Sea Ice–Albedo Feedback and Nonlinear Arctic Climate Change - Arctic Sea Ice Decline: Observations, Projections, Mechanisms, and Implications - page 118

increase is gradual, consistent with the arcing decrease in
effective annual surface albedo (Plate 4). In the MPI model
a sharp increase occurs in the transition to ice-free tempera-
ture range, consistent with the kinked shape of the effective
annual albedo decline for that model. In the MPI model, the
SAF becomes very large (2.3 W m −2 °C −1 ) in the warmest
temperature range.
Figure 1 (bottom) shows the monthly contributions to the
SAF of the two models in the three temperature ranges. As
the warming progresses, there is a shift to earlier months
in the sunlit season. This shift allows the SAF to increase
even as the ice-free season appears and grows. Aside from
seasonal insolation variation, the early months of the sun-
lit season potentially contribute more to SAF than the later
months for two reasons:

1. Surface albedos are initially larger so there is the po-
tential for a larger albedo reduction as the ice is removed
exposing the low-albedo seawater. Plate 4 shows that Sep-
tember albedos are 0.1 to 0.2 lower than those in March at
the beginnings of the runs.
2. Atmospheric transmissivities are largest in the spring
and decline through the summer to a minimum in September
in both models. Ignoring multiple cloud-ground reflection,
the SAF is the product of the downward atmospheric trans-
missivity and the surface albedo change, so these two factors
compound each other.
The pattern of surface albedo decline in the CCSM3 model
(not shown) shows a plume of reduced albedo penetrating into
the half-cap region from the kara Sea, indicating an oceanic
influence in the decline. This interpretation is borne out by

Figure 3. (top) Polar versus Arctic temperature and (bottom) Arctic versus global temperature for MPI ECHAM5 (cir-
cles) and NCAR CCSM3 (plusses). All data have been boxcar filtered over a 5-year period.