ERSST v6¶

NOAA Extended Reconstructed Sea Surface Temperature (ERSST), Version 6

Available

At a glance¶

Note on period: When checked on 2026-10-12, the NCEI monthly files ran through ersst.v6.202609.nc (September 2026), and the NOAA PSL single file also included September 2026. NCEI notes that the data are more reliable after the 1940s.

Overview¶

ERSST is a global monthly analysis of SST derived from the International Comprehensive Ocean–Atmosphere Dataset (ICOADS Release 3.0.2) ship and buoy observations plus Argo profiling floats, starting in 1850. Version 6 fills gaps with artificial neural network (ANN) interpolation to give a spatially complete 2° grid; no satellite data are used. It also upgrades ship-observation quality control (nearest-neighbour check), the large-scale filter and the proxy SST in ice-covered regions, and anomalies are computed against a 1991–2020 climatology. Since 2026 it has been NOAA's official SST dataset for ENSO monitoring and for the ocean component of NOAAGlobalTemp.

Available variables¶

The NCEI original files (ersst.v6.YYYYMM.nc) contain both the SST sst and the anomaly ssta (both °C); the NOAA PSL file (sst.mnmean.nc) contains only sst.

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

Sea surface temperature¶

Climatology (separate PSL files)¶

The full variable list is in the official documentation.

How to get the data¶

• NCEI direct download (monthly NetCDF) — HTTPS (wget / curl) No login needed
  • The official location given on the NCEI product page. One file per month, named ersst.v6.YYYYMM.nc (about 164 KB each).
• NCEI pub directory (monthly files + full record in one file) — HTTPS (wget / curl) No login needed
  • Holds the same monthly files (ersst.v6.185001.nc onward), plus ersst.v6.1850-last.nc in netcdf.one/ (1850 to the latest month in one file, about 272 MB) and v6-based climate index files in index/ (ersst.v6.el_nino.dat, pdo, amo, iod, atl3). Note that it sits under the v5 directory.
• NOAA PSL (full record in one file) — Web download / THREDDS / OPeNDAP No login needed
  • The full SST record is in a single file, sst.mnmean.nc, which is convenient to use directly. Monthly long-term mean files (1981–2010, 1991–2020) are also provided. The THREDDS catalog is at https://psl.noaa.gov/thredds/catalog/Datasets/noaa.ersst.v6/catalog.html.

Example code¶

Open the full-record file through the NOAA PSL OPeNDAP URL and compute the Niño 3.4 area-mean anomaly relative to 1991–2020.

import xarray as xr

url = "https://psl.noaa.gov/thredds/dodsC/Datasets/noaa.ersst.v6/sst.mnmean.nc"
sst = xr.open_dataset(url)["sst"]          # °C, monthly mean; lat runs 88 → -88 (north to south)

# 1991–2020 monthly climatology and anomalies
clim = sst.sel(time=slice("1991-01-01", "2020-12-01")).groupby("time.month").mean("time")
ssta = sst.groupby("time.month") - clim

# Niño 3.4 (5°S–5°N, 170°W–120°W = 190–240°E) area mean
nino34 = ssta.sel(lat=slice(5, -5), lon=slice(190, 240)).mean(["lat", "lon"])
print(nino34.sel(time=slice("2025-01-01", None)).round(2).to_series())

