low cloud feedbacks from cloud- controlling …quet al. (2014, 2015) brientand schneider (2016)...
TRANSCRIPT
Low Cloud Feedbacks from Cloud-Controlling Factors: A Review
Stephen Klein1, Alex Hall2, Joel Norris3 and Robert Pincus4
Surveys in Geophysics, accepted
Modified from IPCC AR5, Figure 7.11
Brighter Mid/High Lat Clouds
Rising High Clouds
Rising High Clouds
Broadening of the Hadley Cell
Poleward Shift of Storms
Less Low Clouds
ThisworkwasperformedundertheauspicesoftheU.S.DepartmentofEnergybyLawrenceLivermoreNationalSecurity,LLC,LawrenceLivermoreNationalLaboratoryunderContractDE-AC52-07NA27344.LLNL-PRES-738459
1LLNL 2UCLA3UCSD4U. Colorado
How to Predict Tropical Low Cloud Feedbacks?
Theory
GlobalClimateModels Large-EddySimulation
Observations
How to Predict Tropical Low Cloud Feedbacks?
Theory
GlobalClimateModels Large-EddySimulation
Observations
Low Cloud Feedbacks from Cloud-Controlling FactorsLow clouds are a fast (~hours-days) response to ”cloud-controlling factors” (CCF) of the environment (StevensandBrenguier 2009)If we assume that these cloud sensitivities are ”time-scale invariant”, and we know how CCF change, then we can predict the cloud feedback:
Climate Change in CCFLow Cloud Feedback
Cloud Sensitivity to CCF (xi)
𝑑𝐶𝑑𝑇$
=&𝜕𝐶𝜕𝑥)
�
)
𝑑𝑥)𝑑𝑇$
from Observations (Usually inter-annual variability)
from Climate Models
𝑥) ∈ {SST,EIS,w,RHtropo ,temperatureadvection}
Recent Articles Using This Approach
Qu etal.(2014,2015)Brient andSchneider(2016)
MyersandNorris(2016)
Zhai etal.(2015)
Recent Articles Using This Approach
Qu etal.(2014,2015)Brient andSchneider(2016)
MyersandNorris(2016)
Zhai etal.(2015)
McCoyetal.(2017)
Sample Results
MyersandNorris(2016)
ISCC
PPa
tmos
MIS
RM
OD
IS
ΔLCCΔSST
∂LCC∂SST EIS
Sensitivity from Natural Variability
Climate Change
(K-1)
(K-1)
Qu etal.(2014,2015)
𝑑𝐶𝑑𝑇$
=&𝜕𝐶𝜕𝑥)
�
)
𝑑𝑥)𝑑𝑇$
Observed cloud sensitivity with CCF change from good, average, or poor models
Observed cloud sensitivity and its error bars with average climate model CCF change
Sample Results
MyersandNorris(2016)
ISCC
PPa
tmos
MIS
RM
OD
IS
ΔLCCΔSST
∂LCC∂SST EIS
Sensitivity from Natural Variability
Climate Change
(K-1)
(K-1)
Qu etal.(2014,2015)
𝑑𝐶𝑑𝑇$
=&𝜕𝐶𝜕𝑥)
�
)
𝑑𝑥)𝑑𝑇$
Observed cloud sensitivity with CCF change from good, average, or poor models
Observed cloud sensitivity and its error bars with average climate model CCF change
Goals of Our Review Article
1. Are the cloud feedbacks predicted by these articles quantitatively consistent among these articles and with LES predicted feedbacks? What is the implication for Equilibrium Climate Sensitivity?
2. By considering issues with this approach, what future steps could be taken to reduce the uncertainties in these predictions?
Goals of Our Review Article
1. Are the predicted cloud feedbacks predicted quantitatively consistent among these articles and with LES predicted feedbacks? What is the implication for Equilibrium Climate Sensitivity?
2. By considering issues with this approach, what future steps could be taken to reduce the uncertainties in these predictions?
Goals of Our Review Article
1. Are the predicted cloud feedbacks predicted quantitatively consistent among these articles and with LES predicted feedbacks? What is the implication for Equilibrium Climate Sensitivity?
2. By considering issues with this approach, what future steps could be taken to reduce the uncertainties in these predictions?
