Data are available through the following servers:
Service Resource Locator
FTP  [A- Alaska,  H- Hawaii, W- West Coast, Z- Partial Disk]

[Please acknowledge "NOAA CoastWatch/OceanWatch" when you use data from our site and cite the particular dataset DOI as appropriate.]

The National Oceanic and Atmospheric Administration's Office of Satellite Data Processing and Distribution are generating operational sea surface temperature (SST) retrievals from the Geostationary Operational Environmental Satellites GOES-East and West. The generation of SSTs began with GOES-8 in 2000 and has continued to be generated through GOES-15. The current operational satellites are GOES-13 and GOES-15. They are situated at longitude 135oW and 75oW, respectively, thus allowing the acquisition of high-temporal-resolution SST retrievals. The algorithm calculates SST by utilizing a fully physical retrieval scheme based on modified total least squares (MTLS, Koner et al., 2015) and a probabilistic (Bayesian) approach for cloud masking (Merchant et al., 2005). A more detailed description can be described in the Algorithms and Bayesian Cloud Mask section below.

The GOES-13 and 15 imagers observe both northern and southern hemisphere sectors every half an hour. These 5-band (0.6, 3.9, 6.7, 10.7, 13.3 µm) images are processed to retrieve SST retrievals at 4-km resolution. The 3.9, 10.7 & 13.3 µm channels are used to determine the SST, while the 0.6, 3.9 and 10.7 µm channels are used to detect cloud contamination. Individual sectors are output as GHRSST Level-2 P SST product files, and retrievals are also remapped, averaged and composited hourly into single-byte per pixel "flat" binary files and posted to a server for user access. The retrievals are available approximately 90 minutes after the nominal epoch of the SST determinations. 3-hour and 24-hour composite files are also made available. CoastWatch Regional Imagery is generated every three hours by combining the 1-hourly SST images for these areas.

The same algorithm approach is used to generate SSTs from Meteosat-10 data (centered at 0¡ longitude), taking a sub-selection of channels from the SEVIRI instrument that corresponds with those of the GOES-Imager. The main differences for the end-user are that the Meteosat SST products are "full-disk", every 15 minutes, and have a resolution at the nadir point of ~3-km, at least for the GHRSST L2P data.


Prior to the implementation of the fully physical retrieval algorithm with GOES-13 and 15, the algorithm retrieval schemes were still based on Radiative Transfer Modeling (RTM), generating skin temperatures rather than bulk temperatures. The form of the prior GOES operational SST equation was:

where i is GOES-Imager channel number (2, 4, 5), S = sec(satellite zenith angle) - 1 and Ti is channel brightness temperature in Kelvin.

The current algorithm is a deterministic physical retrieval scheme based on modified total least squares (MTLS, Koner et al., 2015). The retrieval can be expressed as

where x is the vector of parameters to be retrieved, in this case SST and total column water vapor, i.e. [SST, TCWV]T, y is a vector of channel brightness temperatures, and subscripts r, ig and o denote "retrieved", "initial guess", and "observed" respectively. The initial guess brightness temperatures, yig are calculated using the Community Radiative Transfer Model (CRTM) with NCEP atmospheric profile and surface temperature information on a coarse grid (1¡«1¡). G is the gain matrix, which multiplies the difference between observed and initial guess brightness temperatures (i.e. Dy) to obtain the delta-adjustment in retrieved parameters. This approach is common to all physical retrieval methods – it is the details of the gain matrix formulation where methods differ. The MTLS method formulates G starting from a deterministic paradigm, whereby a true value is assumed to exist, and the inputs have error Acronyms. Starting from the assumption that Dy = KDx, where K is the matrix (Jacobian) of partial derivatives of the components of y with respect to the components of x, the least squares solution (a.k.a. "normal" equation) is

Note that K is obtained as part of the CRTM calculations. MTLS modifies the above equation accordingly

where I is the identity matrix and l is the regularization strength. The difference between this (and the key aspect of MTLS) and other regularization-type methods are that l itself is dynamically calculated on a pixel-by-pixel basis:

where k is the condition number of K, g = max(1, ||Dy||), and s2end is the smallest singular value of the matrix [K Dy]. Thus s2end is a pixel-by-pixel based estimate of the noise in the data, while the log of the condition number accounts for how ill-conditioned the retrieval is, and g reduces the damping towards the initial guess when the required increment is large. One final aspect of the MTLS algorithm is that it is possible to derive pixel-by-pixel estimates of error in the retrieval, and this information is included in the GHRSST L2P output as the Sensor Specific Error Statistics (SSES) bias and standard deviation fields.

