Arctic Ocean Freshening Linked to Anthropogenic Climate Change: All Hands on Deck

Arctic Ocean freshwater storage increased since the mid‐1990s, but the cause was unknown. Jahn and Laiho (2020, https://doi.org/10.1029/2020GL088854) use ensemble runs of a coupled climate model to suggest that the observed increase is anthropogenic. The paper quantifies when the anthropogenic signals should emerge from the noise of natural variability. This result contextualizes research on the Arctic Ocean freshwater system and sketches an unprecedented opportunity. Future work should elucidate mechanisms, seek to observe the anthropogenic freshwater changes, and investigate the impacts on biogeochemistry and the North Atlantic Ocean circulation.

. Climate model projections and observations of the Arctic Ocean freshwater cycle. The left and right subplots show the principal time series of freshwater (FW) inflows and outflows (km 3 yr −1 relative to a salinity of 34.80; positive poleward). The middle subplots show the freshwater storage in the Arctic Ocean as sea ice (solid, top) and liquid (bottom) freshwater (km 3 relative to 34.80). Results from the Community Earth System Model (CESM) control (gray), large ensemble (LE, purple), and low warming (LW, green) experiments are shown in each case, adapted from Jahn and Laiho's (2020) Figure 2. The subplots show the times when the models show emergence of a forced, anthropogenic signal (meaning the time of first permanent departure from the ±3.5 σ envelope of control variability, where σ is the standard deviation; horizontal and vertical lines). The observations synthesized by Haine et al. (2015) are plotted in red (with updates from de Steur et al., 2018, Woodgate, 2018, and Spreen et al., 2020; the liquid storage data are adjusted to match the Jahn & Laiho, 2020 Arctic Ocean control volume by excluding Baffin Bay). For estimates and discussion of the uncertainty in the observations, see Haine et al. (2015). The basemap shows the liquid freshwater content, which is the vertically integrated salinity anomaly relative to 34.80, based on Haine et al.'s (2015) Figure 6. PIOMAS data product Schweiger et al., 2011). The summertime sea ice is also now thinner and younger (Laxon et al., 2013;Lindsay & Schweiger, 2015). These sea ice changes are attributed to anthropogenic effects because they only occur in coupled climate models perturbed by anthropogenic forcing (Notz & Marotzke, 2012).
Overlapping with this decline in sea ice is a conspicuous increase in liquid freshwater volume stored in the Arctic Ocean. Measurements of seawater salinity show a remarkable freshening between 1992 and 2012 (Giles et al., 2012;McPhee et al., 2009;Proshutinsky et al., 2019;Rabe et al., 2011Rabe et al., , 2014. This development is also shown in Figure 1 by the red line in the bottom middle subplot (taken from the synthesis of Haine et al., 2015). It measures the volume of freshwater that dilutes the upper Arctic Ocean to form the halocline (it is the volume-integrated salinity anomaly).

