Using Atrain observations to evaluate clouds in CAM

1
Using A-train observations to evaluate clouds in
CAM

• Jennifer Kay (NCAR/CSU)
• Andrew Gettelman (NCAR)
• Thanks to Hugh Morrison (NCAR)

2
Spaceborne radar and lidar 101

• Active instruments such as radar or lidar
• emit a pulse.
• The pulse is either reflected back to the
  instrument, continues downward, or is absorbed
  and lost.
• The reflected signal is a measure of vertical
  cloud and aerosol structure.

CloudSats 94 GHz (3 mm) radar measures cloud
particles, raindrops, and snowflakes. CALIPSOs
532/1064 nm lidar measures aerosols and thin
clouds.



3
CloudSat and CALIPSO Data Sampling
4
Global Zonal Mean Cloud Fraction(CloudSatCALIPSO
)
More data plots http//www.cgd.ucar.edu/cms/jenka
y/
5
The A-train satellite data provide a unique view
of Arctic clouds.
DJF Low Cloud Maps
CloudSatCALIOP (radarlidar)
ISCCP D2 (infrared)
Warren (surface obs.)
6
How do we use CloudSat data to evaluate CAMs
clouds?

• Important factors to consider
• How do we define a cloud? (radar sensitivity)
• - Are these data representative? (short data
  record)

• Clear advantages of CloudSat data
• first measure of global cloud vertical structure
• measured cloud quantities such as dBZ can be
  directly compared to simulated model cloud
  quantities (w/MG microphysics)

7
The importance of cloud definition
JJA low cloud cover
8
Variability in the short CloudSat record
Western Arctic cloud reductions from 2007 to 2006
are associated with differing atmospheric
circulation patterns.
9
Overall GoalApple-to-Apple ComparisonsCloudSat
vs. CAM-dev
CAM-dev ? CAM 3.6 ? CAM 3.5 MG microphysics
empirical radar reflectivity simulator 3 years,
6-hourly output
Some important cloud definitions cloud ? -30 dBZ
lt cloud lt 10 dBZ cloud fraction ? cloud / total
-cloud fraction can be by-profile or
by-height

• TODAY, preliminary comparisons of
• Global low cloud cover
• Global high cloud cover
• dBZ-ht histograms, cloud profiles in specific
  regions

10
JJA Low Cloud Fraction Maps
CloudSat Observations (1-3 km, by-profile)
CAM 3.6 (1-3 km, by-profile)
11
DJF High Cloud Fraction Maps
CloudSat Observations (7-22 km, by-profile)
CAM 3.6 (7-22 km, by-profile)
12
Tropical Comparison(CFAD, Cloud fraction
by-height)
13
Sub-tropics Comparison(CFAD, Cloud fraction
by-height)
14
Mid-Latitude Storm Track (CFAD, Cloud fraction
by-height)
15
Future Plans
Conclusions

• CloudSat data are a unique tool for evaluating
  the representation of clouds in next-generation
  climate models.
• Cloud definition is key to useful comparisons.
• Much more work to be done

• Add CFMIP ISCCP/CloudSat/CALIPSO simulator to
  CAM
• Use DART to constrain CAM dynamics, look at
  clouds
• Actively engage with model evaluation efforts
  for CAM4

PLUG Does your work incorporate model-obs cloud
comparisons? I can provide cloud data to help you
evaluate model performance E-mail me
(jenkay_at_ucar.edu) or Talk to me later.