This dataset provides harmonized output from multiple Regional Climate Models (RCMs) for the Antarctic and Greenland ice sheets. It includes historical simulations forced with ERA5 reanalysis, as well as historical and future simulations extending to 2100 driven by a selection of Earth System Models (ESMs) under different SSP scenarios.

The RCMs included are MAR, RACMO, and HIRHAM5, forced by ERA5 and by various ESMs, including MPI-ESM, CESM2, UKESM, IPSL-CM6A, CNRM-CM6, EC-Earth3, NorESM, and CNRM-ESM. The dataset covers SSP1-2.6, SSP2-4.5, SSP3-7.0, and SSP5-8.5, with availability depending on the specific model combination.

All variables are provided as monthly accumulated fluxes with units of kg m⁻². The dataset includes the following variables:

prsn – snowfall

prra – rainfall, calculated as total precipitation minus snowfall

melt – melt of snow or ice

mrro – runoff of rainfall or meltwater

subl – surface sublimation and evaporation (>0 = sublimation, <0 = deposition)

smb – surface mass balance

sndivgl – blowing snow transport divergence (>0 = erosion, <0 = deposition). Available only in RACMO2.4p1 and RACMO2.3p2, where it was used to compute the surface mass balance. Note: The blowing snow scheme performs significantly better in RACMO2.4p1 compared to RACMO2.3p2; therefore, we recommend using RACMO2.4p1 outputs when considering blowing snow as a freshwater flux.

sndivsn – blowing snow transport divergence on non-glaciated tiles (<0=deposition). Available only in RACMO2.4p1.

All gridded data are stored in NetCDF4 format, with one file per simulation. Files follow a standardized naming convention:

{domain}_{RCM}{version}_{resolution}_{forcing}_{ssp/historical}_{year_start}_{year_end}.nc

For example 'ANT_MARv311_35km_MPI-ESM_historical_1980_2014.nc' contains the monthly output from MAR version 3.11 for the Antarctica domain, forced with historical MPI-ESM for the period 1980 - 2014

In addition to the model output, the dataset includes mask files for each model domain that provide grid-cell information required for regional and ice-sheet–integrated analyses. Mask files follow the naming convention:

{domain}_{RCM}{version}_{resolution}_mask.nc

For example: GR_RACMO24p1_5.5km_mask.nc

Each mask file includes at least the following variables:

ice_mask – fraction of ice-covered area per grid cell (0–1)

cell_area – grid cell area in m²

ocean_sectors – classification of grid cells into ocean sectors

ice_mask_greenland or ice_mask_antarctica – masks defining the glaciated regions used for computing ice-sheet–integrated mass balance (0-1)

For RACMO2.4p1 and MAR, the ice masks are fractional, whereas RACMO2.3p2 and HIRHAM5 use binary masks.

Two python scripts are given which can be used to compute yearly ice sheet or ocean sector integrated SMB terms in Gt/yr.