Draft instructions and plan for HIRLAM meso-scale modelling 1

Document Sample
Draft instructions and plan for HIRLAM meso-scale modelling 1 Powered By Docstoc

Draft instructions and plan for HIRLAM meso-scale modelling
                                       29 January 2004
                                          Per Unden

   1 Introduction and background
   This document summarises the actions and general direction of the work following the
   discussion and conclusions from the Hirlam Advisory Committee telephone conference
   on 26 January 2004. An outline plan was submitted for that conference and the current
   document builds on parts of that, but with some important modi cations and clari ca-
   The Hirlam Project work for developing the meso-scale non-hydrostatic modelling sys-
   tem is based on the ALADIN NH dynamics and cooperation with Meteo-France and
   There are two options for achieving this goal. 1) A code collaboration around a common
   system with all relevant Hirlam components imported/interfaced or 2) using the Hirlam
   spectral model and transferring the NH science and code from ALADIN into Hirlam.
   Meteo-France is ready to support Hirlam in either alternative.
   The existing hydrostatic Hirlam modelling system will be kept and maintained for a
   number of years into the future. (This will still the basis for operations in Hirlam
   institutes for several years).
   After the HAC meeting on 31 October in Copenhagen the ideas have matured a bit more
   and there have been some re-assessments following internal discussion in a number of
   institutes. A rapid progress in this priority area is essential and the code collaboration
   has the advantage of progressing faster with less risks. This option is certainly now one
   possible alternative and will be investigated. The impact on the working practices of
   Hirlam members and on the Hirlam management structure and decision making needs
   to be worked out.
   The possibilities of a scienti c cooperation in the eld of mesoscale modelling with the
   UK and the Uni ed Model developments should be investigated by the Project at an early
   stage. National installations and general use of the UM in one or more of the institutes is
   outside of the Project. However, to bene t from the scienti c cooperation with the UK,
   parallel tests and other comparisons may be performed. Such tests and experiments that
   are for the purpose of comparing with the Hirlam meso-scale modelling system or are
   for the cooperation between the Hirlam meso-scale group and the Met O ce, will be
   part of the Project.
Draft instructions for HIRLAM meso-scale work, 29 January 2004                       2

A meso-scale group with its assigned Area leader will be appointed as quickly as possi-
ble in order to start the work. A natural starting point is to get acquainted with the
ALADIN/AROME modelling system, to set it up and to introduce/interface Hirlam
physics and get experience for further work.
The Area leader and the group should analyse the alternatives in much more detail and
recommendations will come out of that work. The scienti c and technical advantages
as well as the risks need to be analysed in the two alternatives. The impact on the
management and working practice of Hirlam will also be worked out by the group, the
Management Group and the Project Leader. The Project Leader is ultimately responsible
for the proposals that will come forward (as for all other areas).
There will be sta initially from SMHI, DMI, KNMI, INM and FMI, with 1-3 persons
from each institute. does not have enough sta now, due to the UM work, but
will be able to set up and run comparisons with the UM.
The plan and proposals should be presented to the Hirlam Council on 2 June, and by
implication also discussed in its preliminary form, at the HAC meeting 26-27 April.

2 The two alternatives
The two alternatives are summarised below. The code collaboration alternative is the
generally preferred option, but there are many unknowns and the implications and risks
are to be carefully analysed.

2.1 1. Code collaboration
The code collaboration would mean that Hirlam bases the development on a version
of ALADIN/AROME. Hirlam components and speci cs, rst the Hirlam physics, are
ported and interfaced to this common system. Other components, particularly the dy-
namics, would be used from ALADIN, as in the hydrostatic case it is equivalent with
Hirlam. The Data Assimilation is also based on the same theoretical framework, but
there are technical questions and options that need to be addressed and implemented. A
number of embedding aspects are more advanced in ALADIN whereas others are more
developed in Hirlam and a best synthesis of these will be made.
A code collaboration would mean that
a) the spectral model equivalent to the Hirlam one is used (no principal problems ex-
pected with this)
b) Hirlam complete physics is used, with the option of trying or using AROME physics
at high resolution
Draft instructions for HIRLAM meso-scale work, 29 January 2004                           3

c) a common 3D/4D-VAR is used and developed with the same functionality and formu-
lation as in Hirlam
d) e ort has to be spent on code collaboration in a wider context and implementations
are governed to a certain extent by the collaboration
e) technical code and system constructs would change to the one used and developed in
the code collaboration (also a lot to be gained from that)
Advantages are:
- Speed of implementation is higher
- Low scienti c risks
- New developments from a a wider community can be bene ted from
- Better technical support and standard of code, system, climate data, post-processing,
veri cation etc.
- Support and usage of ECMWF physics readily achievable and maintainable
- Hirlam physics can be used and checked by Meteo-France
Disadvantages are:
- E ort in code coordination
- Learn and implement new system (although easier for people who have been to ECMWF
or MF).
- Some political agreements need to be made on ownership etc.

