Comparison on Cloud and radiation properties at Barrow between ARM/NSA measurements and GCM outputs Qun Miao and Zhien Wang University of Wyoming

1
Comparison on Cloud and radiation properties at
Barrow between ARM/NSA measurements and GCM
outputs Qun Miao and Zhien WangUniversity of
Wyoming
Contacts Qun Miao miao_at_uwyo.edu Zhien Wang
zwang_at_uwyo.edu
2. Data
1. Introduction
Model simulations Four climate models NCAR
CCSM3.0, GFDL CM2.0, GISS ModelE, UKMO-HadCM3
Climate of the 20th Century and committed climate
change experiments Period 01/1990 to 12/2010
https//esg.llnl.gov8443/index.jsp Ground-base
d measurements at Barrow, AK Combined
MMCR and MPL cloud boundaries 05/1998 to
02/2005 MWR liquid water path 01/1998 to
10/2007 SKYRAD downwelling longwave (LW)
radiation 02/1998 to 02/2008 GNDRAD
upwelling LW radiation 04/1998 to 02/2008
Arctic region has drawn lots attention due to its
rapid changes during the past decades and
potentially large impacts on global climate, less
understanding in the interactive feedback
processes, and also relatively poor
representations in models. Arctic clouds are
among the most difficult variables for model
simulations. One of the reasons is that Arctic
clouds are very difficult to validate. In this
study, we utilize ground-based radar and lidar
combined cloud properties at the ARM/NSA site
since 1998 to compare with the simulations of
four selected IPCC climate models. Together with
radiation measurements at the ARM/NSA site, the
dataset provides us an opportunity to examine
model simulated cloud vertical structure and
cloud radiative forcing.
3. Results
3c. Surface LW radiation and LW downwelling Cloud
Forcing (CF)
3a. Cloud occurrence Simulations 01/1990 to
12/2010 Observations 05/1998 to 02/2005

• Models simulate the annual cycles of surface LW
  radiation well, but simulations reach their peaks
  about one month earlier than the observations.
  There is about 20-30 W/m2 difference in
  magnitude.
• CCSM overestimates LW cloud forcing by about 20
  W/m2 GFDL simulations are very close to
  observations HadCM3 underestimates it by 5-10
  W/m2 .

3d. LWP vs LW Cloud forcing

• No model simulates the observed seasonal cycle
  of cloud occurrences well, especially for low
  clouds.
• For most months, CCSM and UKMO underestimate,
  GFDL and GISS overestimate total cloud
  occurrence.
• The seasonal cycle of total clouds is mainly
  determined by low clouds.
• The models simulate less middle level clouds
  than the observations.

• The differences between model simulated and
  observed LWP-cloud forcing relation indicate that
  model simulated cloud vertical structure and
  microphysical properties are different from
  observations.

• 3b. Cloud liquid water path
• Observations 01/1998 to 10/2007
• Models simulate LWP seasonal cycle well, but
  with a large difference in magnitude.
• GFDL and HadCM3 are very close to the
  observations in the cold season, but
  underestimate it in the warm season.
• CCSM and GISS overestimate LWP (up to 150 g/m2).

4. Conclusions

• Large discrepancies of cloud occurrence between
  model simulations and observations exist at
  Barrow, in both the seasonal trend and magnitude.
  
• Large LWP magnitude differences are present (up
  to 150 W/m2). GFDL and UKMO underestimate while
  CCSM and GISS overestimate LWP.
• Models simulate the seasonal cycle of surface
  upwelling/downwelling LW radiations well.
• Significant biases of downwelling LW cloud
  forcing are exist in some models.