A HighLatitude Convective Cloud Feedback

1
A High-Latitude Convective Cloud Feedback

• Dorian S. Abbot
• Harvard University
• 4 December 2008

Future Climate
Equable Climates
Coauthors Eli Tziperman Matthew Huber Ben
Leibowicz Chris Walker Gabriel Bousquet
2
Outline

• Equable Climates Overview
• Convective Cloud Feedback and Equable Climates
• CC Feedback and Future Climate
• Maximumal Seasonal Sea Ice at High CO2
• 4) Preliminary Observational Evidence for CC

3
Equable Climates
When are we talking about anyway???
4
Equable Climates what (we think) we know
Equable climate warm poles, mild winters

• Late Cretaceous and Early Paleogene (100-35
  MaYr)
• High global mean temperature
• Warm deep ocean 12-15C
• Very high polar temperatures
• above freezing winter temperatures _at_ 60N,
  interior of N. America (now -30C)
• No significant ice
• Tropical SSTs gt modern
• Low equator-pole temperature difference 25C
• (compared to 45C modern)
• Weak high-latitude seasonality

5
Why do we think we know what we think we know????
6
Nearest Living Relative

• Find a fossil
• Look for something related
• thats still alive
• 3. Assume ancient climate
• where you found fossil is like
• climate where modern
• plant/animal is

7
Leaf Margin Analysis (LMA)
Wilf, 1997
Eastern Redbud - Untoothed (Entire Margin)
American Elm - Toothed
The physiological basis for the MAT vs.
leaf-margin correlation has never been adequately
demonstrated. Wilf, 1997
8
Planktonic Foraminifera
Now!
50 Million Years ago!
Forams.
Drill
Die and Settle To Bottom
Mud!
9
Eocene Temperature as a function of latitude
Greenwood and Wing, 1995
10
State-of-the-art fully coupled GCMs have
difficulty reproducing Eocene (50Myr) proxy
observations
Coupled GCM study of Huber and Sloan, 2001, for
example. Uses best guess Eocene

• Bathymetry
• Topography
• Land Surface
• Vegetation
• CO2 (560 ppm)

11
Whats up?

• Bad data maybe tropics hotter?
• Bad models missing mechanism

Some GCMs do better at simulating equable climate
data, can we understand why?
12
What could the models be missing?
13
An Equator-to-Pole Hadley cell
?

• Venus
• Low Rotation
• Large Hadley Cell
• Poles warm as Equator

Farrell, 1990
14
Hurricanes?

• Warm climate
• Stronger/ more tropical cyclones
• stronger ocean mixing
• Increased overturning circulation, especially at
  low latitudes
• ?May cool the tropics, perhaps also warm the high
  latitudes

Emanuel, 2002 Korty Emanuel 2007
15
Polar Stratospheric Clouds
Polar stratospheric clouds (PSCs), at 15-25km,
have a strong greenhouse effect! Formed via
methane-moistening of stratos. Sloan 92
Or a complex feedback mechanism due to global
warming? Kirk-Davidoff et al 2002 but maybe
not? Korty Emanuel 2007
16
What about tropospheric clouds?
Cloud parameterizations most uncertain component
of climate models Cess et al., 1990,1996
Soden and Held, 2006
17
Simple Model

• Zonally averaged
• Equator to pole
• Two levels Boundary Layer
• Free Troposphere
• Mixed layer ocean
• Non-linear momentum eqns
• Merid resolution 3/30 cells

• Prognostic dry static energy water vapor
• Simple land surface
• Advection
• Diffusive eddies
• ConvectionPrecipitation
• Radiation SW, LW

200mb
900mb
1000mb
Equator
Pole
18
Model experiments results summary

• Model experiments
• Slowly increase CO2 to extreme values then
  decrease it
• Results
• A qualitatively different climate regime at
  sufficiently high CO2, warm high latitudes and
  low equator-to pole temperature difference.

