Transcript Atmospheric Forcing

Atmospheric Forcing Fields for
Ice-Ocean Models
Axel Schweiger
University of Washington
Applied Physics Lab
Polar Science Center
Missing Atmosphere
• Need to specify energy exchange at the
ice/ocean atmosphere boundary
• All “forced” climate variability needs to be
transferred from the atmosphere to the
ocean
Overview
• Components of atmospheric forcing fields
• Sensitivity of simulations to forcing
• Data sources. Choices to make
• Highlights of some differences
• Problems
• Bibliography
P
S
E

F
QE
QH
Sea Ice
R
F
S
NetR=F -F+S-S
QH
QE
From Lindsay, 1998
How do you do it?
• Flux fields often not available
• Specify near surface state of “atmosphere” from better
•
•
known quantities (pressure, wind, near surface
temperature, humidity, …clouds)
Parameterize heat exchange as functions of near surface
atmospheric state and model SST/IST using bulk
formulae
In the open ocean: Model can be forced/restored to SST
directly
Wind Stress


   C U U
a D
z
z
 Surface Stress
 a : Density of air
C D : Neutral Stability Coefficien t for Drag
U z : Vector difference in between wi nd at height z and surface
Sensible Heat Flux
QH   c C
a p H

 z  TSST / IST  U
QH Sensible Heat Flux
 a : Density of air
c p : Specific heat of air
C H : Neutral Stability Coefficien t for Heat
 z : Potential temperatu re at reference height z
TSST / IST : Air/Surfac e interface temperatu re Sea/Ice/Sn ow Surface Temperatur e
U z : Difference in wind speed between wi nd at height z and surface
z
Latent Heat Flux

QE   C qz  q0  U
a E
z
QE Sensible Heat Flux
 a : Density of air
CE : Neutral Stability Coefficien t for Evaporatio n
q z , q0 : Specific humidity at reference height z and surface
U z : Difference in wind speed between wi nd at height z and surface
Turbulent exchange


   C U U
a D
z
z
QH   c C
a p H

 z  TSST / IST  U

QE   C qz  q0  U
a E
z
z
Depend on:
•Stability
•Roughness
•Reference Height z
Parameterized LW Radiation:
F
Longwave down Fluxes at Sheba
RMS: 12 Wm-2
R^2 : 0.96
Parameterization for LW:
(Efimova, 1961, Jacobs)
F clr
=*Tair4*(0.746+0.0066*q)
FcldFclr*(1+0.26)*Ct
Inputs:
• Ct: Cloud fraction
(Metobs)
• Tair : ETL Tower
• q : specific humidity
(saturated/ice)
Parameterized SW Fluxes
(SHEBA)
May - August
Shine Parm. Inputs:
• C: Cloud Fraction, Sheba Metobs
 a: albedo, ETL/Tower
 : optical depth, Inverted
monthly average
 Solar zenith angle
• e: water vapor pressure
R2
: 0.88
RMS
: 28 Wm2
S cldy 
53.5  1274.5 cos( )
(1  0.139(1.0  0.935a )
S o cos( ) 2
1.2 cos( )  (1  cos( ) * 0.001e _  0.0445)
 S clr (1  C )  S cldyC
S clr 
Stot
See Key et al. 1996 for a systematic evaluation of radiation
parameterizations in the Arctic
Sensitivity Studies:
• Wide range of studies examine sensitivities.
•
(Perturbation/Substitution)
1-D ice models
–
–
–
–
Maykut and Untersteiner, 1971: Snow
Shine and Crane, 1984: Clouds
Curry et al 2002: Various
Makshtas, 2007: Various
–
–
–
–
Harder et al., 2000: Wind
Rothrock and Zhang, 1996: Radiative Fluxes
Holland, 1993: about everything
Hunke and Holland, 2007: Different forcing data sets
• 3-D
Radiation:
Forcing Variable

