1. © Crown copyright Met Office Diurnal cycle Land-Atmosphere Coupling Experiment (DICE) GLASS / GASS joint project Martin Best and Adrian Lock

2. © Crown copyright Met Office Outline of the 3 stages of DICE LSM and SCM stand-alone performance against observations What is the impact of coupling? How sensitive are different LSM and SCM to variations in forcing?

3. © Crown copyright Met Office CASES-99 case study 23-26 October 1999 Field experiment in Kansas, USA We follow Steeneveld et al (2006) 3 day simulation from 2pm local time on 23 rd October 1999 Recall GABLS II ran for from 2pm on 22 nd for 2.5 days Clear skies throughout Gives 3 nights of varying character intermittent turbulence continuous turbulence very stable, almost no turbulent fluxes

4. © Crown copyright Met Office Experimental protocol LSM Soil spin-up: 9 years from saturated using WATCH forcing data 10 th year forcing data from local site Two stage 1a experiments with forcing from 2m and 55m Stage 3a LSM experiments forced with stage 1b SCM data interpolated to 20m SCM Large-scale forcing: Time-varying geostrophic wind (uniform with height) Large-scale horizontal advective tendencies for T, q, u, v estimated from a simple budget analysis of the sondes Subsidence for T, q No relaxation Radiation switched on in all simulations SCM in stage 1b use observed sensible and latent heat fluxes and u * (either directly or via c D ) Stage 3b SCM experiments forced with stage 1a LSM surface fluxes

5. © Crown copyright Met Office Participating Models ModelContactInstitute Levels Sensitivity tests Arome & Arpege (NWP) Eric Bazile Meteo France 60/70 Resolution, soil Arpege (CMIP5) Isabelle Beau Meteo France ECEARTH Reinder Ronda, Bert Holtslag Wageningen University 91 GDPS3.0 Ayrton Zadra CMC 79 Surface properties GFDL Sergey Malyshev, Kirsten Findell Princeton/GFDL 24 GISS_E2 Ann Fridlind, Andy Ackerman GISS 40 WRF (IAP) Bingcheng Wan IAP 119 IFS/HTESSEL Irina Sandu, Gianpaolo Balsamo ECMWF 137 LAI LMDZ, ORCHIDEE Sonia Ait-Mesbah, Marie-Pierre Lefebvre, Frederique Cheruy LMD 70 MESO_NH Maria Jimenez, Patrick LeMoigne, Joan Cuxart IMEDEA, Meteo France, UIB 85 Bare soil UM/JULES Adrian Lock, Martin Best Met Office 70 Vegetation NCEP Weizhong Zheng, Mike Ek NOAA 65 z0 WRF-NOAH Wenyan Huang, Xinyong Shen, Weiguo Wang NUIST 60 Many WRF Wayne Angevine NOAA 119 PBL scheme CAM5, CLM4 David Lawrence, Ben Sanderson NCAR 26

6. © Crown copyright Met Office A challenging surface? October grass was largely dead Rain in September left soil moist Excessive evaporation a feature of the first round of DICE Google streetview Courtesy of Joan Cuxart

7. © Crown copyright Met Office Stage 1a Surface fluxes from 55m-forced LSMs Round 1 data Round 2 data Remember these will be the SCM surface fluxes in Stage 3b Not all LSM provided u * (not compulsory under ALMA convention)

8. © Crown copyright Met Office SCM grids Solid lines = control model Dotted/dashed lines = experiment Lowest grid-levels range from 1.5m to 85m

9. © Crown copyright Met Office Stage 1b SCM forced by observed surface fluxes

10. © Crown copyright Met Office Stage 1b summary (from October workshop) Simulations successfully completed SCM can be forced by observed fluxes and stresses Overall doing a reasonable job Stable 3rd night not stable enough Lots to look at SCM forcing issues: I‘m not happy with the wind forcing Can geostrophic wind be set better? simply use 1-3km average instead of below 3km? Remove fine-scale structure in wind ICs and forcing? Subsidence slightly too weak? Check cirrus on 26 th (is it spurious?) Check SW TOA between models Check temperature budgets for all models What should be happening with near surface night-time moisture (LHF~0)? Tending to ignore 26 th Improved since round 1 See entrainment budgets later Still ? No change to forcing

11. © Crown copyright Met Office Stage 1b near surface evolution SCM driven by observed surface fluxes 20m55m

12. © Crown copyright Met Office Stage 1b to 2 Impact of coupling to surface

13. © Crown copyright Met Office Stage 1b vs 2 Bulk PBL sensitivity (variables at 55m) More spread between coupled models in stage 2 than stand-alone SCM in stage 1b More degrees of freedom Moisture more sensitive than temperature? θ 55m q 55m Stage 2 θ 55m q 55m Stage 1b

14. © Crown copyright Met Office Stage 1b vs 2 Bulk PBL depth sensitivity Some suggestion that PBL depth is less sensitive when coupled (especially in the evening) Stage 1b Stage 2 PBL depth calculated as where Ri B =0.25

15. © Crown copyright Met Office Stage 1a (55m) Impact of coupling on winds

16. © Crown copyright Met Office SBL sensitivity in stage 2 Models with stronger evening cooling in T 2m have Weaker sensible heat flux Stronger soil heat flux Consistent with stronger T gradients GABLS3 (Bosveld et al, 2014) sensitivity tests suggest implies similar soil conductivity?

