Methodology:

The dataset was generated using the MIT General Circulation Model (MITgcm), a numerical model designed to simulate the dynamics of the ocean, atmosphere, and other geophysical fluids. The version of MITgcm used for these simulations includes high-resolution capabilities tailored for regional studies, particularly in polar regions.

- Model Configuration: The model was configured to focus on the Amundsen Sea, with a high-resolution grid that captures the detailed topography and bathymetry of the region as developed and discussed in Naughten et al. (2022) and Naughten et al. (2023). The horizontal resolution was fine enough to resolve key oceanographic features, such as the intrusions of Circumpolar Deep Water (CDW) onto the continental shelf.

- Forcing and Boundary Conditions: The simulations were forced with atmospheric and oceanic boundary conditions derived from global climate models under the Representative Concentration Pathways RCP8.5 scenario. This includes surface wind stress, air temperature, humidity, and radiation. The dataset includes several experimental setups:

o ALL: All anthropogenic changes (both wind and thermodynamic).

o NONE: No anthropogenic changes (pre-industrial conditions).

o THERMO: Anthropogenic thermodynamic changes only.

o WIND: Anthropogenic wind changes only.

- Ensemble Simulations: each experiment comprises 9 ensemble runs. These ensembles provide a range of possible outcomes under the same external forcings, to ensure good representation of variability in the region.

The model was run from pre-industrial conditions (late 19th century) through to the year 2100, with a particular focus on the period 2070-2100 for future projections. Transient boundary conditions were used for some ensemble experiments to assess the sensitivity of the model outputs to boundary conditions.

Data collection:

ARCHER2 supercomputer used to run all the simulations

Python 3.6.3

Below are the packages used in the virtual environment:

Package Version

------------------ ------------

matplotlib 3.3.4

netCDF4 1.2.8

numpy 1.19.5

pandas 1.1.5

Pillow 8.4.0

pytest 7.0.1

scipy 1.5.4

xarray 0.16.2

mitgcm_python (https://github.com/knaughten/mitgcm_python)

UaMITgcm (https://github.com/knaughten/UaMITgcm): repository created by Kaitlin Naughten and Jan De Rydt

Data quality:

For the accuracy and model configuration of MITgcm please see Naughten et al. (2022) and Naughten et al. (2023). The analysis data masks any grid cells not representing the ocean with NaN values. All data was quality controlled and Continuous Integration (CI) testing on the analysis script outputs was implemented where possible.