Date post: | 28-Mar-2015 |
Category: |
Documents |
Upload: | leah-townsend |
View: | 214 times |
Download: | 0 times |
Robin HoganRobin Hogan
(with input from Anthony Illingworth, (with input from Anthony Illingworth, Keith Shine, Tony Slingo and Keith Shine, Tony Slingo and
Richard Allan)Richard Allan)
Clouds and climateClouds and climate
Overview• The importance of clouds feedbacks
– Feedbacks associated with specific cloud types
• Getting clouds right in current climate models– Evaluation of simulated clouds (e.g. using A-train data)– Accurate radiation schemes (e.g. cloud inhomogeneity)
• Tackling feedbacks and model cloud schemes– “Analogues” for global warming– Using new observations as a tight constraint on model
development– Convection and high-resolution modelling
Cloud feedbacks
• Main uncertainty in climate prediction arises due to the different cloud feedbacks in models that are not associated with aerosols!
IPCC (2007)
Key cloud feedbacks• Boundary-layer clouds
– Many studies show these to be most sensitive for climate– Not just stratocumulus: cumulus actually cover larger area– Properties annoyingly dependent on both large-scale divergence
and small-scale details (entrainment, drizzle etc)
• Mid-level and supercooled clouds– Potentially important negative feedback (Mitchell et al. 1989)
but their occurrence is underestimated in nearly all models
• Mid-latitude cyclones– Expect pole-ward movement of storm-track but even the sign of
the associated radiative effect is uncertain (IPCC 2007)
• Deep convection and cirrus– climateprediction.net showed that convective detrainment is a
key uncertainty: lower values lead to more moisture transport and a greater water vapour feedback (Sanderson et al. 2007)
– But some ensemble members unphysical (Rodwell & Palmer ‘07)
Evaluating models
AMIP: massive spread in model water content - need some observations!90N 80 60 40 20 0 -20 -40 -60 -8090S
0.05
0.10
0.15
0.20
0.25
Latitude
Ver
tical
ly in
tegr
ated
cl
oud
wat
er (
kg m
-2)
Delanoe and Hogan (2008)
Observed ice water content
UM ice water content
• A-Train can now provide this via new techniques combining the radar and lidar
July 2006 global IWC comparison
• Too little spread in model
• Better than AMIP comparison implied!
Tem
pera
ture
(˚C
)
A-Train Model
• Much more detailed evaluation of models (including high resolution ones) will proceed within NCEO and CASCADE…
• NCAS should be involved in using these comparisons to improve the model
Cloud structure in radiation schemes
TOA Shortwave CRF TOA Longwave CRF
Tripleclouds minus plane-parallel (W m-2)
Current models:Plane-parallel
Fix only overlap
Fix only inhomogeneity
New Tripleclouds scheme: fix both!
With help from NCAS CMS, Jon Shonk shortly to implement interactively in Met Office climate model
“Analogues” for global warming
• A model that predicts cloud feedbacks should also predict their dependence with other cycles, e.g. tropical regimes– Tropical boundary-layer clouds in
suppressed conditions cause greatest difference in cloud feedback
– IPCC models with a positive cloud feedback best match observed change to BL clouds with increased T (Bony & Dufresne 2005)
• Apply to other cycles (seasonal, diurnal, ENSO phase…)– Can we use such analysis to find
out why BL clouds better represented?
– Novel compositing methods?– Can we “throw out” bad models?
Convective Suppressed
Bony and Dufresne (2005)
Models with most positive cloud
feedback under climate change
Other models
Observations
Mixed-phase clouds• Potentially strong negative feedback
– Warmer climate more clouds in liquid phase more reflective& longer lifetime (Mitchell et al. 1989)
– But mid-level clouds underestimated in nearly all models
− Suggested approach: single column modelling over Chilbolton with different parameterizations
− Evaluate against radar/lidar observations
Radiative transfer
Turbulent mixing
Freezing
Sublimation
Entrainment of nucleating aerosol
Key processes
Further activities required• Using observations in model development
– Climate models in NWP mode (or single column version forced by large-scale tendencies – preferred by Pier Siebesma)
– Re-run many times with different physics and compare to single radar/lidar sites (or A-train observations for global runs)
– Remove unjustified complexity (e.g. double-moment ice?)
• Deep convection– Need to bite the bullet and modify the convection scheme in
the light of cloud-resolving runs (e.g. CASCADE)?– Observational constraint on water vapour detrained from
convection, e.g. combination of AIRS and CloudSat?
• Even more tricky areas– Is there any hope of getting a reliable long-term cloud signal
from historic datasets (e.g. satellites)?– How do we get cloud feedback due to storm-track movement?– Coupling of clouds to surface changes, e.g. in the Arctic?