Influence of fast ice on future ice shelf melting in the Totten Glacier area, East Antarctica

.


Introduction
The Totten Glacier area, located on the Sabrina Coast in East Antarctica, underwent significant grounding-line fluctuations during the recent past. Driven by changes in the ocean (Aitken et al., 2016), these fluctuations are making the region potentially vulnerable to rapid ice sheet collapse (Roberts et al., 2011). There has been some indication of ice shelf thinning during the 20 last decade (Khazendar et al., 2013), although it remains unclear whether this represents a long-term trend (Paolo et al., 2015). Furthermore, the Totten catchment, located in the Aurora Subglacial Basin of East Antarctica, contains 3.5-m sea level rise equivalent and is one of the few sectors of East Antarctica where changes in ice dynamics have been observed recently dynamical :::::::: influence ::: on ::: the ::: ice :::: shelf, ::: as ::: the loss of buttressing from the break-up of seasonal fast ice increases the seasonality of the Totten ice shelf (TIS) basal melt rate close to the ice front (Greene et al., 2018). 40 Large density, temperature, salinity and sea level gradients are found across the Antarctic Slope Front (ASF; Whitworth et al., 1985;Jacobs, 1991), which separates the continental shelf from the open Southern Ocean. A strong pressure gradient is observed across the ASF, mainly caused by the strong easterly winds that drive a sea surface height gradient via Ekman drift (Mathiot et al., 2011), as well as a density gradient, which results from the differences in temperature and salinity of the water masses across the ASF. Additionally, the ASF manifests itself through strong isopycnal doming towards the continental shelf. 45 These lateral gradients across the ASF contribute to establishing the geostrophically balanced, vertically sheared along-slope flows of the Antarctic Slope Current (ASC; Jacobs, 1991;Thompson et al., 2018). The ocean dynamics associated with the ASF and ASC govern along-and across-slope heat transport , and act as a barrier to mixing between shelf and open-ocean waters (Thompson et al., 2018). Shifts in position of the ASF, or changes in the range of densities of waters that occupy the continental shelf, will therefore strongly influence the heat budget of the continental shelf (Thompson et al., 2018). 50 Moorman et al. (2020) suggested that increasing glacial meltwater fluxes strengthens the lateral density gradient associated with the ASF, which reduces cross-slope water exchanges and isolates shelf waters from warm mCDW. Naughten et al. (2018) also found an intensified density gradient across the continental slope which reinforces the Antarctic Coastal Current. In the Totten Glacier region, the ASC modulates the heat intrusion towards the Totten Glacier (Nakayama et al., 2021).
As a consequence, understanding how the ASC will evolve in this region under future climate conditions is key to gain 55 insights on changes in heat intrusion across the continental shelf break. The future changes in ice shelf melt rate under different Representative Concentration Pathway (RCP) scenarios have been studied with both global and regional models (Hellmer et al., 2012;Timmermann and Goeller, 2017). In the Totten Glacier area, Pelle et al. (2021) found that, by the end of the 21st century, the ASC might weaken by 37% compared to its present-day state and the Totten ice shelf melt rate might increase by 56% following a high emission scenario. Those models include representations of ocean-ice shelf interactions, but none of them 60 has an prognostic representation of the fast ice.
This manuscript is organised as follows. The model, regional configuration and experimental design are described in Section 2. In Section 3, we analyse the changes in sea ice and ocean characteristics and ice shelf melt rate between the last decades ::::: recent :::: past and the end of the 21st century simulated by the model. The sensitivity of the ice shelf melt rate to the representation 75 of fast ice is then addressed in Section 4. Conclusions are finally given in Section 5.
2 The model, forcing and experimental design 2.1 Ocean-sea ice model We make use of NEMO 3.6 (Nucleus for European Modelling of the Ocean; Madec, 2008) that includes the ocean model OPA (océan parallélisé) coupled with the Louvain-la-Neuve sea ice model (LIM3; Vancoppenolle et al., 2009;Rousset et al., 80 2015). This combination is hereafter referred to as NEMO-LIM. OPA is a state-of-the-art, finite-difference ocean model based on primitive equations. Our setting includes a polynomial approximation of the seawater equation of state (TEOS-10, IOC, 2010) optimized for a Boussinesq fluid (Roquet et al., 2014). Vertical turbulent mixing is rendered through a Turbulent Kinetic Energy (TKE) scheme (Bougeault and Lacarrere, 1989;Gaspar et al., 1990;Madec et al., 1998). The enhanced vertical diffusion mixing coefficient utilised in this scheme is fixed to 20 m 2 /s. LIM3 uses a five-category subgrid-scale distribution of sea 85 ice thickness (Bitz et al., 2001). The drag coefficient is set to 7.1 × 10 −3 at the sea ice-ocean interface and 2 × 10 −3 at the sea ice-atmosphere one (Massonnet et al., 2014). Ice shelf cavities with explicit ocean-ice shelf interactions are represented by the ice shelf module implemented in NEMO by Mathiot et al. (2017), using the three-equation formulation from Jenkins (1991). Transfer coefficients for heat (γ T ) and salt (γ S ) between the ocean and ice shelves are velocity dependent (Dansereau et al., 2014): γ T,S = Γ T,S × u * . The friction velocity is given by u * = C d × u 2 T M L and constant values of Γ T and Γ S taken 90 from Jourdain et al. (2017) are employed (Γ T = 2.21 × 10 −2 and Γ S = 6.19 × 10 −4 for temperature and salinity, respectively).
C d is the top drag coefficient, set to 14 × 10 −3 ::::::: 3 × 10 −3 , and u T M L is the ocean velocity in the top mixed layer, which is either the top 30 m of the water column or the top model layer (if thicker than 30 m) (Losch, 2008).