H Ice
Downwelling shortwave
radiation S
10 Wm-2
20 cm
Downwelling longwave
radiation F
10 Wm-2
50 cm
Ice Export: 2%/Wm-2 (LW), Lemke et al, 2000
After Rothrock and Zhang, 1996
Wind: Fram Strait Ice Export
From Harder et al. 2000
Tair, p, u, q, c
Control NP forcing
NCEP clouds
Makshtas et al. 2007
Thermodynamic only Model
All NCEP forcing
Temporal Resolution
monthly
Daily
Curry at al. 2002
Hourly and 6-hourly
Not just ice: Changed Ocean Circulation
Hunke and Holland, 2007
Ocean Currents at 466 m depth
a) modified AOMIP b) original AOMIP
Sensitivity Summary
• Small changes in forcings have large
impact
• Smaller components of the heat budget
are still important!
• Thermodynamics-only models more
sensitive. Negative
Thermodynamics/Dynamics feedback
Sources for Forcing Data
• Global/Arctic domain
• In-situ Observations/Climatologies
• Weather Models (Reanalysis, Operational)
(data assimilation to improve initial
conditions for forecast)
• Satellite Data
• Hybrid Data Sets
• Reconstructions
Reanalysis Data Sets
Name
Period
Available
Temporal
Resolution
Spatial
Resolution
Assimilation
Scheme
Comment
NCEP/NCAR
Renalysis (1)
1948-present
6-hourly
T62 (2x2 Deg)
3D Var?
Most widely
used
NCEP/NCAR DOE 2
1979-2008
6-hourly
T62 ( 2x2 Deg)
3D Var
Some bug
fixes
ERA-15
1979-1993
6-hourly
T106 (1.1 Deg)
OI
ERA-40
1958-2002
6-hourlyu
T159 (1.1 Deg)
3D Var
ERA-Interim
1989-present
6-hourly
T255 (0.75
Deg)
4D Var
JRA-25
1979-2007
6-hourly
T106 (1.1 Deg)
3D Var
MERRA (NASA)
1979-present
Hourly
0.5 Lat, 0.77
Lon
3D Var
NCAR/CIRES
20th Century
Renalysis
1908-1958
6-hourly
T62 (2x2 Deg)
EKF
Only surface
pressure
assimilated
On the Horizon
• NCEP: CFSSR: Climate System Reanalysis
and Reforecast, 1979-2009, T382 (35 km),
Interactive Ocean/Sea Ice model. Starting
in 2010?
• JRA-55: 1958-2012. T319, 4D Var.
Production starting in 2009
• ERA-70 ?
Walsh and Chapman, 2009
Cloud Anomaly June-August 2007 ( w.r.t 2000-2006)
NCEP
Schweiger et al. 2008
MODIS
Adjusted Renalysis: NCEP-ADJ
NCEP DSW corrected to ERA-40: 29 Wm-2 RMS daily means (NCEP-ERA-40)
Precip
NCEP R1
MERRA
1-day sample of daily-averaged precip
Buoy locations by type from 1980-2006
TAD Buoy
Not all buoys are the same!
…and they don’t seem to measure the same thing
Annual cycle of
difference between in
situ observation and
NCEP R1 by
Observation Type
OBS – NCEP(R1)
OBS – NCEP (R1)
OBS – ERA40
Contours: ERA40 – NCEP(R1)
Circles: Obs – NCEP(R1)
Hats: NP – NCEP(R1)
Difference between ERA40 and NCEP R1 is largely due to incorporation of buoys!
Satellite-derived Products
Tskin, S , F
• Global:
– ISCCP 1983-2008, global, D: clouds, FD: radiative
fluxes
– SRB: Based on ISCCP, radiative fluxes, Tskin
• Arctic:
– TOVS (1980-2006, N60)
• Path-P: Cloud fraction, Tskin
• Path-P derived: Downwelling Longwave
• Path-P PFLX: Parameterized (SW, LW)
– APP-X: 1983-2004 AVHRR-derived, radiative fluxes,
Tskin (all-sky)
Option: Hybrid Data Sets
• Pick/choose parameters
• Correct/adjust others
– AOMIP Forcings