19
Results two modes of atmospheric dynamics
multiple equilibria at a given co2, hysteresis
Equator to pole temperature difference (EPTD, C)
Convective high cloud fraction, polar column
3 column model
Present-day solution colder, high
EPTD Equable solution warm, low EPTD Arrows
path of solution if CO2 slowly increased then
decreased.
20
Results two modes of atmospheric dynamics
multiple equilibria at a given co2, hysteresis
Equator to pole temperature difference (EPTD, C)
Convective high cloud fraction, polar column
Convecting, High Clouds
3 column model
Not Convecting, Low Clouds
Present-day solution colder, high
EPTD Equable solution warm, low EPTD Arrows
path of solution if CO2 slowly increased then
decreased.
21
Cloud feedbacks

• Low clouds (marine stratus)
• High albedo
• Low altitude (lt1km )
• Cooling effect on climate

• High clouds (cirrus)
• low albedo, high emissivity
• High altitude (gt8 km)
• Warming effect on climate

? high clouds may help explain equable climate
22
Atmospheric convection

• Air parcel in lower atmosphere rises up
• It expands, cools, and water vapor condenses
• Condensation leads to latent heat release, air
  parcel heats
• Parcel becomes warmer, lighter, more buoyant, and
  rises even more ? positive feedback, instability
• Condensation creates clouds, rain

http//apollo.lsc.vsc.edu/classes/met130/
23
Investigated In Hierarchy of ModelsBox Model,
Analytical Work, Single Column Model, Atmospheric
GCM, Coupled GCMs
The Convective Cloud Feedback

• Further investigation revealed
• Importance of sea ice for this feedback.
• Feedback should occur during winter.

warmer surface ? unstable air column ? deep
convection ? high clouds ? greenhouse effect
warmer surface
24
Winter (DJF) Cloud Radiative Forcing GCMs

• CAM Atmospheric GCM in Modern Configuration
• CCSM Coupled GCM in Eocene Configuration (Thanks
  Huber!)

• Winter CRF increases as CO2 increases.
• High CRF occurs at lower CO2 for Eocene boundary
  conditions

25
DJF CRF North of 60N Increases Over Land and Ocean

• Changes in surface temperature due to changes in
  boundary conditions produce similar CRF changes
  as those due to changes in CO2
• CRF increases over land similar to those over
  ocean

Eocene, Ocean
Each data point represents a GCM run in at a
different CO2 and in a different configuration.
Modern, Ocean
Eocene, Land
Modern, Land
26
Different Processes Produce Similar Results Over
Land and Ocean
Difference between zonal average of winter cloud
characteristics in modern config. CAM at CO2280
ppm and CO22240 ppm
27
Different Processes Produce Similar Results Over
Land and Ocean
Difference between zonal average of winter cloud
characteristics in modern config. CAM at CO2280
ppm and CO22240 ppm
Ocean
Land
0 Latitude 90
0 Latitude 90

• Convective cloud feedback produces CRF changes
  over Ocean
• Over land CRF increase is due to layered clouds,
  which are parameterized diagnostically based on
  relative humidity

28
Different Processes Produce Similar Results Over
Land and Ocean
Difference between zonal average of winter cloud
characteristics in modern config. CAM at CO2280
ppm and CO22240 ppm
Ocean
Land
0 Latitude 90
0 Latitude 90

• Convective cloud feedback produces CRF changes
  over Ocean
• Over land CRF increase is due to layered clouds,
  which are parameterized diagnostically based on
  relative humidity

29
Conclusions from Eocene Work

• Convective cloud feedback active in hierarchy of
  climate models at high enough CO2
• May help explain warm high latitudes during
  equable climates
• Whether or not CC feedback is active could play a
  role in determining the ability of different GCMs
  to simulate equable climates.

30
Relevant for Future Climate??More on Sea Ice
31
CC Feedback and Maximum Seasonal Sea Ice at High
CO2

• Evidence for CC feedback in IPCC AR4 GCM runs
• CC feedback may help determine maximum seasonal
  (spring) sea ice in these runs
• CAM snapshot sensitivity runs establish
  importance of CC feedback for spring sea ice

32
CC Feedback Active During Winter in GCMs
NCAR GFDL 3D Coupled Ocn-Atm GCMs x4 CO2 (1120
ppm) minus preindustrial (280 ppm)
NCAR
Cloud CRF up unchanged
Convection up unchanged
Srfc Temp up unchanged
Sea ice gone unchanged
GFDL
33
CC Feedback in Arctic for all IPCC Models
CC Feedback associated with winter sea ice loss
r20.57 p0.004
34
CC Feedback in Arctic for all IPCC Models
Effect of CC feedback on spring sea ice?
r20.52 p0.008
35
CAM runs to Compare CC OHT Feedback

• Increase CO2 to 1120 ppm
• and do four runs
• 1. LO OHT and LO CRF
• 2. LO OHT and HI CRF
• 3. HI OHT and LO CRF
• 4. HI OHT and HI CRF

