Expectations for the responses - PowerPoint Presentation

1. Expectations for the responses of low clouds to aerosolsRobert Wood, Atmospheric Sciences, University of Washington

2. MotivationAerosol impacts on clouds are not simply explained by Twomey’s argumentsChanges in macrophysical cloud properties produce radiative impacts of same order as those from Twomey (e.g. Lohmann and Feichter 2005, Isaksen et al. 2009)Conceptually, it is useful to divide AIEs into two types: primary or quasi-instantaneous effects (e.g. Twomey effect, dispersion effect);effects that require an understanding of the system feedbacks on timescales comparable to or longer than the cloud element lifetime.

3. Theoretical expression for AIEResponse of cloud optical thickness t to change in some aerosol characteristic property A primary feedback (secondary)

4. (Mostly) regulating feedbacks in stratocumulus

5. Twomey’s hypothesis Increases in the number of aerosol particles will lead to increases in the concentration Nd of cloud dropletsFor a given LWC, greater Nd implies smaller droplets (since droplet radius r  {LWC/Nd}1/3)Greater Nd  total surface area will increase ( Nd r2h, so   Nd1/3h5/3) and clouds reflect more solar radiationd(ln)/d(lnNd) = 1/3

6. Albrecht’s hypothesis (1989)A greater concentration of smaller drops (Twomey) suppresses precipitation because the coalescence efficiency of cloud droplets increases strongly with droplet size.Reduced precipitation leads to increased cloud thickness, liquid water content, coverage  more reflective cloudsfrom Wood (2005)cloudbase drizzle rate [mm d-1]Nd [cm-3]

7. TwomeyAlbrecht

8. Model estimates of the two major aerosol indirect effects (AIEs)Pincus and Baker (1994) – 1st and 2nd AIEs comparableGCMs (Lohmann and Feichter 2005) 1st AIE: -0.5 to -1.9 W m-2 2nd AIE: -0.3 to -1.4 W m-2 Large scale models problematic, so need observational and process model constraints

9. Shiptrack surprises!Liquid water content in shiptracks is typically reduced compared with surrounding cloudClear refutation of Albrecht’s hypothesiscourtesy Jim Coakley, see Coakley and Walsh (2002)3.7 m

10. Precipitation suppression in ship tracksChristensen and Stephens (submitted JGR) First observed in MAST (Ferek et al. 2000, JAS) CloudSat brings new perspective: Marked suppression of drizzle in most shiptracks in closed cellular Sc. Enhancement of precipitation in some open cell cases Poor data below 800 m Reliable statistics?CloudSat

11. LES resultsCloud droplet concentration [cm-3]LWP [g m-2]P0 [mm d-1]we [cm s-1]Impact of aerosols simulated by varying NdIncreased Nd  Reduced precipitation  increased TKE  increased entrainment weChanges in we can sometimes result in cloud thinning (reduced LWP)Also noted by Jiang et al. (2002)Ackerman et al. (2004)

12. Precipitation reduces TKE

13. short timescaleslong timescalesshort timescales

14. Different processes, different timescalesRIE - ratio of secondary to primary AIEs short term thinning

15. Transient response of an equilibrated mixed layer PBL model to Nd increases Ratio of 2nd to 1st AIE RIE (right) is a strong function of cloud base height More elevated cloud base heights zcb lead to Albrecht effects which partly cancel those due to Twomey effect Elevated zcb associated with dry FT and less surface drizzle, consistent with LES results, but with a far less sophisticated model  hope for the representation in climate modelsWood (J. Atmos. Sci., 2007), also seen in LES (Chen et al. 2011)2nd AIE cancels 1st2nd AIE = 1st2nd AIE = 0

16. Sedimentation of cloud dropletsBretherton, Blossey and Uchida, GRL, 2007Cloud droplet sedimentation removes water from the (10 m thick) entrainment interface, lowers LWC there, reduces evaporative cooling, and suppresses entrainment, resulting in thicker clouds Since increased Nd reduces sedimentation  pollution can lead to thinner clouds

17. Entrainment enhancement raises inversion/cloud top height....in some casesChristensen and Stephens, J. Geophys. Res. (2011).....also Taylor and Ackerman, QJRMS, 1999 (MAST Sanko Peace case study)

18. Marked deepening appears to occur only in previously collapsed MBLssurrounding cloud remains radiatively activelittle or radiatively inactive surrounding cloudChristensen and Stephens, J. Geophys. Res. (2011)

