CARRA¶

Copernicus Arctic Regional Reanalysis (CARRA1)

Being replaced

At a glance¶

Note on period: The CDS entries are titled "from 1991 to present", but the reanalysis starts in September 1990, so the WMO reference period 1991–2020 is fully covered. As of 2026-10-12, the sub-daily data run to 2026-07-31 and the daily and monthly data (reanalysis-carra-means) to 2026-02-28. Updated monthly with a 2–3 month latency; timely updates end in March 2027.

Overview¶

The C3S Arctic Regional Reanalysis (CARRA) provides 3-hourly analyses and hourly short-range forecasts of atmospheric, surface and sea-ice variables at 2.5 km resolution for two Arctic domains: CARRA-West (Greenland and Iceland) and CARRA-East (Svalbard, the Barents Sea and Arctic Scandinavia). It is built on the non-hydrostatic HARMONIE-AROME model (cycle 40h1) with 3D-Var data assimilation on a 3-hourly cycle and ERA5 as lateral boundary conditions. Compared with ERA5 it offers finer resolution of orography, coastlines and glaciers, more local observations, and better surface characterisation from satellite and physiographic data. It will be superseded by the second-generation pan-Arctic reanalysis, CARRA2.

What changes with the successor¶

CARRA → CARRA2 Planned¶

Transition: 2027-03

The second-generation pan-Arctic regional reanalysis, CARRA2, released its full historical period (September 1985 to 2025) on the CDS in May 2026; timely CARRA1 updates stop at the end of March 2027. Grid spacing stays at 2.5 km, but a single grid now covers the whole Arctic instead of two domains, and the model cycle, land-surface scheme, output levels, forecast length and CDS layout all change.

Sources: C3S — Copernicus Arctic Regional Reanalysis service (CARRA1 and CARRA2 comparison) · Copernicus pan-Arctic Regional Reanalysis (CARRA2): Data User Guide · 9th C3S General Assembly (June 2026), CARRA presentation (Schyberg) — CARRA1 not extended beyond March 2027 · ECMWF Forum — CARRA2 full historical period from 1986 to 2025 now available on the CDS

Compare the two side by side

Available variables¶

Main parameters from the tables in the CARRA Data User Guide. Names in backticks are GRIB short names; CDS requests use long names such as 2m_temperature and total_precipitation. The CDS single-levels entry offers about 67 variables.

instantaneous mean accumulated maximum minimum static — instantaneous: value at the time stamp · mean/accumulated/maximum/minimum: over the indicated interval · static: does not change with time

Surface and single levels (analysis and forecast, instantaneous)¶

Surface and single levels (forecast only, accumulations and extremes)¶

Pressure levels (23 levels, 1000–10 hPa; analysis and forecast)¶

Height levels (11 levels, 15–500 m above the surface; CDS request names)¶

Invariant fields¶

The full variable list is in the official documentation.

How to get the data¶

• CDS — Arctic regional reanalysis on single levels (reanalysis-carra-single-levels) — Web download / CDS API (cdsapi)
  • Requires an ECMWF account and acceptance of the dataset licence (CC-BY). A request selects the domain (west_domain or east_domain), the product type (analysis or forecast) and the level type (surface_or_atmosphere or soil).
• CDS — pressure levels (reanalysis-carra-pressure-levels) — Web download / CDS API
• CDS — height levels (reanalysis-carra-height-levels) — Web download / CDS API
• CDS — model levels (reanalysis-carra-model-levels) — Web download / CDS API
• CDS — daily and monthly data (reanalysis-carra-means) — Web download / CDS API
  • Released October 2025. A single entry with daily and monthly means, extremes and some totals for all level types; it lags behind the sub-daily data.

Example code¶

Put the CDS API URL and your personal key in ~/.cdsapirc first. Request keys and values are taken from the CDS download form (API request).

