孔 (2026) ERA5-PrecipSR

Identification

• Journal: Mendeley Data
• Year: 2026
• Date: 2026-03-30
• Authors: 令志 孔
• DOI: 10.17632/7c93v7mbcj.1

Research Groups

• Nanjing University of Information Science and Technology

Short Summary

This paper introduces ERA5-PrecipSR, a regional NetCDF dataset derived from ERA5 hourly data, specifically designed to facilitate precipitation super-resolution and spatial downscaling research. It provides high-resolution meteorological fields and dynamically generated low-resolution precipitation to support physics-informed precipitation reconstruction.

Objective

• To provide a comprehensive, structured dataset (ERA5-PrecipSR) for developing and evaluating models for precipitation super-resolution and spatial downscaling, particularly for physics-informed precipitation reconstruction.

Study Configuration

• Spatial Scale: Regional, with high-resolution meteorological fields and low-resolution precipitation generated by a spatial downsampling factor of 6 from the high-resolution fields. Specific grid resolutions are not provided.
• Temporal Scale: From 1940 to present (as of 2026), with hourly time steps.

Methodology and Data

• Models used: No specific models are used in this dataset generation; the dataset is designed for use with precipitation super-resolution and spatial downscaling models.
• Data sources: Copernicus Climate Data Store (CDS) ERA5 hourly reanalysis data on single levels.

Main Results

• The ERA5-PrecipSR dataset was constructed, comprising 17,892 training, 2,237 validation, and 2,236 test samples.
• Each sample is a multivariate meteorological field at a specific time and region, including high-resolution total precipitation (tp), surface pressure (sp), 2-meter air temperature (t2m), relative humidity (rh), wind speed (wind), and land-sea mask (land_mask), along with corresponding low- and high-resolution latitude and longitude coordinates.
• Low-resolution precipitation is not stored independently but is dynamically generated by spatially downsampling the high-resolution precipitation fields by a factor of 6, reducing data redundancy.

Contributions

• Provides a ready-to-use, structured NetCDF dataset specifically tailored for precipitation super-resolution and spatial downscaling research.
• Offers a comprehensive set of multivariate meteorological fields alongside precipitation, enabling physics-informed reconstruction studies.
• Optimizes data storage by dynamically generating low-resolution precipitation from high-resolution data.
• Leverages the extensive temporal coverage (1940-present) and quality of ERA5 reanalysis data.

Funding

• Not specified in the provided text.

Citation

@article{孔2026ERA5PrecipSR,
  author = {孔, 令志},
  title = {ERA5-PrecipSR},
  journal = {Mendeley Data},
  year = {2026},
  doi = {10.17632/7c93v7mbcj.1},
  url = {https://doi.org/10.17632/7c93v7mbcj.1}
}