Central tropical Pacific convection drives extreme high temperatures and surface melt on the Larsen C Ice Shelf, Antarctic Peninsula

The Larsen Ice Shelf on the eastern Antarctic Peninsula has undergone substantial retreat and collapse since the mid-1990s, notably the disintegration of Larsen A (1995) and collapse of Larsen B (2002), with Larsen C now experiencing thinning, rifting, and major calving (e.g., A-68 in 2017). These collapses are linked to anomalously warm summer air temperatures and widespread surface melting that promote hydrofracture. Foehn winds, which produce intense lee-side warming over the eastern AP and Larsen Ice Shelf, are a key mechanism for extreme temperatures and melt. Prior work often connected increased melt and warming to strengthened circumpolar westerlies and positive Southern Annular Mode (SAM). However, the quantitative link between SAM and extreme melt over Larsen C has been unclear, and recent studies suggested a role for tropical forcing. This study tests the hypothesis that deep convection in the central tropical Pacific (CPAC) triggers a distinctive South Pacific circulation pattern featuring a cyclonic anomaly and a strong anticyclone over Drake Passage that advects warm, moist air and promotes foehn warming and surface melt over Larsen C, thereby driving both interannual variability and extremes more strongly than SAM.

Previous research documented dramatic losses of Larsen ice shelves and subsequent acceleration and thinning of tributary glaciers, enhancing Antarctica’s sea-level contribution. Studies attribute extreme eastern AP surface temperatures and melt to foehn winds and link increased melt events to strengthened summer circumpolar westerlies and positive SAM, influenced by ozone depletion and greenhouse gases. Yet, robust quantification of SAM’s role in Larsen C melt variability and extremes has been lacking. Teleconnections from the tropics to the Antarctic Peninsula are known, often via the Amundsen Sea Low affecting the western AP, but are typically weaker in summer due to unfavorable jet configurations. Some recent analyses indicated that tropical variability, including MJO-related events and Rossby wave teleconnections, can contribute to AP temperature extremes and surface melt, and that atmospheric rivers (ARs) are significant triggers of West Antarctic surface melt. The present work builds on these findings by identifying CPAC convection within the SPCZ as the dominant tropical driver of the asymmetric South Pacific wave train that produces foehn warming and melt on Larsen C.

The study combines observation-based reanalyses, regional modeling, teleconnection diagnostics, case studies, and a targeted climate model sensitivity experiment. Data sources include ERA5 reanalysis (monthly means for seasonal correlations; 6-hourly fields for synoptic pentad composites), NOAA ERSSTv5 SSTs, and NOAA Interpolated OLR for convection. ENSO is characterized by the Southern Oscillation Index (SOI), and SAM by the Marshall (2003) index. CPAC convection is quantified by area-mean OLR over 10–15°S, 170–165°W. A Drake Passage mid-tropospheric circulation index (Drake Z500) is defined as area-mean ERA5 500 hPa geopotential height over 57–62°S, 81–71°W. Surface melt over the Larsen C Ice Shelf is computed using Polar-WRF (version 3.9.1) simulations for December 1991–March 2015 at 15 km resolution over an Antarctic Peninsula nested domain, with ERA-Interim boundary conditions, Bootstrap sea-ice concentrations, and standard parameterizations (RRTMG radiation, Morrison double-moment microphysics, MYJ PBL, Grell-Freitas cumulus, Noah-MP land surface). Melt is obtained as liquid-equivalent melted snow, aggregated to monthly and seasonal totals within a Larsen C polygon mask. Statistical analysis: seasonal (e.g., DJFM) anomalies are detrended and correlated using least-squares; significance assessed via two-tailed Student’s t-tests (n=25, dof=22) for correlations and trends at p thresholds; model differences are tested via two-sample t-tests (dof=58). Stationary Rossby wave activity flux is diagnosed following Takaya and Nakamura. Atmospheric rivers: Interannual AR activity is derived from AR catalog V3.0 (ERA-Interim based, 1979–2019), converted to daily frequency; extreme AP landfalling ARs in DJFM (1991–2014/15) are days when an AR intersects 61.5–72°S, 75–63°W and daily mean AP-averaged IVT exceeds the 95th percentile. For the two case studies (24 Mar 2015; 6 Feb 2020), a bespoke AR detection algorithm using ERA5 6-hourly IVT at 0.25° resolution identifies landfalling AR objects based on intensity thresholds, geometry (length >2000 km, length/width >2), poleward IVT component, and directional coherence. Sensitivity experiment: CESM1.2 (CAM5, atmosphere-only, 1.9°×2.5°, 30 levels) runs two 30-year DJF climatologies with prescribed climatological SST/sea ice and pre-industrial GHG/ozone; the perturbed run applies a +2 °C SST anomaly centered at 168°W, 13°S (6°×6° sine-damped box) to generate enhanced CPAC convection. Differences (perturbed minus control) diagnose the direct atmospheric response, including wave fluxes, geopotential height, precipitation, and surface air temperature.

- Larsen C summer (DJFM) surface melt correlates strongly with enhanced deep convection in the central tropical Pacific (CPAC OLR; r = -0.63, p < 0.01), but shows no significant relationship with ENSO (SOI) and only weak, non-significant relationships with SAM during summer. Spatially, CPAC convection aligns with an off-equatorial SPCZ-like band (10–20°S) conducive to Rossby wave generation. - Positive Larsen C melt anomalies co-occur with a zonally asymmetric circulation featuring an elongated cyclonic anomaly across the South Pacific and a strong anticyclonic anomaly over Drake Passage (Drake Z500), yielding moist southwesterly flow across the AP and a foehn pattern with moisture flux divergence over the eastern AP. Correlations with Drake Z500 are strong (DJF/DJFM r ≈ 0.67/0.65; p < 0.01). - Extreme melt events: Of the top 15 DJFM and monthly melt extremes (90th percentile), 13 occurred with enhanced CPAC convection (−CPAC OLR ≤ −0.5). ENSO phase was mixed (9 El Niño, several La Niña), and SAM showed no consistent sign (4 negative, 4 positive, 7 neutral), underscoring CPAC’s primacy. - Case studies: 24 March 2015 (+17.5 °C at Esperanza) and 6 February 2020 (+18.3 °C, record temperature; >50% AP surface melt) each exhibited deep CPAC convection, a downstream wave train producing a South Pacific cyclone and Drake Passage anticyclone, strong poleward heat/moisture transport, and landfalling atmospheric rivers hours before the records. - Atmospheric rivers: Total summer Larsen C melt correlates with the number of extreme AP landfalling ARs (r = 0.79, p < 0.01). Extreme AR frequency correlates with Drake Z500 (r = 0.64, p < 0.01) and CPAC OLR (r = -0.70, p < 0.01). Days with strong CPAC convection exhibit a 15–20% increase in AR frequency over the central AP near Larsen C. - CESM sensitivity experiment: Imposed CPAC heating reproduces the observed wave-train: an elongated South Pacific cyclone (~120°W, 50°S) and Drake Passage anticyclone, increased precipitation and near-surface warming along the southwest AP and West Antarctica, and two primary Rossby wave propagation pathways linked to Indo-Atlantic and South Pacific jet streak configurations. - Drivers of CPAC convection: Synoptic analyses show CPAC convection is primarily triggered by subtropical cyclones and northward-advancing cold fronts that organize a diagonal SPCZ-like convective band; background central Pacific SST anomalies and MJO phases modulate conditions but are not the primary trigger.

The study demonstrates that deep convection in the central tropical Pacific initiates a robust Rossby wave teleconnection that establishes an elongated South Pacific cyclone and a strong anticyclone over Drake Passage. This configuration advects warm, moisture-laden air toward the Antarctic Peninsula and favors foehn winds and atmospheric rivers, resulting in extreme high temperatures and extensive surface melt over Larsen C. The findings directly address the hypothesis that CPAC convection, rather than SAM, is the dominant summer driver of Larsen C melt variability and extremes. The mechanism is consistent across interannual statistics, extreme events, and a controlled climate model perturbation, providing strong causal support. The identified link between CPAC convection and AR frequency further explains how moisture transport contributes to thermodynamic and dynamic preconditioning for foehn warming. The results refine the understanding of AP climate drivers in summer, shifting emphasis from zonally symmetric SAM variability to asymmetric, tropically forced wave trains that specifically impact the central and southern AP and Larsen C.

CPAC convection is a key driver of extreme high temperatures and surface melt over the Larsen C Ice Shelf by forcing an asymmetric South Pacific circulation pattern culminating in an anticyclone over Drake Passage and associated southwesterly, moisture-rich flow that enhances foehn warming and atmospheric river landfalls. Statistically robust relationships between CPAC OLR, Drake Passage geopotential height, AR frequency, and Larsen C melt, together with case studies and a targeted CESM perturbation experiment, support a causal teleconnection. Given projected strengthening of circumpolar westerlies/SAM, these results indicate that future variability in CPAC convection may be more critical for Larsen C stability than SAM trends. Future research should focus on improving understanding and prediction of CPAC convective variability, the jet-stream configurations that enable wave propagation into high latitudes, and the occurrence of asymmetric cyclone–anticyclone patterns and AR activity to better constrain projections of Antarctic temperature extremes, surface melt, and ice-shelf stability.

Surface melt estimates are based on a regional model (Polar-WRF) for 1991–2015 and depend on model physics, boundary conditions, and the Larsen C mask; observational melt constraints are indirect. Statistical relationships are computed over a relatively short record, with detrending and p-value thresholds that may affect significance assessments. Reanalysis products and OLR datasets carry uncertainties, especially at high latitudes. The CESM sensitivity experiment uses an idealized, localized SST perturbation and atmosphere-only configuration with prescribed climatological SST/sea ice and pre-industrial forcings; results demonstrate mechanism but do not represent coupled feedbacks or future forcing scenarios. AR detection involves algorithm choices and reanalysis dependence; results for 2020 rely on a custom method. The study does not provide quantitative projections of future CPAC variability or explicit attribution of long-term trends in Larsen C melt.