Bayesian Cloud Mask

The previous cloud detection scheme was based on a series of threshold tests. The new methodology applies Bayes' theorem to estimate the probability of a particular pixel being clear of cloud given the satellite-observed brightness temperatures, a measure of local texture and channel brightness temperatures calculated for the given location and view angle using NCEP GFS surface and upper air data and the CRTM fast radiative transfer model. The method is described in detail in a paper by Merchant et al. (2005).

Validation Methodology

NOAA/NESDIS STAR/SOCD generates a matchup data base for validation of the GOES-SST retrieval algorithms. This is important for the maintenance and improvement of the GOES-SST products.

The global drifting buoys and the TOGA TAO moored buoy array are matched with GOES-SST retrievals within one hour and 5 km. The buoys used are extracted by the Climate Prediction Center (CPC), which is one of the National Weather Service's National Centers for Environmental Prediction (NCEP). The buoys are quality controlled using the Reynolds Optimum Interpolation Sea Surface Temperature (OISST) Analysis and NCEP Atmospheric Analysis Fields before being matched with the GOES-SST retrievals. Matchup files are stored in the NOAA Satellite Active Archive (SAA) for user access.

An automated validation system has been in place since the inception of the operational GOES-SST retrievals. This system computes the GOES-SST accuracy statistics (monthly) from the satellite-buoy matches. The program uses the match up file as an input to calculate the number of matches, bias, and standard deviation. The statistical results are continually updated with the latest matches.


NRT Geostationary SST


GOES-E/W / Imager Meteosat / SEVIRI



Short Name


Sample Filename


Dataset Type


Processing Level


Spatial Coverage


Temporal Coverage

CW Regions


Geographic / WGS84


Near real-time + 3 days



Swath Width

Near real-time

Sample Frequency


Temporal Repeat


Orbital Period



~24 hours Geo-synchronous, altitude 35780 km, 1436 min, inclination ~0 (0.180087)

Data Provider

Creator: NOAA OSPO
Release Place: Suitland, MD, USA




NOAA, Imager, sea surface temperature, SST, CoastWatch

  • Koner, P. K., A. Harris, and E. Maturi. "A Physical Deterministic Inverse Method for Operational Satellite Remote Sensing: An Application for Sea Surface Temperature Retrievals." IEEE Transactions on Geoscience and Remote Sensing 53, no. 11 (November 2015): 5872-88. doi:10.1109/TGRS.2015.2424219♦.
  • Kurihara, Yukio, Hiroshi Murakami, and Misako Kachi. "Sea Surface Temperature from the New Japanese Geostationary Meteorological Himawari-8 Satellite." Geophysical Research Letters 43, no. 3 (February 16, 2016): 2015GL067159. doi:10.1002/2015GL067159♦.
  • Maturi, Eileen, Andy Harris, Jon Mittaz, Chris Merchant, Bob Potash, Wen Meng, and John Sapper. "NOAA's Sea Surface Temperature Products From Operational Geostationary Satellites." Bulletin of the American Meteorological Society 89, no. 12 (December 1, 2008): 1877-88. doi:10.1175/2008BAMS2528.1♦.
  • Merchant, Christopher J., and Pierre Le Borgne. "Retrieval of Sea Surface Temperature from Space, Based on Modeling of Infrared Radiative Transfer: Capabilities and Limitations." Journal of Atmospheric and Oceanic Technology 21, no. 11 (November 1, 2004): 1734-46. doi:10.1175/JTECH1667.1♦.
  • Merchant, C. J., A. R. Harris, E. Maturi, and S. Maccallum. "Probabilistic Physically Based Cloud Screening of Satellite Infrared Imagery for Operational Sea Surface Temperature Retrieval." Quarterly Journal of the Royal Meteorological Society 131, no. 611 (October 1, 2005): 2735-55. doi:10.1256/qj.05.15♦.
  • Merchant, C. J., A. R. Harris, E. Maturi, O. Embury, S. N. MacCallum, J. Mittaz, and C. P. Old. "Sea Surface Temperature Estimation from the Geostationary Operational Environmental Satellite-12 (GOES-12)." Journal of Atmospheric and Oceanic Technology 26, no. 3 (March 1, 2009): 570-81. doi:10.1175/2008JTECHO596.1♦.
  • Wick, Gary A., John J. Bates, and Donna J. Scott. "Satellite and Skin-Layer Effects on the Accuracy of Sea Surface Temperature Measurements from the GOES Satellites." Journal of Atmospheric and Oceanic Technology 19, no. 11 (November 1, 2002): 1834-48. doi:10.1175/1520-0426(2002)019<1834%3ASASLEO>2.0.CO%3B2♦ (2002).
  • Wu, Xiangqian, W. Paul Menzel, and Gary S. Wade. "Estimation of Sea Surface Temperatures Using GOES-8/9 Radiance Measurements." Bulletin of the American Meteorological Society 80, no. 6 (June 1999): 1127-38.

(♦ - non-government website)