HNMS contribution to CONSENS

1. HNMS contribution to CONSENS Petroula Louka & Flora Gofa Hellenic National Meteorological Service louka@hnms.gr, fgofa@hnms.gr

2. HNMS involvement • Task 2.1.Test new parameter perturbations, including combination of parameters. • 2.1.1 select new parameters to be perturbed, select parameter combinations • 2.1.3 analyse the impact with respect to the control • Task 2.2.Develop perturbations for the lower boundary fields. • 2.2.1 develop techniques for lower boundary fields perturbation. 11th COSMO General Meeting

3. Physics perturbations • 2 suites • CSPERT: 16 members of the same initial conditions with different values for selected physics parameters. • New: 16 members of different initial conditions with different combinations of physics perturbations. • 3 testing periods of investigation and statistical analysis • Summer 2008 and Spring 2009 with CSPERT suite • May – August 2009 with the new suite • The focus of the study is mainly on the continuous parameters (T, U and Td) rather than precipitation. 11th COSMO General Meeting

4. CSPERT suite • ECMWF initial conditions 11th COSMO General Meeting

5. 2m Temperature Summer 2008 Spring 2009 11th COSMO General Meeting

6. 10m Wind Summer 2008 Spring 2009 11th COSMO General Meeting

7. 2m Dew point temperature Summer 2008 Spring 2009 11th COSMO General Meeting

8. = - -- + ++ • Based on Summer case study

9. Remarks from the CSPERT suite • The effect of perturbing each physics parameter on improving or worsening the statistical values of the results in comparison to the corresponding default was investigated. • The spread of the results is larger for the summer case for temperature and dew point mainly during the hours of intense solar radiation (as expected). • It seems that the particular parameter perturbations do not influence greatly the mean horizontal wind apart from a few exceptions. Possibly looking at the vertical wind component would make the effects more apparent. 11th COSMO General Meeting

10. Remarks (cont) • For the summer case, looking separately at each parameter perturbation compared to the control run: • rlam_heat (scaling factor of the laminar layer depth): Changing its value affects mainly temperature. This is an important parameter since it defines the layer with non-turbulent characteristics (molecular diffusion effects only). Therefore, the depth of the laminar layer will influence the vertical fluxes from the surface. • rat_sea (ratio of laminar scaling factor for heat over sea): Changing its value affects only temperature, as it has to do with laminar heat flux. • crsmin (minimal stomata resistance): This seems to be an important parameter for all results with its largest value producing better statistics. 11th COSMO General Meeting

11. Remarks (cont) • tur_len (maximal turbulent length scale): Changing its value affects the wind only during daytime. This parameter is used mainly in the calculation of the characteristic length scale for vertical mixing and thus into the calculation of the vertical transport momentum coefficient. • pat_len (length scale of thermal surface patterns): Changing its value affects only temperature. This was expected as this parameter is mainly used in the calculation of the heat flux. • c_lnd (surface area density of the roughness elements over land): affects mainly temperature but also slightly the wind values. 11th COSMO General Meeting

12. First discussion • tur_len (maximal turbulent length scale): This parameter is related to vertical rather than the horizontal transport. It is possible that if the vertical wind component were examined separately the effect would have been more evident.(?) Theoretically, an increased value of tur_len would lead to increasing the coefficient of vertical transport of momentum, which in turn may imply a more turbulent surface layer (mainly due to shear production of turbulent kinetic energy). • pat_len (length scale of thermal surface patterns): As this parameter is directly related to the horizontal sub-grid scales perhaps the value of 10000m that has been chosen for the perturbation (default=500) is too large for the horizontal grid size of the ensemble (=10km, i.e. equal to pat_len). Maybe a value less than 10000m is more appropriate to extract useful information for the sub-grid heat flux. (?) 11th COSMO General Meeting

13. New CONSENS suite • Random combination of parameter perturbation and initial model conditions. • The results are investigated in a different manner as no control run is available. • Reminder: The period is summer! 11th COSMO General Meeting

14. Major discrepancies for T2m 11th COSMO General Meeting

15. Major discrepancies for U10m 11th COSMO General Meeting

16. Major discrepancies for Td2m 11th COSMO General Meeting

17. Effects of physics perturbations rlam_heat: Scaling factor of the laminar layer depth While rlam_heat increases the depth of the laminar layer increases and produces the largest errors especially during the daytime underestimating T (suppresses the exchange of heat from the surface ?). 11th COSMO General Meeting

18. rlam_heat: Scaling factor of the laminar layer depth Although there are differences among the members, they do not obtain a clear behaviour. 11th COSMO General Meeting

19. rlam_heat: Scaling factor of the laminar layer depth While rlam_heat increases the depth of the laminar layer increases and produces the largest errors especially during the daytime underestimating largely Td (even greater than T). 11th COSMO General Meeting

20. rat_sea: Ratio of laminar scaling factor for heat over sea It seems that there is a diurnal pattern here with the largest rat_sea producing less error during daytime than the smaller value, while the opposite occurs during the night hours. 11th COSMO General Meeting

21. rat_sea: Ratio of laminar scaling factor for heat over sea It seems that the largest rat_sea produces less error for wind? 11th COSMO General Meeting