2.2 2. Introduction of ALADIN NH dynamics into Hirlam
The other alternative, which was the compromise agreed in the HAC meeting in Copen-
hagen, was to only introduce the non-hydrostatic part of the dynamics into the Hirlam
spectral model. This may be the least "invasive" approach, and in principle possible, and
there will be many practical problems in coding in these changes and with interactions
with the existing hydrostatic part, e.g. the changes of variables and interaction with the
semi-implicit scheme. It is believed that this is feasible, but it requires dynamics exper-
tise. Meteo-France is o ering advanced expertise to help Hirlam and Hirlam would
develop its expertise in the process.
This approach means:
a) to use the spectral model (in the H case it would be exactly the Hirlam original one)
b) to use the Hirlam physics, and with a developed interface it would also be possible
Draft instructions for HIRLAM meso-scale work, 29 January 2004                          4

to use AROME physics
c) the Hirlam Data assimilation would be used as it is
d) the Hirlam code and system environment can remain as it is (for better or worse).
The advantages are:
- Keeps Hirlam integrity or brand in a clearer way, even though important inputs from
- Independent code development and no need to wait for coordination, unless wanted
- No transition to a new system needed (if no radical overhaul)
- Less need for political agreements or new associations
The disadvantages are:
- More development time to get the NH meso-scale working well
- More scienti c risk
- No direct and easy possibility to "upgrade" and bene t from the common developments
- Existing de ciencies in the Hirlam system and support, incl. climate system likely to
persist and get even worse due to the shortage of sta for this

3 Outline instructions for the meso-scale group
A full code collaboration with Meteo-France and ALADIN is a possible alternative to be
analysed carefully. It would be the fastest way to a full meso-scale modelling system, but
there is a great transition e ort and there are the political implications and agreements
that will be needed in some form.
The way of working and making decisions about plans and developments and speci c
releases in Hirlam and the collaborating partners will be a ected and scenarios for this
should be investigated. There is an unavoidable overhead in a collaboration but much to
be gained for all partners. It is probably most advantageous to have internal coordination
with existing Hirlam members and another top-level coordination, but detailed proposals
of this should be worked out by the Area Leader and the Project Leader (incl. the
Management Group).
The meso-scale group needs to be formed, with the appointment of an Area Leader and
with a limited number of sta , but each working most of their time with meso-scale
Draft instructions for HIRLAM meso-scale work, 29 January 2004                            5

The group needs to be acquainted with the ALADIN modelling system and learn how to
install it and operate it. A learning week will be organised by Meteo-France, quite soon,
and a couple of sta in the group should go there to learn about the system. Probably this
work will be extended to build Hirlam physics interface to AROME (the new interface
structure) to gain experience and for reference for the Hirlam implementation work.
This can be decided by the group, Area leader and MG, depending on the time schedule
and sta ng availability and pro le etc.
The ALADIN system will be installed for the use of the group in one or two of the
institutes or at ECMWF, whatever is most convenient for the group. This ALADIN
installation will be used as a reference.
The Hirlam spectral model will be tried and analysed. The physics may be updated
to the current Reference ( especially if to be used in 4D-VAR instead of interfacting the
gridpoint model ).
The group will take up contacts with the Met O ce scientists in the meso-scale area at
an early stage, and subjects for scienti c collaboration will be identi ed. Experiments
with the Uni ed Model may be a base for the collaboration.
There will be a code, system and tools responsible person, both for the prototype AL-
ADIN/AROME system and the development version with Hirlam.
The group studies and analyses the code collaboration as well as the NH dynamics imple-
mentation strategies in Hirlam and develops a proposal together with the Project Leader
and the MG for the most feasible one, following the general agreements, but with a lot of
freedom for the detailed choices. The feasibility in combination with the best choices for
the future must be deciding factors. The solutions which gives the most choices for the
future should be preferred.
The Project Leader together with the Area leader and the rest of the MG work out scenar-
ios or proposals with Meteo-France and ALADIN-2 on how to organise code collaboration
and management of common structures and developments.
The development of the Hirlam version of the meso-scale modelling system then proceeds
following agreements in the HAC and Council. The group should work a signi cant part
of their time together, for maximum speed and e ciency. Extra support for travel and
extended visits is available from the Hirlam Budget, as soon as the amount of travel
exceeds the "normal" amount, not to put unfair nancial burden on particular institutes.