17. © Crown copyright Met Office Stage 3b variability Perspective from the atmosphere 1) Daytime

18. © Crown copyright Met Office Stage 3b daytime sensitivities All models have more variability in the PBL moisture than temperature But (by 25 th) ) SCM differ more for theta than q How strongly is this driven by the surface fluxes? 24 th and 25th Flavours of WRF

19. © Crown copyright Met Office Stage 3b temperature sensitivity to SHF Typically weak correlation between daytime PBL temperature and surface sensible heat flux

20. © Crown copyright Met Office Stage 3b moisture sensitivity to LHF Much stronger correlation between daytime PBL moisture and surface latent heat flux

21. © Crown copyright Met Office Controls on daytime PBL Why should q correlate more with LHF? or T less with SHF? SHF dominates deepening of PBL through entrainment Stronger SHF implies greater PBL warming but also entrainment and PBL deepening but that implies further enhanced warming so still ought to correlate Look further at differences in PBL growth…

22. © Crown copyright Met Office Daytime sensitivity: θ profiles (1400 CDT 25th) NUIST warm outlier

23. © Crown copyright Met Office Daytime sensitivity: q profiles (1400 CDT 25th)

24. © Crown copyright Met Office Stage 3b sensitivities Some models have much greater sensitivity in PBL depth than others Diagnosed consistently using Ri B =0.25 24 th and 25th

25. © Crown copyright Met Office Does PBL depth depend on EF? Not really!

26. © Crown copyright Met Office Does PBL depth depend on surface buoyancy flux? In most models, yes

27. © Crown copyright Met Office Entrainment sensitivity Estimate entrainment fluxes from PBL θ v budget Integrate over the boundary layer: Rearranging gives: Horizontal advection and radiation (~small) I’m ignoring vertical advection (small in PBL)

28. © Crown copyright Met Office Variability in entrainment buoyancy flux Note A ent <0 implies PBL warming less than expected from surface heating+forcing Could imply surface heating distributed above h pbl Hence sensitive to definition of h pbl “Expected value” ~ 0.2 Generic h pbl (Ri B ) Models’ internal h pbl

29. © Crown copyright Met Office Variability in entrainment buoyancy flux Note A ent <0 implies PBL warming less than expected from surface heating+forcing Could imply surface heating distributed above h pbl Hence sensitive to definition of h pbl “Expected value” ~ 0.2 Generic h pbl (Ri B ) Models’ internal h pbl Take SCMs 3 and 10 on 25 th as examples

30. © Crown copyright Met Office Examples: GISS and NUIST (Exp2=MYNN level 3) Models’ h pbl are significantly higher than Ri B predicts Using the higher level adds warming within the inversion to the mixed layer budget and gives a larger A ent Generic h pbl (Ri B )Models’ internal h pbl

31. © Crown copyright Met Office Variability in entrainment buoyancy flux Met Office SCM has A ent =0.2 with very little variability Slight odd because the code sets A ent =0.23 and also includes a contribution from u * “Expected value” ~ 0.2 Generic h pbl (Ri B ) Models’ internal h pbl

32. © Crown copyright Met Office Entrainment sensitivity to u * Sensitivity to u * goes the wrong way!

33. © Crown copyright Met Office Variability in moisture entrainment Entrainment flux of moisture depends on PBL to free-troposphere moisture difference Most models (not NUIST) have mixed out dry slot above inversion on 24 th Hence moist air is entrained and A entq > 0 On 25th free-atmosphere is robustly drier and A entq is more similar (<0) Use models’ internal h pbl

34. © Crown copyright Met Office Stage 3b Stable boundary layers

35. © Crown copyright Met Office Stable boundary layer As in daytime, more spread in moisture than temperature SCM fairly consistent, particularly for moisture 23 rd, 24 th and 25 th

36. © Crown copyright Met Office Stable boundary layer More negative SHF generally implies colder 50m temperature Heat lost to surface

37. © Crown copyright Met Office Stable boundary layer Larger u * implies warmer 50m temperature on 24th (+) More turbulent SBL with more mixing of heat downwards

38. © Crown copyright Met Office Remaining data issues Some stage 3b files do not have correct u * forcing Were given observed u* by (my) mistake Can anyone face rerunning stage 3b again? Could all groups provide u* and Tskin from LSMs? Notes for future intercomparisons Decide on grid indexing from surface or TOA Height as height above surface (not sea-level) Make sign convention clear

39. © Crown copyright Met Office Potential discussion points Stage 1a: should be straightforward to understand reasons for excessive u* in some LSMs Simply excessive z 0 ? If so why (eg canopy height, LAI)? Stage 1a: u* distributions very different (fall into 2 groups) Could all groups provide u* from LSMs Stage 1a: some LSMs have different daytime fluxes forced by 2m and 55m Similarity theory not holding? Stage 2 coupling: some positive some negative feedbacks on surface fluxes (cf stage 1a) Stage 2 coupling: net LW enhanced by day (increased upward LW) with less SHF  Change in near surface T gradients? Interesting soil heat flux “hysteresis” between day and night in some models Why this difference and what is the impact? Stage 1b,3b: why do some SCM mix out the inversion? Dependent on parametrization structure (eg non-local vs EDMF vs higher order)? Does it matter? Affects PBL budget so can effects be seen in going from stage 1a 

40. © Crown copyright Met Office Next steps Further iteration (eg corrected u* in stage 3b)? Ensure everyone has submitted all data (eg u* and Tskin, 2m and 55m forcing datasets) Start on writing intercomparison papers Overview paper on intercomparison stages 2nd paper on detailed LSM analysis and coupling? 3rd paper on SCM sensitivities (eg entrainment, SBL)? Special issue for DICE related studies? Make intercomparison model data (and CASES 99 obs) available for others to do more analysis Clean up datasets first?