Geoscience Reference
In-Depth Information
Figure 5. The 20C3M values for the energy budget residual in (5), which includes ocean heat flux convergence and
latent heat transfer due to sea ice motion.
model at different resolutions), outgoing longwave radiation
(36 and 37 W m -2 , respectively) exceeds absorbed shortwave
radiation (19 and 21 W m -2 ) by 15.5 and 18 W m -2 . Part of
the compensation comes from the budget residual (figure
5), for which models 1 and 2 rank fourth (7.0 W m -2 ) and
second (8.4 W m -2 ), respectively.
The remainder of the budget deficit for models 1 and 2
must be compensated by the sensible and latent heat fluxes,
of which the annual mean sensible heat flux dominates, with
values of 10.4 W m -2 and 11.5 W m -2 for the two models,
compared to about -1.8 W m -2 for annual mean latent heat
flux for both. Figure 6b shows the sensible heat flux for the
17 models, and it is apparent that models 1 and 2 (thick lines)
have the highest values of sensible heat flux in all months.
It is difficult to infer physical relationships between the high
sensible heat gain and low insolation from available model
outputs. One factor determining sensible heat flux is the tem-
perature difference between air at the surface and the sur-
face itself, plotted in figure 6c. Models 1 and 2 are the only
models for which surface air temperature exceeds surface
temperature throughout the melt season. It is possible that
the low values of melt season insolation in these models are
responsible for the fact that the surface remains colder than
the air at the surface, so that the shortwave flux deficiency
induces a compensating enhancement in sensible heat flux.
of several flux-related MJJA quantities. Table 4 shows that
the spread of S is most strongly correlated with the spread in
the net shortwave flux ( F SW , r = 0.95) which explains 90% of
its variance. the high correlation with albedo (a, r = -0.91)
and much lower correlation with the incoming shortwave
flux ( r = 0.43) suggests that albedo differences account for a
large part of the dependence of S on F SW . Longwave fluxes,
which include the effects of longwave cloud forcing, are less
important for the spread in the net MJJA surface flux.
The quantities specified externally in our diagnostic cal-
culation are not, of course, truly external, and most of the
variables in table 4 have substantial correlations with the
MJJA surface temperature, T M . the high correlation with
F LW ¯ is consistent with the idea that upwelling longwave flux
warms the overlying atmosphere which reradiates to the sur-
face (e.g., l98).
S and a are also closely correlated with T M ( r = 0.72 and
-0.77, respectively), as colder ice surfaces are expected to
have more snow and fewer melt ponds and hence higher
albedo. As the surface warms, albedo becomes smaller and
S increases, leading to the standard sea ice-albedo feedback
[e.g., Ebert and Curry , 1993]. In contrast to these results, the
radiation-albedo-thickness correlations inferred by EuW's
albedo tuning would be positive: to maintain a given thick-
ness under increased downwelling longwave radiation, the
albedo must increase (see their figure 3). Along the same
lines, one might expect albedo tuning to lead to a positive
correlation between albedo and downwelling shortwave
flux, since larger (smaller) values of albedo could be used to
counteract the effects of higher (lower) levels of insolation.
But the correlation of F SW ¯ and a found here is negative and
very small ( r = -0.09). Instead, the correlations in table 4
are consistent with the idea that the sea ice-albedo feedback
4.3. Correlations in the Ensemble Spread of S
the spread of S values among the coupled climate models
(table 3) is quite large (13 to 61 W m -2 ), and our results
suggest that a characterization of this spread may be helpful
in understanding the spread of h D and ultimately h . table 4
shows the correlations between the spread of S and the spread
Search WWH ::




Custom Search