4 Planned list of actions
The following items will be necessary to initiate the work in the meso-scale modelling:
Draft instructions for HIRLAM meso-scale work, 29 January 2004                         6

     Announce and advertise the new position as Area leader for meso-scale modelling
     with closing date 12 February 2004. Decision to be taken by the HAC via e-mail
     as soon as possible after this date (by 24 February). Council will then approve via
     e-mail within 2 weeks. Bearing in mind that candidates are known already from the
     12th or 13th, a deadline of 5 March for the Council approval of the appointment
     should be su cient.
     The Scienti c plan is updated to among other things include the new direction and
     priorities and remove or reduce others of lower priorities. The Area leader is to be
     involved in the planning.
     The sta for the meso-scale modelling are allocated jointly with the Project Leader,
     Area leader and the Heads of Research.
     A more detailed plan is established in a meeting with the Project Management, the
     sta in the group including appropriate Meteo-France and ALADIN representation.
     Scenarios and a proposal for the management and coordination in Hirlam and with
     the other partners are worked out, for the code collaboration alternative.
     A work group is established, in close contacts and possibly organised around a
     virtual centre, in one of the institutes.
     The Project takes contacts with the Met O ce scientists in particularly the meso-
     scale modelling area and identi es subjects for scienti c collaboration.

5 Announcement of the Position of Area Leader for
  meso-scale modelling in Hirlam-6
The Hirlam-6 Project is establishing a meso-scale modelling group in order to develop
the meso-scale modelling system as speci ed as one of the priorities in the Memorandum of
Understanding. The work in the area is based on the non-hydrostatic dynamics developed
by Meteo-France and ALADIN. Meteo-France is part of Hirlam, in terms of research
development, and a special agreement exists. In addition the meso-scale group should
identify subjects for scienti c collaboration with the Met O ce scientists in the eld.
The cooperation with Meteo-France (and ALADIN) may be in the form of a common code
collaboration or in the form of importing parts of the ALADIN dynamics into the Hirlam
system. The Area leader should together with the group analyse scienti c and technical
consequences, advantages as well as risks and problems. The impact on management and
decision making within Hirlam and between cooperating partners will be investigated
and proposed together with the Project Leader and the Management Group.
Draft instructions for HIRLAM meso-scale work, 29 January 2004                  7

The position is on 50 % of full time and funded by the Project to that degree. It is
expected that also most of the remaining time is spent on Hirlam work or other meso-
scale modelling work.
The Area leader is part of the Management Group, on the same terms as the other
members, following the Memorandum of Understanding. This position replaces the one
assigned to Model Dynamics, since it is vacant and moreover, the activity the dynamics is
expected to be mostly related to the meso-scale modelling. The meso-scale modelling Area
leader will lead the work in all aspects of the Hirlam meso-scale modelling - dynamics,
physical parameterisation, data assimilation, use of observations and system aspects that
are relevant for the meso-scale. There will thus be quite a lot of overlap with other
Management Group members and this is natural and should only be stimulating. The
responsibilities for the meso-scale developments in the di erent areas lies with its Area
leader, in rst instance, but also shared with the rest of the Management Group and
ultimately with the Project Leader.
The particularly important role will be the coordination and management of the group of
people and the work on the emerging meso-scale modelling system.
The candidate is expected to have a good scienti c background and experience of meteo-
rological NWP models, both in dynamics and parameterisation and a good understanding
and experience of code structures in general. General knowledge of the other aspects of
NWP including data assimilation and observations is of advantage, but experience in all
areas can not be expected. The modelling aspects are initially the most important ones
and the Management Group will serve to complement the Area leader where necessary.
Scienti c management experience and good communication skills are needed.
The position should be appointed as soon as possible and the application should reach the
Project Leader ( and the Hirlam Council Chairman (
by Thursday 12 February 2004. The application may contain a short declaration of inter-
est and a short CV. Application by e-mail is su cient, due to the time constraint.
It is anticipated that the position will be appointed early in March and it is of de nite
advantage if the candidate can start the work as soon as possible and already in March.
A learning week at Meteo-France will probably take place in March.

Shared By: