Role of atmospheric rivers in shaping long term Arctic moisture variability

The study investigates how atmospheric rivers (ARs) contribute to recent Arctic moistening trends and disentangles the relative roles of anthropogenic forcing versus low-frequency, large-scale circulation variability. Background context highlights Arctic Amplification (AA), the Clausius–Clapeyron-driven increase in specific humidity, and model projections of a wetter Arctic. Observations indicate notable summertime circulation changes, including a trend toward a barotropic high-pressure anomaly over Greenland, likely with substantial internal variability contributions. ARs, key drivers of poleward moisture transport into the Arctic and closely linked to extratropical storm tracks, have shown increasing summer frequency in regions such as western Greenland. However, historical climate model ensembles forced by anthropogenic emissions fail to reproduce the observed circulation and AR patterns, suggesting a strong role for internal atmospheric variability. The research aims to quantify how observed circulation trends regulate AR activity and, in turn, how AR changes have contributed to the long-term increase in Arctic atmospheric moisture, with a focus on boreal summer (JJA).

Prior work demonstrates that the Arctic has warmed more than twice the global average due to AA, with increased atmospheric moisture, cloudiness, and precipitation. The CC relationship largely explains projected uniform increases in temperature and humidity, and models project a future shift toward a rain-dominated Arctic. Nonetheless, observed summer trends also show strong modulation by large-scale circulation variability, including persistent high-pressure anomalies over Greenland linked to internal variability and implicated in recent sea ice and Greenland Ice Sheet (GrIS) melt. ARs account for over 90% of poleward water vapor transport into the Arctic annually and are most frequent in summer due to the poleward-shifted jet. Studies suggest AR frequency and intensity are sensitive to thermodynamic effects of warming and aerosol forcing, with increased moisture and eddy energy potentially enhancing AR occurrence. However, AR variability may also arise from internal atmospheric processes, and climate models often struggle to capture observed tropical–extratropical teleconnections and Arctic circulation trends. Case studies (e.g., 2012, 2014, 2021) link extreme GrIS melt events to AR intrusions coupled with anomalous anticyclonic flow, underscoring the need to separate forced and internal contributions to AR-driven moistening.

Data and detection:

• Reanalysis: ERA5 (primary) and ERA-Interim at 6-hourly and monthly resolution (1979–2019), regridded to 1.5°×1.5°.
• Variables: 200 hPa geopotential height (Z200), winds, specific humidity (surface–500 hPa average), temperature (surface–200 hPa average), clouds, radiation; NSIDC monthly sea ice concentrations.
• AR detection: IVT-based algorithm (85th percentile monthly IVT threshold at each grid, integrated 1000–300 hPa). Geometric criteria: length ≥2000 km and length-to-width ratio ≥2. ERA5 6-hourly fields on 0.25° grids used to compute IVT; AR masks regridded to 1.5° and converted to monthly AR frequency (days/month).

Statistical analyses:

• Linear trends via least-squares regression (1979–2019).
• Maximum Covariance Analysis (MCA/SVD) between fields (e.g., AR frequency and Z200) to identify coupled modes; leading mode variance explained and associated wind regressions assessed.
• Daily/synoptic MCA for 2012 JJA over western Greenland between daily AR frequency and specific humidity anomalies (climatological cycles and interannual variability removed).
• Composite analysis: 6-hourly composites of specific humidity anomalies from 4 days before to 5 days after onset for strong AR events centered over western Greenland, northern Europe, and eastern Siberia (no AR on the day prior; continuous activity ≥24 h).
• AR-removal humidity attribution: For each 6-hourly field within JJA, reconstruct two seasonal-mean humidity fields per year: (1) AR-related (non-AR grids set to zero) and (2) AR-unrelated (AR grids set to zero). Compute 1979–2019 trends for each to quantify AR contributions.
• Significance tests consider effective sample sizes for autocorrelated data.

Model experiments:

• CESM1 nudging experiments (40-yr integrations each): • CTL: Constant CO2 at year-2000 level (370 ppm), no wind nudging. • WIN: Add observed 3-D wind trend tendencies (ERA5 long-term JJA trends from surface to top; applied north of 60°N at each timestep from June 1–Aug 31), CO2 fixed at 370 ppm. • WIN+CO2: Same wind nudging plus constant CO2 at year-2020 level (415 ppm) to emulate combined effects of anthropogenic forcing and observed wind trends. Other forcings identical between runs.
• CESM2-Large Ensemble (CESM2-LEN): 40 members (1979–2014) to separate forced response (ensemble mean) from internal variability (member deviations). Implement a fingerprint analysis by selecting subgroups with strong AR trend increases over regions (e.g., western Greenland) and examining corresponding Z200 trends.
• CMIP6 multi-model ensemble: Monthly outputs from 34 models and daily outputs from a 10-model subset (1979–2014) used to evaluate forced responses of ARs, Z200, temperature, and humidity; all regridded to 1.5°×1.5°.

