Dynamics SRNWP Expert Team Workplan for 2008-2009
1) LAM modelling toward higher (sub-kilometric) resolutions
An increasing demand is emerging for small-area and very short-range NWP around specific locations (cities, airports, sport-events areas ...). For some of these applications, the expected horizontal resolution is clearly sub-kilometric and even hectometric. Experiments in the ALADIN group are currently performed at 500m horizontal resolution, on Alpine domains, in order to identify the main possible sources of problems at these scales. The plan for SRNWP, at this preliminary stage, is to share the experience of consortia in this field in order to have a more comprehensive view of the obstacles.
The ALADIN, HIRLAM and ECMWF consortia will devote a significant research effort to investigate more in details the feasability of a vertical finite-element (VFE) discretisation for the Euler Equations system, which brings an increase of accuracy compared to the finite-difference discretisation. A first version using VFEs is delivered, but is not totally stable when all source terms are discretised with VFEs. Investigations may explore the possibility of using alternative formulations for the basis functions and the spatial operators, as well as alternative choices for prognostic variables. This action is a medium-term one, which can not be easily cut in pieces fitting in a annual research workplan.
3) Conservation issues
The use of conceptually simple conservative semi-Lagrangian scheme is studied in UKMO and HIRLAM consortia with different but relatively close strategies, and the ALADIN consortium is interested to take adavantage to the knowledge aquired in this field, in the framework of the code cooperation between HIRLAM and ALADIN.
4) The spectral technique for 'new generation' LAMs
The question of the relevance of the spectral technique for 'new generation' LAMs is opened, because the constraints linked to the spectral technique are often felt as an obstacle for developments of new functionalities. This is a quite long-term action which of course involve mainly the ALADIN and HIRLAM groups. Since there is no actual emergency, the plan for the short-terms would be first to identify the parts or functionalities of the model for which the spectral technique is really an handicap, and for which no alternative solution is seen. The question of the choice of an alternative technique for the horizontal discretisation is also still open.
5) Numerical diffusion, and links of Dynamics with 3D-Turbulence
In prediction models with a "column" physics, a 3D-Turbulence scheme is not always easily feasible. The plan here is to examine the possibility to benefit from the 3D stencil of the SL scheme interpolators to define physically-based 3D diffusion operators which might alleviate this weakness. For this, the SLHD scheme of ALADIN is considered as a starting framework. The sensitivity of the Euler system behaviour to the Horizontal Diffusion will be studied in the ALADIN group, with an interest from the ECMWF group.
6) Large domains
The consequences of using a large domain with a uniform linearisation of the local map factor in the SI scheme, for conformal projection geometries will be studied. For the Euler equation system, intrinsically less stable than the HPE system, this has been identified as a potential problem for the robustness of the model. The necessity of using a non-uniform linearisation will be examined, and if necessary, this functionality will be implemented and tested, mostly in the HIRLAM group.
7) Equation sets for multiphase aspects.
The part of this topic involving the vertical part of the momentum equation has not led to clear and definitive conclusions, and it is wishable to progress on this. However, this is a difficult but not urgent subject for which the availability of manpower is not garanteed.
8) Physics/Dynamics interface.
This topic is actively studied in ECMWF, HIRLAM, UKMO and ALADIN consortia, especially in the context of NH models, which are more sensitive than their hydrostatic counterpart from this point of view. However, it is not clear how can this variety of studies be unified or at least shared between the involved consortia.
9) Equation set for NH models
The HIRLAM group will continue to examine the behaviour of non fully-compressible sets of equations, which, in principle, may be a less demanding alternative (in terms of overall complexity) than the Euler system, also potentially allowing an increased stability. However these expectations are not garanteed to be fulfilled in the state of the knowledge.
10) Lateral Boundary Conditions (LBC)
This topic has received a lot of interest in the past years in the HIRLAM and ALADIN consortia, especially in view of well-posed and transparent LBC formulations. However, the traditional "Davies" approach has revealed difficult to beat in terms of efficiency for a given quality of results. Moreover, new approaches would require big developments in the existing codes, and therefore, the situation is not yet very favourable for a practical application in the short-term. The problem will continue to be examinated, mostly in the ALADIN group.
11) Academic experiments.
Various experiments in more or less academic frameworks are performed in SRNWP consortia, mainly at ALADIN, UKMO, and ECMWF. These experiments mainly involve non-hydrostatic kernel, but also occasionally, the way the physics is interfaced to the dynamics. The consortia remain open to organizing intercomparison experiments. ECMWF is oriented toward small planet experiments, while ALADIN group focuses on 2D vertical-plan and 3D academic cases.
On the practical point of view, one of the specific aims of the ET is to continue the SRNWP-NT meetings every two years, in a small and quite unformal format, as previously, and it is envisaged that most of the exchange, information and shared-interest works would still be achieved at these meetings.