Skip to content

CFSR / CFSv2

NCEP Climate Forecast System Reanalysis (CFSR) + CFS Version 2 Operational Analysis

Available

Get notice alertsCompare with other datasetsReport an error

Current status

The CFSR reanalysis (1979 to March 2011) is a fixed record; it is continued by the analyses of CFSv2, operational since 30 March 2011. As of 2026-10-11, the CFSv2 6-hourly data at NCAR GDEX are updated daily and run to 11 October. NOAA is developing a new coupled Seasonal Forecast System (SFS) to replace CFSv2, but as of 2026-10-11 no implementation date and no end date for the CFSv2 analyses had been announced.

At a glance

Item Details
Section Atmosphere › Reanalyses
Type Reanalysis
Domains Atmosphere, Ocean, Sea ice, Land
Provider NOAA NCEP Environmental Modeling Center (EMC) (distributed by NOAA NCEI and NSF NCAR GDEX)
Period 1979-01 ~ present
Coverage Global
Horizontal resolution Atmospheric analysis CFSR T382 (~38 km) → CFSv2 T574 (27 km) / output on 0.5°, 1.0° and 2.5° latitude-longitude and Gaussian grids / ocean 0.25° at the equator to 0.5° beyond the tropics
Vertical resolution Atmosphere 64 levels / ocean 40 levels / soil 4 levels / sea ice 3 layers
Temporal resolution 6-hourly analyses (00, 06, 12, 18 UTC) with 1–6 (9) hour forecasts, hourly time series, monthly means
Update frequency CFSv2 analyses updated daily (NCAR GDEX d094000)
Licence NCAR GDEX copies under CC BY 4.0 (Creative Commons Attribution 4.0 International) / NOAA NCEI data are U.S. Government public data
Account required No

Note on period: CFSR from January 1979 to March 2011 plus the CFSv2 operational analysis from April 2011 to the present (NCEI split). NCAR GDEX splits the record differently: CFSR through December 2010 (d093000) and CFSv2 from January 2011 (d094000).

How to read the resolution
  • Horizontal resolution lists both the native model grid and the grid you download, as given in the official documentation. The same dataset can come on different grids from different providers or products, so check the grid of your downloaded file.
  • Grid spacing is not the effective resolution. A provided grid that is finer than the native one carries no extra information.
  • Intervals and whether a value is instantaneous, mean or accumulated differ between variables — see 'Available variables' below.
  • The 'provided resolution' in lists and the comparison table is a simplified km/degree value for quick comparison. Use this page for exact values.

Overview

Combines NCEP's third-generation reanalysis, CFSR, made with a coupled atmosphere-ocean-land-sea ice model, and the analyses of CFSv2, which has run operationally with the same model since 2011. CFSR couples atmosphere and ocean during the 6-hour guess forecast and assimilates satellite radiances directly, and it had the highest resolution of any NCEP reanalysis of its time. Because full 3-D ocean fields are included, it is widely used for air-sea interaction studies and for boundary conditions for regional models.

What changes with the successor

CFSR (reanalysis) → CFSv2 operational analysis

Transition: 2011-04

The CFSR reanalysis was produced only through March 2011; later periods come from the real-time analyses of CFSv2, which became operational on 30 March 2011. NCAR describes the CFSv2 products as an extension of CFSR from the same model, but production changed, including an increase in atmospheric analysis resolution from T382 to T574. The two can be used together from 1979 to the present, but they do not form a homogeneous reanalysis across 2011.

