Investigating mechanisms of cloud
feedback differences.
Mark Webb, Adrian Lock (Met Office)
Thanks also to Chris Bretherton (UW), Sandrine Bony (IPSL),Jason Cole (CCCma),
Abderrahmane Idelkadi (IPSL), Sarah Kang (UNIST), Tsuyoshi Koshiro (MRI), Hideaki Kawai
(MRI), Tomoo Ogura (NIES), Romain Roehrig (CNRM), Yechul Shin (UNIST), Thorsten
Mauritsen (MPI), Steve Sherwood (UNSW), Jessica Vial (IPSL), Masahiro Watanabe (AORI),
Matthew Woelfle (UW), Ming Zhao (GFDL).
WCRP Grand Challenge Workshop: Earth's Climate Sensitivities, Ringberg Schloss, 2015
Selected Process On/Off Klima
Intercomparison Experiment (SPOOKIE)
Aims
•  Establish the relative contributions of different processes to inter-model
differences in cloud feedbacks
Approach
•  Repeat CFMIP-2 AMIP/AMIP Uniform +4K SST perturbation experiments
•  Switch off or simplify various model schemes/processes in turn
Pilot Experiments
•  Start by switching off convective parametrization
© Crown copyright Met Office
Global Mean Cloud Feedback (Wm-2K-1)
(Including cloud masking)
© Crown copyright Met Office
amip4K Cloud Feedbacks over 30ON/S Oceans
Stable/
Shallow
Precipitating/
Deep
Stable/
Shallow
Precipitating/
Deep
Stable/
Shallow
Stable/
Shallow
We sort cloud regimes in the tropics using a composite index based on
precipitation rate and Lower Tropospheric Stablity (LTS). We call this
the Angular LTS/Precipitation Index (ALPI).
Stable/
Shallow
Precipitating/
Deep
Stable/
Shallow
Precipitating/
Deep
Stable/
Shallow
amip4K Cloud Feedbacks over 30ON/S Oceans
Precipitating/
Deep
Stable/
Shallow
Precipitating/
Deep
Stable/
Shallow
Stable/
Shallow
Black diamonds
mark significant
correlations with
the net cloud
feedbacks
in same regime
Stable/
Shallow
Stable/
Shallow
Precipitating/
Deep
Stable/
Shallow
Precipitating/
Deep
Stable/
Shallow
Grey squares
mark significant
correlations with
the net cloud
feedback area
averaged
over tropical
oceans
Why might the amount of mid level cloud regulate
cloud feedback?
•  Sherwood et al. 2014 and Zhao 2014 argue that convective precipitation
efficiency plays an important role in cloud feedback.
•  Sherwood et al. 2014 defines precipitation efficiency in terms of the amount
of precipitation for a given vertical transport of water vapour from the
boundary layer to the free troposphere.
•  Might large-scale precipitation efficiency be related to mid-level clouds?
•  Models with more mid level cloud might rain out the surface more easily
than models which form cloud at higher levels. So they might need to
transport less water vapour vertically to produce a given precipitation rate.
•  These might have a stronger precipitation efficiency, and so might according
to the arguments of Sherwood et al. dry the PBL less in the warmer climate
than models with less mid-level clouds and stronger vertical transports.
•  We could test these ideas by preventing formation of mid-level clouds in
models – this would be predicted to strengthen the cloud feedbacks
© Crown copyright Met Office
Standard Models
Precipitating/
Deep
Stable/
Shallow
ConvOff Models
Precipitating/
Deep
Stable/
Shallow
Stable/
Shallow
Stable/
Shallow
Stable/
Shallow
Black diamonds
mark significant
correlations with
the net cloud
feedbacks
in same regime
Grey squares
mark significant
correlations with
the net cloud
feedback area
averaged
over tropical
oceans
Why might the MSE near the top of the PBL be
correlated with cloud feedback?
•  Stronger lower tropospheric mixing by models with weaker precipitation
efficiences would be expected to remove MSE at a faster rate from the PBL
(as in Sherwood et al 2014).
