1. The document analyzes the role of clouds in modifying estimates of the direct radiative effect of aerosols using satellite observations from CALIPSO and CloudSat.
2. It finds that the global mean direct radiative effect is -1.9 W/m^2, agreeing with prior estimates. However, there are significant regional differences when compared to estimates from the CESM climate model.
3. These differences may be partly due to biases in the model's representation of cloud cover, as patterns in cloud fraction biases correspond to patterns in direct radiative effect biases.
ONLINE SCALABLE SVM ENSEMBLE LEARNING METHOD (OSSELM) FOR SPATIO-TEMPORAL AIR...
ccstm14_matus
1. University of Wisconsin–Madison
Dept. of Atmospheric and Oceanic Sciences
amatus@wisc.edu
The Role of Clouds in Modifying
Global Aerosol Direct Radiative Effects
Alex Matus, T. L’Ecuyer, J. Kay, C. Hannay, J. Lamarque
CALIPSO/CloudSat STM
Alexandria, Virginia
5 November 2014
2. What are aerosol direct effects? Intro > Objective > Approach > Results > Recap
Eyjafjallajokull,
Iceland
Direct Radiative Effect:
Net SW flux perturbation at TOA
due to the presence of aerosol
DRE<0: net cooling effect
(aerosol brighter than underlying surface)
DRE>0: net warming effect
(aerosol darker than underlying surface)
3. What are aerosol direct effects? Intro > Objective > Approach > Results > Recap
Eyjafjallajokull,
Iceland
DRE<0
DRE>0
Direct Radiative Effect:
Net SW flux perturbation at TOA
due to the presence of aerosol
DRE<0: net cooling effect
(aerosol brighter than underlying surface)
DRE>0: net warming effect
(aerosol darker than underlying surface)
4. Challenges and opportunities Intro > Objective > Approach > Results > Recap
Challenge: Assessing DRE requires knowledge of the vertical structure
of clouds and aerosols, which passive sensors lack.
Opportunity: Satellite-based estimates of DRE may be improved by
leveraging observations from active sensors aboard the A-Train!
Aqua
CloudSatCALIPSO
PARASOL
Aura
5. Intro > Objective > Approach > Results > Recap
Inputs
Cloud properties:
Aerosol properties:
Surface albedo:
Temp., humidity:
CloudSat, CALIPSO
CALIPSO
MODIS, AMSR-E
ECMWF
Algorithm
Two-stream, adding-doubling RTM
• Optical properties assigned to CALIPSO
aero layers using values in SPRINTARS
• SW/LW fluxes at 125 vertical levels
• Fluxes computed twice:
1. Once with aerosol included
2. Once with aerosol excluded
Outputs
Profiles of radiative flux (Wm-2)
(DRE = difference in net SW
flux with and without aerosol)
CloudSat 2B-FLXHR-LIDAR
CloudSat Radiative Fluxes and Heating Rates (2006–2011)
Thin cloudsThick clouds Aerosols
30ºN 20ºN 10ºN 0º40ºN
0
z (km) 8
16
6. Intro > Objective > Approach > Results > RecapAddressing uncertainty
Overall, FLXHR-LIDAR agrees well with CERES
• Comparisons of TOA fluxes against collocated CERES indicate
biases of <4 Wm-2 on global/monthly scales.
Uncertainty estimated through rigorous sensitivity studies
• We have investigated possible sources of error due to assumed
aerosol optical properties and retrieved CALIPSO AOD:
• All aerosols except marine set to dust
• All aerosols except marine set to smoke
• Smoke changed to dust
• Factor of 2 error assumed in retrieved AOD
• A conservative estimate of the product’s accuracy
7. Intro > Objective > Approach > Results > RecapAnnual mean DRE (global, all-sky)
Warming effectCooling effect
Estimates using similar methods All-sky DRE (Wm-2)
Present study -1.9 ± 0.6
Su et al., 2012 -1.92
Heald et al., 2013 -2.03
Hatzianzianastassiou et al., 2007 -1.62
Yu et al., 2006 -1.9
8. Intro > Objective > Approach > Results > RecapComparison with CESM
Good agreement globally,
but poor agreement regionally!