Aspect CFSR (reanalysis) CFSv2 operational analysis
Period 1979-01-01 to 2011-03-31 (NCEI; NCAR d093000 ends in 2010-12) 2011-04-01 to present (NCEI; NCAR d094000 starts in 2011-01)
Production Reanalysis computed for the past period in one campaign Operational real-time analysis that initialises the CFSv2 forecasts (updated daily at NCAR)
Atmospheric analysis horizontal resolution T382 (~38 km) T574 (~27 km)
Vertical levels 64 levels 64 hybrid sigma-pressure levels (unchanged)
Highest-resolution output grid Gaussian ~0.312° (1152 x 576) Gaussian ~0.204° (1760 x 880)
Surface and radiative flux (FLX) grid Gaussian T382 Gaussian T574
Model Coupled CFS model (atmosphere, ocean, land, sea ice) Same model as CFSR (per NCAR GDEX); ocean MOM4 with GODAS
Reforecasts CFSRR reanalysis and reforecasts available NCEP states that the CFSRR will not be extended beyond March 2011; later forecasts exist only as operational output
Archived initial conditions and observation input High- and low-resolution initial conditions and full observation dumps for the whole period High-resolution initial conditions and observation dumps archived at NCEI only for 2011-04 to 2014-12; low-resolution initial conditions to the present
Datasets and DOIs NCAR d093000 (6-hourly, doi:10.5065/D69K487J), d093002 (monthly, doi:10.5065/D6DN438J) / NCEI CFS Reanalysis NCAR d094000 (6-hourly, doi:10.5065/D61C1TXF), d094002 (monthly, doi:10.5065/D69021ZF) / NCEI CFS Operational Analysis

Things to watch when switching

  • Resolution and production change around January–April 2011, so check for a step change there before joining 1979–present into one series. Take particular care with resolution-sensitive quantities such as precipitation, surface fluxes and extremes indices.
  • NCAR and NCEI split the record differently. NCAR ends CFSR in December 2010 and starts its CFSv2 dataset (d094000) in January 2011, whereas NCEI runs CFSR to March 2011 and the operational analysis from April. Mixing the two archives can duplicate or drop January–March 2011.
  • The operational analysis is not recomputed for long-term homogeneity the way a reanalysis is. For trends after 2011, check your results against long reanalyses such as ERA5 or JRA-3Q.
  • The high-resolution output grid changes (1152 x 576 → 1760 x 880), so files with the same name have different grid definitions. Regrid both periods to a common grid (e.g. 0.5°) before combining them.

Sources: Saha et al. (2010), The NCEP Climate Forecast System Reanalysis, BAMS, 91, 1015–1057 · Saha et al. (2014), The NCEP Climate Forecast System Version 2, J. Climate, 27, 2185–2208 · NWS Technical Implementation Notice 10-55, Amended (CFS changes effective 2011-03-30) · NOAA NCEI — Climate Forecast System (CFSR and CFSv2 product table) · NSF NCAR GDEX — CFSR 6-hourly (d093000) · NSF NCAR GDEX — CFSv2 6-hourly (d094000) · NCEP CFS version 2 page (CFSRR will not be extended beyond March 2011)

Available variables

Key fields from the NCAR GDEX d093000 GRIB2 parameter table and the NCEP GRIB2 code tables. Names are GRIB2 abbreviations (as shown by wgrib2 and cfgrib). Pressure-level fields are mostly in the PGB files (pgbh, pgbl) and surface and radiative fluxes in the FLX files (flxf); not all parameters are available at all resolutions.

Time-stamp convention — instantaneous, mean or accumulated? — click to show

CFSR and CFSv2 make an analysis every 6 hours (00, 06, 12, 18 UTC) and run 1–6 hour forecasts from it. Pressure-level analyses (pgbh etc., f00) are instantaneous at the valid time. Averaged and accumulated fields such as precipitation, heat fluxes and radiation come only from the forecasts, averaged or accumulated from the start of each 6-hour cycle to the forecast hour (0–1, 0–2 … 0–6 h). Some fields, such as 2 m temperature and 10 m winds, are not analysed, so always check the time description of each GRIB record (anl / N hour fcst / 0-N hour ave).

  • Averaging and accumulation restart at the beginning of every 6-hour cycle. The f01–f05 averages in the hourly time series are 0–N hour averages, not instantaneous or one-hour values; for example, the 2–3 h average is 3 × (0–3 h average) − 2 × (0–2 h average) (NCEI CFSRR Known Issues).
  • The hourly time series are built by concatenating the analysis with 1–5 hour forecasts, or 1–6 hour forecasts. The f00 record is a spin-up of the model physics lasting several minutes — it is not the analysis.
  • PRATE, LHTFL, DSWRF and similar fields are mean rates (kg m-2 s-1, W m-2); APCP is an accumulation (kg m-2 = mm). Multiply a mean rate by the interval length in seconds to get an amount.
  • TMAX and TMIN are the maximum and minimum within the forecast interval.
  • Monthly means (d093002, d094002, NCEI Monthly Means) average the same cycle and forecast step over the month. What is averaged depends on the cycle in the file name (e.g. .06Z); see the wgrib2 -fix_CFSv2_fcst notes.

