SRNWP Expert Team on Data Assimilation and Use of Observations (ETDA)
Draft plan for 2008-2009
The interests of the ETDA cover data assimilation algorithms, developments for new observation types, OSE and OSSE experiments and ensemble data assimilation. Developments in the consortia are listed and the potential for improving collaborative activity is outlined.
Developments in the consortia
Aladin and Hirlam are developing a common 4-year R&D plan for the 2008-12 period. The main aims currently are as follows:
- promote a more general installation of 3DVAR assimilation. Since at present only France, Hungary and Morocco run their own 3DVAR system, there are plans to extend it to the LACE countries, Turkey, Algeria and Portugal.
- pursue the convergence process with Hirlam, which in practice means some further efforts in order to provide a LAM 4DVAR system suitable for the meso-alpha/beta scales
- provide a convective scale 3DVAR assimilation scheme within the AROME project, suitable for radar data assimilation, cloudy and rainy satellite radiances, dense usage of cloud-cleared radiances (mw and IR) mostly relying on Arpege experience, maximum use of mesoscale surface network observations.
- develop radar data assimilation (both winds and reflectivities)
- engage in research into background error covariance statistics. This will involve work on LAM wavelets, an ensemble approach to capture flow and situation dependency and application of objective a posteriori criteria to the assimilation system
- Algorithms: development of some form of ensemble data assimilation (EnDA) to replace the existing nudging scheme is envisaged in the medium term. Two separate efforts are under way in this direction: KENDA, a COSMO Priority Project led by Christoph Schraff at DWD to build a km scale EnDA system, and a project led by Massimo Bonavita at CNMCA, Italy, which focuses on a similar system at the 10-14 km scale.
- Observations: given the intrinsic difficulties in the direct ingestion of radiances in the nudging algorithm, considerable effort has been and is still being devoted to deriving 1DVar temperature and humidity retrievals from SEVIRI, ATOVS, AIRS and IASI radiances.
- work is under way both at DWD and MeteoSwiss towards refining the current operational algorithm for the ingestion of radar-derived precipitation fields through the latent heat nudging (LHN) technique.
- other work is also being carried out towards the operational use of scatterometer winds, assimilation of Doppler radar radial winds (whose quality control is an important issue), and GPS tomography data for humidity analysis. There are also plans to improve the current use of screen level observations and to make use of METEOSAT data for a cloud analysis to be ingested in the nudging analysis (direct use of cloudy radiances will eventually be tried in the Ensemble Data Assimilation framework)
UK Met Office plans
- introduce in 2009 a new version of the 12km North Atlantic & European (NAE) 4DVAR system to include multiple outer loops, variational cloud and precipitation assimilation to replace nudging, new moist control variable (after Holm)
- introduce in 2009 a 1.5km model with variable resolution, interfaced to an hourly 3DVAR system, still with nudging for assimilation of cloud and precipitation
- introduce in 2009 a high resolution nowcasting system to act as a test bed for convective scale 4DVAR research
- improve the performance in the boundary layer of the linear model in 4DVAR document the impact of assimilating radar rain rate composites on a Europe wide scale (currently testing the impact of the OPERA composite alongside that used operationally at the Met Office)
- improve covariance statistics by deriving them from ensembles
- optimise the relative weight given to observations in assimilation by statistical analysis of innovations
- introduce Doppler radial wind assimilation to the convective scale model
- introduce reflectivity assimilation to the convective scale model
The main research themes in each consortium are convective scale data assimilation, and ensemble data assimilation, while a key observation type requiring new developments is radar. These 3 challenging topics seem to offer most opportunity for further collaboration.
Developments at slightly larger scale are still taking place but the usual exchange of knowledge and experience which happens through annual EWGLAM/SRNWP gatherings and other conferences is sufficient to support these.
Steps to improve collaboration between consortia:
- when a new technical report is written, or external paper published, in one of the consortia on some topic of relevance to ETDA, we invite consortia contact points for ETDA to circulate it to the other members of ETDA
- if one consortium holds a significant planning or review meeting on some aspect of data assimilation, they might consider whether to invite representatives from other consortia
- at the annual EWGLAM/SRNWP meeting, ETDA members present will meet informally to plan more specific collaboration
- an SRNWP workshop was held in the UK in 2004 on the subject "High resolution data assimilation: towards 1-4km resolution". A second workshop was held in Sweden in 2007 on 'High resolution data assimilation with emphasis on the use of moisture-related observations'. A joint workshop with ETEPS on the topic of ensemble data assimilation might be worthwhile in, say, 2010. For radar assimilation specifically, the best forum might be a special session at one of the biannual ERAD conferences where the radar community are present eg at the next ERAD in 2010 in Romania.
- ETDA will also study the plans of other expert teams and see where they can contribute expertise eg concerning surface data assimilation, observational format issues in interoperability, ensemble prediction.
- EMETNET fund a number of observational programmes such as EUCOS, E-AMDAR, E-ASAP, WINPROF, OPERA, SURFMAR, E-GVAP. These programmes need expertise in data assimilation, so ETDA members are encouraged to interact with their national representatives in these programmes and to consider where general action by ETDA may be beneficial. For example, in the next 12 months, ETDA will aim to compile a summary of observation impact studies with Doppler radial winds and submit it to OPERA to encourage it in efforts to establish international exchange of these data.