FLXHR-
LIDAR
CESM-
CAM5
CESM (Community Earth System Model)
• Fully-coupled atmosphere/ocean
• Community Atmospheric Model version 5
• AMIP-style run with climatological SST/sea ice
• Aerosol effects computed in a similar way to
2B-FLXHR-LIDAR
9. Intro > Objective > Approach > Results > RecapComparison with CESM
FLXHR-
LIDAR
CESM-
CAM5
CESM (Community Earth System Model)
• Fully-coupled atmosphere/ocean
• Community Atmospheric Model version 5
• AMIP-style run with climatological SST/sea ice
• Aerosol effects computed in a similar way to
2B-FLXHR-LIDAR
Model – obs DRE
10. Intro > Objective > Approach > Results > RecapComparison with CESM
Why are there regional biases?
Model – obs DRE
11. Intro > Objective > Approach > Results > RecapAre DRE biases due to aerosol?
Model – obs DREModel – obs AOD
AOD explains some but not all DRE biases.
CESM-CAM5 overestimates AOD of absorbing aerosol over the Sahara and
underestimates AOD of scattering aerosol over SE Asia (Shindell et al., 2013).
12. Model – obs cloud fraction
Intro > Objective > Approach > Results > RecapAre DRE biases due to clouds?
Model – obs DRE
Over subtropical ocean, we see similar
patterns in cloud fraction and DRE biases.
Observed cloud fraction: CALIPSO/CloudSat
Model cloud fraction: CALIPSO-GOCCP cloud simulator
13. Model – obs cloud fraction
Intro > Objective > Approach > Results > RecapAre DRE biases due to clouds?
Model – obs DRE
Over subtropical ocean, we see similar
patterns in cloud fraction and DRE biases.
Observed cloud fraction: CALIPSO/CloudSat
Model cloud fraction: CALIPSO-GOCCP cloud simulator
14. More cloud,
weaker DRE
Less cloud,
stronger DRE
Intro > Objective > Approach > Results > RecapMarine aerosol in the SE Pacific
FLXHR-
LIDAR
CESM-
CAM5
Bias in cloud cover likely contributes to
some of the bias in aerosol direct effects.
Low clouds are defined
by CTT>0ºC
CALIPSO-GOCCP
CALIPSO/CloudSat
15. Biomass burning over the SE Atlantic Intro > Objective > Approach > Results > Recap
DJF MAM JJA SON
FLXHR-
LIDAR
CESM-
CAM5
Dark smoke over bright cloud produces a positive DRE.
The seasonal cycle in biomass burning
results in a seasonal cycle in DRE.
16. Biomass burning over the SE Atlantic Intro > Objective > Approach > Results > Recap
Increased cloudiness enhances the warming effect
exerted by absorbing aerosol over cloud (Chand et al., 2009).
Low cloud
fraction (%)
DRE
(Wm-2)
AOD
Biomass burning
17. Intro > Objective > Approach > Results > Recap
Model is skewed more
toward clear-sky, resulting
in stronger DRE.N
Cloud fraction
Aerosol DRE sensitive to cloud cover
Global ocean
DRE
(Wm-2)
Both datasets show similar
sensitivity of DRE to cloud
fraction.
18. Intro > Objective > Approach > Results > Recap
Model is skewed more
toward clear-sky, resulting
in stronger DRE.
