Satellite-Derived Sea Surface Temperature Data Available From the NOAA/NASA Pathfinder Program

A new high-resolution, global satellite-derived sea surface temperature (SST) data set called "Pathfinder," from the Advanced Very High Resolution Radiometer (AVHRR) aboard the NOAA Polar Orbiters, is now available from the Jet Propulsion Laboratory Physical Oceanography Distributed Active Archive Center (JPL PO.DAAC). Suitable for research as well as education, the Pathfinder SST data set takes advantage of archived Earth science data. Where necessary, sensors have been intercalibrated, algorithms improved, and processing procedures revised, to produce long time series, global measurements of ocean, land, and atmospheric properties necessary for climate research.

Many Pathfinder data sets are available to researchers now, nearly a decade before the first launch of NASA's Earth Observing System (EOS). The lessons learned from the Pathfinder programs will facilitate the processing and management of terabytes of data from EOS. The Oceans component of Pathfinder reprocessed all Global Area Coverage (GAC) data acquired since 1981 by the 5-channel AVHRRs. The resultant data products are consistent, calibrated [Rao, 1993a, Rao, 1993b, Brown et al., 1993] Earth-gridded SST fields at a variety of spatial and temporal resolutions. The following homepages on the World Wide Web (WWW) provide detailed information on the NOAA/NASA Pathfinder Program as a whole, including how to access the data:

http://pegasus.nesdis.noaa.gov/pathfinder.html or
http://xtreme.gsfc.nasa.gov/hq/path_sites.html

Algorithms

Algorithm coefficients for Version 2 of the Pathfinder NLSST are also derived for three regimes but on a monthly, rather than yearly basis. The Version 2 Pathfinder NLSST implementation better allows for low-frequency temporal trends as well as rapid changes in atmospheric conditions [Evans et al., 1996]. The AVHRR Pathfinder Oceans Matchup Database is used to estimate Versions 1 and 2 Pathfinder algorithm coefficients and is a unique by-product of the Pathfinder SST data set development [Podesta et al., 1995]. This matchup database is a multiyear, multisatellite database of approximately cotemporal, colocated in situ SST and AVHRR measurements; moored and drifting buoys are included. The database records, called "matchups," can be used to develop and test algorithms for the estimation of sea surface temperature using AVHRR data and for validation exercises. A complete description of the AVHRR Pathfinder Oceans Matchup Database can be found on the WWW (http://www.rsmas.miami.edu/~gui/matchups.html).

Cloud detection techniques applied by NOAA/NESDIS [McClain et al., 1985] were adopted for the Pathfinder SST retrievals. These techniques include visible and visible/near-infrared bidirectional thresholds, spatial coherence thresholds using visible and infrared data, and multichannel infrared intercomparison tests. Pathfinder SST cloud detection tests are two-tiered, however, and in addition to the suite of satellite tests described above, comparisons of the Pathfinder SST retrievals to a reference SST field are also performed [Reynolds and Smith, 1994].

We stress that multichannel algorithms for estimating SST from the AVHRR, such as the MCSST, NLSST and Pathfinder NLSST, do not take into account the difference between the ocean's skin temperature (what the AVHRR actually sees) and underlying mixed layer SST, which is referred to as the "bulk" temperature. The "bulk" temperature is the quantity measured by most buoys. The uppermost millimeter of the ocean, or skin, can be as much as 0.7°C cooler than the water just below [Ewing and McAlister, 1960] due to evaporative or radiative cooling. Conditioning the satellite-derived SST estimates using buoys, as is done in the Pathfinder SST algorithms, results in a measurement that more closely resembles a "bulk" SST estimate, not a skin estimate [Schluessel et al., 1990]. Hence, biases in satellite-derived SST estimates that arise because of this difference between the ocean's skin and bulk temperatures are being closely scrutinized in the Pathfinder SST data sets.

Validation

Fig. 1.The top image is the average weekly data density, shown as a percent, for the Pathfinder SST 18-km gridded data (daytime) for 1987. The bottom panel is the same quantity for the NLSST data set for the same time period and resolution.

Figure 2 shows how the Pathfinder SST data set compares globally with the Miami MCSST and the optimum interpolation SST (OISST) analysis produced by NMC [Reynolds and Smith, 1994], for daytime (2a) and nighttime (2b) satellite retrievals from 1987. Mean temperature as a function of latitude is plotted, and it is clear that the Pathfinder SST data set is slightly cooler at all latitudes than either the Miami MCSST data or the OISST data. This cool bias is more pronounced at higher latitudes for both the daytime and the nighttime SST retrievals, and is most likely the result of cloud contamination in the Pathfinder SST data set.

Fig. 2. Temperature as a function of latitude, for three different SST data sets, for daytime (a) and nighttime (b) retrievals. The solid line is the average weekly temperature for the Miami MCSST data set; the dotted line represents the Pathfinder SST data set and the dashed line, an optimum interpolation SST data (OISST) produced at the National Meteorological Center.

The Pathfinder Oceans Matchup Database described earlier provides a convenient opportunity for initial validation of the gridded SST fields. The root mean square difference (RMSD) between the Pathfinder SST 9-km gridded fields and the global set of moored and drifting buoys compiled in the Pathfinder Matchup Data Base for 1987 is 0.96°C. Considering daytime and nighttime satellite retrievals separately, the RMSD is 0.94°C between the daytime gridded SST and the buoy SST, and 0.97°C for the nighttime matchups. A WWW site for validation of the Pathfinder SST products is available at http://www.ccpo.odu.edu/~lizsmith.

Products and Availability

Fig. 3. Example of daily, 54-km Pathfinder SST products. This is the "best" SST product, which includes only the best quality pixels.

Fig. 4. This image is an example of a daily, 54-km Pathfinder SST field in which no cloud mask has been applied.

One of the primary goals of the NOAA/NASA Pathfinder Program is to provide easy access to large, satellite-derived data sets. Pathfinder AVHRR SST data products are available electronically via either anonymous FTP or the WWW from the JPL PO.DAAC. Users may easily extract regional subsets from the daily, global SST files via a WWW interface. The URL for the JPL PO.DAAC homepage is http://podaac-www.jpl.nasa.gov/. In addition to the data subsetting and extraction capability, users may also look at global "browse" images to quickly make a determination about cloudiness for a specific region and time period. Via anonymous FTP, users may obtain large volumes of data over the Internet. For information about obtaining Pathfinder SST data via the EOS Information Management System, FTP, or on magnetic tape, please contact the JPL PO.DAAC at podaac@podaac.jpl.nasa.gov.

The pathfinder data set is the result of a collaboration between the National Oceanographic and Atmospheric Administration (NOAA), NASA, and investigators at several universities. NOAA and NASA are the sponsors of the Pathfinder Program.

References

Walton, C. C., E. P. McClain, and J. F. Sapper, Recent changes in satellite- based multichannel sea surface temperature algorithms, Rep. MTS 90, Mar. Technol. Soc., Washington, D.C., September 1990.

EES Contents