• Observed AR frequency trends: Summer (JJA) AR frequency into the Arctic increased by ~0.1 days/month/decade (1979–2019), with strong regional increases over western Greenland, northern Europe, and eastern Siberia, and decreases over parts of the North Atlantic, western Alaska, and central Siberia.
• Circulation consistency: Trends in Z200 show significant increases over western Greenland, northern Europe, and eastern Siberia, aligning with AR frequency and with increased specific humidity and temperature in those regions.
• MCA evidence: The leading MCA mode (MCA1) between detrended JJA Z200 and AR frequency explains 41.3% of squared covariance, exhibiting a zonal wave-2 structure around 70°N. Southerly anomalies enhance AR frequency; northerly/easterly anomalies suppress it.
• Synoptic coupling: In JJA 2012 over western Greenland, the leading daily MCA mode explains 58.3% of covariance; AR and humidity mode time series are tightly coupled (r≈0.70), with a mid-July AR preceding a basin-wide moisture surge and exceptional GrIS melt. Composites across summers show humidity increases begin ~1 day before AR onset, peak ~1 day after, then decay.
• AR contribution to moisture trends: Using AR-removal reconstruction, Arctic-wide (≥60°N) JJA specific humidity trends partition into AR-related 0.017 g kg⁻¹ decade⁻¹ and AR-unrelated 0.03 g kg⁻¹ decade⁻¹, implying AR activity trends contributed ~36% of the overall Arctic summer atmospheric moisture increase since 1979. Regional AR contributions are substantial: western Greenland 57.1%, northern Europe 47.1%, eastern Siberia 67.8% of local increases.
• Forced vs internal variability: CMIP6 and CESM2 ensemble means show more uniform Z200/temperature/humidity increases and an AR increase centered near the Arctic Ocean close to the North Pacific, failing to reproduce observed regional AR/circulation patterns—indicating a large internal variability role.
• Nudging experiments: CESM1 WIN vs CTL reproduces AR frequency increases over northwestern Greenland and northern Eurasia (+0.5 days/month) and decreases over southern/eastern Greenland and Bering Strait (−0.5 days/month), consistent with imposed Z200/wind anomalies. WIN+CO2 yields similar AR/circulation differences to WIN, suggesting minor direct anthropogenic-forcing effects on AR patterns in this configuration.
• Fingerprint analysis (CESM2-LEN): Subgroups of members with strong AR increases over western Greenland also show Greenland high-pressure Z200 trends; two members with AR increases over both western Greenland and eastern Siberia exhibit a chain of high pressure in both regions, mirroring observations. This supports internal circulation variability as a key modulator of low-frequency AR changes.

The findings demonstrate that summertime Arctic AR variability is strongly governed by low-frequency large-scale circulation trends, which steer AR pathways and modulate poleward moisture transport. The observed regional AR frequency increases and associated humidification and warming are linked to a zonal wave-2 circulation trend pattern, providing a mechanistic pathway by which internal atmospheric variability induces long-term changes in AR activity and Arctic moisture. Quantitatively, AR activity accounts for roughly one-third of the Pan-Arctic summer moisture trend and more than half in key regions such as western Greenland, northern Europe, and eastern Siberia, helping explain observed amplified regional moistening and related impacts (e.g., GrIS melt). Model ensemble means under historical forcing fail to replicate the spatial structure of observed AR and circulation trends, indicating that anthropogenic forcing alone cannot account for recent AR changes; instead, internal variability in the large-scale flow exerts primary control. Nudging experiments confirm that introducing observed wind trends alone reproduces AR responses consistent with observations, while adding elevated CO2 yields little additional pattern change, underscoring the steering role of circulation. Nonetheless, anthropogenic influences may indirectly affect ARs via their impacts on circulation and background thermodynamics, and model dependence remains possible. The study highlights the importance of ARs in Arctic moisture feedbacks—enhancing clouds and downward longwave radiation, thereby promoting surface warming and ice melt—and emphasizes the need to consider AR-driven variability when interpreting long-term Arctic hydroclimatic changes.

This work shows that low-frequency, large-scale circulation variability has decisively regulated recent summertime Arctic AR activity, thereby contributing substantially to long-term atmospheric moistening—about 36% Pan-Arctic and over 50% regionally in western Greenland, northern Europe, and eastern Siberia. Observational analyses, MCA, synoptic composites, AR-removal attribution, CESM1 wind-nudging experiments, and CESM2-LEN fingerprinting coherently support that internal circulation trends with a zonal wave-2 structure steer AR frequency and pathways, while anthropogenic forcing alone (as represented by ensemble means) does not reproduce the observed spatial patterns. These results identify ARs—often viewed as stochastic synoptic phenomena—as a key mechanism in shaping multi-decadal Arctic moisture variability and related cryospheric impacts. Future research should: (1) clarify AR responses to projected weakening and poleward shifts of the midlatitude jets; (2) better quantify indirect anthropogenic effects on ARs via circulation changes; (3) improve model representation of tropical–extratropical teleconnections and Arctic circulation; and (4) explore the roles of ARs in high-latitude moisture feedbacks and extreme melt events, particularly in regions like eastern Siberia where circulation–AR coupling remains less clear.

• Model representation limitations: CMIP6/CESM ensemble means do not reproduce observed circulation/AR spatial patterns, reflecting biases in teleconnections and Arctic circulation variability.
• Nudging design: Imposes constant JJA wind trend tendencies north of 60°N without adding synoptic variability; may shift regional responses (e.g., eastern Siberia) and does not capture all dynamical pathways.
• Data constraints: AR detection relies on 6-hourly ERA5, but only daily outputs are available for many CMIP6 models, potentially affecting model–observation comparisons of AR characteristics.
• Attribution scope: While WIN vs CTL isolates circulation effects and fingerprinting isolates internal variability, model dependence remains, and indirect anthropogenic impacts on circulation cannot be fully excluded.
• Regional uncertainties: Eastern Siberia shows discrepancies between nudging results and fingerprint analysis, indicating incomplete understanding and need for further study.