Cloud-Controlling Factor Predictions from Observations
Q15
M16
B16
Z15
M17
Global Climate Models
Wm-2 K-1
TropicalLow-CloudFeedbacks
Large-Eddy Simulations
ISCCPPATMOS-x
MISRMODIS
De-SeasonalizedIntra-Annual
Inter-AnnualSeasonal
This Study’s Meta-analysis
Latitude BandsTrade Cumulus
Stratocumulus + Trade Cumulus
ISCCPCERES
Trade Cumulus Stratocumulus
-1 0 1 2
CLOUDSAT-CALIPSO
CERES-EBAF
MODIS
Cloud-Controlling Factor Predictions from Observations
Q15
M16
B16
Z15
M17
Global Climate Models
Wm-2 K-1
TropicalLow-CloudFeedbacks
Large-Eddy Simulations
ISCCPPATMOS-x
MISRMODIS
De-SeasonalizedIntra-Annual
Inter-AnnualSeasonal
This Study’s Meta-analysis
Latitude BandsTrade Cumulus
Stratocumulus + Trade Cumulus
ISCCPCERES
Trade Cumulus Stratocumulus
-1 0 1 2
CLOUDSAT-CALIPSO
CERES-EBAF
MODIS
Cloud-Controlling Factor Predictions from Observations
Q15
M16
B16
Z15
M17
Global Climate Models
Wm-2 K-1
TropicalLow-CloudFeedbacks
Large-Eddy Simulations
ISCCPPATMOS-x
MISRMODIS
De-SeasonalizedIntra-Annual
Inter-AnnualSeasonal
This Study’s Meta-analysis
Latitude BandsTrade Cumulus
Stratocumulus + Trade Cumulus
ISCCPCERES
Trade Cumulus Stratocumulus
-1 0 1 2
CLOUDSAT-CALIPSO
CERES-EBAF
MODIS
Cloud-Controlling Factor Predictions from Observations
Q15
M16
B16
Z15
M17
Global Climate Models
Wm-2 K-1
TropicalLow-CloudFeedbacks
Large-Eddy Simulations
ISCCPPATMOS-x
MISRMODIS
De-SeasonalizedIntra-Annual
Inter-AnnualSeasonal
This Study’s Meta-analysis
Latitude BandsTrade Cumulus
Stratocumulus + Trade Cumulus
ISCCPCERES
Trade Cumulus Stratocumulus
-1 0 1 2
CLOUDSAT-CALIPSO
CERES-EBAF
MODIS
Cloud-Controlling Factor Predictions from Observations
Q15
M16
B16
Z15
M17
Global Climate Models
Wm-2 K-1
TropicalLow-CloudFeedbacks
Large-Eddy Simulations
ISCCPPATMOS-x
MISRMODIS
De-SeasonalizedIntra-Annual
Inter-AnnualSeasonal
This Study’s Meta-analysis
Latitude BandsTrade Cumulus
Stratocumulus + Trade Cumulus
ISCCPCERES
Trade Cumulus Stratocumulus
-1 0 1 2
CLOUDSAT-CALIPSO
CERES-EBAF
MODIS
Cloud-Controlling Factor Predictions from Observations
Q15
M16
B16
Z15
M17
Global Climate Models
Wm-2 K-1
TropicalLow-CloudFeedbacks
Large-Eddy Simulations
ISCCPPATMOS-x
MISRMODIS
De-SeasonalizedIntra-Annual
Inter-AnnualSeasonal
This Study’s Meta-analysis
Latitude BandsTrade Cumulus
Stratocumulus + Trade Cumulus
ISCCPCERES
Trade Cumulus Stratocumulus
-1 0 1 2
CLOUDSAT-CALIPSO
CERES-EBAF
MODIS
Cloud-Controlling Factor Predictions from Observations
Q15
M16
B16
Z15
M17
Global Climate Models
Wm-2 K-1
TropicalLow-CloudFeedbacks
Large-Eddy Simulations
ISCCPPATMOS-x
MISRMODIS
De-SeasonalizedIntra-Annual
Inter-AnnualSeasonal
This Study’s Meta-analysis
Latitude BandsTrade Cumulus
Stratocumulus + Trade Cumulus
ISCCPCERES
Trade Cumulus Stratocumulus
-1 0 1 2
CLOUDSAT-CALIPSO
CERES-EBAF
MODIS
Cloud-Controlling Factor Predictions from Observations
Q15
M16
B16
Z15
M17
Global Climate Models
Wm-2 K-1
TropicalLow-CloudFeedbacks
Large-Eddy Simulations
ISCCPPATMOS-x
MISRMODIS
De-SeasonalizedIntra-Annual
Inter-AnnualSeasonal
This Study’s Meta-analysis
Latitude BandsTrade Cumulus
Stratocumulus + Trade Cumulus
ISCCPCERES
Trade Cumulus Stratocumulus
-1 0 1 2
CLOUDSAT-CALIPSO
CERES-EBAF
MODIS
Cloud-Controlling Factor Predictions from Observations
Q15
M16
B16
Z15
M17
Global Climate Models
Wm-2 K-1
TropicalLow-CloudFeedbacks
Large-Eddy Simulations
ISCCPPATMOS-x
MISRMODIS
De-SeasonalizedIntra-Annual
Inter-AnnualSeasonal
This Study’s Meta-analysis
Latitude BandsTrade Cumulus
Stratocumulus + Trade Cumulus
ISCCPCERES
Trade Cumulus Stratocumulus
-1 0 1 2
CLOUDSAT-CALIPSO
CERES-EBAF
MODIS
Implications for ECS
Calculating ECS = – F2xCO2 / S l, ECS = 3 K
Term Value
Radiative forcing for doubling of CO2 concentration (F2CO2) 3.43 W m-2