22. rat_sea: Ratio of laminar scaling factor for heat over sea It seems that the largest rat_sea largely underestimates Td especially during daytime (opposite than T and U). 11th COSMO General Meeting

23. crsmin: Minimal stomata resistance It seems that there is a smaller diurnal pattern here with the largest crsmin producing more error during daylight (photosynthesis and evapotranspiration hours), while the opposite occurs during the night hours. 11th COSMO General Meeting

24. crsmin: Minimal stomata resistance Although there are some differences among the members, they do not obtain a clear behaviour. 11th COSMO General Meeting

25. crsmin: Minimal stomata resistance It seems that the largest crsmin produces more error during daylight (photosynthesis and evapotranspiration hours). 11th COSMO General Meeting

26. tur_len: Maximal turbulent length scale The largest value of tur_len using AVN seems to produce larger errors during the daytime. 11th COSMO General Meeting

27. tur_len: Maximal turbulent length scale Not a clear effect although it seems that for the first forecast day the largest value of tur_len using AVN produces larger errorsduring the daytime. 11th COSMO General Meeting

28. tur_len: Maximal turbulent length scale Not a clear effect. 11th COSMO General Meeting

29. Remarks • It is difficult to identify the effect (small or not) of the combined physics perturbations altogether. • Some of the members of the new CONSENS suite produce large errors compared to the CSPERT suite. • The physical importance of the value of each perturbed parameter is apparent. Based on that it could be suggested • large values of rlam_heat produce large errors (independently of the crsmin and iconv). • rat_sea affects differently the parameters with its largest value producing less error during daytime than its smaller value (for T and U), while for Td its largest value leads to large errors. • the largest crsmin produces more error during daylight (photosynthesis and evapotranspiration hours) for T and Td. • the largest value of tur_len using AVN produces larger errors during the daytime for T and U although the effect is not as clear as the other perturbations (but it is difficult to study w). 11th COSMO General Meeting

30. Suggestions for next steps on physics perturbations Based on HNMS results: • The perturbation of rlam_heat, tur_len, crsmin should be further considered although the largest value of rlam_heat producing the worst results should be re-examined. • The perturbed values of pat_len should be carefully examined (consider smaller values than 10km?). • New physics perturbations? • The dissipation rate for turbulent kinetic energy (TKE) is defined by using a “tuning constant” cε ( d_heat or d_mom). Obviously this is an important parameter that affects the dissipation rate and therefore the consumption of TKE. Could we “touch” this value? 11th COSMO General Meeting

31. Developing perturbations for the lower boundary Task 2.2.1 techniques AIM Implement a technique for perturbing soil moisture conditions and explore its impacts Reasoning The lack of variability is typically worse near the surface rather than higher in the troposphere. Also soil moisture is of primary importance in determining the partition of energy between surface heat fluxes, thus affecting surface temperature forecasts 11th COSMO General Meeting

32. Soil Perturbation method Based on the method proposed by Sutton and Hamill (2004) • Select a period that provides variability in soil moisture e.g. spring • Use of data from a land–surface model analysis for the defined period for a few years in order to create some “climatology” • Implement the EOF (Empirical Orthogonal Function – Principal Component Analysis) to the data in order to generate random perturbations • Define the number of perturbations that will be initially used • Decide to which member of the CONSENS suite (control or other?) the method will be applied. 11th COSMO General Meeting

33. EOF analysis This analysis will generate random perturbations that will have the same spatial structure as the daily deviations of soil moisture from a running mean climatology. Based on the selected period data, a 30-day period running mean will be calculated. This mean will be subtracted from the daily soil moisture analysis to get daily deviations. Using these deviations to EOF, the perturbations will be calculated and ordered from largest to smallest variability. Where: εj is the jth perturbation ri is the standard normally distributed random number ui is the ith singular vector σi is the ith singular values The perturbations will be added to the soil moisture data to create the final perturbed field to be adopted in the ensemble run 11th COSMO General Meeting

34. 11th COSMO General Meeting

36. = - -- + ++ • Based on Spring case study

37. 2m Temperature 11th COSMO General Meeting

38. 10m Wind 11th COSMO General Meeting

39. 2m Dew point 11th COSMO General Meeting

40. CSPERT precipitation (deterministic approach) 11th COSMO General Meeting

41. CSPERT precipitation (probabilistic approach) • Small sample! 11th COSMO General Meeting

42. Analytical results of CSPERT for summer 2008

43. T2m – Summer 2008tur_len Maximal turbulent length scale 11th COSMO General Meeting

44. U10m – Summer 2008tur_len Maximal turbulent length scale 11th COSMO General Meeting

45. Td2m – Summer 2008tur_len Maximal turbulent length scale 11th COSMO General Meeting

46. T2m – Summer 2008pat_len Length scale of thermal surface patterns 11th COSMO General Meeting

47. U10m – Summer 2008pat_len Length scale of thermal surface patterns 11th COSMO General Meeting

48. Td2m – Summer 2008pat_len Length scale of thermal surface patterns 11th COSMO General Meeting

49. T2m – Summer 2008rat_sea Ratio of laminar scaling factor for heat over sea 11th COSMO General Meeting

50. U10m – Summer 2008rat_sea Ratio of laminar scaling factor for heat over sea 11th COSMO General Meeting