Transcript PowerPoint Presentation - Changing the albedo in CSIM4

Changing the albedo in CSIM4
Julie Schramm
CCSM Software Engineering Group
[email protected]
http://www.cgd.ucar.edu/csm/models/ice-csim4/index.html
www.ccsm.ucar.edu
What is in the albedo parameterization?
• Surfaces: dry snow, melting snow, bare ice, open ocean
• Properties
– two spectral intervals: VIS, NIR
– Direct and diffuse
– Melt ponds are crudely approximated using (Ts) for the bare
ice albedo
• Dependencies
– Snow thickness, ice thickness, surface temperature
• Calculated at each ocean grid point, for each thickness
category, then area-weighted
www.ccsm.ucar.edu
What’s not in the parameterization?
•
•
•
•
Zenith angle dependence (direct = diffuse)
Variations due to surface topography
Snow age
Explicit melt ponds
www.ccsm.ucar.edu
Control Run Albedo
Surface VIS
NIR
BB
Dry
Snow
0.98 0.70
0.85
Wet
Snow
0.88 0.55
min min
0.72
min
Bare
Ice
0.78 0.36
0.58
Melting
Ice
0.71 0.29
min min
0.51
min
Open
water
NA
0.06
NA
www.ccsm.ucar.edu
Source code residence
• $CSMROOT/models/ice/csim4/src/source
• Source code is in module ice_albedo.F
– 185 lines
• Contains subroutine albedos
• Modified source code should be put in a separate
directory called src.ice in $SCRIPTS/$CASE/
• Albedo parameterization described in Scientific Guide
available at:
http://www.ccsm.ucar.edu/models/ccsm2.0/
• User’s Guide also available here
www.ccsm.ucar.edu
Base albedos
Simplest way to modify the albedo:
real (kind=dbl_kind), parameter ::
& albocn = 0.06
! Ocean albedo
&, albicev = 0.78
! Visible ice albedo for h > ahmax
&, albicei = 0.36
! Near-ir ice albedo for h > ahmax
&, albsnowv = 0.98
! Cold snow albedo, visible
&, albsnowi = 0.70
! Cold snow albedo, near IR
..
.
..
.
www.ccsm.ucar.edu
Other albedo parameters
real (kind=dbl_kind), parameter ::
& snowpatch = 0.02
! Parameter for fractional snow area
• Fractional area covered with snow : fs = hs/(hs + snowpatch)
• Represents that for relatively thin snow cover, there will be some
areas of bare ice
• Value of 0.02 provides best match with SHEBA data
www.ccsm.ucar.edu
Other albedo parameters
real (kind=dbl_kind), parameter ::
& ahmax = 0.5
! Thickness above which ice albedo is constant (m)
www.ccsm.ucar.edu
Temperature dependence parameters
&, dT_mlt = 1
! change in temp to give dalb_mlt albedo change (>0)
&, dalb_mlt = -0.075 ! albedo change per dT_mlt change in temp for ice
&, dalb_mltv = -0.100 ! albedo vis change per dT_mlt change in temp for snow
&, dalb_mlti = -0.150 ! albedo nir change per dT_mlt change in temp for snow
www.ccsm.ucar.edu
Beyond Basics
• Putting modifications into the
do loop in ice_albedo.F
do j=jlo,jhi
do i=ilo,ihi
.
.
enddo
enddo
• Global grid is decomposed
into local subdomains
• The physical portion of a
subdomain is dimensioned
(ilo:ihi, jlo:jhi)
• When adding arrays to this
loop, check the dimensions!
jmt_local
• Subdomains are dimensioned
imt_local x jmt_local
ilo:ihi
imt_local
ilo:ihi
imt_global
www.ccsm.ucar.edu
Troubleshooting
• The model won’t compile. For example, in ice.log file
“ice_albedo.F”, 1516-036 (S) Entity fsnow has undefined type.
– Verify that any arrays you’ve added to ice_albedo.F are
accessible via one of the use statements.
• I changed some parameters in ice_albedo.F, but the
answers aren’t changing…
– Be sure to delete all files in /ptmp/$LOGNAME/$CASE/ice/obj
and recompile.
• I’m getting funny lines in the albedo field…
– Make sure that any arrays you add to the do loop are
indexed as (ilo:ihi, jlo:jhi) or (imt_local, jmt_local), not
(imt_global, jmt_global).
www.ccsm.ucar.edu