Around 40 persons (3/4 of visitors during this overbooked quarter when it was impossible to find a free seat in the GMAP offices) have exceeded a common effort of 60 people.month. PhD students (D. Banciu, I. Gospodinov, M. Janiskova and F. Vana), LACE representants (R. Bubnova and M. Siroka), phasers (I. Gospodinov, P. Neytchev, C. Soci and M. Zitouni, see 1.), stays on embassy support (see 2.) were present in GMAP this quarter.
1. Main events in Toulouse this quarter
The phasing for the creation of AL10 with CY20 begun this quarter.
- ECTOplasme nox also computes tvorticity, divergence and kinetic energy; also,the difference of two fields is available (a new document is available on the public ftp : docECTO.v1.10;
- MAKDO, a tool for creating domains is being developed : different parameters such as projection type, projection view, scale, rotation, ... can be chosen and you will get the EGGX parameters of your domain.
2. Other visitors research or development studies that ended during this quarter
A 10-km resolution version of ALADIN was used to simulate two events of convective systems in the Balaton Lake area, including the latest developments in the parameterization of convection. A good overall description of the precipitation patterns is noticed, when compared to radar reflectivities. Small scale wind features are missed due to the insufficient resolution which prevents from a satisfactory representation of Lake Balaton itself.
Some options were examined to solve the problems encountered when the land-sea masks strongly differ between model and data (especially over islands, lakes, along coasts, ...), while keeping the consistency between embedded models.
The previous work was pursued and some unexpected problems revealed.
The workstation version of configuration E923 now works, using data covering only the target domain. Tools for extracting a subset of global databases were designed. A procedure to mix low and high resolution topographic data was also proposed and validated. It will be quite useful for model resolutions below 5 km.
The idea was to replace the variational low resolution analysis of upperair fields by a CANARI O.I. one, within the frame of the existing operational ARPEGE incremental assimilation suite. The experiments with O.I. incremental assimilation showed the incremental CANARI cannot compete incremental 3d-Var. There is a positive impact when including the stretched geometry of the model at low resolution in CANARI incremental, but it is necessary to reconsider the CANARI statistical model.
The computation of the coefficients required by the parameterization of ozone was added to configuration 923. They can now be stored either as grid-point (as previously in ARPEGE) or as spectral fields. Afterwards some preliminary modifications were brought to configurations 927 and 001.
This last stay is devoted to the production of the last results and the redaction of the report on the intercomparison of precipitation forecasts over the Alps. An article will present the results in the next Newsletter.
A modification of the calculation of grid-point fields in the extension zone has been introduced. The procedure is now the same in both E923 and E927 configurations and ensures reasonable values of the constants used by physics. Besides some problems related to non-hydrostatic dynamics have been investigated.
The use of digital filtering to design a 3.5d-Var (or poor man's 4d-Var) was investigated, but the results are disappointing. The procedure of calculation of background errors by the NCEP method was afterwards adapted to ALADIN-Maroc (errors are estimated as the deviations between two forecasts valid for the same time but with different integration ranges). They are expected to differ from that already computed for ALADIN-France or ALADIN-LACE due to the quite different locations of these models.
Some preliminary sensitivity experiments with respect to the initial conditions were carried out. Was posed the question of the impact of the sensitivity of a forecast aspect with respect to a small perturbation of the model initial conditions when the size of the domain is relatively small comparatively with a large-scale perturbation.
Installation of CANARI and associated tools on the SUN workstation, with J.M. Audoin. Some assimilation experiments were also performed.
see article in this Newsletter
3. Other research or development studies by the Toulouse permanent staff
Preliminary study of the lateral boundary conditions for variational-type of configurations. Proper implementation of the gradient test in 3D-Var.
In configuration 923, it is now possible to split the ALADIN domain into rows of latitudes (which is necessary to obtain tracable memory sizes for global ALADIN BDAP grids).
 |
Home |