radiometric determination of the Gulf Stream boundary. The ship 

 data were shifted 1/2° westward, and the aircraft data were shifted 

 1° westward with respect to the HRIR coordinates to obtain the best 

 fit of the data. The marked agreement between the ship data and 

 the HRIR data demonstrate the persistence of this major Gulf Stream 

 pattern over a period of several days. 



Table 1 presents the results of the comparison of the ship and 

 aircraft data with the NIMBUS II radiation temperature data. Surface 

 temperature data from 64 BT casts, taken between 15 and 18 June 1966, 

 were compared with the corresponding satellite grid-point data. 

 Comparison of the average temperature values from the two sets of 

 data showed that the satellite radiation temperatures were biased 

 3.6°C higher than the ship surface temperature data. Upon removal 

 of the 3.6°C bias from the satellite data, there remained a mean 

 absolute error of 1.5°C between the ship and the aircraft data. 



Comparison of the satellite and the aircraft data showed the 

 satellite data to be biased 2.4°C higher than the aircraft data. 

 Removal of the bias from the satellite data results in a mean 

 absolute error of 2.0°C between the satellite and aircraft data. 

 One hundred and thirty-four one-minute (3nm) averages of the ART 

 record were compared with the corresponding grid-point data values. 



The primary sources of bias and mean absolute errors between 

 the satellite radiation temperatures and the conventional sea 

 surface temperatures are instrumental noise, atmospheric absorptions, 

 computer averaging of the resolution elements into the grid-point 

 values, extraneoiis reflections of solar radiation into the radiometer 

 windows, and field contamination by subresolution element clouds. 

 The instrumental system errors in the satellite radiation tempera- 

 tures have been discussed by Wamecke, McMillin, and Allison (1968). 

 The problem of removal of subresolution element clouds is being 

 addressed by Meteorology International, Inc., in cooperation with 

 FNWC, Monterey, under contract from Project FAMOS (Fleet Applications 

 for Meteorological Observations from Satellite) . 



SST FIELD FROM SATELLITE DATA 



Table 2 presents the summary data comparing the corrected 

 satellite temperature field measurements with the aircraft and 

 ship ground- truth data. It is seen that, within the clear-sky 

 portion of the NIMBUS scan, the satellite data provide an adequate 

 characterization of the temperature structure of the Gulf Stream 

 boundary and of the surrounding water masses. Ihe temperature 

 gradient across the boundary, as determined from the satellite 

 data, is approximately 5°C/11 kilometers between 71°W and 75°W. 



368 