Planck feedback -3.15 W m-2 K-1
Water vapor feedback 1.69 W m-2 K-1
Lapse rate feedback -0.53 W m-2 K-1
Surface albedo feedback 0.38 W m-2 K-1
High cloud altitude feedback 0.20 W m-2 K-1
Tropical low cloud feedback 0.25 W m-2 K-1
Forcingandnon-cloudfeedbackvaluesfromCaldwelletal.(2016);High-cloudaltitudefeedbackfromZelinka etal.(2016);
TropicalLow-cloudfeedback(thisstudy)
Fundamental and Implementation Issues Discussed
F1. Are cloud sensitivities time-scale invariant?F2. Are clouds responding to the controlling factors?F3. Uncertainty in the climate change prediction of cloud-controlling factorsF4. Time-dependency of cloud-controlling factors during a climate change
I1. Imperfect observations of clouds and their controlling factorsI2. Limited duration of the observational recordI3. Limited spatial sampling of the observationsI4. Imprecise statistical modelingI5. Incomplete set of cloud-controlling factors
Fundamental and Implementation Issues Discussed
F1. Are cloud sensitivities time-scale invariant?F2. Are clouds responding to the controlling factors?F3. Uncertainty in the climate change prediction of cloud-controlling factorsF4. Time-dependency of cloud-controlling factors during a climate change
I1. Imperfect observations of clouds and their controlling factorsI2. Limited duration of the observational recordI3. Limited spatial sampling of the observationsI4. Imprecise statistical modelingI5. Incomplete set of cloud-controlling factors
Fundamental and Implementation Issues Discussed
F1. Are cloud sensitivities time-scale invariant?F2. Are clouds responding to the controlling factors?F3. Uncertainty in the climate change prediction of cloud-controlling factorsF4. Time-dependency of cloud-controlling factors during a climate change
I1. Imperfect observations of clouds and their controlling factorsI2. Limited duration of the observational recordI3. Limited spatial sampling of the observationsI4. Imprecise statistical modelingI5. Incomplete set of cloud-controlling factors
Fundamental and Implementation Issues Discussed
F1. Are cloud sensitivities time-scale invariant?F2. Are clouds responding to the controlling factors?F3. Uncertainty in the climate change prediction of cloud-controlling factorsF4. Time-dependency of cloud-controlling factors during a climate change
I1. Imperfect observations of clouds and their controlling factorsI2. Limited duration of the observational recordI3. Limited spatial sampling of the observationsI4. Imprecise statistical modelingI5. Incomplete set of cloud-controlling factors
Fundamental and Implementation Issues Discussed
F1. Are cloud sensitivities time-scale invariant?F2. Are clouds responding to the controlling factors?F3. Uncertainty in the climate change prediction of cloud-controlling factorsF4. Time-dependency of cloud-controlling factors during a climate change
I1. Imperfect observations of clouds and their controlling factorsI2. Limited duration of the observational recordI3. Limited spatial sampling of the observationsI4. Imprecise statistical modelingI5. Incomplete set of cloud-controlling factors
Fundamental and Implementation Issues Discussed
F1. Are cloud sensitivities time-scale invariant?F2. Are clouds responding to the controlling factors?F3. Uncertainty in the climate change prediction of cloud-controlling factorsF4. Time-dependency of cloud-controlling factors during a climate change
I1. Imperfect observations of clouds and their controlling factorsI2. Limited duration of the observational recordI3. Limited spatial sampling of the observationsI4. Imprecise statistical modelingI5. Incomplete set of cloud-controlling factors
Fundamental and Implementation Issues Discussed