19. LES modeling of open-closed cell boundaryBerner et al. (2011, ACP), and Bretherton et al. (2010, JAMES)Wang and Feingold (2009), Wang et al. (2010)Nd=10 cm-3Nd=60 cm-3secondary circulation above MBLinversion rises at samerate in both regions

20. (Mostly) regulating feedbacks in stratocumulus

21. Response pathwaysWoodAlbrechtAckermanWoodArnason and Greenfield (1972), Kogan and Martin (1995), Lee et al.(2009)

22. ProspectsTurbulence, entrainment and drizzle absolutely central to how STBL clouds respond to aerosol injectionsWell-defined and testable hypotheses exist for how the STBL should respondProcess models becoming capable of handling all relevant aerosol-cloud interactionsNew in-situ and spaceborne measurements offer tremendous possibilities to go beyond previous studies (e.g. airborne doppler radar and lidar  turbulence/drizzle)

23. A proposalA limited area perturbation experiment to critically test hypotheses related to aerosol indirect effectsCost $30M

24.

25. Mixed layer modelLW/SW radiation and bulk surface flux (LHF/SHF) parameterizationsEntrainment closure (Turton and Nicholls 1986) Precipitation: For standard runs use formulation derived from shipborne radar observations in SE Pacific stratocumulus (Comstock et al. 2004). – cloud base precipitation PCB  h3.5/Nd1.75 – treatment of evaporation below cloud to give surface precipitation

26. Effects of drizzle vs effects of sedimentation of cloud droplets GCSS DYCOMS-2 RF02 drizzling Sc case studyAckerman et al. (MWR, 2009)With drizzle, without sedimentationWith drizzle and sedimentationEffect of drizzle Effect of sedimentation.....in this case, sedimentation dominates over drizzle impact on cloud LWP

27. Indirect effect ratio RIE 1st AIE 2nd AIE Define RIE = 2ndAIE / 1st AIE Relative strength of the Albrecht effect compared with Twomey For adiabatic cloud layers,   Nd1/3 LWP5/6

28. Suite of simulationsSurface divergence [10-6 s-1]: {2, 3, 4}Sea Surface Temp. [K]: {288, 292, 296}Moisture above MBL [g kg-1]: {1, 3, 6}700 hPa potential temperature set to 312 KNo advective termsMLM is run to equilibrium twice:(control) Nd=Nd,control(perturb) Nd=1.05Nd,controlRIE is calculated – examine dependence of RIE upon forcings and parameterizationssee Wood (2007), J. Atmos. Sci., 64, 2657-2669.

29. Base case, Nd,control=100 cm-3For most forcing conditions 2nd AIE > 1st AIERIE scales with surface precipitation in the controlLittle dependence of scaling upon forcing conditions

30. Base case, Nd,control=200 cm-3Lower values of RIE because surface precip. is lowerSame RIE scaling with surface precip

31. Different drizzle parameterizationsBASE (Comstock)VanZanten et al. (2005)

32. Fixed entrainmentOnly surface moisture/energy budget important (Albrecht effect)Entrainment important in determining the nature of the feedback response

33. Non-equilibrium responseTimescale for 2nd AIE is long – due to long zi adjustment timescaleOn short timescales RIE can be negative (noted in Ackerman, 2004) Important to understand timescales of aerosol evolution

34. Timescales But what is the timescale N for evolution of Nd? N=Nd {dNd/dt}-1Coalescence scavenging (removal of CCN by coalescence of cloud/drizzle drops): N  (cloudbase precip rate)-1 ≈ 0.5-1 days Coalescence scavenging timescale is comparable to entrainment drying timescale  aerosol perturbation is largely removed before MBL reaches equilibrium

35. Short timescale cloud responseCloud base height determined by a balance between surface precipitation moistening (P) and entrainment drying (E)Derive expression for cloud thickness change dh/dt ≈ - dzCB/dt using moisture and energy budgets for MBL  is relative importance of entrainment drying compared with surface precipitation moistening

36. What determines  ?Ackerman showed RHFT importantBut cloudbase height dominates over wider range of phase space

37. Annual mean LCL 400-600 m over much of the subtropical and tropical oceans  cancellation of aerosol indirect effects?

