Arctic Amplification of Marine Heatwaves under Global Warming

Marine heatwaves (MHWs) are extreme ocean warming events lasting days to months and already pose severe ecological and socioeconomic risks. They occur globally, including in the Arctic, where over the past four decades Arctic marginal seas experienced MHW intensities comparable to lower latitudes. As global warming increases ocean temperatures, species distributions shift poleward into the Arctic, yet the future evolution of Arctic MHWs and their risks remain uncertain. The study asks: how will MHWs and total heat exposures (THEs)—extremes defined relative to a fixed historical baseline—change in the Arctic under continued warming, and what processes drive these changes? By distinguishing MHWs (relative to a shifting, detrended baseline) from THEs (relative to a fixed historical baseline), the work aims to clarify risks to Arctic marine ecosystems and inform adaptation and conservation strategies.

Prior studies documented widespread MHWs, their intensification under global warming, and substantial ecological and socioeconomic impacts across regions including the Arctic. Observational analyses and modeling efforts have shown prolonged Arctic MHWs since the 1980s and identified sea ice retreat as a key factor influencing surface temperature variability and extremes. Earlier CMIP-based assessments noted model biases and uncertainties in reproducing MHW metrics, as well as the importance of clear definitions separating transient extremes from long-term warming contributions (THEs). Work on Arctic amplification and Barents Sea warming highlights enhanced oceanic and atmospheric warming at high latitudes, while studies emphasize potential improvements from higher model resolution and better representation of Arctic processes. This study builds on these foundations by systematically contrasting future changes in MHWs and THEs in the Arctic using CMIP6 across multiple SSP scenarios and attributing drivers to sea ice decline and long-term SST trends.

• Datasets: Historical observations from OISST v2.1 (daily SST and SIC; 0.25° resolution). Model outputs from CMIP6 daily SST and SIC: 18 models (historical), 17 (SSP126), 16 (SSP245), 15 (SSP370), and 18 (SSP585).
• Periods and domains: Historical baseline 1985–2014; future projection 2071–2100; Arctic Ocean defined as ocean area north of 65°N. Multi-model mean (MMM) computed per model/scenario first, then averaged.
• Event definitions: Following Hobday et al. with modifications. Events occur when temperature exceeds the 90th percentile threshold for at least 5 consecutive days, allowing gaps <3 days. Baseline and threshold for each calendar day are computed using all daily data within an 11-day window centered on that day from a 30-year period, then smoothed with a 31-day moving average.
• MHW vs THE calculation: MHWs computed on detrended SST relative to a shifting baseline (trend removed within each period). THEs computed on original SST relative to a fixed historical baseline (1985–2014). Thus, MHWs exclude long-term warming; THEs include both long-term warming and transient extremes.
• Sea ice treatment: OISST sets SST in ice-covered grid cells (SIC >35%) to freezing point; the study uses 35% SIC as threshold. MHWs/THEs are considered absent under sea ice (SIC >35%). Open-water determination uses daily SIC.
• Metrics: Mean intensity (average anomaly above baseline over event days), annual total days (sum of event days per year), mean frequency (events per year), and mean duration (days per event). Monthly mean intensities for seasonality are computed by averaging daily intensities over event days within each month (to handle events spanning months).
• Change definition: Future changes are differences between 2071–2100 and 1985–2014 unless otherwise noted. Statistical significance testing across models indicates most Arctic changes in mean intensity and annual total days for MHWs and THEs exceed the 95% confidence level.
• Scenario forcing: SSP126/245/370/585 representing low/medium/high/high-end forcing pathways, with effective radiative forcing ~2.6/4.5/7.0/8.5 W m−2 by 2100.

