Preliminary Simulations of the CFMIP/GCSS LES Intercomparison andColumn Modeling of Sc Feedbacks in SP-CAM
Peter Blossey and Chris Bretherton (Univ. of Washington, USA)
with help from Matt Wyant and funding from CMMAP/NSF.Thanks to Marat Khairoutdinov for SP-CAM runs, SAM and his drizzle scheme.
Wednesday, June 10, 2009
Outline
• Column modeling of stratocumulus cloud feedbacks in SP-CAM:
• Background on column modeling approach, LTS-binning.
• Focus on stratocumulus region composite using 90-100th LTS percentiles.
• Column model results: sedimentation, diurnal cycle.
• Negative cloud feedbacks with most configurations.
• Preliminary simulations of GCSS/CFMIP LES intercomparison (s11 case):
• Temperature drift above inversion allows additional BL deepening.
• Omega feedback
• Drizzle feedback reverses sign of ΔSWCF.
• Boundary layer structure and cloud feedback are sensitive to configuration.
Wednesday, June 10, 2009
Column Modeling Approach
• Large-scale context for column model runs is extracted from GCM runs of control and perturbed climates. GCM is SP-CAM. CRM/LES is SAM 6.5, 6.7.
• Large-scale context: SST, soundings, omega, large-scale horizontal advection.
• Monthly SP-CAM output over tropical oceans composited based on deciles of lower tropospheric stability (LTS).
• Column LES/CRM use omega-feedback to simulate the effect of stratified adjustment in the tropics, limits T drift from composite soundings.
• Old work: Column LES/CRM simulations based on SP-CAM’s 70-80th & 80-90th percentile composites based on LTS.
• New work: LTS 90-100 from SP-CAM, CFMIP LES intercomparison Months per year in given LTS decile
Cu
Cu-Sc
Sc
Wednesday, June 10, 2009
Omega Feedback
• Simulates effect of stratified adjustment: local temperature anomalies are removed by large-scale vertical motion.
!
!p
!
f2 + a2m
am
!"!
!p
"
!
k2Rd
pT !
v
• Here, Tv’ and ω’ are anomalies from specified profiles.
• A lengthscale (which fixes k) and a profile of momentum damping rate (am) must be chosen.
200
400
600
800
1000Tv’
p, h
Pa
200
400
600
800
1000!’
Wednesday, June 10, 2009
Column Modeling of Composite SP-CAM Trade Cumulus Regimes (Blossey et al. JAMES to appear)
• Column CRM/LES simulations using composite forcings from SP-CAM’s 70-80th and 80-90th percentiles of lower tropospheric stability (LTS).
• CRM/SP-CAM: Δx=4km, Nz=30. LES: Δx=50-400m, Δz=20-160m in BL.
• Column CRM’s cloud climatology and +2K cloud response for 70-80th and 80-90th deciles of LTS is broadly similar to SP-CAM.
• Column LES has less cloud and weaker SWCF & ΔSWCF. ΔSWCF<0 for LTS 80-90, ΔSWCF~0 for LTS 70-80.
• Diurnal cycle important.
• CRM/SP-CAM: Radiatively-driven Cu increase.
• LES: more Sc under stronger inversion in +2K runs.
Wednesday, June 10, 2009
• Column LES has less cloud and weaker SWCF & ΔSWCF. ΔSWCF<0 for LTS 80-90, ΔSWCF~0 for LTS 70-80.
• Diurnal cycle important.
• CRM/SP-CAM: Radiatively-driven Cu increase.
• LES: more Sc under stronger inversion in +2K runs.
• Column CRM/LES simulations using composite forcings from SP-CAM’s 70-80th and 80-90th percentiles of lower tropospheric stability (LTS).
• CRM/SP-CAM: Δx=4km, Nz=30. LES: Δx=50-400m, Δz=20-160m in BL.
• Column CRM’s cloud climatology and +2K cloud response for 70-80th and 80-90th deciles of LTS is broadly similar to SP-CAM.
Column Modeling of Composite SP-CAM Trade Cumulus Regimes (Blossey et al. JAMES to appear)
Wednesday, June 10, 2009
SP-CAM climatology in 90-100th percentile of LTS
• Insufficient resolution leads to imperfect simulation of Sc clouds in SP-CAM.
• SP-CAM shows +2K increase in low cloud and radiative cooling in BL in LTS90-100, also across subsidence regions.
• Subsidence little changed between CTRL and +2K runs. HADVS confined largely to BL.
• Column LES run to equilibrium using steady forcings: no transient variability included in forcings.
Wednesday, June 10, 2009
290 300 310
800
850
900
950
1000
!, KPr
essu
re, h
Pa
LTS: 19.1 (19.5) K"LTS: 1.2 (1.3) K
EIS: 4.2 ( 4.8) K"EIS: 0.5 (0.6) K
CTRL+2K
0 25 50 75 100RH, %
CLDLOW: 0.25 (0.32)"CLDLOW: 0.01 (0.06)
Precip: CLB/SRFCTRL: 0.01/0.00
+2K: 0.01/0.00mm d!1
0 0.25 0.5 0.75 1Cloud Fraction
SWCF: !22.4 (!52.2)"SWCF: 3.6 (!8.0)
W m!2
Column LES Results for LTS 90-100 Bin
• LES: 2D, Lx=25.6km, Δx=25m, Δz=5-25m in BL
• Large diurnal cycle in cloud thickness related to daytime insolation leads to weak SWCF compared to SP-CAM.
