1. Federal Department of Home Affairs FDHA Federal Office of Meteorology and Climatology MeteoSwiss PP COLOBOC Status report Jean-Marie Bettems 06.09.2010 Moscow (RU)

2. 2 COSMO GM, PP COLOBOC / Moscow, September 2010 Review – COLOBOC, task 0 Observation sets for SVAT model validation. New permanent instruments at Payerne Measurement of turbulence @ 10m, in activity since spring 2009. Soil moisture and temperature. Documentation of various data sets on the COSMO web site. Data pool Collect soil / surface / BL observations at selected sites, on behalf of the C-SRNWP Programme. Operational data with time lag from: Lindenberg (D), Payerne (CH), Toulouse (F), San Pietro (I), Sodankylaa (Finland), Cabauw (Netherland). Data of 6 participating sites for 2006-2008 soon available on-line on the COSMO web site. Year 2009 will follow afterwards. Common xls format. Special side meeting on this topic at the EMS conference in Zurich (Wed. Sept. 15th, 14h00 - 15h30).

3. 3 COSMO GM, PP COLOBOC / Moscow, September 2010 Review – COLOBOC, task 0 Proposal for extension JM. Bettems/MCH, C.Heret, F.Beyrich/DWD, T.Andreadis/HNMS (0.3 FTE) Extend on-line documentation (e.g. site characteristics, effective data availability of participating sites) Check usability of collected data (Master student at ETHZ, in group of S.Seneviratne) Collect 2009 data, include additional observatories (e.g. Valday/Russia, Cardington/UK) Make this effort sustainable

4. 4 COSMO GM, PP COLOBOC / Moscow, September 2010 Review – COLOBOC, task 0 Open questions Use SRNWP data pool for routine diagnostic of the soil, surface, and boundary layer of all COSMO models (a similar action was started by G.Bonafe and G.Vogel)

5. 5 COSMO GM, PP COLOBOC / Moscow, September 2010 Review – COLOBOC, task 1 Consolidate externalized TERRA module. Experience from G. Vogel has shown comparable results between full COSMO and offline experiment (3 months, soil moisture); more tests soon available from MeteoSwiss Updated package including tests and documentation soon available (within the next weeks) Package will be integrated with the COSMO code (compiler flag to chose the full COSMO or Terra standalone) Code remains fragile when used in a non tested configuration; code limitations will also be documented

6. 6 COSMO GM, PP COLOBOC / Moscow, September 2010 Review – COLOBOC, task 1 Task is terminated, but open questions remain Support after end of project, in particular with respect to the introduction of GRIB2 and the new physics interface Provision of COSMO temporal mean values for all driving fields (e.g. T_2M) Code re-write to achieve a production standard

7. 7 COSMO GM, PP COLOBOC / Moscow, September 2010 Review – COLOBOC, task 2 Consolidate software for generating external parameters. Software Requirement Specifications (SRS) is written Raw data files are converted into netcdf format New code for the aggregation and interpolation of the raw data to the target grid is available (Fortran 90), technical documentation is available → Status to check at COSMO GM Implement GRIB output Activate reference system at DWD, accessible through a Web interface Extend documentation (for each raw data set : included in EXTPAR, or experimental, or not available in EXTPAR)

8. 8 COSMO GM, PP COLOBOC / Moscow, September 2010 Review – COLOBOC, task 3.1 Document external parameters data set. Documentation of datasets available on the COSMO web site (http://www.cosmo-model.org/content/model/modules/externalParams/default.htm)http://www.cosmo-model.org/content/model/modules/externalParams/default.htm → Status to check at COSMO GM Update documentation to include latest developments

9. 9 COSMO GM, PP COLOBOC / Moscow, September 2010 Review – COLOBOC, task 3.2 Consolidate external parameters data set. New external parameters available for any domain: monthly NDVI climatology, minimum stomatal resistance, bare soil emissivity, deep soil temperature, lake fraction and lake depth, urban fraction, SSO (4 parameters, with proper scale separation for z 0 ), monthly climatology for aerosols optical thickness (4 species) Only over Germany: BUEK1000 high resolution soil data with vertical soil structure External parameters for orographic radiation correction will not be available before COSMO GM (bug correction required)

10. 10 COSMO GM, PP COLOBOC / Moscow, September 2010 Review – COLOBOC, task 3.2 Proposal for extension (DWD, MCH, Roshydromet, CLM community) (1.0 – 1.5 FTE) Consolidate new MODIS derived solar albedo ("background albedo" derived from MODIS data at MPI Hamburg) Consolidate alternative sets of soil types in Europe (Harmonized World Soil Database, European Soil DataBase) Add support for vertically dependent soil information (e.g. depth of water reservoir or inactive layer, texture) Alternative vegetation characteristics using MODIS derived phenology model (work by R.Stöckli) Add parameters for orographic radiation correction (as developed at MeteoSwiss; code exists but crashes when running at DWD). Allow correct representation of scale separation for SSO / resolved scales (option for topo smoothing in EXPAR instead of INT2LM) Provide topography at higher resolution than GLOBE when possible (e.g. SRTM or ASTER GDEM) Consolidate Root depth

