Deep Learning Regional Climate Model Emulators: Antarctic Ice Sheet Data (1980–2100)

Antarctic ice sheet data from the MAR regional climate model and ACCESS1-3 GCM, used to compare deep learning downscaling frameworks. The dataset contains monthly outputs for surface mass balance and its components, spanning the period from 1980 to 2100. It was produced by researchers at ETH Zurich and accompanies a 2022 paper published on Authorea.

Use Cases

• Training deep learning emulators for regional climate model downscaling based on the described model outputs.
• Comparing different downscaling training frameworks using the provided MAR and ACCESS1-3 data.
• Analyzing long-term surface mass balance trends over the Antarctic ice sheet based on the 120-year monthly time series.
• Upscaling regional model outputs to GCM resolution for validation studies, as suggested by the description.

Strengths

• Covers a long temporal range from 1980 to 2100, enabling analysis of past and projected climate.
• Includes data from both a regional climate model (MAR) and a general circulation model (ACCESS1-3), facilitating comparison.
• Associated with a published scientific paper and has a citable DOI (10.5281/zenodo.7875967).

Limitations

• Column-level documentation is absent; field semantics must be inferred after download.
• Row count and file sizes are unknown, which may limit suitability assessment.
• Last update date is unknown; freshness unverified.

Provenance

Source

ETH Zurich

Collection Method

Outputs from the MAR regional climate model and the ACCESS1-3 general circulation model.

Time Range

1980–2100

Geography

Antarctic ice sheet