Insights into low- la.tude cloud feedbacks from high- resolu.on models

Transcription

1 Insights into low- la.tude cloud feedbacks from high- resolu.on models Christopher S. Bretherton University of Washington Department of Atmospheric Sciences Thanks to: Peter Blossey, Minghua Zhang, CGILS contributors SubmiDed to Phil Trans Roy Soc A. 1

2 Global cloud feedbacks IPCC AR5 SPM: The net radia.ve feedback due to all cloud types combined is likely posi.ve. - CMIP GCMs - Robust cloud feedbacks - Process understanding Soden and Vecchi 211, based on CMIP3

3 IPCC AR5 SPM: Uncertainty in the sign and magnitude of the cloud feedback is due primarily to con.nuing uncertainty in the impact of warming on low clouds. Do high- resolu.on (CRM and LES) limited- area and global models add insight into cloud feedbacks: In given regimes? Globally? IPCC AR5 3

4 Use of high- resolu.on models to study cloud feedbacks High- resolu.on = cloud- resolving (CRM) or large- eddy simula.on Simulate cloud- turbulence- convec.on- precipita.on interac.ons with less subjec.ve subgrid parameteriza.ons than GCMs. Easier to compare with small- scale cloud observa.ons Help iden.fy cloud response mechanisms to climate change, Help test plausibility of possible emergent constraints. But simula.on of full global to turbulence range of scales over climate.mescales is computa.onally infeasible. This talk looks at: LES of subtropical low cloud responses to idealized imposed climate change Local and global cloud- resolving simula.ons of deep convec.ve response 4

5 Large- eddy simula.on of low cloud climate- change response LES with sufficiently fine grids simulate cloud- topped boundary layers more accurately and robustly than GCM parameteriza.ons LES domain is a few km on a side; large- scale drives the LES Hence, use LES to explore cloud response to prescribed large- scale changes expected from GCMs, and compare with available observa.ons and the GCMs themselves. Do LES support the GCM consensus of less subtropical marine low cloud in a greenhouse- warmed climate? If so, do they provide a physical explana.on for this? 5

6 CGILS LES cloud feedback intercomparison (Blossey et al. 213) Mean summer.me forcing at 3 NE Pacific loca.ons: S12: Shallow coastal stratocumulus (Sc) S11: Decoupled Cu rising into Sc S6: Shallow Cu LES control run 1 d to near steady- state with diurnal- mean insola.on, 1 cm - 3 cloud droplet conc. LES rerun to steady state with climate perturba.ons added to forcing (temperature, subsidence, CO 2 etc.) Difference in cloud, BL gives climate change response S6 S11 S12 CGILS forcings:zhang et al. 212

7 S12 Cloud Response MISR Cloud Top Height PDF Courtesy of J. Karlsson. Δx = Δy = 25 m, Δz = 5 m at inv. P2: Cloud thins with little change in inversion height or entrainment. P2S: Cloud layer thickens as inversion deepens in response to weakened subsidence.

8 Different LES give similar cloud responses DALES LaRC MOLEM MOLEMA SAM SAMA UCLA 1 a) b) c) d) e) f) g).8 z, km cloud frac.5 1 cloud frac.5 1 cloud frac 8-1 day means.5 1 cloud frac.5 1 cloud frac.5 1 cloud frac CTL P2 P2S.5 1 cloud frac 1 h) i) 95 j) S12 k) l) m) n).8 9 z, km w w, m 2 s 2 Inversion Height, m P2S w w, m 2 s 2 w w, CTL m 2 s P2 w w, m 2 s 2 Blossey et al. 213 JAMES.2.4 w w, m 2 s 2 DALES LaRC MOLEM MOLEMA SAM.2.4 w w, SAMA m 2 s 2 UCLA.2.4 w w, m 2 s SWCRE, W m 2

9 dcmip3 forcing changes GILS S12: UW LES Sensitivity Studies S12 CGILS Scaled S12: by CMIP3 UW LES Composite Sensitivity Changes Studies Inversion Height, m P2S Perturba.on δsst δω(5 hpa) δeis δrh CTL δ(wind speed) 2SFT P2 dws dcmip3 2.5 ±.5 K - 5 ± 3 %.6 ±.2 K ± 1 % drh ± 1.5 CTLD % 4xCO2 P2D Cloud P2S reduc.ons from ΔSST, ΔCO 2 overwhelm increase from Δω è posi.ve dcmip3 low cld feedbk P2SFT EIS+.5K dc3 SWCE, W m 2 SWCE, W m 2 65 CGILS S12: LES Intercomparison CTL SWCRE, W m 2 4x Inversion Height, m SAMA DALES LaRC MOLEMA UCLA CTL CTL ω 5% drh SST+2.5K dcmip3 2xCO2 P2 WS 1.5% RH 1.5% dws 4xCO2 CTLD P2D Bretherton et al. 213 JAMES 9

10 Inversion Height, m 16 CGILS S11: UW LES Sensitivity Studies S11: Decoupled P2S Cu under Sc drh 145 CTL P2 14 dws Inversion Height, m CGILS Scaled S11: by UW CMIP3 LES Sensitivity Composite StudiesChangesScaled b Cloud reduc.ons P2S from ΔSST, ΔCO 2 overwhelm increase from Δω, ΔEIS è posi.ve dcmip3 low cld feedbk drh EIS+.5K WS 1.5% CTL P2 CTL dws ω 5% SST+2.5K C RH 1.5% dcmip3 2xCO2 Inversion Height, m P2SFT 4xCO SWCE, W m Δx = Δy = 5 m, Δz = 5 m at inv. CGILS S6: UW LES Sensitivity Studies S6: Cu P2S drh CTL P2 4CO2 P2SFTh dws P2SN25 N25 Inversion Height, m P2SFT 4xCO SWCE, SWCE, W m 2 W m CGILS Scaled S6: by UW CMIP3 LES Sensitivity Composite StudiesChangesScaled b P2S dcmip3 ω 5% CTL drh 2xCO2 SST+2.5K CTL P2 RH 1.5% EIS+.5K 4CO2 WS 1.5% P2SFTh dws Cloud reduc.ons from P2SN25 ΔSST è weak posi.ve dcmip3 fdback Δx = Δy = 1 m, Δz = 4 m at inv. SWCRE, W m 2 SWCRE, SWCRE, W m 2 W m 2 N25