F1. Are cloud sensitivities time-scale invariant?F2. Are clouds responding to the controlling factors?F3. Uncertainty in the climate change prediction of cloud-controlling factorsF4. Time-dependency of cloud-controlling factors during a climate change
I1. Imperfect observations of clouds and their controlling factorsI2. Limited duration of the observational recordI3. Limited spatial sampling of the observationsI4. Imprecise statistical modelingI5. Incomplete set of cloud-controlling factors
Fundamental and Implementation Issues Discussed
F1. Are cloud sensitivities time-scale invariant?F2. Are clouds responding to the controlling factors?F3. Uncertainty in the climate change prediction of cloud-controlling factorsF4. Time-dependency of cloud-controlling factors during a climate change
I1. Imperfect observations of clouds and their controlling factorsI2. Limited duration of the observational recordI3. Limited spatial sampling of the observationsI4. Imprecise statistical modelingI5. Incomplete set of cloud-controlling factors
Fundamental and Implementation Issues Discussed
F1. Are cloud sensitivities time-scale invariant?F2. Are clouds responding to the controlling factors?F3. Uncertainty in the climate change prediction of cloud-controlling factorsF4. Time-dependency of cloud-controlling factors during a climate change
I1. Imperfect observations of clouds and their controlling factorsI2. Limited duration of the observational recordI3. Limited spatial sampling of the observationsI4. Imprecise statistical modelingI5. Incomplete set of cloud-controlling factors
Fundamental and Implementation Issues Discussed
F1. Are cloud sensitivities time-scale invariant?F2. Are clouds responding to the controlling factors?F3. Uncertainty in the climate change prediction of cloud-controlling factorsF4. Time-dependency of cloud-controlling factors during a climate change
I1. Imperfect observations of clouds and their controlling factorsI2. Limited duration of the observational recordI3. Limited spatial sampling of the observationsI4. Imprecise statistical modelingI5. Incomplete set of cloud-controlling factors
Fundamental and Implementation Issues Discussed
F1. Are cloud sensitivities time-scale invariant?F2. Are clouds responding to the controlling factors?F3. Uncertainty in the climate change prediction of cloud-controlling factorsF4. Time-dependency of cloud-controlling factors during a climate change
I1. Imperfect observations of clouds and their controlling factorsI2. Limited duration of the observational recordI3. Limited spatial sampling of the observationsI4. Imprecise statistical modelingI5. Incomplete set of cloud-controlling factors
We believe that the limited duration of the observational record is the largest source of uncertainty in the prediction. Once records are long enough to more precisely determine ,-
,../,the
uncertainty in the climate change prediction of cloud-controlling factors will become more relevant.
Fundamental and Implementation Issues Discussed
F1. Are cloud sensitivities time-scale invariant?F2. Are clouds responding to the controlling factors?F3. Uncertainty in the climate change prediction of cloud-controlling factorsF4. Time-dependency of cloud-controlling factors during a climate change
I1. Imperfect observations of clouds and their controlling factorsI2. Limited duration of the observational recordI3. Limited spatial sampling of the observationsI4. Imprecise statistical modelingI5. Incomplete set of cloud-controlling factors
Most studies target regions of stratocumulus or cumulus-under-stratocumulus. More observational studies targeting trade cumulus are needed.
Still, the existing evidence for trade cumulus regions is supportive of similar cloud sensitivities.