Practical tips¶

• v5 and v6 use different anomaly base periods (v5 1971–2000, v6 1991–2020). Do not compare or splice anomalies from the two versions as they are.
• v6 changes the interpolation method (EOT to ANN) and drops v5's 3-month running mean. Values and spatial patterns differ from v5 even for the same month, and month-to-month variability can be larger, so do not mix the versions in one time series.
• Latitude in the PSL file runs north to south (88 → -88). Selecting with sel(lat=slice(south, north)) returns an empty result; put the northern bound first, e.g. slice(5, -5).
• The PSL file's global attributes (summary, source, etc.) still carry v5-era text such as a '1971–2000 climatology' and 'NCEP GTS'. Follow the NCEI documentation for the base period and input data.
• A Niño 3.4 anomaly you compute yourself will not match NOAA's official ENSO indices (RONI, etc.), which use different definitions and smoothing. If you need to agree with the official ENSO status, use the index values NOAA distributes.
• Gridded uncertainty (ensemble) estimates have not yet been formally released. The developer guidance on the NCAR Climate Data Guide (August 2025) said they were expected in 2026; in October 2026 RONI uncertainty files appeared in the unc/ folder of the NCEI pub directory, but we have not found documentation for them.
• Before 1945, when observations are sparse — especially before 1900, during the two World Wars and south of 40°S — uncertainties are large and spurious variability is possible (NCAR Climate Data Guide). Check trends in these periods against other datasets such as HadSST or COBE-SST.
• Being 2° monthly means, the data cannot represent extremes or daily variability. When comparing with models, regrid the model to the 2° grid; the developers recommend comparing absolute SST rather than anomalies, because climatologies differ between products.
• The dataset DOI for v6 is 10.25921/z7cw-es89. Some catalogues (data.gov, and the 'landing page' link in the NCEI metadata) still show the v5 DOI (10.7289/V5T72FNM); do not use it to cite v6.

Notices and change history¶

References for your paper¶

Dataset citation¶

NCEI dataset DOI (v6) — DOI 10.25921/z7cw-es89

@misc{ersst_v6_dataset,
  author    = {Huang, Boyin and Yin, Xungang and Boyer, Tim and Liu, Chunying and Menne, Matthew J. and Rao, Yuhan Douglas and Smith, Thomas M. and Vose, Russell S. and Zhang, Huai-Min},
  title     = {{NOAA} Extended Reconstructed Sea Surface Temperature ({ERSST}), Version 6},
  year      = {2025},
  publisher = {NOAA National Centers for Environmental Information},
  doi       = {10.25921/z7cw-es89},
  note      = {Accessed DD MMM YYYY}
}

Reference papers¶

ERSST v6 paper, Part I (artificial neural network interpolation) — DOI 10.1175/JCLI-D-23-0707.1

@article{huang2025ersstv6a,
  author  = {Huang, Boyin and Yin, Xungang and Boyer, Tim and Liu, Chunying and Menne, Matthew and Rao, Yuhan Douglas and Smith, Thomas and Vose, Russell and Zhang, Huai-Min},
  title   = {Extended Reconstructed Sea Surface Temperature, Version 6 ({ERSSTv6}). Part {I}: An Artificial Neural Network Approach},
  journal = {Journal of Climate},
  volume  = {38},
  number  = {4},
  pages   = {1105--1121},
  year    = {2025},
  doi     = {10.1175/JCLI-D-23-0707.1}
}

ERSST v6 paper, Part II (quality control and large-scale filter) — DOI 10.1175/JCLI-D-24-0185.1

@article{huang2025ersstv6b,
  author  = {Huang, Boyin and Yin, Xungang and Boyer, Tim and Liu, Chunying and Menne, Matthew and Rao, Yuhan Douglas and Smith, Thomas and Vose, Russell and Zhang, Huai-Min},
  title   = {Extended Reconstructed Sea Surface Temperature, Version 6 ({ERSSTv6}). Part {II}: Upgrades on Quality Control and Large-Scale Filter},
  journal = {Journal of Climate},
  volume  = {38},
  number  = {4},
  pages   = {1123--1136},
  year    = {2025},
  doi     = {10.1175/JCLI-D-24-0185.1}
}

Acknowledgements¶

The NCEI metadata ask users to cite the dataset DOI and to reference the Huang et al. (2025) Part I and Part II articles. The NCEI product page still carries the editorial note "(we may need a data citation such as for v5 after data DOI is generated)", but DOI 10.25921/z7cw-es89 has already been issued and resolves to the NCEI metadata page. The author name "Chunying Kiu" in the NCEI cite-as text is a typo; the DOI registration and the papers give Liu. If you obtained the data from NOAA PSL, also include the acknowledgement requested on the PSL 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¶

• NCEI ERSST product page (v6 and v5)
• NCEI ERSST v6 metadata and citation page
• NOAA PSL ERSST v6
• NOAA PSL THREDDS — noaa.ersst.v6
• NCAR Climate Data Guide — ERSST v6
• NCEI News — Key NOAA Dataset Upgraded Using AI (2025-12-04)
• NCEI — Cloud Migration Offers Bright Future for NOAA's Data