38. Sensitive to size of drizzle dropsReducing mean radius of drizzle drops leads to more evaporation, and different ratio of surface moistening and entrainment drying Representation of evaporation criticalrdriz=49 mrdriz=65 m

39. SEP stratocumulus in GCMsPoor representation of the vertical structure of stratocumulus-topped boundary layers – surface moisture budget is completely out to lunchBretherton et al. 2004, BAMS

40. Analogies in trade cumulus cloudsaerosol concentration [cm-3]CFLWPXue and Feingold (2006)LES model of trade CuBoth cloud fraction (CF) and LWP decrease with increasing CCN conc.Effect attributed to more rapid evaporation of smaller cloud droplets (higher N) during entrainment events resulting in more rapid cloud dissipation

41. ConclusionsRelative strength of 2nd AIE strongly dependent upon balance between precipitation suppression moistening and entrainment dryingRIE reduced by ~50% by changing drizzle parameterization  need to understand climatology of precipitation and its dependency on LWP and NdOver timescales comparable with aerosol lifetime in the MBL, 1st and 2nd AIEs may cancel – implications for sensitivity of low clouds to aerosolsUnlikely that current global models can capture the essential physics (evaporation/entrainment)

42. Shiptracks only seen in shallow boundary layersDurkee et al. (2000)

43. Durkee et al. (2000)Track width as a function of age

44. CloudSat observes drizzleSE Pacific

45.

46. The End

47. Low clouds in climate models- change in low cloud amount for 2CO2from Stephens (2005)GFDLCCM model number

48. Re-examination of Klein and Hartmann data

49. Williams et al. (2006)Change in LTS (K)Low cloud amount in an ensemble of 2xCO2-control GCM simulations is poorly estimated using LTS’ (for which a general increase is predicted)Much better agreement with change in saturated stability (≈EIS’)

50. Precipitation parameterizationsBASE: Comstock et al. (2004): PCBh3.5/Nd1.75Van Zanten et al. (2005): PCBh3/NdRange typical in MBL clouds

51. Weak temperature gradient

52. Minimalist approach

53. No entrainment drying/warmingEntrainment only allowed to influence zileads to stronger LWP feedback

54. Doubling and halving entrainment efficiency Enhanced entrainment counteracts (by drying) the increased LWP caused by reduced precip. efficiency… ….but data fall on same curve 2 /2

55.

56. Cloud feedbacks remain the largest uncertainty in the prediction of future climate changefrom Cess et al. 1989, 1996

57. Sensitivity of cloud optical depth  to increasing Nd in the MLM 1st AIE 2nd AIE constant LWP feedback on LWP For the MLM,   Nd1/3 LWP5/6

58. Indirect effect ratio RIE 1st AIE 2nd AIE Define RIE = 2ndAIE / 1st AIE Relative strength of the Albrecht effect compared with Twomey

59. Relationship between LTS and EIS is not uniqueFor a given value of LTS, EIS decreases with surface (or 700 hPa) temperature

60. (2) Aerosols and low cloudsCloud droplet concentrationWood et al. (2006)Chiquicamata, Chile

61. Primary effect (Twomey)Seminal papers in 1974, 1977 hypothesizing an important potential brightening of clouds subject to increased aerosol concentrationImportance of fixed cloud macrophysical properties (LWP) – difficult to test in practice

62. Aerosol-cloud microphysical observationsObserved cloud droplet concentration [cm-3]Predicted droplet concentrationfrom aerosol spectrum [cm-3]From Twomey and Warner, J. Atmos. Sci., 24, 704-706, 1967Aerosol concentration [cm-3]Cloud droplet concentration [cm-3]From Martin et al., J. Atmos. Sci., 51, 1823-1842, 1994….first measurements (1967)….recent measurements (1994)+ wind from land wind from sea

63. Aerosol loading and cloud dropletradius….….35 years onfrom Breon et al. 2002, Science, 295, 834-838 6 8 10 12 14Cloud droplet radius [micron]0.0 0.1 0.2 0.3 0.4 Aerosol index

64. Remote sensing estimatesfrom Feingold et al. 2003, GRLIE = dlnre/dlna  dln/dlnNa = (dln/dlnNd)(dlnNd/dlnNa)(dlnNd/dlnNa)  0.5 to 0.7Expect IE =1/3(0.5 to 0.7)  0.17 to 0.23IE (Observations)0.02-0.16 (Feingold et al.)0.085 (Breon, ocean)0.04 (Breon, land)0.06 (Nakajima et al., ocean)