Using GERB and CERES data to
evaluate NWP and Climate models
over the Africa/Atlantic region
Richard Allan, Tony Slingo, Ali Bharmal
Environmental Systems Science Centre, University of Reading
Sean Milton, Malcolm Brooks, Mark Ringer
Met Office, Exeter
Outline
• Use of GERB/CERES in model evaluation
– Evaluation of cloud in NWP model
– OLR anomaly over the W Sahara
– Current activities at the Hadley Centre
• Combining surface ARM data and top of
atmosphere radiation data to evaluate models
60S-60N clear ocean
ΔAlbedoc
ΔOLRc
Time series (Global, 1200, Model-GERB)
2003
2004
2005
2006
2007
2008
60S-60N clear ocean
Time series (Global, 1200, Model-GERB)
Major model updates
MS7
SEV2;V004/5
ΔAlbedoc
ΔOLRc
Edition 1
2003
2004
2005
2006
2007
2008
Clear-sky
Biases
2006
Shortwave
Longwave
All-sky
Allan et al. (2007)
QJRMS
All-sky
Biases
2006
Surface
albedo
Shortwave
Marine
stratocumulus
Mineral
dust
aerosol
Longwave
Cirrus outflow
Clear-sky
Convective
cloud
Allan et al. (2007)
QJRMS
Model evaluation: near-real time
• Change in model minus
GERB flux differences
• Relate to change in model
physics implementation
Model
SW
albedo
2005
2006
13th March | 14th March 2006
• Modified
convective
parametrization
•  Increased
convective
cloud cover
Tropical land: modelGERB albedo difference
Allan et al. (2007) QJRMS
• Comparisons with GERB and
CloudSat show improved simulation
of tropical convective cloud
60S-60N land
Time series (Global, 1200, Model-GERB)
Major model updates
SEV2;V004/5
Edition 1
ΔAlbedo
ΔOLR
MS7
Convective decay time-scale
Changes in boundary layer and convection scheme parametrization
Diurnal cycle of convection
6.3um BT < 225 K
- Diagnosing convective cloud fraction: (OLRc-OLR)/cloud fraction
- Comparable to narrowband method from SEVIRI
- Model delay in convective cloud? Peak in SEVIRI total cloud at
midday – spurious?
Gulf of Guinea
Model
CloudSat
5th
July
2006
19th
July
2006
Allan et al. (2007) QJRMS
Namibian Stratocumulus
Namibian Stratocumulus
Cloud liquid
water path
Reduction in
model bias
from June to
July 2006 relates to cloud
liquid water?
Allan et al. (2007)
QJRMS
Bias: model minus GERB; SSM/I; SEVIRI
Albedo
Liquid Water Path
Cloud
2004
JJA
12:00
2005
JJA
12:00
2006
JJA
12:00
2007
JJA
12:00
2008
JJA
12:00
60S-60N clear land
Time series (Global, 1200, Model-GERB)
ΔOLRc
MS7
Major model updates
SEV2;V004/5
ΔAlbedoc
Edition 1
2003
2004
2005
Increased Sahara albedo
2006
2007
2008
Improved Sahara albedo
Multi-model ensemble mean minus CERES EBAF 2000
LW difference
SW difference
• Compare AMIP climate model ensemble
– CCSM3, INMCM3, MIROC high/low, CGCM2.3, CNRM_CM3, HadGEM1
• Biases due to cloud and surface albedo
• Also models overestimate OLR over W. Sahara
OLR bias
• OLR bias
present for
clear-sky also
• Smaller for
EBAF
(MODIS
enhanced)
OLR bias
• OLR bias
present for
clear-sky also
• Smaller for
EBAF
(MODIS
enhanced)
• Bias present
for ERBE-like
processing
Use of CERES/GERB at the Met
Office/Hadley Centre (Mark Ringer)
• HadGEM1 was model used for IPCC AR4
• HadGEM2 will be submitted to AR5
– Basic physics as HadGEM1 but will include many ‘Earth System’
processes such as C-cycle, chemistry, etc
• HadGEM3 is currently being developed
– Focus on regional climate and decadal prediction
– Includes completely new prognostic cloud scheme
• CERES data fully integrated into model development and evaluation
– Evolving model versions and tests of parameterizations are
systematically evaluated using CERES products
• GERB: Monthly mean and monthly mean diurnal cycle
would be of most use initially
• Also note that these data are an important part of our work
on feedbacks and attempts to constrain climate sensitivity
Development of HadGEM3:
Use of CERES SW cloud forcing,
impact of removing new cloud scheme.
EXPERIMENT
EXPERIMENT
minus
CONTROL
CONTROL
EXPERIMENT
minus
minus
CERES
CERES
Comparison of GERB with HadGEM1:
Model forced with climatological SSTs
OLR
GERB
HadGEM1
RSW
Simulating radiative fluxes using ground-based retrievals
Nicky Chalmers, Robin Hogan
Lindenberg
Ice Water Content
Liquid Water Content
retrieval
GERB
TOA LW up
TOA SW up
0
3
6
9
12
Time [Hours]
15
18
21
24
SURFACE
TOA
Clear
Cloudy
High cloud
Lindenberg
CERES
clear
Using methodology to evaluate
NWP clouds
-40°C
0°C
-40°C
0°C
ECMWF: surface LW comparison
Clear
Cloudy
Cloud at all levels
Comparison of radiance and flux: Edwards/Slingo vs GERB
N. Ali Bharmal (ESSC)
Angular conversion factor for the Niamey GERB ARG pixel shows that
aerosol alone cannot explain the discrepancy between ES96 & GERB.
Ali Bharmal in collaboration with Helen Brindley
Skin T > sfc T
Time of agreement
Preliminary results comparing SBDART and ES96 RT radiances at TOA
using aerosol-free profiles suggests a disagreement in relation to the
surface/skin temperature difference.
Conclusions
• Continued near-real time comparisons between
GERB-1/SEV2 and Met Office forecast model
• Large model biases have been reduced
– Surface albedo (MODIS bare soil and seasonal vegetation)
– Convective cloud (convective decay time-scale)
– Seasonal dust climatology likely
• Remaining problems
– Marine stratocumulus too reflective
• Appears to relates to bulk water properties
– Detrainment of cirrus cloud over Gulf of Guinea
– Continued monitoring using GERB data
•
•
•
•
• www.nerc-essc.ac.uk/~rpa/GERB
Paper detailing previous Edition 1 comparisons published in QJRMS
www.nerc-essc.ac.uk/~rpa/PAPERS/allan_et_al07QJ.pdf
Initial comparisons between GERB/CERES and climate models
Combining surface and top of atmosphere radiation data to evaluate
models