Data Access

NCEP Sea Surface Temperature
NCEP Sea Surface Temperature Anomalies
NCEP Sea Surface Temperature Climatology

Readme Contents

Data Set Overview

Sponsor

Original Archive

Future Updates

The Data

Characteristics

Source

The Files

Format

Name and Directory Information

Companion Software

The Science

Theoretical Basis of Data

Processing Sequence and Algorithms

Scientific Potential of Data

Validation of Data

Data Access and Contacts

FTP Site

Points of Contact

References

This data set contains the National Center for Environmental Prediction (NCEP), formally the National Meteorological Center (NMC) Optimal Interpolation (OI) monthly mean sea surface temperature (SST) data (Reynolds and Smith, 1994) which starts with November 1981 and continues to the present. Also included are climatological values for each of the twelve months (Reynolds and Smith, 1995). The temperatures were derived over water areas on a 1x1-degree latitude and longitude world grid. An Optimal Interpolation analysis was used to combine measurements from satellite-borne instruments and in situ ship and buoy platforms. The monthly SST values are presented in two forms: the actual values, and as differences (anomalies) from the climatological temperatures for the month. Additional months of data will be made available from this site as they become available from NCEP. These SST measurements provide a means of monitoring both the long- and short-term variability of the dynamic and thermodynamic properties of the ocean on a global scale. On this site, this relatively new OI SST data set replaces the original Climate Analysis Center "blended" SST data set (Reynolds, 1988). The OI analysis, derived on a 1x1-degree grid, shows greater regional detail than the "blended" analysis which was derived on 2x2-degree grid.

Sponsor

The collection and distribution of this data set at this site are being funded by NASA's Earth Science enterprise. The data are not copyrighted; however, we request that when you publish data or results using these data :

Please reference Reynolds and Smith (1994) and Reynolds and Smith (1995) as appropriate and thank the Distributed Active Archive Center (Code 902) at Goddard Space Flight Center, Greenbelt, MD, 20771, for putting the data in its present format and distributing them. Goddard's share in these activities was sponsored by NASA's Earth Science enterprise.

Original Archive

These data were obtained from the Climate Modeling Branch of NCEP/NOAA where the data in its original format is available. Richard W. Reynolds leads the group responsible for its calculation. The sst anomalies are easily calculated as (sst-climatological sst). The anomalies are also available from the Integrated Global Ocean Services System (IGOSS).

In setting up the sea surface temperature data as part of the Goddard Interdiscipline Collection (IDC), two adjustments were made. The original data ran from south to north. To conform with the other IDC data sets the temperature data was flipped so that here it now runs from north to south. The original data also had unreal temperature data over land which had been put in by Cressman interpolation. This was done to provide complete global data for those desiring to do interpolation. To avoid possible confusion by some general users a land mask was inserted so that no land temperatures appear in this IDC data set.

Future Updates

The Goddard DAAC will update this data set as new data are processed and made available by NCEP.

The Data

Characteristics

• Parameters, Units

  Parameter	Range	Units
  monthly Sea Surface Temperature (SST)	-1.8 to 35	Degrees Celsius
  Climatological SST	-1.8 to 33
  Monthly SST Anomalies	-9.5 to 7.6

• Temporal Coverage:

  Start	End
  November 1981	July 1997

• Temporal Resolution:

  All gridded values are monthly means

• Spatial Coverage:

  Global

• Spatial Resolution:

  1-degree x 1-degree

Source

The NCEP Sea Surface Temperature data set is derived from both in situ (ocean based) measurements as well as global satellite observations. The in situ data consist of ship and buoy observations, while the satellite data are collected from the Advanced Very High Resolution Radiometer (AVHRR) flown aboard the NOAA-7, NOAA-9, NOAA-11, and NOAA-14 polar orbiting platforms.

For the more recent period, 1990-present, the in situ data were obtained from radio messages carried on the Global Telecommunication System. The satellite observations were obtained from operational data produced by the National Environmental Satellite, Data and Information Service (NESDIS).

During the period 1981-1989, the in situ data were obtained from the Comprehensive Ocean Atmosphere Data Set (COADS) for the 1980s. These data (see Slutz, et al., 1985 , and Woodruff, et al., 1993) consist of logbook and radio reports. The satellite data were obtained from analyses of NESDIS data produced at the University of Miami's Rosenstiel School of Marine and Atmospheric Sciences.