•  In the absence of convective parametrization this mixing would be done by
the resolved flow or by the turbulence scheme.
•  This could explain lower MSE in the cloud layer in models with stronger
lower tropospheric mixing.
•  As the climate warms, the hydrological cycle strengthens, increasing lower
tropospheric mixing and MSE loss from the PBL, reducing cloudiness.
•  This effect will be stronger in models with weaker precipitation efficiencies,
stronger lower tropospheric mixing, and less MSE in the cloud layer in the
control climate.
•  This effect could be tested by adding an artificial MSE mixing term between
the PBL and the free troposphere tied to the surface precipitation rate.
© Crown copyright Met Office
amip4K Cloud Feedbacks over 30ON/S Oceans
Precipitating/
Deep
Stable/
Shallow
Precipitating/
Deep
Stable/
Shallow
Stable/
Shallow
Black diamonds
mark significant
correlations with
the net cloud
feedbacks
in same regime
Stable/
Shallow
Stable/
Shallow
Precipitating/
Deep
Stable/
Shallow
Precipitating/
Deep
Stable/
Shallow
Grey squares
mark significant
correlations with
the net cloud
feedback area
averaged
over tropical
oceans
How might low cloud fraction regulate shallow
cloud feedback?
•  Brient and Bony (2012) proposed a ‘beta feedback’ in which radiative
cooling from low clouds increases relative humidity and enhances cloud
amount. A stronger beta effect would be expected to increase low level
cloud fraction and also to amplify any low cloud feedback.
•  Brient and Bony (2012) tested this idea in the IPSL models by making
clouds transparent to radiation.
•  Experiments to make clouds transparent to longwave radiation are proposed
for CMIP6/CFMIP-3. These will allow importance of the beta effect for the
shortwave cloud feedback to be assessed across models.
© Crown copyright Met Office
amip4K Cloud Feedbacks over 30ON/S Oceans
Precipitating/
Deep
Stable/
Shallow
Precipitating/
Deep
Stable/
Shallow
Stable/
Shallow
Black diamonds
mark significant
correlations with
the net cloud
feedbacks
in same regime
Stable/
Shallow
Stable/
Shallow
Precipitating/
Deep
Stable/
Shallow
Precipitating/
Deep
Stable/
Shallow
Grey squares
mark significant
correlations with
the net cloud
feedback area
averaged
over tropical
oceans
How might cirrus cloud regulate shallow cloud
feedback?
•  Models with weaker precipitation efficiencies and stronger vertical transports
of water vapour might produce more or thicker cirrus clouds above shallow
cloud regimes.
•  Could this somehow encourage the breakup of low level clouds in the
warmer climate by affecting the circulation or the downwelling longwave
flux?
•  We can test the idea that the radiative effects of cirrus influence the low
cloud feedback by making cirrus transparent to longwave radiation.
© Crown copyright Met Office
HadGEM2-A amip4K Shortwave Cloud Feedback
Standard HadGEM2-A
HadGEM2-A with invisible cirrus
HadGEM2-A with invisible cirrus
minus Standard version
Precipitating/
Deep
Making cirrus clouds transparent to radiation
decreases the shortwave cloud feedback
parameter by just 0.1 Wm-2K-1 – not that big
an effect compared to the overall range of
1.1 Wm-2K-1
Stable/
Shallow
Precipitating/
Deep
Stable/
Shallow
Conclusions
•  Differences in parametrized convection can have a substantial impact on
cloud feedback in models, but do not explain the overall range in the ten
models examined here.
•  Models with smaller amounts of mid level cloud and moist static energy at
the top of the boundary layer tend to have more positive cloud feedbacks.
•  Building on the ideas of Sherwood et al. 2014, this could be explained by
regulating effects of large scale precipitation efficiency and/or lower
tropospheric mixing by resolved or turbulent processes.
•  Such relationships could be used to develop ‘emergent constraints’.
•  More work is required however to understand and test the underlying
mechanisms.
•  We can test competing hypotheses by suppressing processes or
interactions in climate model experiments – ie further ‘mechanism denial’
experiments.
© Crown copyright Met Office