11. 11 COSMO GM, PP COLOBOC / Moscow, September 2010 Review – COLOBOC, task 3.2 Real-time phenology Historical records of vegetation characteristics reveal a substantial inter-annual variability of the start of season, which may limit the usefulness of a climatology based data set. A framework has been developed by R.Stöckli et al., using a prognostic phenology model with parameters constraint by MODIS data, which can provide an offline gridded forecast of the vegetation characteristics taking into account the actual evolution of the weather [Stöckli 2008]. Basically a statistical approach is used, relying on an ensemble Kalman filter to define the optimal parameters of the phenology model, for a specified set of meteorological data predictor (e.g. from a NWP model).

12. 12 COSMO GM, PP COLOBOC / Moscow, September 2010 Review – COLOBOC, task 3.2 Real-time phenology The prognostic penology model includes the three main climatic controls of seasonal phenological processes: minimum daily temperature, mean daily vapor pressure deficit, mean daily global radiation (Growing Season Index, Jolly et al. 2005). The model derives LAI and FPAR (Fraction of Absorbed Photosynthetically Active Radiation) A 1000 members EKF is used to derives the optimal parameters of the prognostic phenology model; one set of parameters is computed for each type of vegetation. ‘… the model reproduces the inter-annual variability of start of season with correlations ranging between 0.6-0.9 when compared to independent ground observations.’ [Stöckli et al, Journal of Geophysical Research 2008]

13. 13 COSMO GM, PP COLOBOC / Moscow, September 2010 Review – COLOBOC, task 3.2 Open questions Use of alternative land use data "GLOBCOVER" (higher resolution than GLC2000) Evaluate usage of crop life cycle information (e.g. CLM AgroIBIS, J.L.Roujean/MeteoFrance product)

14. 14 COSMO GM, PP COLOBOC / Moscow, September 2010 Review – COLOBOC, task 4 Revision of TERRA and the associated look-up tables. Developments of TERRA have been integrated in official COSMO code Verification of COSMO-EU experiments with latest TERRA results in undesirable increase in T2M bias. COSMO sensitivity experiments showed a large impact of the PRS_MIN / LAI ratio on the results. Experiments are running to build a consolidated set of look-up tables to avoid this problem, in particular by calibrating PRS_MIN. → Status to check at COSMO GM Tuning of look-up tables and recommended configuration of new system

15. 15 COSMO GM, PP COLOBOC / Moscow, September 2010 Review – COLOBOC, task 4 Proposal for extension (DWD, MCH, CLM community ) (1.0 FTE) Revision of TERRA rainfall interception and surface water treatment Implementation of an orography dependent surface runoff Detailed comparison of COSMO/TERRA with COSMO/CLM in weather mode (Master student at ETHZ, in group of S.Seneviratne)

16. 16 COSMO GM, PP COLOBOC / Moscow, September 2010 Review – COLOBOC, task 4 Open questions Introduction of a leaf temperature to replace T_2M (T_2M may be tuned to best match SYNOP observations, which is not necessarily the most adequate value for representing the temperature of the plant environment) Optimal stretching factor of the soil mesh (a study by A.Will on the characteristics of the numerical solution seems to indicate that the current stretching factor is too large; this could explain the inexistent daily cycle of low cloudiness)

17. 17 COSMO GM, PP COLOBOC / Moscow, September 2010 Review – COLOBOC, task 5.1 Verify and consolidate the new multi-layer snow model Available in official COSMO code, documentation being completed Tests with terra standalone at MeteoSwiss show a significantly improved quality of the snow depth as compared with the old snow model Tests with full model at MeteoSwiss show some issues with very low temperature within the snow pack at some points Tests with full model at DWD show a degradation of the T2M scores Tests at Rosydromet points to some significant issues with the density of the fresh snow

18. 18 COSMO GM, PP COLOBOC / Moscow, September 2010 Review – COLOBOC, task 5.1 Proposal for extension (E. Machulskaya/DWD, Roshydromet, CLM community) (1.0 – 1.5 FTE) Consolidate multi-layers snow model (snow pack temperature …) Invite E. Machulskaya at MeteoSwiss for working on complex topography issues Adapt snow roughness length in transfer scheme Correct fresh snow density and feedback effects on soil temperature Investigate partial snow cover issue, in particular by using the mosaic approach and by considering the complex topography situation Improve representation of snow in forest covered area (in particular albedo) Tests in climate mode

