Using A-train observations to evaluate clouds in CAM

Using A-train observations to evaluate clouds in CAM

Jennifer Kay (NCAR/CSU)

Andrew Gettelman (NCAR) Thanks to Hugh Morrison (NCAR)

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.

CloudSat’s 94 GHz (3 mm) radar measures cloud particles, raindrops, and snowflakes.

CALIPSO’s 532/1064 nm lidar measures aerosols and thin clouds.

Spaceborne radar and lidar 101

CloudSat and CALIPSO Data Sampling

Example radar ‘quicklook’ showing tropical convection - February 7, 2008

1400 km

30 km

Global Zonal Mean Cloud Fraction(CloudSat+CALIPSO)

QuickTime™ and a decompressor

are needed to see this picture.

More data plots: http://www.cgd.ucar.edu/cms/jenkay/

The A-train satellite data provide a unique view of Arctic clouds.

CloudSat+CALIOP(radar+lidar)

ISCCP D2(infrared)

Warren(surface obs.)

DJF Low Cloud Maps

How do we use CloudSat data to evaluate CAM’s 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)

The importance of cloud definition

ISCCP D2Warren Surface Obs.

CloudSat/CALIOP CAM 3.6 (CAM 3.5 + MG)

JJA low cloud cover

Variability in the short CloudSat record

Western Arctic cloud reductions from 2007 to 2006 are associated with differing

atmospheric circulation patterns. New record sea ice

extent minimum, Sept. 16 2007

Credit: NSIDC

2007 cloud reductions contributed to dramatic sea ice loss.

Kay et al. (submitted to GRL)

Overall Goal:Apple-to-Apple Comparisons

CloudSat vs. “CAM-dev”

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

Some important cloud definitions…cloud -30 dBZ < cloud < 10 dBZcloud fraction cloud #/ total # -cloud fraction can be “by-profile” or “by-height”

“CAM-dev” CAM 3.6 CAM 3.5 + MG microphysics + empirical radar reflectivity simulator; 3 years, 6-hourly output

JJA Low Cloud Fraction Maps

CloudSat Observations (1-3 km, “by-profile”)

CAM 3.6 (1-3 km, “by-profile”)

CAM 3.6 (from standard diagnostics)

CAM 3.6 (7-22 km, “by-profile”)

DJF High Cloud Fraction Maps

CloudSat Observations (7-22 km, “by-profile”)

CAM 3.6 (from standard diagnostics)

Tropical Comparison(CFAD, Cloud fraction “by-height”)

Sub-tropics Comparison(CFAD, Cloud fraction “by-height”)

Mid-Latitude Storm Track (CFAD, Cloud fraction “by-height”)

Future Plans

- 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@ucar.edu) or Talk to me later.

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…