments and questions at the end.
10:30 Meeting room (Coffee)
11:00 Introduction: Stephen Mobbs: Background to the project
11:15 Alan Gadian: " Structure of the workshop. What we are trying to achieve? "
11:20 Andy White: "Equation sets"
11:35 Nigel Wood: " Validity of anelastic and other equation sets as inferred from normal-mode analysis"
11:50 Andy Brown: " Blasius and orography issues "
12:00 Tito Toro: "High order non-oscillatory schemes for advection-reaction equations"
12:10 Ian Roulstone: "Data Assimilation"
Short discussion to highlight outstanding issues. To be continued over ...

12:30 to 13:15 Sandwich lunch

13:15 Rex Britter & Bill Dawes: " Microscale Modelling from an Engineering perspective"
13:35 Mike Cullen: " Issues about a fine scale model " (Not able to be present; presented by Alan Gadian)
13:50 Christiane Montavon & Alan Burns: " Experience of modelling atmospheric flows with a commercial CFD package"

14:10 General discussion of presentations with specific attention to the General Comments and Specific Issues raised below.
14:30 What strategy should we employ? Can we say what is a "good strategy" , or what a "good strategy to formulate a good strategy" would be?
Would it be useful to examine specific test cases? To provoke discussion in this area, some possible examples are listed below. Are these useful examples and how should they be modified to be of use? Any other test examples? Please propose before or during the workshop.

15:20 Summary and outcomes?. Next workshop? . Date? .
15:30 Tea
16:00 Close

Specific Issues that need addressing:

Grids and grid structures.

Equation sets e.g. compressible, anelastic etc… LES capability? Discretisation of equations.

Vertical co-ordinate systems and representation of orography

Horizontal co-ordinate systems, e.g. variable, 2 way nested grids.

Compatibility with other models and Data Assimilation issues

Pressuer Solvers, numerical advection and time stepping schemes

Turbulence closure and 1+ order schemes

Structure of the initial tests.

General Comments:

Engineering codes with unstructured grids are the ways forward. Why not develop a "relevant equation set"? What about microphysics and moisture?

Why not develop a currently existing compressible model, e.g the M.O. Unified Model New Dynamics for small scales?

Why not develop one of the good anelastic models currently available?

We want something designed that is easy to run and can be used on a variety of platforms. A research rather than a forecasting tool is required.

How "steep" is steep terrain? Should we even be aiming at street canyon type flow?

With only limited resources, what "new" can be achieved?

How can we maximise current UK knowledge and experience to make informed decisions about the advantages and disadvantages of current schemes?

Should we conduct any inter-comparisons /experiments on current models to test limitations, and if so what?

Possible test cases. (many thanks for those who have contributed). These are only suggestions for discussion as examples for discussion foci.

0. Basic tests to look at the pressure solver (offline) and pure advection test for any new code.

For existing codes:

1. Flow over a small hill. This is a gravity wave case. Linear solutions are available (e.g. R. Smith. 1980, TELLUS, 32, 348-364) dx = dy = 1.0km , dz = 200m, Hill = hm / (1 + (r/a)²)^1.5, hm = 100.0m, a=10.0km, u =10 m/s. Damping above 10km? ???nx=ny=nz=128??? B=1.3 * 10^-5 m^-1 . T₀ = 280K

2. Flow over a small hill. This is a gravity wave case. Solutions are available (e.g. G.Shutts et al. 1999 Q.J.R Met Soc, 125 2743-67) dx = dy = dz = 1.0km, Hill = hm / (1 + (r/a)²)^1.5, hm = 100.0m, a=10.0km B=1.3 * 10^-5 m^-1 . T₀ = 280K u₀ = 10 ms^-1 and u = 10 [ cos(pie*z/6.333) , sin(pie*z/6.333)] where z is in km No damping needed.

3. Smolarkiewicz & Rotuno, 1989, JAS, 46, 1154-64) Low Froude Number Flow, with a bell shaped hill. Flow of a uniformly stratified fluid past a three dimensional obstacle.

4. Flow over large steep hill. This is a gravity wave case. (e.g. G.Shutts et al. 1999 Q.J.R Met Soc, 125 2743-67) , dx = dy = dz = 0.250km, Hill = hm / (1 + (r/a)²)^1.5, hm = 1.5km, a=1.0km B=1.3 * 10^-5 m^-1 . T₀ = 280K
u₀ = 10 ms^-1 and u = 10 [ cos(pie*z/6.333) , sin(pie*z/6.333)] where z is in km. Either a radiation or damped side boundary conditions

5. Insertion of a large steep hill in a stationary stable atmosphere. This should be a basic test of any non-hydrostatic atmosphere. No references for this case, but use the same hill as 4. T₀ = 280K Zero wind throughout the model.
a. . . . Isothermal atmosphere, T₀ = 280K
b. . . . Isentropic atmosphere.

6. Neutrally stratified flow over periodic hills. This is similar to 3. except as inviscid, no roughness required , and no corilis i required. Numerical dissipative effects will be apparent for a long steady state response. The flow response is local in the vertical and therefore there is no need to worry about sponges. Also bell shaped hills have their analytic virtues but my view is that in a finite domain they can cause problems at the lateral boundaries and their mix of scales or width of spectrum (from the grid scale to the half-width) can cause problems with interpretation of results. Therefore in general perhaps the use of cosine hills - either isolated (in the form of cosine-squared hills) or periodic. would be better for the above.

7. For viscous (ie with a turbulence closure) flows, a good test is the simple 1D pressure-gradient driven planetary boundary layer - This tests a no-slip lower boundary condition. If that works then look at neutral viscous flow over a periodic hill driven by a pressure gradient now with Coriolis (to give a steady state finite depth boundary layer response).

8. ( Later stages ) Neutral Flow, B=0, Coriolis Force included. As example 1, but with free slip and z₀ = 0.1m?

CFD type test cases.

9. Flow around a square house (see CEDVAL below) and / or over a steep cliff .....

10. Flow over a street canyon. (to be finalised .....

11. Preset QNET CFD or other test examples. e.g. Standard flow over hill example, which is similar to 1. http://labo.univ-poitiers.fr/informations-lea/Workshop-Ercoftac-2002/case9.2/case9.2.html
Another (not very suitable example) taken from the CEDVAL Data sets is the study of flow past a square building is a cross-section or a plan view
Another (even less suitable example) is a study of the flow past a square cylinder. http://vortex.mech.surrey.ac.uk/LESig/les2/

Attendees:

Stephen Belcher , Alan Burns , Rex Britter , Andy Brown , Xiaoming Cai , Ken Carslaw , Ian Castro ? , Peter Clark , Terry Davies , Bill Dawes , Alan Gadian , Sue Gray , Brian Golding , Martin Juckes , Roy Kershaw , Paul Mason , Christiane Montavon , Stephen Mobbs , Martin Miller , Ian Renfrew , Ian Roulstone , Chris Smith , Piotr Smolarkiewicz , Andrew Staniforth , Glenn Shutts ? , John Thuburn , Tito Toro , Simon Vosper , Neils Wedi , Andy White , Nigel Wood.

Unable to attend:

Mike Cullen , Doug Parker ,

Back to microscale home page

Further details email Alan Gadian