Nominal orbital parameters for NOAA satellites
	NOAA-7	NOAA-9	NOAA-11	NOAA-14
Launch Date	06/23/81	12/12/84	9/24/88	12/30/94
Orbit	Sun Synchronous
Nominal Altitude	833 km
Inclination	98.8 degrees
Orbital Period	102 minutes
Equatorial daytime crossing time at launch:	1430 LST	1420 LST	1340 LST	1330 LST
Nodal Increment	25.3 degrees

The NOAA-series satellites carry the AVHRR instruments. The orbital period of about 102 minutes produces 14.1 orbits per day. Because the daily number of orbits is not an integer, the sub orbital tracks do not repeat daily, although the local solar time of the satellite's passage is essentially unchanged for any latitude. The 110.8 degrees cross-track scan equates to a swath of about 2700 km. This swath width is greater than the 25.3 degrees separation between successive orbital tracks and provides overlapping coverage (side-lap).

The spectral band widths and Instantaneous Field of View (IFOV) of the AVHRR instrument are given in the following table.

Channel	Wavelength (micrometer)	IFOV (milliradians)
1	0.58 - 0.68	1.39
2	0.73 - 1.10	1.41
3	3.55 - 3.93	1.51
4	10.3 - 11.3	1.41
5	11.5 - 12.5	1.30

At a nominal altitude of 833 km, the IFOV values result in a resolution of 1.1 km at the ground for each measurement at nadir. Over sampling occurs in all the channels--e.g. in channel 5 there are 1.362 samples per IFOV. However global coverage is available only in the Global Area Coverage (GAC) data set. GAC data contains only one out of three original AVHRR scan lines and along a scan line every four adjacent samples are averaged and the fifth sample is skipped. The GAC data thus has a 4 km resolution. A more detailed, comprehensive description of the NOAA series satellites and the AVHRR instrument can be found in the NOAA Polar Orbiter Data User's Guide (Kidwell, 1991 & 1995).

The Files

Format

File Size	259200 bytes (64800 data values)
Data Format	IEEE floating point
Headers	none
Trailers	none
Delimiters	none
Land/water mask	Land	-99.999
Orientation	Start	179.5W, 89.5N
	End	179.5E, 89.5S

Name and Directory Information Naming Convention

The file names for the Sea Surface Temperature data set conform to the Interdisciplinary Data Collection template

xxxxxxxx.pppppp.llctgrr.[yymm].ddd

which is interpreted as follows:

Substring	Function		NCEP SST Indicator	Specific meaning
xxxxxxxx	data product designator		ncep_sst	NCEP sea surface temperature products
pppppp	parameter name		sst	Sea surface Temperature
			anom	Sea Surface Temperature Anomalies
			clim	Sea Surface Temperature Climatology
llctgrr	ll	number of levels	1
	c	vertical coordinate	n	not applicable
	t	temporal period	m	monhtly
	g	horizontal grid resolution	e	1 x 1-degree
	rr	spatial coverage	go	global
yymm	yy	year	81 - 96	range of years
	mm	month	01 - 12	range of months
	[dd]	day	not used
ddd	data format		bin	IEEE 32-bit
			ctl	GrADS control file
Note:Indicators in bold are constant. Others refer to variable values, i.e., ncep_sst.anom.1nmego.9206.bin

Directory Paths to Data Files

/data/surf_temp_press/ncep_sst/sst/yyyy/

/data/surf_temp_press/ncep_sst/anom/yyyy/

/data/surf_temp_press/ncep_sst/clim/yyyy/

where yyyy refers to year.

Companion Software
Several software packages have been made available on the CIDC CD-ROM set. The Grid Analysis and Display System (GrADS) is an interactive desktop tool that is currently in use worldwide for the analysis and display of earth science data. GrADS meta-data files (.ctl) have been supplied for each of the data sets. A GrADS gui interface has been created for use with the CIDC data. See the GrADS document for information on how to use the gui interface.