• Arctic amplification of MHWs: By 2071–2100 under SSP585, the increase in Arctic MHW mean intensity is about 7.6 times the global average (Arctic +0.38°C vs global +0.05°C).
• THE amplification: Arctic THE mean intensity change exceeds global by a factor of ~1.5 under SSP585 (Arctic +1.82°C vs global +1.21°C).
• Absolute intensities: In 2071–2100, Arctic MHW mean intensity is ~2°C and THE mean intensity ~3°C, exceeding or matching hotspot regions at lower latitudes.
• Spatial patterns: Largest MHW mean intensity increase occurs in the Arctic deep basin (up to +1.4°C), while marginal seas and many mid/low-latitude regions show <+0.4°C change. THE intensification is strongest in the Arctic, especially the Barents Sea (+4°C), where MHW intensity change is minimal and THE change is dominated by long-term warming.
• Event days: MHW annual total days in the Arctic rise from near-zero in the central basin historically to ~20 days by 2071–2100; Arctic increase is +9.45 days vs +0.47 days globally (SSP585). THE annual total days exceed 300 across ~80% of the global ocean, but are ~200 in the central Arctic; increases are smaller in the Arctic (+223.58 days) than globally (+301.06 days).
• Drivers: Arctic MHW intensity changes are strongly anti-correlated with sea ice area decline across scenarios (r ≈ −0.56 to −0.81, p ≤ 0.05–0.01), with seasonality featuring peaks in summer (solar radiation maximum) and December (sea ice decline maximum). THE intensity changes correlate tightly with SST changes (r ≈ 0.73–0.96, p < 0.01), indicating dominance of long-term warming.
• Frequency vs duration: In the Arctic, increased MHW annual total days are primarily due to higher frequency, whereas increased THE annual total days are mainly due to longer duration. Frequency changes outside the Arctic are small or negative.
• Significance: Most models show statistically significant (>95%) Arctic increases in mean intensity and annual total days for both MHWs and THEs; changes in many mid/low-latitude regions are insignificant in most models under SSP585.
• Regime shift expectation: As the Arctic approaches seasonally or perennially ice-free conditions under high forcing, further MHW intensity and event-day increases diminish, resembling lower-latitude behavior by late century.

The study addresses the key question of how Arctic MHWs and THEs will evolve under warming and why. Results demonstrate a pronounced Arctic amplification of MHW intensity driven primarily by sea ice retreat, which enhances air–sea interactions and allows greater seasonal heat uptake, fostering stronger, more frequent MHWs. THE intensity increases are dominated by the long-term SST warming trend. Seasonality of MHW changes is set by the interplay of maximum solar input (summer) and peak sea ice decline (early winter). While the Arctic exhibits strong increases relative to the global ocean, once sea ice becomes minimal year-round under high forcing, additional increases in MHW intensity and days taper, mirroring mid/low-latitude oceans. Ecologically, amplified Arctic MHW/THE intensities and more frequent or longer events imply heightened risks for marine species, communities, and fisheries, with the Barents Sea standing out for especially large THE increases. These findings clarify mechanisms and quantify relative risks, supporting targeted adaptation and conservation planning in the Arctic.

This work introduces and quantifies Arctic MHW Amplification: future increases in Arctic MHW mean intensity that far exceed global averages, alongside strong THE intensification. Using CMIP6 projections and refined MHW/THE definitions that separate transient extremes from long-term warming, the study attributes MHW amplification to sea ice decline and THE amplification to sustained SST warming. It highlights distinct seasonality, spatial hotspots (deep Arctic basin for MHW increases; Barents Sea for THE increases), and differing roles of frequency (MHWs) versus duration (THEs) in increasing event days. Future research should pursue improved Arctic process representation and resolution in climate models, reduce inter-model spread, assess ecosystem-specific thresholds and cumulative thermal stress, and develop quantitative risk frameworks to guide conservation and adaptive management under varying warming scenarios.

• Model spread and biases: CMIP6 models show substantial spread in simulated and projected MHW/THE metrics, reflecting differences in Arctic temperature, sea ice, and surface flux representation.
• Resolution constraints: While higher resolution can improve MHW representation, resolution alone does not eliminate key Arctic biases.
• Sea ice treatment: Assuming no MHW/THE occurrence under SIC >35% and setting SST to freezing under ice may omit under-ice thermal variability and introduces dependence on sea ice parameterizations.
• Detection choices: Event detection thresholds (90th percentile, 5-day minimum, gap allowance) and baseline windows influence metrics; alternative definitions could yield different magnitudes.
• Significance outside Arctic: Projected MHW changes in many mid/low-latitude regions are not significant in most models, limiting broader generalization.
• Scenario dependence: Results depend on SSP forcing pathways and late-century periods; earlier periods or mitigation scenarios may show different magnitudes and timing.