Aerosol DRE sensitive to cloud cover
Model – obs cloud fraction Model – obs DRE
N
Cloud fraction
19. New observational estimates of aerosol direct effects
• Improved detection of clouds and aerosols using A-Train sensors
• Capability to assess DRE over bright surfaces (e.g. cloud, desert)
Our global estimate agrees well with CESM-CAM5
• FLXHR-LIDAR (-1.9) vs. CESM-CAM5 (-1.7)
• However, there are significant regional differences
Cloud cover may explain some of the DRE biases
• Other biases may include aerosol sources and optical properties
• Results may help further evaluate anthropogenic forcing in models
Summary Intro > Objective > Approach > Results > Recap
21. Intro > Objective > Approach > Results > RecapWhy do obs and model differ?
FLXHR-
LIDAR
CESM-
CAM5
AOD alone cannot account for these differences.
22. Intro > Objective > Approach > Results > RecapWhy do obs and model differ?
FLXHR-
LIDAR
CESM-
CAM5
1. Model is not as cloudy.
2. Similar patterns in model DRE and CF.
CALIPSO-GOCCP
CALIPSO/CloudSat
23. Intro > Objective > Approach > Results > RecapAOD by CALIPSO aerosol type
a) Marine b) Smoke
c) Dust d) Polluted dust
e) Polluted continental f) Total
24. a) Marine b) Smoke
c) Dust d) Polluted dust
e) Polluted continental f) Total
Intro > Objective > Approach > Results > RecapAOD by CALIPSO aerosol type
25. Intro > Objective > Approach > Results > RecapSensitivity of DRE to cloud fraction
SE Pacific SE AtlanticGlobal ocean
Model is overall less cloudy, which results in
stronger (more negative) DRE over ocean.
FLXHR-LIDAR
N
Cloud fraction Cloud fraction Cloud fraction
DRE
(Wm-2)
26. Model Low Cloud Bias
Kay et al., 2012
Intro > Objective > Approach > Results > Recap
27. Intro > Objective > Approach > Results > RecapTable of DRE estimates
Clear-sky
Thin
cirrus
Other
clouds
All-sky
Land -2.8 -2.6 0.1 -1.5
Ocean -2.9 -2.7 -1.3 -2.0
Global -2.9 -2.7 -0.9 -1.9
Clear-sky Cloudy-sky All-sky
Land -2.0 1.0 -0.5
Ocean -3.8 -0.9 -2.1
Global -3.3 -0.4 -1.7
CESM-CAM5 annual mean DRE (Wm-2)
FLXHR-LIDAR annual mean DRE (Wm-2)
28. Intro > Objective > Approach > Results > RecapTable of DRE estimates
Clear-sky
Thin
cirrus
Other
clouds
All-sky
Land -2.8 -2.6 0.1 -1.5
Ocean -2.9 -2.7 -1.3 -2.0
Global -2.9 -2.7 -0.9 -1.9 ± 0.6
FLXHR-LIDAR annual mean DRE (Wm-2)
Clear-sky
Thin
cirrus
Other
clouds
All-sky
Land --- --- --- ---
Ocean -5.4 ± 0.91 --- --- -1.8 ± 0.22
Global --- --- --- ---
IPCC satellite-based estimates of annual mean DRE (Wm-2)
1 IPCC AR4, 2007
2 Manalo-Smith, 2005
29. Reference Instrument(s) DRE (clear-sky, ocean)
Bellouin et al. (2005)
MODIS, TOMS,
SSM/I
-6.8
Loeb and Manalo-Smith (2005) CERES, MODIS -5.5
Remer and Kaufman (2006) MODIS -5.7
Christopher and Zhang (2004) CERES, MODIS -5.3
Previous estimates of DRE Intro > Objective > Approach > Results > Recap
Observation-based DRE (Wm-2)
Model-based DRE (Wm-2)
IPCC AR4, 2007
Reference Model DRE (clear-sky, ocean)
Chin et al. (2002) GOCART -4.1
Takemura et al. (2002) SPRINTARS -1.6
Koch and Hansen (2005) GISS -3.5
Reddy et al. (2005) LMDZ-LOA -2.3
7
Editor's Notes
----- Meeting Notes (10/24/14 15:03) -----
all surfaces novel
consistent with similar studies, measurements