Snow and sea ice thermodynamics in the Arctic: model validation against CHINARE2003 and SHEBA data

1
Snow and sea ice thermodynamics in the Arctic
model validation against CHINARE2003 and SHEBA
data Bin Cheng1) and Timo Vihma2) 1)Finnish
Institute of Marine Research 2)Finnish
Meteorological Institute

• Study the variability of snow on sea ice mass
  balance
• Study the effect of model resolution
• Study the effect of model forcing
• in situ observations,
• ECMWF operational analyses and forecasts
• NCEP/NCAR re-analysis

2


C
Ice with
q(z )
T(z )
V(z )
a
a
a
snow cover
z
SBL
T
a
h
sfc
D
s
h
T
s
snow
T
Surface
Q
si
sfc
T
h
D
in
x
i
surface layer
F
si
T
i
Ice
0
z
h
D
i
T
f
Ocean
-Surface heat balance (with air-ice interaction
- Penetrating solar radiation
in snow/ice - Sub-surface
melting - Snow to ice
transformation (snow-ice and superimposed ice
formation) - Heat and mass balance
at ice-ocean interface

Structure of HIGHTSI model



0?C
0?C
0?C
0?C
3
CHINARE2003 results
The FY1-D satellite image of ice concentration in
the western Arctic on 28, August, 2003. White
dots indicate the initial position of Argos buoys
deployed during CHINARE03. The white cycle shows
one of the biggest ice floes where the Ice Camp
was set up.
The distribution of measured snow and ice
thickness during CHINARE03 (early Aug- early
September). Most of the snow and ice thickness
samples were collected during the Ice Camp
period.
4
CHINARE 2003 carried out in the Arctic
summer-fall season (July September,
2003) CHINARE 2008 to be carried out in the Arctic
5
(No Transcript)
6
The forcing data for CHINARE2003 Ice Camp
modelling The in situ measurements green lines
The ECMWF forecasts red lines The NCAR/NCEP
re-analysis blue lines
7
The surface temperature Observed Tsfc green
lines Modeled Tsfc high resolution model
(red) Coarse resolution model (blue) ECMWF,
NACR/NCEP broken line
Observed snow thickness (red cirles)
High/coarse resolution External forcing of
1) in situ measurement (red) 2) ECMWF forecast
(blue) 3) NCEP re-analysis (green). The
observations represent daily mean values around
the Ice Camp, and the spatial standard deviation
is indicated by vertical bars. The black line
from day 238 to 247 is the snow surface elevation
read by the position sensor at solely one
location near the Ice Camp.
8
Observations of snow and upper part of ice
temperature field during the Ice Camp period.
Snow surface elevation was interpolated by
diurnal measurements. The surface temperature was
derived from surface emitted longwave radiation.
The dotted line indicates a period of missing
surface temperature. The vertical coordinate is
zero at snow/ice interface. The initial ice
thickness was 1.8m.
Time series of modeled and observed snow
temperatures at (a) 2 cm, (b) 6 cm, (c) 10 cm,
and (d) 20 cm below the surface. Ob green High
resolution red Coarse resolution blue
9
The comparison of observed and modelled vertical
snow temperature profiles. In each sub-plot, the
gray line is the observed temperature profile
linked between 5 sensors (circles) linearly. The
thick solid line is the result of coarse
resolution model (3-layers), while the thin solid
line is the result of fine resolution model
(30-layers). Normalized coordinate (vertical
depth / snow thickness) is used in the y-axis
10
The ECMWF forecasts(red) and NCAR/NCEP
re-analysis for CHINARE 2003 seasonal period
from beginning of May to end of September.
11
The surface temperature calculated by large scale
models and HIGHTSI. The a) and c) give the Tsfc
from ECMWF and NCEP (black lines), and from
HIGHTSI (red lines). The b) and d) refer to the
Tsfc of the Ice Camp period, the green line is
the observation
12
Time series of modelled surface albedo and
various components of snow mass balance. The
modelling cases are X (red), Y (green) and Z
(blue).
13
Time series of modelled ice mass balance
components for experiments X (thick line), Y
(gray line) and Z (thin black line) (a)
accumulated surface and sub-surface ice melting,
(b) superimposed ice thickness, (c) accumulated
ice bottom melting, and (d) the total ice
thickness. The dot and vertical bar in (d) refers
to the mean ice thickness and ? standard
deviation measured during the Ice Camp period.
14
(No Transcript)
15
The SIMIP2 external forcing for thermodynamic sea
ice modelling. All the data are derived from
local in situ measurements
16
HIGHTSI modeled snow and ice mass balance against
SHEBA observation. Forcing data SIMIP2(Huwald,
et al. 2005). Precipitation x 1.5 Observed albedo
(melt pond effect?) Estimated oceanic heat flux
11W/m2 average Too much melting with coarse
resolution Improved results with superimposed ice
formation.
SHEBA Observed ice temperature field From
Perovich, et al. (2003)
17
(B)
Initial ice formation
(C)
Freezing season

(A)
Cooling procedure
0?C
T


(D)
(F)
Thermal equilibrium stage
Melting season (II)
z
(E)
Melting Season (I)
18

• Summary of CHINARE2003 and SHEBA modelling
• HIGHTSI provides good results compared with the
  CHINARE Ice Camp measurements, especially for a
  model run with a high vertical resolution.
• For the model run with coarse resolution, the
  temperature profile shows much less spatial
  variability than the measurements.
• The results for snow thickness strongly depend
  on the precipitation. In the seasonal scale, the
  ECMWF operational forecasts produced
  precipitation, which yielded realistic snow
  accumulation, while the precipitation in
  NCEP/NCAR reanalysis was unrealistically large.
• A time dependent surface albedo is critical for
  a seasonal sea ice modelling. A good albedo
  parameterisation scheme is even more essential
  than high accuracy in the external forcing.
• For seasonal model run, the coarse model
  resolution provide delay of onset ice melting
  compared with the high-resolution model. During a
  cold period, however, the model resolution does
  not affect much the results.