Barthel et al. (2023): Processed Climate Model Simulation Data for ML Correction

Processed data from simulations of a 2-layer Quasi-Geostrophic model and DOE's E3SM Atmosphere Model Version 2 (EAMv2), along with ERA5 reanalysis data, used to generate results for a 2023 study on machine learning correction operators for under-resolved climate models. The archive was created by Benedikt Barthel of the Massachusetts Institute of Technology. The data is post-processed for plotting and analysis as described in the associated paper.

Use Cases

• Training machine learning models for climate model error correction based on nudged simulation data.
• Benchmarking correction operator performance using processed Quasi-Geostrophic model outputs.
• Analyzing climate simulation biases by comparing EAMv2 model data with ERA5 reanalysis.

Strengths

• Includes data from multiple sources: an in-house Quasi-Geostrophic model solver, the DOE's E3SM Atmosphere Model Version 2 (EAMv2), and ERA5 reanalysis.
• Data is post-processed and accompanied by plotting scripts for result reproduction as per Barthel et al. (2023).

Limitations

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

Provenance

Source

Massachusetts Institute of Technology (Benedikt Barthel) via paperswithcode.

Collection Method

Data processed from in-house model simulations, EAMv2 simulations, and ERA5 reanalysis.