11 Radia%ve( Mechanisms of Sc Cloud Response More%emissive%FT% (more%co 2 %or%h 2 )% Less%turbulence% produc<on%by%top% longwave%cooling.% Less%entrainment.% Sc%lowers,%thins.% % Thermodynamic( warmer%sst% or%drier%rh% Larger%surface% %FT% moisture%difference% allows%thinner%cloud% to%sustain%same% entrainment.%% Sc%thins.% Dynamic( Inversion(strength( Observa(onal support: Thermodynamic: Qu et al. 213, Clim Dyn Radia.ve: Christensen et al. 213, JAS Dynamic: Myers and Norris 213, J Clim Inversion strength: Klein and Hartmann 1993, J Clim Less%subsidence% Sc%top%rises.%More% entrainment%lies% cloud%base.%% Sc%thickens.% FT%warms%more%than%SST% Stronger%inversion% reduces%entrainment.% Sc%top%and%base%lower.%% Sc%thickens.% Bretherton et al. 213 JAMES

12 a) (Bretherton and Blossey 214) z, km GASS Lagrangian case z, km z, km.3 c).1 1 slight Sc thinning vs. CTL SWCRE = -63 W m-2 P4 2 z, km 1 SWCRE = -83 W m-2 4CO2 1 d) more Sc thinning & breakup SWCRE = -45 W m-2 P4CO z, km b) 2 Entrainment liquid flux feedback Mechansim for thermo Sc thinning cloud fraction 1 Δx =Δy = 35 m, Δz = 5 m, 4.5x4.5 km doubly- periodic diurnal insola.on Nd = 1 cm- 3 SWCRE = -92 W m-2 CTL 2 e).3 SWCRE = -93 W m-2 EIS 1 most Sc thinning & breakup no Sc thinning time, day Like CGILS, Sc cloud thins in 4CO2, more in P4, most in P4 4x. deis simula.on recovers CTL cloud due to stronger inversion.

13 dsc/dsst across inversion jump phase space Dussen et al. 215 JAMES Idealized steady Sc BLs with different inversion jumps. Apply 2 K ΔSST w fixed RH + All cases show Sc thinning, consistent with CGILS thermodynamic mechanism 13

14 Rieck et al. (212 JAS) Idealized nonprecipita.ng BOMEX trade Cu. Slight cloud cover decrease with warming; more pen. entrainment 14

15 How about deep convec.ve response to ΔSST in CRMs? Limited- area simula.ons of radia.ve- convec.ve equilibrium (Tompkins and Craig 1999 JClim) RCE for SST = 298, 3, 32 K; 45 days, 6 x 6 km x 21 km, Δx = 2 km, L35. Clouds rise following isotherms in a warmer climate with very slight reduc.on in horizontal extent. Kuang and Hartmann (27, J Clim) found similar results (FAT è posi.ve cloud height feedback) increases (Tompkins and Craig 1999) 15

16 NICAM global cloud- resolving model (Tsushima et al. 214 JAMES) Δx = 14 km for 9 days/δx = 7 km for 3 days; 4 ver.cal levels. With specified SST increase, large tropical cirrus increase è strong posi.ve longwave cloud feedback, Results are sensi.ve to subgrid turbulence parameteriza.on (MYNN), snow fall speed, Δx è uncertain.es in GCRM, too! 16

17 SP- CCSM4 superparameterized climate model Latitude [deg N] a) 4x: global mean =.49 W m 2 K 1 Abrupt 4xCO 5 2 : ECS of 2.8 K and global cld feedback in CMIP5 5 range Posi.ve SW 3 feedback mainly due to low cloud decrease over land Latitude [deg N] 9 Low- lat marine subtropical Sc feedback slightly nega.ve (opposite of LES! under- resolu.on?) b) 4x: global mean =.3 W m 2 K 1 4x: global mean =.19 W m 2 K 1 c) Bretherton 9 et al. 214 JAMES Net Cld Fdbk [W m 2 K 1 ] LW Cld Fdbk [W m 2 K 1 ] 1 ] 17

18 Comparison of GCM and hi- res cloud feedbacks Sc (mixed) Sc (decoupled) Cu (precip) Sc-Cu trans Cu (nonprec) RCE NICAM GCRM SPCAM3 SPCCSM4 LES/CRM cloud feedback synthesis LES (local) CRM (local) CRM (global) CMIP3/5 GCM C l C s Cloud feedback strength [W m -2 K -1 ] 18

19 Synthesis: LES and CRM of cloud response to climate change Do LES support the GCM consensus of less subtropical marine low cloud in a greenhouse- warmed climate? Yes Do they provide a physical explana.on for this? Yes (thermodynamic and radia.ve mechanisms, but compe.ng with ΔEIS, Δw) Limited- area and global CRMs suggest that deep convec.on has a posi.ve cloud height feedback, show no evidence for an iris effect, and show insensi.vity of associated shallow Cu to SST. These results support the GCM consensus of posi.ve low- la.tude cloud feedbacks on climate change. However, high- resolu.on models have not yet constrained the CMIP5 GCM- simulated range of global cloud feedbacks.