10.1029/2020GL090678
Independent measurements of the ocean freshwater fluxes into and out of the Arctic over the last 20 years corroborate this finding. In the presence of large variability and uncertainty, they show increasing freshwater inflow through Bering Strait and nearly unchanged outflows through Davis and Fram Straits (with some shifts between the individual flux terms Haine et al., 2015). These data are shown by the red lines on the left and right subplots in Figure 1. The meteoric freshwater flux to the Arctic Ocean is also likely increasing since the 1980s and 1990s. The integrated effect of the changes in these freshwater fluxes, plus the loss of freshwater stored in sea ice, plausibly matches the increase in liquid freshwater stored in the Arctic (Haine et al., 2015). The question is what causes these freshwater changes?
Now Alexandra Jahn and Rory Laiho of the University of Colorado at Boulder have provisionally answered this question (Jahn & Laiho, 2020). They show that the Arctic Ocean liquid freshwater storage increase is anthropogenic, like the sea ice decline. They examine the simulated Arctic Ocean freshwater cycle from the 21st century projections of the Coupled Earth System Model (CESM), a climate model from the National Center for Atmospheric Research (thin colored lines in Figure 1). Jahn and Laiho (2020) use output from an ensemble of CESM runs, which is a set of many model projections that differ only in their natural, unforced climate variations. They consider two CESM ensemble experiments for the 21st century: the large ensemble (CESM-LE, Kay et al., 2015, purple lines) and the low warming ensemble (CESM-LW, Sanderson et al., 2017, green lines). The CESM-LE uses the IPCC RCP8.5 high-emission scenario and the CESM-LW uses a reduced emission scenario that stabilizes global warming at 2 • C for several decades before 2100.
Jahn and Laiho (2020) define a metric to quantify the forced (anthropogenic) signal relative to background noise. They use the CESM preindustrial control ensemble (gray lines) to characterize the natural variability in the Arctic freshwater system. Departures outside this variability envelope (horizontal lines) reveal the forced response. They define emergence as a permanent fluctuation away from the range of control variability, which is almost certainly anthropogenic (vertical purple and green lines). Applying this metric to the Arctic Ocean freshwater content, Jahn and Laiho (2020) find that all CESM ensemble runs show a permanent freshening (emergence of an anthropogenic signal) by the early 2020s. They conclude therefore that the observed increase in liquid freshwater storage in the real Arctic "is likely already driven by climate change." These are state-of-the-science methods that exploit ensembles of control and projection climate model experiments. Still, the results are provisional. The detection of forced changes in the Arctic freshwater system need to be confirmed in other coupled climate models and at higher spatial resolution. The CESM freshwater fluxes are biased compared to the observations in Figure 1, which affects the emergence metric in uncertain ways. For example, the model exports too much sea ice through Fram Strait and not enough liquid freshwater.
Despite these caveats the paper by Jahn and Laiho (2020) is an important advance. It clearly points to future research priorities because it frames the debate about the nature of the Arctic freshwater system. It contextualizes exploration of the processes at work, which are poorly-known. It projects how the system will change in the coming years and decades. And it provides a rationale to investigate downstream impacts on the North Atlantic Ocean. Most importantly, it revitalizes the arduous task of observing changes in the Arctic freshwater system.
The dynamics of the freshwater system emerge from the accumulation and interaction of many diverse mechanisms in the Arctic Ocean. Most of these mechanisms are poorly observed, poorly understood, and poorly modeled, for example, by the CESM. Arctic liquid freshwater is stored predominantly in the Canadian Basin (Figure 1 basemap), in particular in the Beaufort Gyre. The Beaufort Gyre is thought to be driven by a balance between the stress from anticyclonic winds encircling the Beaufort High in sea level pressure and ocean eddies (see, for instance, Manucharyan et al., 2016). Changes in the Beaufort High are believed to contribute to the observed Beaufort Gyre freshwater increase (Cornish et al., 2020).  Figure 1 for locations). The forced signals at Bering and Barrow Straits do not occur before 2100 in the CESM experiments because of smaller signals relative to the natural variability.
Although the CESM forced signal is strong at Nares Strait, observing the freshwater flux there is logistically challenging (Melling, 2011) and the extant time series is short (Melling et al., 2008). At Davis Strait the CESM forced signal emergence is imminent, the logistics are easier, and the records are longer. The CESM forced signal is weak at Bering Strait, but it is least challenging to observe because Bering Strait is narrow and shallow and the time series is relatively long and uninterrupted. In designing and interpreting data from a holistic Arctic freshwater measurement network, tradeoffs such as these must be carefully weighed. The Jahn and Laiho (2020) projections provide a rational basis to do so.
Anthropogenic change in the Arctic freshwater system also has broader impacts. For example, we know that Arctic freshwater affects ocean biogeochemistry, like phytoplankton community composition, primary production, and ocean acidification (Brown et al., 2020;Carmack et al., 2016). It also lowers the density of downstream surface waters in the sub-Arctic deep water formation sites. The Arctic outflows freshen the Greenland, Iceland, Irminger, and Labrador Seas, which are the source regions for surface waters entering the deep limb of the Atlantic meridional overturning circulation (AMOC). The outflows thereby tend to slow the AMOC, which is principally driven by temperature contrasts (the water sinks because it is cold; see Weijer et al., 2019 for a recent review). We know that the AMOC fluctuates naturally on time scales from days to centuries with widespread implications for climate variability (Zhang et al., 2019). To date, the observed AMOC variations appear to be natural and unforced (Haine, 2016). Nevertheless, we also know that the AMOC is very likely to weaken in the 21st century due to anthropogenic climate change (Collins et al., 2013). Jahn and Laiho (2020) illuminate the link between these ideas. They point to the prospect of observing and understanding changes in the Arctic freshwater system and its downstream effects. This chance to spectate on wholesale shifts in the climate system is an unprecedented scientific opportunity. It deserves an unprecedented scientific response.

Data Availability Statement
Alexandra Jahn provided the data and code to make Jahn and Laiho's (2020) Figure 2, which is used in Figure 1 here. Gunnar Spreen, Rebecca Woodgate, and Laura de Steur provided the data to extend the Haine et al. (2015) observational time series in Figure 1. The PIOMAS data are taken from this site (psc.apl.uw.edu/research/projects/arctic-sea-ice-volume-anomaly). The code and data to make Figure 1 are at this site (github.com/hainegroup/Arctic-Ocean-Freshwater-Synthesis).