Decompression software for PC and Macintosh platforms have been supplied for datasets which are compressed on the CIDC CD-ROM set. For additional information on the decompression software see the aareadme file in the directory:

software/decompression/

Sample programs in FORTRAN, C and IDL languages have also been made available to read these data. You may also acquire this software by accessing the software/read_cidc_sftwr directory on each of the CIDC CD-ROMs

Theoretical Basis of Data
Two primary methods exist for determining sea surface temperature on regional and global scales. The first involves traditional, direct measurements of SST using ship-borne instrumentation and both fixed and drifting buoys. A rather large network of these observation platforms was set up over the past several decades to provide near continuous measurements of sea surface characteristics linked through a global telecommunication system. However, sufficient coverage of the sea surface temperature is generally available for in situ observations only from 60 degrees North to 30 degrees South and even there gaps occur.

The second method uses satellite observations to indirectly infer SST from radiance measurements in a set of discrete spectral channels sensitive to emission of electromagnetic energy from Earth's surface. Assuming constant surface emissivity and negligible effects from the overlying atmosphere, the amount of energy received by the satellite sensor will vary with fluctuations in the surface temperature in accordance with the Planck function. The modulation of the surface- emitted radiation by the intervening atmosphere can be minimized by judicious selection of the spectral channels. For the AVHRR instrument, the 3.5-4.0 micron and 10-12 micron channels (channels 3,4,5) are used in the retrieval of SST. These channels are located in the so-called "window" regions of the spectrum in which there is relatively little absorption and emission of infrared radiation by the atmosphere (except in humid regions like the tropics where water vapor in the lower atmosphere plays an increasingly important role in absorbing the surface-emitted radiation). In addition to the effects of water vapor, clouds are effectively opaque to radiation across the infrared spectrum (including the window regions) and therefore tend to mask radiation emitted by the surface from the satellite sensor. Finally, small but non negligible variations in the ocean surface emissivity within the window channels must be taken into account when attempting to retrieve reliable estimates of SST.

The basis of satellite retrievals of sea surface temperature involves developing regression formulas relating SST to a combination of multiple satellite window measurements. The coefficients of the channel measurements are originally calculated using theoretical simulations, but then Reynolds and Smith (1994) state that

"the satellite SST retrieval algorithms are 'tuned' by regression against quality controlled drifting buoy data using the multichannel SST technique of McClain et al. (1985) and Walton (1988). The tuning is done when a new satellite becomes operational or when verification with the buoy data shows increasing errors."

The algorithms are global in nature and are not time dependent . This tuning procedure has the effect of removing global biases that exist between the satellite-derived SST and the "ground truth" SST as measured by buoys. The use of multiple channels in the regression essentially provides corrections for the effects of varying atmospheric water vapor and surface emissivity in this type of retrieval. However, before attempting to retrieve SST with these algorithms, it must first be determined that no contamination from clouds exist in the observed radiances. This screening procedure is reviewed briefly in McClain et al. (1985). The 4 km resolution GAC data are formed into 2x2 or larger blocks which are subjected to a number of tests to see if any clear scenes are present. The tests are based on the assumptions that: clouds are brighter and colder than the clear ocean; the clear ocean will be nearly uniform from one scene to its neighbors but clouds will often not be; and at night the channel 3 (3.7 micrometers) black body temperature will behave differently, when compared to those from channels 4 & 5, over clear ocean than over cloudy scenes. There are separate day and night tests. In the open ocean the GAC data is organized into "targets" with a nominal spacing of 25 km. These targets are processed one at a time by the satellite SST algorithm. A target consists of five 11x11 arrays, one corresponding to each of the five AVHRR channels. If some clear ocean scenes are identified within the target a single sea surface temperature value is calculated for the target, otherwise the target is rejected as cloud covered. Details concerning the cloud screening and AVHRR SST algorithms are given by McClain et al. (1985) and Walton (1988).