– Large and Yeager, 2004, 2008
– Modified AOMIP
– Roeske, 2006 (OMIP) based on ERA-15, NYF
• Adjusted NCEP radiation
AOMIP forcing set
Wind
From SLP NCEP/NCAR-R1
Air Temperature
NCEP/NCAR- R1
Humidity
Fixed to 90%
Clouds
OMIP v.2 (ERA-15 climatology)
Shortwave Radiation
Downwelling:
Parkinson and Washington, 1979 (Zillman. 1972) using OMIP clouds. fixed humidity
Longwave
Radiation
Net: Rosati and Miyakoda, 1988, OMIP clouds, NCEP/NCAR Tair
Model SST/IST, fixed humidity
Precipitation
Serreze and Hurst, 2000 Climatology
Large and Yeager, 2004, 2008
Variable
Source
Adjustment
Wind
NCEP/NCAR-R1
Spatially varying ncreased globally to
match Scatterometer
measurements (most in
tropics/high lats
Air Temperature
NCEP/NCAR-R1
Corrected to IABP/POLES in the
Arctic, Antarctic Minimum
Specified
Humidity
NCEP/NCAR-R1
Reduced spatially varying, 3%
minimum
Clouds
Not Needed
Shortwave
Radiation
ISCCP-FD
Reduced by 5% between 50S and
30N
Longwave
Radiation
ISCCP-FD
Reduced by 5 Wm-2 in the Arctic to
increase sea ice thickness
Precipitation
Multi-Source (Xie and Arkin)
Defaults to NCEP/NCAR in the Arctic
LY 2008=> CORE-1 (NYF) , CORE-2 Forcing (Inter-annual)
RÖSKE, 2006 (P-OMIP)
• Based on ERA-15 (1,2) /ERA-40 (version 36)
• NYF (representative annual cycle)
• Budget closure, adjustments to match
measured oceanic transport
http://www.omip.zmaw.de/omip/overview.php
Hunke and Holland, 2007
Variable
Source
Adjustment
Wind
LY-04, NCEP/NCAR-R1
Spatially varying increased globally to
match Scatterometer
measurements (most in
tropics/high lats
Air Temperature
LY-04, NCEP/NCAR-R1
AS LY04, Corrected to IABP/POLES in
the Arctic, Antarctic Minimum
Specified
Humidity
LY-04, NCEP/NCAR-R1 (additional reduction)
Limit to 100% over ice
Clouds
P-OMIP (Roeske, Version 2, ERA-15)
Shortwave
Radiation
As original AOMIP, Parkinson and Washington, P-OMIP
clouds
Same as AOMIP
Longwave
Radiation
As original AOMIP, Rosati and Miyakoda, 1988
Slightly different from AOMIP
because of different
Temperature
Precipitation
LY04 Normal Year
Defaults to NCEP/NCAR in the Arctic
Issues with Hybrids
• Choices are difficult to make
• Inconsistent Forcing Fields
• Adjustments may alter sensitivities
Sensible Heat Flux: Affected by Radiation
ISCCP-FD
Radiation
Observed
(Lindsay,1998)
Modified AOMIP
From Hunke and Holland, 2007
AOMIP
Model Validation
Ice Extent
NET Cloud Forcing (Radiative Effect) at SHEBA (1998)
Data from Intrieri 2002, Key and Wang, 2005
Albedo is low enough
Problems lack of Atmosphere
• Lack of feedback
Initial State
Cold Air
Ice
Coupled “Real” World
Cold Air
Ice
Warm Air
Ocean
Ice-Ocean Model World
Warm Air
Ocean
Model Ice Edge Movement
Cold Air
Ice
Warm Air
Ocean
Model Ice Edge Movement
From Griffies et al. 2008
Conclusions
• Models are highly sensitive to forcing
parameters
• Accuracy of forcing fields is still lacking
• Tuning of models remains a necessity
– But: Are sensitivities to climate variations
maintained?
• Keep in mind:
Lack of atmosphere ocean feedback