Source: NOAA NCEI CFSRR Known Issues technical document (August 2011) + NCAR GDEX d093000 and d094001 descriptions

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

Pressure levels (PGB)

Variable Short name Units Time convention Notes
Temperature TMP K instantaneous
Geopotential height HGT gpm instantaneous
U- and V-component of wind UGRD, VGRD m s-1 instantaneous
Vertical velocity (pressure) VVEL Pa s-1 instantaneous
Specific humidity SPFH kg kg-1 instantaneous
Relative humidity RH % instantaneous
Potential temperature POT K instantaneous

Surface and single level

Variable Short name Units Time convention Notes
Pressure reduced to MSL PRMSL Pa instantaneous
Surface pressure PRES Pa instantaneous
Precipitable water PWAT kg m-2 instantaneous
Dew point temperature DPT K instantaneous Values such as 2 m may be unanalysed forecasts (NCAR: 2 m temperature and 10 m winds are not analysed)
Maximum / minimum temperature TMAX, TMIN K Forecast interval (0–N h) maximum TMAX is the interval maximum and TMIN the interval minimum (kind is set for the maximum). Reset every 6-hour cycle
Ice cover ICEC fraction instantaneous
Planetary boundary layer height HPBL m instantaneous Diagnosed during the forecast step. Check the record's time description to see whether it is instantaneous or averaged

Fluxes, precipitation and cloud (FLX)

Variable Short name Units Time convention Notes
Precipitation rate PRATE kg m-2 s-1 Forecast start to valid time (up to 6 h) mean
Total precipitation APCP kg m-2 Forecast start to valid time (up to 6 h) accumulated Accumulation restarts from zero every 6-hour cycle
Latent and sensible heat net flux LHTFL, SHTFL W m-2 Forecast start to valid time (up to 6 h) mean
Downward short-wave and long-wave radiation flux DSWRF, DLWRF W m-2 Forecast start to valid time (up to 6 h) mean
Upward short-wave and long-wave radiation flux USWRF, ULWRF W m-2 Forecast start to valid time (up to 6 h) mean
Momentum flux, u and v components UFLX, VFLX N m-2 Forecast start to valid time (up to 6 h) mean
Total cloud cover TCDC % Forecast start to valid time (up to 6 h) mean We could not confirm from the CFSR documentation which levels are averaged and which are instantaneous. This follows the general rule for FLX files (interval mean), so check the record's time description

Land and snow

Variable Short name Units Time convention Notes
Volumetric soil moisture content SOILW fraction instantaneous
Water equivalent of accumulated snow depth WEASD kg m-2 instantaneous
Snow depth SNOD m instantaneous

The full variable list is in the official documentation.

How to get the data

Practical tips

  • NCAR notes that some parameters, such as 2 m temperature and 10 m winds, are not analysed but come from the forecast. Choose between the analysis and forecast steps (+00, +01 to +06) deliberately.
  • NCEI notice (18 February 2010): some CFSR time-series files downloaded before 22 UTC on 18 February 2010 (28 of the 73 file types) have wrong GRIB metadata that makes averages read as instantaneous values. Download those files again.
  • The f00 records in the CFSR time series are model spin-up, not the analysis (NCEI Known Issues). The same document says not to use tmpsfc from the ipv files and to take it from the flx files instead.
  • Resolution and dataset change at the 2011 boundary, so pick either the NCAR or the NCEI split and align your periods to it before any long-term analysis.

Notices and change history

2011-08 (day not given) · Info · CFSRR Known Issues technical document released

NCEI provides the August 2011 document listing known issues in the CFSR and reforecast files: the f00 time-series records are not the analysis, a precision problem in some monthly means, tmpsfc in the ipv files should not be used, and mislabelled isotherm levels in the ocean files, among others.