• Time average too short for trustworthy ΔSWCF.
• Boundary layer shows signs of decoupling.
• Too little Sc due to LES overentrainment?
Stats: LES (SP-CAM)
LES Avg: Days 5-10
Wednesday, June 10, 2009
Sensitivity to Additional Sedimentation
• Many LES models are acknowledged to produce excessive entrainment at sharp inversions like those atop Sc boundary layers.
• Different effects could play a role in this over-entrainment: numerical diffusion, excessive subgrid-scale diffusivity and unmodeled physics (e.g. cloud droplet sedimentation, finite droplet evaporation timescale, partial cloudiness).
• Additional sedimentation (wsed=7.5 cm/s) is added to all cloud water to counteract this. This results in well-mixed boundary layer w/persistent cloud.
• With this amount of sedimentation, SAM simulations of DYCOMS RF01 at this vertical resolution (Δz=5m) settle into an LWP consistent with observations.
Wednesday, June 10, 2009
290 300 310
800
850
900
950
1000
!, KPr
essu
re, h
Pa
LTS: 19.4 (19.5) K"LTS: 1.3 (1.3) K
EIS: 4.6 ( 4.8) K"EIS: 0.6 (0.6) K
CTRL+2K
0 25 50 75 100RH, %
CLDLOW: 0.86 (0.32)"CLDLOW: 0.01 (0.06)
Precip: CLB/SRFCTRL: 0.41/0.05
+2K: 0.44/0.06mm d!1
0 0.25 0.5 0.75 1Cloud Fraction
SWCF: !99.3 (!52.2)"SWCF: !6.0 (!8.0)
W m!2
Column LES Results w/Extra Sedimentation
• Persistent cloud w/diurnal cycle in cloud thickness.
• +2K inversion shallower, has larger LWP during day.
• Negative ΔSWCF is comparable to SP-CAM, even though SWCF is about twice as strong.
Stats: LES (SP-CAM)
LES Avg: Days 10-30
Wednesday, June 10, 2009
Sensitivity to Diurnal Cycle of Insolation
• Perpetual (diurnally-averaged) insolation has little effect on mean LWP.
• With diurnal cycle, SWCF weaker by >30% due to daytime thinning of Sc cloud.
• Similar phenomena seen in LTS 70-80 and 80-90 bins.
Diurn Avg: Days 10-30Perpet Avg. Days 10-20
0.9
0.95
1
cld
frac
T"#$!a&'( *+,-d f0a*CTRL d"-04: 0(98
+2K d"-04: 0(98CTRL p$0p$t: 0(98
+2K p$0p$t: 0(98
0 5 10 15 200
50
100
150
200
Local Time of Day
LWP
[g m!2
]
T"#$!a&'( L?PCTRL d"-04: 113(1
+2K d"-04: 118(0CTRL p$0p$t: 108(2
+2K p$0p$t: 114(2
!300
!200
!100
0
SWC
F [W
m!2
]
T"#$!a&'( S?CECTRL d"-04: !99(3+2K d"-04: !105(3
CTRL p$0p$t: !138(0+2K p$0p$t: !139(9
0 5 10 15 200
0.1
0.2
0.3
0.4
Local Time of Day
Clo
ud A
lbed
o
T"#$!a&'( C+d A+HCTRL d"-04: 0(24
+2K d"-04: 0(25CTRL p$0p$t: 0(33
+2K p$0p$t: 0(33
CTRL diurn+2K diurnCTRL perpet+2K perpet
Wednesday, June 10, 2009
CFMIP LES Intercomparison (s11) vs. LTS 90-100
• Forcings for CFMIP LES intercomparison are intended to represent a particular location, while LTS90-100 forcings are a composite over 10% of tropical oceans
• Control sounding stability larger for s11 forcings in both LTS and EIS.
• CTRL→+2K increment in LTS and EIS is smaller for s11 forcings (ΔEIS = 0 K).
CFMIP s11 LTS 90-100
Wednesday, June 10, 2009
CFMIP LES Intercomparison (s11) vs. LTS 90-100
• CFMIP LES forcings have stronger omega than base LTS 90-100 (similar w/ω-feedback)
• Cool advection much stronger in s11 forcings for deep BL.
• Dry advection profiles similar.
• LES setup same as LTS 90-100: 2D, Lx=25.6km, Δx=25m, Δz=5-25m in BL.
CFMIP s11
LTS 90-100
Wednesday, June 10, 2009
Temperature Drift Above Inversion
• Temperature drifts cold above inversion, allows boundary layer to deepen substantially. (TH BIAS plotted starting 200m above cloud.)
• Sc layer persists with additional sedimentation, decouples without.
• Drift more prominent in +2K runs.
