Journal of Geophysical Research: Oceans

Processing and analysis of large volumes of satellite-derived thermal infrared data


  • Peter Cornillon,

  • Craig Gilman,

  • Lothar Stramma,

  • Otis Brown,

  • Robert Evans,

  • James Brown


Reducing the large volume of TIROS-N series advanced very high resolution radiometer-derived data to a practical size for application to regional physcial oceanographic studies is a formidable task. Such data exist on a global basis for January 1979 to the present at approximately 4-km resolution (global area coverage data, ≈2 passes per day) and in selected areas at high resolution (local area coverage and high-resolution picture transmission data, at ≈1-km resolution) for the same period. An approach that has been successful for a number of studies off the east coast of the United States divided the processing into two procedures: preprocessing and data reduction. The preprocessing procedure can reduce the data volume per satellite pass by over 98% for full-resolution data or by ≈84% for the lower-resolution data while the number of passes remains unchanged. The output of the preprocessing procedure for the examples presented is a set of sea surface temperature (SST) fields of 512 × 1024 pixels covering a region of approximately 2000 × 4000 km. In the data reduction procedure the number of SST fields (beginning with one per satellite pass) is generally reduced to a number manageable from the analyst's perspective (of the order of one SST field per day). This is done in most of the applications presented by compositing the data into 1- or 2-day groups. The phenomena readily addressed by such procedures are the mean position of the Gulf Stream, the envelope of Gulf Stream meandering, cold core Gulf Stream ring trajectories, statistics on diurnal warming, and the region and period of 18°C water formation. The flexibility of this approach to regional oceanographic problems will certainly extend the list of applications quickly.