Read the original

2011-04-18 · Info · Amended: Climate Forecast System (CFS) changes effective March 30, 2011 (TIN 10-55)

NWS confirmed that CFSv2 became operational beginning with the 1200 UTC cycle on 30 March 2011 and that the original CFS will continue to run through June 2012. The analysis moves to T574 (27 km) with 64 hybrid sigma-pressure levels, the atmospheric forecast model to T126 (100 km), and the ocean to the global MOM4.

Read the original

2010-02-18 · Warning · GRIB2 metadata issue in CFSR time-series files corrected

A format conversion issue affected the metadata in non-instantaneous CFSR time-series files (28 of the 73 file types), so averaging and accumulation intervals were recorded as instantaneous. Corrected files were in place by 22 UTC on 18 February 2010; download again any files obtained earlier (the data values themselves are valid).

Read the original

References for your paper

Double-check before using in a paper

The citations below were compiled by the maintainer and may contain errors. Always double-check DOIs, authors, years and access-date formats against the provider's official dataset page and your journal's guidelines.

Dataset citation

NCAR GDEX CFSR 6-hourly products (d093000) — DOI 10.5065/D69K487J

NCEP Climate Forecast System Reanalysis (CFSR) 6-hourly Products, January 1979 to December 2010. NSF NCAR Geoscience Data Exchange (GDEX). https://doi.org/10.5065/D69K487J (Accessed DD MMM YYYY) — copy the exact format, including authors and year, from the Citation tab on the GDEX dataset page.

NCAR GDEX CFSv2 6-hourly products (d094000) — DOI 10.5065/D61C1TXF

NCEP Climate Forecast System Version 2 (CFSv2) 6-hourly Products. NSF NCAR Geoscience Data Exchange (GDEX). https://doi.org/10.5065/D61C1TXF (Accessed DD MMM YYYY)

Reference papers

CFSR reference paper — DOI 10.1175/2010BAMS3001.1

Saha, S., Moorthi, S., Pan, H.-L., Wu, X., Wang, J., Nadiga, S., Tripp, P., Kistler, R., Woollen, J., Behringer, D., et al. (2010): The NCEP Climate Forecast System Reanalysis. Bulletin of the American Meteorological Society, 91(8), 1015–1057.

Show BibTeX 
@article{saha2010cfsr,
  author  = {Saha, Suranjana and Moorthi, Shrinivas and Pan, Hua-Lu and Wu, Xingren and Wang, Jiande and Nadiga, Sudhir and Tripp, Patrick and Kistler, Robert and Woollen, John and Behringer, David and others},
  title   = {The {NCEP} Climate Forecast System Reanalysis},
  journal = {Bulletin of the American Meteorological Society},
  volume  = {91},
  number  = {8},
  pages   = {1015--1057},
  year    = {2010},
  doi     = {10.1175/2010BAMS3001.1}
}

CFSv2 paper (when using analyses from 2011 onwards) — DOI 10.1175/JCLI-D-12-00823.1

Saha, S., Moorthi, S., Wu, X., Wang, J., et al. (2014): The NCEP Climate Forecast System Version 2. Journal of Climate, 27(6), 2185–2208.

Show BibTeX 
@article{saha2014cfsv2,
  author  = {Saha, Suranjana and Moorthi, Shrinivas and Wu, Xingren and Wang, Jiande and others},
  title   = {The {NCEP} Climate Forecast System Version 2},
  journal = {Journal of Climate},
  volume  = {27},
  number  = {6},
  pages   = {2185--2208},
  year    = {2014},
  doi     = {10.1175/JCLI-D-12-00823.1}
}

Acknowledgements

Cite Saha et al. (2010) if you used only 1979–2010 (or to March 2011), and add Saha et al. (2014) if you also used the analyses from 2011 onwards. Give the DOI of each NCAR dataset you used, or, for NCEI downloads, the NCEI product page and your access date.

If this page helped, please acknowledge the Korean Climate Data Navigator

Please add the sentence below to the Acknowledgements of your paper.

Acknowledgements

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).