Processing Sequence and Algorithms
The methodology called optimal interpolation (OI) is used to produce the sea surface temperatures on a 1x1-degree world grid, with land areas excluded. Four data measurement sets are collected and screened for bad data. These are: in situ ship measurements, in situ buoy measurements, daytime satellite measurements, and nighttime satellite measurements. There is no reflected solar radiation at night hence the day and night satellite SST algorithms differ. After the data preparation, Optimal Interpolation analysis is used to calculate the final SST values. Operationally both daily and weekly OI analyses are made. For this study weekly OI analyses were used. The monthly OI fields are derived by a linear interpolation of the weekly OI fields to daily fields then averaging the daily values over a month. The monthly fields are in the same format and spatial resolution as the weekly fields. The discussion is broken into four parts:

• data preparation
• Optimal Interpolation (OI) analysis
• the Climatology
• DAAC modifications

Data preparation

Reynolds and Smith (1994) list the preliminary steps. Before the SST data are used in the OI they must pass the quality control procedures. These procedures include the use of programs that track ships and buoys by their identification codes and eliminate observations with unlikely position changes. All in situ observations that pass the tracking tests and all satellite retrievals are tested for accuracy of the SST. All observations are discarded if the SST is less than -2-degree C or greater than 35-degree C or if the SST anomaly lies outside + -3.5 times the climatological standard deviations. These tests were designed to eliminate some of the worst data."

To improve the SST near the ice edge, sea ice information is used. It became available at the National Center for Environmental Prediction (NCEP) in real time starting in 1988. The data have a 2-degree spatial and a weekly temporal resolution.

"If a grid box was ice covered (concentration of 50% or greater), an SST value was generated with a value of -2-degree C. The freezing point temperature of sea water with a salinity of 33 to 34 psu is -1.8-degree C. This range of salinity is typical near the ice edge in the open ocean. Thus, -2-degree C is slightly too negative. After the SST field has been computed, any SST gridded value less than -1.8-degree C is set to -1.8-degree C. The use of simulated SSTs of -2-degree C over ice-covered regions allows the analysis to reach its fixed minimum more robustly."

The OI analysis assumes that the data contains errors but is not biased. Reynolds et al. (1989) showed that regional, time dependent biases exist between the satellite and in situ SST, and that most of the problem lay with the satellite measurements. Further, daytime and the nighttime satellite measurements had different biases. These perturbations arise from volcanic aerosol clouds and other causes (Reynolds, 1993; Reynolds and Smith, 1994). Before the OI analysis, separate adjustments are made in the satellite daytime and nighttime measurements using the Poisson technique of Reynolds (1988) and Reynolds and Marsico (1993). The procedure uses Poisson's equation to form preliminary weekly blends of the in situ and satellite measurements on a 4x4-degree grid. The blended results have a resolution of about 12-degrees. The blended analysis adjusts any large scale satellite biases and gradients relative to the boundary conditions defined by the in situ analysis. The blended results are then interpolated to the original satellite measurement positions to determine regional, time dependent bias corrections (Reynolds and Smith, 1994).

Optimal Interpolation (OI) Analysis

The OI analysis procedure is discussed in Reynolds and Smith (1994). The analysis is computed in terms of increments rather than the actual temperatures so that the first guess is preserved in regions with little or no data. the OI analysis is now produced both daily and weekly on a 1- degree grid. Since local conditions tend to persist for a time, the previous OI analysis is used as a first guess for the next analysis. This was found to be more accurate than using climatology as a first guess. To reduce the number of observations used in the OI, averages over 1-degree squares are computed. These 'super observations' are computed independently for each ship and buoy identification code and for day and night satellite retrievals. Ships normally make only one, 6-hourly report in a given grid box. Thus the chief effect of this averaging is to reduce the number of buoy and satellite measurement values used in the analysis. The analysis increment is defined as the difference between the analysis and the first guess; the data increment is defined as the difference between the data and the first guess. The analysis increment, r(k), is given by

r(k) = sum[w(ik)q(i)]

where q(i )are the data increments and w(ik) are the least square weights.

The subscript (k) ranges over the grid points where the solution is required and the subscript (i) ranges over the data points. When there is little or no data in a region the weights, w(ik), approach zero. The OI is only optimal when the correlations and variances needed to calculate the w(ik) are known for the analysis increment and for each type of data increment. The calculation of the w(ik) involve matrix inversions which become unstable if too many data points are involved. This is one reason that averages over the grid squares are formed before the analysis. The ship measurements are noisier than the buoy and satellite measurements and therefore have smaller weight values in the analysis.

Reynolds and Smith 1994 Monthly AOI SST climatology

Reynolds and Smith (1995) also produced an adjusted optimal interpolation (AOI) climatology for each calendar month on a 1x1-degree grid. The following description is taken from the abstract of their paper.

