Synthesis and Evaluation of Meridional Heat Transport from Mid-latitudes towards the Arctic

Meridional Energy Transport (MET), both in the atmosphere (AMET) and ocean (OMET), has significant
impact on the sea ice in the Arctic. We quantify the AMET & OMET from six reanalysis datasets and
investigate their relations with sea ice variation at annual to interannual scales. The results indicate that,
although the mean transport in all datasets agree well, the spatial distribution and temporal variation of
AMET and OMET deviate substantially. This leads to large difference in the low frequency variability of
AMET and OMET in subpolar regions. Only after 2010 the low frequency signals from ocean reanalysis
products agree well. A comparison with the observation from RAPID ARRAY and a high resolution NEMO
ORCA hindcast suggests the overall trend captured by all the ocean reanalysis products converge well.
For the atmosphere, the deviations between data sets mainly originate from temperature transport.
Presumably, the assimilated observations are not sufficient to constrain the MET. A more detailed
analysis on the linkages between the Arctic sea ice concentration (SIC) and AMET & OMET identifies a
strong anticorrelation between AMET/OMET and SIC at Greenland Sea and Barents Sea. However, with
respect to AMET-SIC relationships there are large differences among the datasets. Only ERA-Interim
data corroborates physical explanations hypothesized from climate models. With respect to the ocean
data, ORAS4 and SODA3 agree well on the relation between OMET and SIC, while GLORYS2V3 differs
from them. All the reanalysis products find strong links between AMET/OMET at subpolar latitudes and
ENSO, which hints towards teleconnections between MET at high latitudes and climate variability at
lower latitudes. As a result, our study suggests, since the reanalysis products are not designed for the
quantification of energy transport, the AMET and OMET estimated from reanalysis should be used with
great carefulness, especially when studying low frequency variability.