Comparison and Optimization of AVHRR Sea Surface Temperature Algorithms

I. J. Barton CSIRO Division of Atmospheric Research, Aspendale, Victoria, Australia

Search for other papers by I. J. Barton in
Current site
Google Scholar
PubMed
Close
and
R. P. Cechet CSIRO Division of Atmospheric Research, Aspendale, Victoria, Australia

Search for other papers by R. P. Cechet in
Current site
Google Scholar
PubMed
Close
Restricted access

Abstract

Satellite measurements of sea surface temperature (SST) are regularly available from data supplied by the AVHRR instruments on the NOAA meteorological satellites. In cloudless areas SST is derived from the infrared data using a differential absorption technique to correct for the effect of the atmosphere. For the AVHRR data a multichannel (multiwavelength) approach is used and global operational algorithms are in use. During 1990 a new instrument that has been specifically designed to measure SST will be launched on the European satellite, ERS-1. The Along Track Scanning Radiometer (ATSR) will provide six infrared measurements for each pixel on the earth's surface. Using the same differential absorption techniques, a multitude of algorithms for providing SST will then be possible. In this note a technique is described that will enable the comparison and optimization of SST algorithms and will also aid in the selection of the most appropriate algorithm for ATSR data analysis.

To demonstrate the technique mosaic images were constructed from small areas of cloud-free infrared images of the sea surface as seen by the NOAA-9 AVHRR. Each area was approximately 55 km by 55 km and, by arranging them in order of decreasing mean temperature and increasing mean zenith angle, it was possible to use an image analysis system to compare the relative performance of different algorithms for deriving surface temperature. The images were also used to compare some NOAA-7 SST algorithms.

A second set of mosaic images was constructed using NOAA-10 AVHRR data collected on the same night and for the same surface location. Images of SST derived with theoretical NOAA-10 algorithms were compared with those from an operational NOAA-9 algorithm. Then a simple optimization technique was used to obtain a new algorithm for deriving SST from channels 3 and 4 of the NOAA-10 instrument. This optimization scheme, using an ordered mosaic image that covers a wide range of conditions (location, local zenith angle, or some other parameter), should be applicable to the comparison and optimization of other satellite data products.

Abstract

Satellite measurements of sea surface temperature (SST) are regularly available from data supplied by the AVHRR instruments on the NOAA meteorological satellites. In cloudless areas SST is derived from the infrared data using a differential absorption technique to correct for the effect of the atmosphere. For the AVHRR data a multichannel (multiwavelength) approach is used and global operational algorithms are in use. During 1990 a new instrument that has been specifically designed to measure SST will be launched on the European satellite, ERS-1. The Along Track Scanning Radiometer (ATSR) will provide six infrared measurements for each pixel on the earth's surface. Using the same differential absorption techniques, a multitude of algorithms for providing SST will then be possible. In this note a technique is described that will enable the comparison and optimization of SST algorithms and will also aid in the selection of the most appropriate algorithm for ATSR data analysis.

To demonstrate the technique mosaic images were constructed from small areas of cloud-free infrared images of the sea surface as seen by the NOAA-9 AVHRR. Each area was approximately 55 km by 55 km and, by arranging them in order of decreasing mean temperature and increasing mean zenith angle, it was possible to use an image analysis system to compare the relative performance of different algorithms for deriving surface temperature. The images were also used to compare some NOAA-7 SST algorithms.

A second set of mosaic images was constructed using NOAA-10 AVHRR data collected on the same night and for the same surface location. Images of SST derived with theoretical NOAA-10 algorithms were compared with those from an operational NOAA-9 algorithm. Then a simple optimization technique was used to obtain a new algorithm for deriving SST from channels 3 and 4 of the NOAA-10 instrument. This optimization scheme, using an ordered mosaic image that covers a wide range of conditions (location, local zenith angle, or some other parameter), should be applicable to the comparison and optimization of other satellite data products.

Save