Abstract In response to the development of a new higher resolution sea surface temperature (SST) analysis at the National Center for Environmental Prediction (NCEP), a new monthly 1-degree global sea surface temperature climatology was constructed from two intermediate climatologies: the 2-degree SST climatology presently used at NCEP and a 1-degree SST climatology derived from the new analysis. The 2-degree SST climatology used a 30-year 1950-79 base period between roughly 40S and 60N based on in situ (ship and buoy) SST data supplemented by 4 years (1982-85) of satellite SST retrievals. The 1-degree SST climatology was based on monthly analyses using in situ SST data, satellite SST retrievals, and sea-ice coverage data over a 12-year period (1982-93). The final climatology was combined from these two products so that a 1-degree resolution was maintained and the base period was adjusted to the 1950-79 period wherever possible (approximately 40S and 60N). Compared to the 2-degree climatology, the 1-degree climatology resolves equatorial up welling and fronts much better. This leads to a better matching of the scales of the new analysis and climatology. In addition, because the magnitudes of large-scale features are consistently maintained in both the older 2-degree and the new 1-degree climatologies, climate monitoring of large-scale anomalies will be minimally affected by the analysis change. The use of 12 years of satellite SST retrievals makes this new climatology useful for many additional purposes because its effective resolution actually approaches 1-degree everywhere over the global ocean and because the mean SST values are more accurate south of 40S than climatologies without these data.

Reynolds and Smith (1995) point out that this new climatology is an improvement over both the Climate Analysis Center Climatology (Reynolds, 1988) and that developed by Shea et al. (1992). A major advantage is that this climatology has a 1x1-degree resolution while both of the other two have a 2x2 degree resolution.

Reynolds and Smith also formed an SST anomaly data set by subtracting climatological SST from the monthly OI analysis SST. All three data sets ( OI analysis SST, the AOI Climatology, and the SST monthly anomalies) are included at these site.

Interdisciplinary Data Collection Changes

The Interdisciplinary Data Collection version of the NCEP OI SST varies in two points from the data set in the IGOSS Products Bulletin.

• Originally the data progressed from the South Pole northward. We reversed the data so that it now progresses from North to South. This was done to match the other data sets in our collection.
• In the original data set, land grids are filled by Cressman interpolation (Cressman, 1959). This was done to allow easier interpolation by those desiring to do so. Some inexperienced analysts might mistake these interpolated values over the continents for real temperatures. To prevent this we followed the ISLSCP (International Satellite Land Surface Climatology Project) convention and put a mask over the land areas.

Scientific Potential of Data
Njoku et al. (1985) provide a comprehensive description of the importance and applications of accurate knowledge of the sea surface temperature over both local and global scales. Some of the applications are

• quantification of the heat, moisture, and radiative fluxes between the ocean and atmosphere, which defines the surface energy balance at the air-sea interface Liu, 1988, Liu, 1990, Liu and Gautier, 1990)
• initialization and validation of mesoscale and large-scale general circulation models
• studies of the climatological warming or cooling of the global oceans (Reynolds et al. 1989)
• understanding of the dynamics of the ocean (e.g., ocean currents) as influenced by the gradients of sea surface temperature and forcing by atmospheric winds
• correlation of oceanic primary productivity (phytoplankton blooms) with SST, especially during transient events or periodic events such as El Nino
• understanding the importance of oceans as a sink for atmospheric carbon dioxide and how the relevant mechanisms are affected by SST (Moore and Bolin 1986).

Validation of Data
As described in Reynolds (1988), the sea surface temperature monthly mean files are subjected to objective quality controls. As reported there for the "blended" analysis, the global monthly average bias error is less than 0.1 degree C. The global monthly average RMS error is less than 0.8 degree C. However, errors at individual grid points could be larger. Reynolds and Smith (1994) show that the higher spatial resolution of the new OI analysis gives better regional detail than did the "blended" analysis.

