Yang et al. (2026) DANRA: the kilometer-scale Danish regional atmospheric reanalysis

Identification

• Journal: Earth system science data
• Year: 2026
• Date: 2026-03-26
• Authors: Xiaohua Yang, Carlos Peralta, Bjarne Amstrup, Kasper S. Hintz, Søren Borg Thorsen, Leif Denby, Simon S. Christiansen, Hauke Schulz, Sebastian Pelt, Michael Schreiner
• DOI: 10.5194/essd-18-2251-2026

Research Groups

• Danish Meteorological Institute (DMI)

Short Summary

This paper introduces DANRA, a novel 2.5-kilometer resolution regional atmospheric reanalysis dataset covering Denmark and its surrounding regions from 1990 to 2023. DANRA demonstrates superior performance compared to global reanalyses like ERA5 in representing essential climate variables and extreme weather events, providing unprecedented detail for climate adaptation and impact modeling.

Objective

• To develop and validate a high-resolution (2.5 km) regional atmospheric reanalysis dataset (DANRA) for Denmark and its surrounding regions, covering 1990-2023, to overcome limitations of coarser global reanalyses in capturing local weather and climate features, especially extremes.

Study Configuration

• Spatial Scale: 2.5 km horizontal grid spacing (800 x 600 grid), 65 vertical levels, covering Denmark and surrounding regions.
• Temporal Scale: 34-year period (1990–2023), with 3-hourly analysis data.

Methodology and Data

• Models used: HARMONIE-AROME-40h1.1 Numerical Weather Prediction (NWP) model suite, utilizing a three-dimensional variational (3DVar) data assimilation scheme for upper-air states and multivariate optimal interpolation for near-surface parameters.
• Data sources:
  • Conventional in-situ observations (synoptic weather stations, ships, drifting buoys, aircraft, radiosondes), including those used by ERA5 and additional local Danish observations.
  • Satellite remote sensing data (microwave and infrared radiances, atmospheric motion vectors, scatterometer readings, GPS radio occultation measurements from ROM SAF Interim Climate Data Record).
  • Hourly ERA5 analyses for lateral boundary conditions.
  • Extensive quality assurance and data rescue efforts for in-situ observations, including high-frequency (hourly/10-minute) data and corrections for errors.

Main Results

• DANRA consistently shows superior performance compared to ERA5 and CERRA in representing near-surface weather parameters (e.g., 2 m temperature, 10 m wind speed) during both ordinary and extreme conditions.
• For 2 m temperature, DANRA exhibits a tighter distribution and reduced standard deviation error against observations, mitigating the systematic warm bias seen in ERA5.
• For 10 m wind speed, DANRA significantly reduces the persistent underestimation observed in ERA5, particularly at higher wind speeds, and shows a clear advantage in standard deviation error.
• Case studies of extreme events:
  • December 1999 hurricane-force storm: DANRA accurately reproduced the storm's intensity with a minimum pressure of 954 hPa and peak wind speeds approaching 40 m/s, closely matching observations, while ERA5 and CERRA significantly underestimated it (ERA5 max 28.5 m/s).
  • July 2022 national temperature record (heatwave): DANRA simulated temperatures up to 37.8 °C, closely aligning with the observed 36.7 °C, and showed realistic spatial temperature variations. ERA5 exhibited a cold bias of approximately 5 °C near the event's peak.
  • August 2007 cloudburst in South Jutland: DANRA successfully reproduced the signature of intense convective activity, simulating maximum total precipitation exceeding 75 mm, whereas coarser reanalyses (ERA5 and CERRA) largely missed or severely underestimated the event.

Contributions

• Creation of DANRA, a novel kilometer-scale (2.5 km) regional atmospheric reanalysis dataset for Denmark, providing unprecedented spatial detail and fidelity over a 34-year period.
• Extensive local observation data acquisition, quality assurance, and data rescue efforts, maximizing the use of high-frequency and precise in-situ data for the region.
• First analysis-ready, cloud-optimized reanalysis product for the region, distributed as CF-compliant Zarr datasets via an Amazon S3 object store, enhancing accessibility and usability for various applications, including machine learning.
• Demonstrates the significant added value of kilometer-scale resolution and dense observation assimilation for accurately representing local weather features and extreme events crucial for climate adaptation, impact modeling, and renewable energy planning.

Funding

• Danish National Center for Climate Research

Citation

@article{Yang2026DANRA,
  author = {Yang, Xiaohua and Peralta, Carlos and Amstrup, Bjarne and Hintz, Kasper S. and Thorsen, Søren Borg and Denby, Leif and Christiansen, Simon S. and Schulz, Hauke and Pelt, Sebastian and Schreiner, Michael},
  title = {DANRA: the kilometer-scale Danish regional atmospheric reanalysis},
  journal = {Earth system science data},
  year = {2026},
  doi = {10.5194/essd-18-2251-2026},
  url = {https://doi.org/10.5194/essd-18-2251-2026}
}