We here define heat content as the vertically integrated temperature, relative to a reference value (taken was the domain mean) from the top to the bottom of the ocean. In the model, 28% of the heat content variability is driven by seasonal and interannual fluctuations of the atmospheric heat flux, which impacts the perennial and seasonally ice-free regions. On the other hand, 44% of the variability is driven by oceanic exchanges with the North Atlantic Ocean, via northward-moving ocean currents that enter the Arctic Ocean through Fram Strait and the Barents Sea Opening. A small contribution comes from heat entering the Arctic Ocean via Bering Strait.
So, how well is our autonomous observing system doing in monitoring this variability? First, we consider the combination of satellite sea surface temperature observations (in ice-free waters) and ice-tethered buoy profilers (in the upper 800 m of ice-covered waters). We find that these two types of observations reproduce most, but not all, of the total variability. A sensitivity experiment indicates the crucial need for more observations in the Eurasian Basin of the Arctic Ocean. Next, we consider the radius of influence of ocean mooring data, from both within the Arctic Ocean and at its inflow/outflow straits. We find surprisingly long-range influence for many, but not all, sites.