Contacts

Points of Contact

For information about or assistance in using any DAAC data, contact

EOS Distributed Active Archive Center (DAAC)
Code 902
NASA Goddard Space Flight Center
Greenbelt, Maryland 20771
Internet: daacuso@daac.gsfc.nasa.gov
301-614-5224 (voice)
301-614-5268 (fax)

Technical inquiries about the NCEP OI SST data set should be addressed to:

Dr. Richard W. Reynolds
Internet: wd01rr@sgi11.wwb.noaa.gov
301-763-8000 ext 7580(voice)
301-763-8125(fax)

Diane C. Stokes
Internet: wd01dm@sgi26.wwb.noaa.gov
301-763-8000 ext 7581(voice)
301-763-8125(fax)

Climate Modeling Branch W/NP24
NCEP/NWS/NOAA
5200 Auth Road, Room 807
Camp Springs, MD 20746, USA

References

Cressman, G.P. 1959. An operational objective analysis system. Mon. Wea. Rev., 87:367-374.

Kidwell, K.B.,compiler and editor, July 1991: NOAA Polar Orbitor Data Users Guide (Tiros-N, NOAA-6,7,8,9,10,11 and -12), NOAA/NESDIS, National Climate Data Center, Washington, DC.

Kidwell, K.B., editor 1995: NOAA Polar Orbitor (POD) Data Users Guide NOAA/NESDIS, National Climate Data Center, Washington, DC: http://www2.ncdc.noaa.gov/POD/

Liu, W.T. 1988. Moisture and latent heat flux variabilities in the tropical Pacific derived from satellite data. J. Geophys. Res., 93:6749-6760.

Liu, W.T. 1990. Remote sensing of surface turbulent flux. In Surface Waves and Fluxes, Vol. II (G.L Geenaert and W.J. Planr, eds.), Kluwer Academic press, pp. 293-309.

Liu, W.T., and C. Gautier. 1990. Thermal forcing on the tropical Pacific from satellite data. J. Geophys. Res., 95:13209-13217.

McClain E.P., W.G. Pichel, and C.C. Walton. 1985. Comparative performance of AVHRR-based multichannel sea surface temperature. J. Geophys. Res., 90:11585-11601.

Moore, B., and B. Bolin. 1986. The oceans, carbon dioxide, and global climate change. Oceanus, 29 (4).

Njoku, E.G., T.P. Barnett, R.M. Laurs, and A.C. Vastano. 1985. Advances in satellite sea surface temperature measurements and oceanographic applications. J. Geophys. Res., 90:11573- 11586.

Reynolds, R.W. 1988. A real-time global sea surface temperature analysis. J. Climate, 1:75-86.

Reynolds, R. W., 1993: Impact of Mount Pinatubo aerosols on satellite-derived Sea Surface Temperatures. J. Climate, 6, 768-774.

Reynolds, R.W., C.K. Folland, and D.E. Parker. 1989. Biases in satellite-derived sea surface temperature data. Nature, 341:728-731.

Reynolds, R. W. and D. C. Marsico, 1993: An improved real-time global sea surface temperature analysis. J. Climate, 6, 114-119.

Reynolds, R. W. and T. M. Smith, 1994: Improved global sea surface temperature analyses. J. Climate, 7, 929-948.

Reynolds, R. W., and T. M. Smith, 1995: A high-resolution global sea surface temperature climatology, J. Climate, 8, 1571-1583.

Shea, D.J., K.E. Trenberth, and R.W. Reynolds. 1992. A global monthly sea surface temperature climatology, J. Climate, 5, 987-1001.

Slutz, R. J., S. J. Lubker, J. D. Hiscox, S. D. Woodruff, R. L. Jenne, D. H. Joseph, P. M. Steuer, J. D. Elms, 1985: Comprehensive Ocean-Atmosphere Data Set: Release 1. NOAA Environmental Research Laboratory, Boulder, CO, 268 pp.

Walton, C. C., 1988: Nonlinear Multichannel Algorithms for Estimating Sea Surface Temperature with AVHRR Satellite Data, J. Appl. Meteor., 27, 115-124.

Woodruff, S.D., S. J. Lubker, K. Wolter, S.J. Worley, and J.D. Elms, 1993: Comprehensive Ocean-Atmosphere Data Set (COADS) Release 1a: 1980-1992. Earth System Monitor, Vol. 4, No. 1, September 1993, NOAA.

NASA	Goddard	GDAAC	CIDC