Conclusions
1. The reviewed cloud-controlling factor observational studies all agree that the tropical low cloud feedback is positive
2. From these studies, we formed a meta-estimate for the local low-cloud feedback of 1 ± 0.7 W m-2 K-1
3. This estimate is similar in magnitude to those of Large-Eddy Simulations
Conclusions
1. The reviewed cloud-controlling factor observational studies all agree that the tropical low cloud feedback is positive
2. From these studies, we formed a meta-estimate for the local low-cloud feedback of 1 ± 0.7 W m-2 K-1
3. This estimate is similar in magnitude to those of Large-Eddy Simulations
Conclusions
1. The reviewed cloud-controlling factor observational studies all agree that the tropical low cloud feedback is positive
2. From these studies, we formed a meta-estimate for the local low-cloud feedback of 1 ± 0.7 W m-2 K-1
3. This estimate is similar in magnitude to those of Large-Eddy Simulations
Conclusions
1. The reviewed cloud-controlling factor observational studies all agree that the tropical low cloud feedback is positive
2. From these studies, we formed a meta-estimate for the local low-cloud feedback of 1 ± 0.7 W m-2 K-1
3. This estimate is similar in magnitude to those of Large-Eddy Simulations
Implications for the Future
1. Longer observational records should help narrow the sensitivity of cloud to SST, which is currently the largest uncertainty in predictions with this approach
2. Additional confirmation of the sensitivity of trade-cumulus portions of the tropics would be helpful
3. If climate models were in greater agreement with observations for the cloud sensitivities, there would be less inter-model spread in their predicted cloud-feedbacks
Implications for the Future
1. Longer observational records should help narrow the sensitivity of cloud to SST, which is currently the largest uncertainty in predictions with this approach
2. Additional confirmation of the sensitivity of trade-cumulus portions of the tropics would be helpful
3. If climate models were in greater agreement with observations for the cloud sensitivities, there would be less inter-model spread in their predicted cloud-feedbacks
Implications for the Future
1. Longer observational records should help narrow the sensitivity of cloud to SST, which is currently the largest uncertainty in predictions with this approach
2. Additional confirmation of the sensitivity of trade-cumulus portions of the tropics would be helpful
3. If climate models were in greater agreement with observations for the cloud sensitivities, there would be less inter-model spread in their predicted cloud-feedbacks
Implications for the Future
1. Longer observational records should help narrow the sensitivity of cloud to SST, which is currently the largest uncertainty in predictions with this approach
2. Additional confirmation of the sensitivity of trade-cumulus portions of the tropics would be helpful
3. If climate models were in greater agreement with observations for the cloud sensitivities, there would be less inter-model spread in their predicted cloud-feedbacks
Tropical Low Cloud Feedbacks
• Climate models predict low-cloud amount to decrease on average– Largest single feedback on average
• The response of tropical low clouds to warming is the single cloud type with the strongest correlation to equilibrium climate sensitivity – Due to albedo changes not
compensated by longwave radiation trapping changes
Average Low Cloud Amount Feedback in Climate Models
Equilibrium Climate
Sensitivity
r = –0.73
Observations
Global Climate Models
Wm-2 K-1
TropicalLowCloudFeedbacks
Large-Eddy SimulationsTrade Cumulus Stratocumulus
-1 0 1 2
Observations
Global Climate ModelsLarge-Eddy Simulations
Tropical Low Cloud Feedbacks• Observations and high-
resolution modeling agree that tropical low cloud feedbacks should be positive
Zelinka etal.(2016)
Brient andSchneider(2016)
Albedo temperature sensitivity (% K-1)
Kleinetal.(submitted)
Physical Basis for Cloud Sensitivities
• Physical bases for cloud sensitivities to EIS, w, RHtropo , and temperature advection are well established
• Increasedverticalmoisturegradientinawarmerworldincreasestheamountentrainmentdryingoftheboundarylayer(Sherwoodetal.2014)
• Increasedverticalmoisturefluxinawarmerworldpromotesmoreefficiententrainmentdryingoftheboundarylayer[“entrainmentliquid-fluxadjustment”](BrethertonandBlossey 2014,Rieck etal.2012)
BrethertonandBlossey (2014) Sherwoodetal.(2014)
Climate Modeling Challenge• Models should better simulate ,-
,01– Low-cloud sensitivities in
stratocumulus regions are highly variable and rarely correct (Quetal.2015)
– Trade cumulus cloud sensitivities are just as problematic (Nuijens etal.2015)
• How to improve models?– Obviously, mixing parameterizations
matter a lot (Brient etal.2015)– Vertical resolution still important – Cloud fraction parameterizations tied
to EIS explained some of the negative low cloud feedbacks found in (older) models (Quetal.2014,Geoffroy etal.2017)
Quetal.(2015)
𝜕𝐶𝜕𝐸𝐼𝑆
𝜕𝐶𝜕𝑆𝑆𝑇
(%K-1)
(%K-1)