36
March Sea Ice, CO21120 ppm Both Feedbacks
Disabled
37
CC Feedback Alone Ice Reduction
38
OHT Feedback Alone Somewhat Larger Ice Reduction
39
BOTH Feedbacks Needed to Eliminate March Sea Ice!!
40
Preliminary Investigation Into Reanalysis Data
Thanks to Ben Leibowicz!!!!
Anomalies averaged over 3 high-sea ice winters
(79,82,99)
Anomalies averaged over 3 high-sea ice winters
(79,82,99)
Anomalies averaged over 3 high-sea ice winters
(79,82,99)
Anomalies averaged over 3 high-sea ice winters
(79,82,99)
Sea ice Concentration
Cloud Radiative Forcing W m-2
Sea ice Concentration
Cloud Radiative Forcing W m-2
Sea ice Concentration
Cloud Radiative Forcing W m-2
Sea ice Concentration
Cloud Radiative Forcing W m-2
41
Anomalies For High-Sea Ice Winters
NCEP Reanalysis averaged for 1979,1982,1999
Sea Ice Concentration
Cloud Radiative Forcing W m-2
Sea Ice? ?Cloud Raditive Forcing ?
Anomalies averaged over 3 high-sea ice winters
(79,82,99)
Anomalies averaged over 3 high-sea ice winters
(79,82,99)
Onset of convection in this region as well.
Anomalies averaged over 3 high-sea ice winters
(79,82,99)
Anomalies averaged over 3 high-sea ice winters
(79,82,99)
Sea ice Concentration
Cloud Radiative Forcing W m-2
Sea ice Concentration
Cloud Radiative Forcing W m-2
Sea ice Concentration
Cloud Radiative Forcing W m-2
Sea ice Concentration
Cloud Radiative Forcing W m-2
42
Winter Anomalies off Novaya Zemla
Cloud Radiative Forcing W m-2
Sea Ice Concentration
Anomalies averaged over 3 high-sea ice winters
(79,82,99)
Anomalies averaged over 3 high-sea ice winters
(79,82,99)
Anomalies averaged over 3 high-sea ice winters
(79,82,99)
R-0.75, plt0.001 accounting for temporal
autocorrelation
Anomalies averaged over 3 high-sea ice winters
(79,82,99)
Sea ice Concentration
Cloud Radiative Forcing W m-2
Sea ice Concentration
Cloud Radiative Forcing W m-2
Sea ice Concentration
Cloud Radiative Forcing W m-2
Sea ice Concentration
Cloud Radiative Forcing W m-2
43
Extending Observational Analysis

• Perform EOF analysis and compare spatial patterns
  of sea ice and cloud radiative forcing anomalies
• Repeat analysis using ECMWF reanalysis data
• Perform similar analysis on IPCC models in
  control climate to determine the quality with
  which they simulate this feedback.
• Any suggestions?

44
Central Conclusions

• Found a simple, interesting unexpected
  convective cloud feedback on high-latitude
  climate change. May help to explain difficulty
  simulating equable climates.
• Feedback may also be important for forecasting
  future spring (maximum seasonal) sea ice under
  global warming.

45
Thank you for your attention! Thanks for advice
John Dykema, Ian Eisenman, Kerry Emanuel, Brian
Farrell, John Higgins, Peter Huybers, Zhiming
Kuang, Richard Lindzen, Ray Pierrehumbert, Dan
Schrag, Jacob Sewall Thanks for helpful reviews
Rob Korty, Adam Sobel, Richard Seager, and many
anonymous.
46
Temperature from Isotopic Analysis
Temperature Fractionation
Erez and Luz, 1983
30
Can use oxygen isotopic ratio in the shells of
foraminifera to figure out ancient ocean
temperatures!
T C
14
-3.0
0.0
?18Occ- ?18Osw
47
Evidence for CC Feedback in Modern Climate
Winter Arctic CRF seems tied to sea ice in
modern climate
Yearly sea ice anomalies tied to yearly CRF
anomalies
Can this constrain feedback so that models are
more useful?
National Snow and Ice Data center Harrison et
al. (1990)
48
Speculation on continental interiors
Continental interiors not addressed in our
study. Speculation Drifting moisture high
clouds from above-ocean convection may provide
greenhouse effect over continental interiors
weaker eddies in equable climate ? less drying
by vertical eddy motions.
Greenwood and Wing, 95

• - palms
• cycads, gingers, tree ferns
• no frost intolerant plants

• - lowlands
• uplands
• higher uplands

49
Different Processes Produce Similar Results Over
Land and Ocean
50
? in CAMs DJF Climate Per CO2 Doubling
CRF Increase Over BOTH Ocean and Land
CO2 Doubling
1
2
3
4