The distribution
and variation of cloud attributes in spring play important roles in
preconditioning September Arctic sea ice change. However, given the background
of increased warming both globally and in the Arctic, the characteristics of
potential connections between springtime clouds and September sea ice in different
areas of the Arctic Sea should be updated. In this paper, we analyze the impact
of springtime clouds on September Arctic sea ice using ERA5 radiation data,
MODIS cloud fraction (CF) and cloud water path data, and sea ice concentration
(SIC) data from the National Snow and Ice Data Center (USA). First, the climatological
spatial distribution characteristics of cloud microphysics properties (i.e., CF
and total water path (TWP)) and cloud radiation properties (i.e., longwave
cloud radiation effects (LWCRE) and shortwave cloud radiation effects (SWCRE))
in spring (2000–2017) in the Arctic region are presented. Then, the correlation
between cloud macroscopic properties and cloud radiation is presented, and the
response of sea ice to springtime cloud properties in different areas of
interest is discussed. Results show that the CF distribution decreases as the
SIC increases, and that the TWP distribution increases as the latitude
increases. The distribution of LWCRE is discontinuous over the Arctic Ocean and
no significant regularity is observed. In areas other than the Barents and
eastern Greenland seas and the Arctic Ocean to the north, the difference in
SWCRE is small. Additionally, Arctic CF and TWP are correlated positively
(negatively) with LWCRE (SWCRE). The correlation (both positive and negative)
between TWP and cloud radiation effects is not as notable as that with CF in
terms of significance and range. Over areas with a higher proportion of the marginal
ice zone, i.e., the Laptev and Kara seas and the Arctic Ocean to the north
(ROI1), and the Beaufort, Chukchi, and eastern Siberia seas and the Arctic
Ocean to the north (ROI4), the warming effect of spring LWCRE tends to enhance
sea ice melting in September, whereas the cooling effect of SWCRE exhibits the
opposite effect on September sea ice with a time lag of approximately 4 months.
The coefficient of determination in the multiple regression model, which can be
used to characterize the degree of explanation of the independent variable to
the dependent variable, indicates that CF variability of ROI1 significantly
explains 18.53% of the cause of sea ice loss. In ROI4, no significant
connection is found between springtime CF (and TWP) and sea ice loss in
September.