19. 19 COSMO GM, PP COLOBOC / Moscow, September 2010 Review – COLOBOC, task 5.2 Improved snow analysis DWD and MeteoSwiss codes have been merged (sfcana 3.0), test are on-going Altitudinal interpolation is missing Issue with temporal stability of the analysis (at least in complex topography)

20. 20 COSMO GM, PP COLOBOC / Moscow, September 2010 The DWD snow analysis package is extended at Meteo Suisse to include the MSG snow mask and to use additional SLF snow data. Before including the code in the present DWD official version the binaries are crosschecked in default configuration (no use of MSG, same data) by rerunning the winter 2007/2008 analysis and validation on Cosmo-7 domain. As expected there is only little difference between MCH V3_0 and DWD V1_24, as shown by timeseries of domain average snow height. Fig. 1 Domain average snow height in COSMO-7 model compared with MCH V3_0 and DWD V1_24*. *figure kindly provided by Guy de Morsier Review – COLOBOC, task 5.2 MCH V3_0 versus DWD V1_24 (M.Lange)

21. 21 COSMO GM, PP COLOBOC / Moscow, September 2010 Only Small difference of snow height over the Cosmo-EU domain at Feb. 15, 2010 Modifications from Meteo Suisse can be included in the code without an expected impact on the forecast. Code must be merged with latest DWD SFCANA version that contains modifications for prognostic ice model and Flake model. Use of MSG snowmask at DWD is still an open task. Review – COLOBOC, task 5.2 MCH V3_0 versus DWD V1_24 (M.Lange)

22. 22 COSMO GM, PP COLOBOC / Moscow, September 2010 Review – COLOBOC, task 5.2 Proposal for extension (JM. Bettems/MCH, M. Lange/DWD, Roshydromet) (0.3 FTE) Include altitudinal interpolation, check/improve temporal stability Adapt analysis for the new multi layers snow model

23. 23 COSMO GM, PP COLOBOC / Moscow, September 2010 Review – COLOBOC, task 6 Urban module. Code and documentation are ready. Code has already been used (S.Schubert PIK/Potsdam and B.Sändig IfT/Leipzig ) Will soon be included in official COSMO code

24. 24 COSMO GM, PP COLOBOC / Moscow, September 2010 Review – COLOBOC, task 7 Parameterization of land surface heterogeneity Modifications implemented in COSMO v4.8 Documentation has been adapted Except for one configuration, control experiments produce now correct results (ctrl vs. std, ctrl vs. twin) → Status to check at COSMO GM Not all branches/options of the COSMO code have been updated (e.g. multi-layers snow model, recent soil model developments) Evaluation of code by Uli S. for merging these modifications in official COSMO code is missing Link with development in PP UTCS

25. 25 COSMO GM, PP COLOBOC / Moscow, September 2010 Review – COLOBOC, task 7 Proposal for extension (J.Helmert/DWD, A.Mazur/IMGW) (1.0 FTE) Implement tile approach (%) Full support of tile and mosaic in official COSMO code Impact studies (e.g. tile with nature/urban/lake/sea, mosaic for partial snow in complex topo)

26. 26 COSMO GM, PP COLOBOC / Moscow, September 2010 Summary What has been achieved Task 0: Document observations sets available for SVAT-model validation Task 1: Consolidation of TERRA standalone code > Code remains fragile when used in a non tested configuration > Some work still to be done within the next weeks Task 2: Tools – Software for generating external parameters Task 3: Revision of external parameters > External parameters for orographic radiation correction: software needs debugging Task 4: Revision of TERRA and the associated look-up tables Task 5.1: Snow model > Promissing, but still some weaknesses (temperature in thin layers, snow density, roughness length over snow) Task 5.2: Snow analysis > Altitudinal interpolation is missing > Issue with temporal stability of the analysis (at least in complex topography) Task 6: Urban model Task 7: Parameterization of land surface heterogeneity by the tile/mosaic approach > Tile approach is missing > Not integrated in official COSMO release

27. 27 COSMO GM, PP COLOBOC / Moscow, September 2010 Summary Proposed extension Task 0 [EXTENDED]: Consolidate data exchange action Task 3 [EXTENDED]: Revision of external parameters Task 4 [EXTENDED]: Revision of TERRA and the associated look-up tables Task 5.1 [DELAYED, EXTENDED]: Snow model Task 5.2 [DELAYED]: Snow analysis Task 7 [DELAYED]: Parameterization of land surface heterogeneity

28. 28 COSMO GM, PP COLOBOC / Moscow, September 2010 COSMO GM Offenbach | postprocessing tool jean-marie.bettems@meteoswiss.ch Thank you for your attention!