Diurnal Cycle
Diurnally-Averaged Insolation
Wednesday, June 10, 2009
290 300 310
800
850
900
950
1000
!, K
Pres
sure
, hPa
LTS: 22.9 (23.0) K"LTS: 0.1 (0.7) K
EIS: 8.5 ( 8.6) K"EIS: !0.5 (0.0) K
CTRL+2K
0 25 50 75 100RH, %
CLDLOW: 0.82"CLDLOW: !0.05
Precip: CLB/SRFCTL: 1.05/0.38+2K: 1.14/0.53
mm d!1
0 0.25 0.5 0.75 1Cloud Fraction
SWCF: !151.4 W m!2
"SWCF: 13.8 W m!2
Asymmetric Temperature DriftAbove Inversion
• Simulations without omega feedback, with additional sedimentation.
• +2K run drifts cold above inversion, leads to smaller ΔLTS than base state.
• +2K run slightly decoupled.
• Strong SWCF, strong positive ΔSWCF.
Time Avg: Days 5-20
290 300 310
800
850
900
950
1000
THETA [K]
p [h
Pa]
LES: THETA(p,t=tf)
LES CTRLLES +2Ks11 CTRLs11 +2K
0 20 40
LES: !0
hPa d-1
Wednesday, June 10, 2009
290 300 310
800
850
900
950
1000
THETA [K]
p [h
Pa]
LES: THETA(p,t=tf)
LES CTRLLES +2Ks11 CTRLs11 +2K
0 20 40 60OMEGA [hPa d!1]
LES: !0 + !’
290 300 310
800
850
900
950
1000
THETA [K]
p [h
Pa]
LES: THETA(p,t=tf)
LES CTRLLES +2Ks11 CTRLs11 +2K
0 20 40
LES: !0Effect of omega feedback
• Omega feedback eliminates drift above inversion. LTS, EIS, Δ’s consistent w/specs
• Results in more cloud, stronger SWCF than without omega feedback.
• ΔSWCF still positive.
Time Avg: Days 5-15
No ωFeedback
With ωFeedback
290 300 310
800
850
900
950
1000
!, K
Pres
sure
, hPa
LTS: 23.0 (23.0) K"LTS: 0.8 (0.7) K
EIS: 8.6 ( 8.6) K"EIS: 0.1 (0.0) K
CTL 100+2K 100
0 25 50 75 100RH, %
CLDLOW: 0.95"CLDLOW: !0.06
Precip: CLB/SRFCTL: 0.46/0.10+2K: 0.71/0.19
mm d!1
0 0.25 0.5 0.75 1Cloud Fraction
SWCF: !179.9 W m!2
"SWCF: 14.9 W m!2
Wednesday, June 10, 2009
290 300 310
800
850
900
950
1000
!, K
Pres
sure
, hPa
LTS: 22.8 (23.0) K"LTS: 0.8 (0.7) K
EIS: 8.3 ( 8.6) K"EIS: 0.1 (0.0) K
CTL 200+2K 200
0 25 50 75 100RH, %
CLDLOW: 0.98"CLDLOW: !0.00
Precip: CLB/SRFCTL: 0.00/0.00+2K: 0.00/0.00
mm d!1
0 0.25 0.5 0.75 1Cloud Fraction
SWCF: !197.8 W m!2
"SWCF: !5.6 W m!2
290 300 310
800
850
900
950
1000
!, K
Pres
sure
, hPa
LTS: 23.0 (23.0) K"LTS: 0.8 (0.7) K
EIS: 8.6 ( 8.6) K"EIS: 0.1 (0.0) K
CTL 100+2K 100
0 25 50 75 100RH, %
CLDLOW: 0.95"CLDLOW: !0.06
Precip: CLB/SRFCTL: 0.46/0.10+2K: 0.71/0.19
mm d!1
0 0.25 0.5 0.75 1Cloud Fraction
SWCF: !179.9 W m!2
"SWCF: 14.9 W m!2
Drizzle Feedback Changes Sign of ΔSWCF
• Sensitivity to cloud droplet number concentration Nc0.
• Drizzle-free, Nc0=200 simulations have thicker, deeper cloud.
• SWCF sensitivity to +2K change reverses sign with Nc0 change.
• Diurnal cycle might make drizzle feedback less prominent...
N100 Avg: Days 5-15N200 Avg: Days 5-15
Nc0=100
Nc0=200
Wednesday, June 10, 2009
Summary
• Column modeling of stratocumulus cloud feedbacks in SP-CAM:
• Additional sedimentation required for cloud to persist through daytime.
• Negative cloud feedbacks with most configurations (ΔEIS>0?).
• Diurnal cycle of insolation has strong impact on time-mean SWCF.
• Preliminary simulations of GCSS/CFMIP LES intercomparison (s11 case):
• Temperature drift above inversion allows additional BL deepening.
• Omega feedback provides consistent environment for evaluating cloud changes.
• Drizzle feedback reverses sign of ΔSWCF.
• Boundary layer structure and cloud feedback is sensitive to configuration.
Wednesday, June 10, 2009