# pip install --upgrade cdsapi
import cdsapi

client = cdsapi.Client()

# 1) Analysis: instantaneous 2 m temperature (00 and 12 UTC of the 3-hourly analyses)
client.retrieve(
    "reanalysis-carra-single-levels",
    {
        "domain": "west_domain",              # or "east_domain"
        "level_type": "surface_or_atmosphere",
        "product_type": "analysis",
        "variable": ["2m_temperature"],
        "year": ["2024"], "month": ["01"], "day": ["01"],
        "time": ["00:00", "12:00"],
        "data_format": "grib",
    },
    "carra_west_t2m_an_20240101.grib",
)

# 2) Forecast: total precipitation accumulated from forecast start; (+18 h) - (+6 h) = 12-hour total
client.retrieve(
    "reanalysis-carra-single-levels",
    {
        "domain": "west_domain",
        "level_type": "surface_or_atmosphere",
        "product_type": "forecast",
        "variable": ["total_precipitation"],
        "year": ["2024"], "month": ["01"], "day": ["01"],
        "time": ["00:00", "12:00"],
        "leadtime_hour": ["6", "18"],
        "data_format": "grib",
    },
    "carra_west_tp_fc_20240101.grib",
)

Practical tips¶

• The West and East domains overlap around Svalbard. The user guide recommends taking Svalbard from the East domain and Jan Mayen from the West domain; uncertainty estimates in the overlap also follow the East domain.
• Since production moved from Reading to Bologna on 2022-07-01, GRIB files carry a different shapeOfTheEarth, so latitudes and longitudes appear shifted by typically less than 1 km. ECMWF says the pre-July-2022 grid is the correct one; when combining periods, match by grid-point index or reuse the earlier coordinates. On the same date the level type of snow depth water equivalent (sd) changed from surface to heightAboveGround.
• The Lambert conformal grids do not overlay directly on a latitude–longitude grid. We could not confirm in the documentation whether 10u/10v are grid-relative or earth-relative; if direction matters, also download 10wdir and 10si and check against them.
• Grid points near the domain edges are influenced by the ERA5 lateral boundaries. ECMWF staff agreed in a forum reply that boundary effects deserve clearer documentation; dropping a few rows along the edges is the safe choice.
• The user guide warns that runoff over glaciers is often low and unrealistic because glacial ice melt is not represented. Do not use it directly for glacier mass-balance work.

Notices and change history¶

References for your paper¶

Dataset citation¶

Single levels — DOI 10.24381/cds.713858f6

@misc{carra_single_levels,
  author    = {{Copernicus Climate Change Service}},
  title     = {Arctic regional reanalysis on single levels from 1991 to present},
  year      = {2021},
  publisher = {ECMWF, Copernicus Climate Change Service (C3S) Climate Data Store (CDS)},
  doi       = {10.24381/cds.713858f6},
  note      = {Accessed on DD-MMM-YYYY}
}

Pressure levels — DOI 10.24381/cds.e3c841ad

Height levels — DOI 10.24381/cds.8679900d

Daily and monthly data — DOI 10.24381/43a81317

Reference papers¶

Near-surface temperature and wind evaluation (CARRA team) — DOI 10.33265/polar.v41.8002

@article{koltzow2022carra,
  author  = {K{\o}ltzow, Morten and Schyberg, Harald and St{\o}ylen, Eivind and Yang, Xiaohua},
  title   = {Value of the {Copernicus Arctic Regional Reanalysis (CARRA)} in representing near-surface temperature and wind speed in the north-east {European} {Arctic}},
  journal = {Polar Research},
  volume  = {41},
  year    = {2022},
  doi     = {10.33265/polar.v41.8002}
}

Sea-ice cover in CARRA — DOI 10.5194/tc-18-1157-2024

@article{batrak2024carra,
  author  = {Batrak, Yurii and Cheng, Bin and Kallio-Myers, Viivi},
  title   = {Sea ice cover in the {Copernicus Arctic Regional Reanalysis}},
  journal = {The Cryosphere},
  volume  = {18},
  number  = {3},
  pages   = {1157--1183},
  year    = {2024},
  doi     = {10.5194/tc-18-1157-2024}
}

Acknowledgements¶

Acknowledge the use of C3S data and cite the DOI and access date of each CDS entry you actually used (single, pressure, height or model levels, or daily and monthly data). There is no peer-reviewed overview paper for CARRA; the full system description is the Copernicus Knowledge Base "Full system documentation" (Yang et al., 2020). Check the current wording under "Citation and attribution" on the CDS dataset page.

Information on the datasets used in this study was compiled with the help of the Korean Climate Data Navigator (KCDN) (Sim, 2026; https://doi.org/10.5281/zenodo.23300923).

Links¶

• CDS — CARRA single levels dataset page
• CARRA Data User Guide (Copernicus Knowledge Base)
• C3S — Copernicus Arctic Regional Reanalysis service (CARRA1 and CARRA2)
• CARRA: list of references (Copernicus Knowledge Base)
• Successor CARRA2 — CDS reanalysis-pan-carra
• ECMWF Forum — carra tag