The Totten24 model configuration 95
Here, we use a regional configuration of NEMO-LIM, referred to as Totten24, which is described in detail in Van Achter et al.
(2022). The horizontal grid is a 1/24 • refinement (less than 2 km grid spacing) of the eORCA1 tripolar grid, centered on the continental shelf in front of the TIS, East Antarctica, and covering an area between 108-129 • E and 63-68 • S (Fig. 1). The NEMO and LIM time steps are 150 s and 900 s, respectively. The vertical discretisation has 75 levels, with level thickness increasing with depth and partial cells used for better representing bedrock and ice shelf bases (Adcroft et al., 1997 (Morlighem et al., 2020).
The ocean lateral boundary conditions and initial conditions are taken from a 1979-2014 simulation with an eORCA025
Annual cycles of the EC-Earth3 climate anomalies are computed as the differences between 2081-2100 and 1995-2014, and are added to all the fields of the atmospheric and oceanic forcings used for the 1995-2014 period in REF and nFST (for the atmosphere: wind velocity, temperature, specific humidity, surface downward radiation and precipitation; for the ocean: current 150 velocity, temperature, salinity, sea surface height, sea ice concentration, sea ice thickness and snow thickness). Figure 3 shows the annual mean ocean temperature, salinity and zonal ocean velocity anomalies at the eastern boundary condition, and the mean near-surface (2 m) air temperature and atmospheric zonal wind (10 m) velocity anomalies. We show the ocean anomalies at the eastern lateral boundary condition as they are very similar to those at the western lateral boundary condition, and also because the ocean eastern boundary condition is one of the drivers of the ocean dynamic over the continental shelf in regional 155 6 modelling (Nakayama et al., 2021). The ocean temperature anomaly is positive everywhere, with values from 0 to 0.5 • C over the continental shelf and in the deep ocean, and from 1 to 1.5 • C in the upper ocean outside of the shelf. The seawater salinity anomaly is mostly negative (down to -0.4 g/kg), with the lower values above the continental shelf. Oceanic zonal velocity anomalies at the eastern boundary are westward over the shelf and eastward off the shelf. The EC-Earth3 anomaly applied at the zonal wind component is mostly eastward over the ocean, increasingly towards the north. Westward winds anomaly :::: wind 160 :::::::: anomalies : also occur, but only over a small part of the shelf and over the continent. The surface air temperature anomaly is positive everywhere (Fig 3e), with values larger than 1 • C and up to 1.8 • C near the coast.  WARM : -:::::::: atmospheric ::::::: anomalies ::::: derived :::: from ::::::: EC-Earth3 ::::: climate ::::: change ::::::: projection :::::::::: WARM_noOce : :: yes ::::: WARM : -:::: ocean :::::: velocity ::::::: anomalies ::::: derived ::: from :::::::: EC-Earth3 ::::: climate ::::: change ::::::: projection Table 1. Names and descriptions of the simulations used in this study.
The TIS and MUIS basal melt rates present a different sensitivity to fast ice. This is explained by both the unchanged MUIS basal melt rate in WARM compared to REF, and the higher MUIS basal melt rate in nFST_WARM compared to WARM.

265
Combined, these two effects contribute to a much larger basal melt rate increase between the simulations with and without fast ice for the MUIS than for the TIS (difference of 37% in melt rate increase for MUIS and 17% for TIS). The unchanged MUIS basal melt rate in WARM compared to REF is attributed to the limited effect of the ocean warming over the MUIS cavity, whereas the warmer ocean masses reaches the TIS cavity (Fig. 7d). This is explained by the differences in bathymetry in front of each ice shelf cavity. which results in an enhanced TIS basal melt rate and a lower MUIS melt rate. In the same way, in WARM, the warmer water masses reach the TIS, but are limited outside of the MUIS cavity, which limits the MUIS basal melt rate changes between REF and WARM. Finally, the higher MUIS basal melt rate in nFST_WARM compared to WARM is attributed to the changes affecting the sea ice in WARM and nFST_WARM. In nFST_WARM, the absence of fast ice allows strong sea ice formation along the coast, with a deep mixed layer depth (mld) :::: MLD : in front of the MUIS cavity (Fig. 10c). In contrast, in WARM, the 275 presence of fast ice allows for sea ice formation along the coast but also at the off-shore polynya created on the west side of fast ice patches in front of the MUIS cavity, ::: but :: it :::: also ::::: allows :::::: strong ::: sea ::: ice ::::::::: production ::::: along ::: the :::: coast ::::: since ::: the ::: fast ::: ice ::::: there : is :::::: largely ::::::: reduced :: in :::: area ::: and ::::::::: frequency. This combination of sea formation both off-shore and along the coast contributes to : a broader area of deep mld :::: MLD : in front of the MUIS cavity in WARM (Fig. 10d), which decreases the amount of warm water able to cross the continental shelf and to reach the MUIS cavity in WARM compared to nFST_WARM ( Fig. 10a and 10b). As 280 a consequence, the MUIS basal melt rate in WARM is lower than in nFST_WARM.
Author contributions. GVA designed the science plan with TF and HG, ran the simulations, produced the figures, analysed the results and wrote the manuscript based on insights from all co-authors. EMC provided the EC-Earth3 dataset.
Competing interests. The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.