Methodology:

Simulations were performed with the Massachusetts Institute of Technology general circulation model (MITgcm) version 67s, over the Amundsen Sea region (140 degrees W:80 degrees W, 76 degrees S:62 degrees S). Resolution is 1/10 degrees in the horizontal, with 50 levels in the vertical. Bathymetry and ice shelf topography were generated using BedMachine version 2. Atmospheric forcing is given by the CESM1 Large Ensemble (1920-2005 historical, 2006-2100 RCP 8.5, first 10 members), the CESM1 Medium Ensemble (2006-2080 RCP 4.5, first 10 members), the CESM1 Low Warming 2C Ensemble (2006-2100, first 10 members), and the CESM1 Low Warming 1.5C Ensemble (2006-2100, first 5 members). Lateral boundary conditions were derived from the corresponding CESM1 ensembles, except for the ''Fixed BC'' experiments which used the World Ocean Atlas 2018 for temperature and salinity, and the Biogeochemical Southern Ocean State Estimate iteration 122 for all other variables. Ice shelf buttressing diagnostic simulations were performed using the Ua ice flow model on the same domain.

Data collection:

The raw MITgcm output was converted to NetCDF using the xmitgcm python package, and then processed into timeseries, Hovmollers, and trends using the mitgcm_python library (https://github.com/knaughten/mitgcm_python).

Data quality:

Values which are identically zero or masked represent the land and ice shelf mask.