the MRIR's resolution was low, it is sometinies difficult to 

 differentiate between low clouds and the sea's surface when their 

 radiational temperatures are similar. 



Therefore, in the examination of single -day HEIR data it is 

 necessary to use clear-sky data.* Such information is possible hy 

 assigning threshold temperature values so high as to guarantee the 

 exclusion of cloud data. Since some sea surface temperature 

 information may he lost hy this method, a second procedure of examining 

 single-day data may be preferred. This method utilizes a minimum 

 threshold grid made of historical minimum temperatures to exclude cloud 

 data, (a maximum ceiling made of maximum temperatures may also be used 

 to help exclude the more obvious of the electronic "noise"). Both of 

 these methods are difficult to use in analyzing high latitude data 

 where cloud and water temperature values are similar. 



A method formulated by LaViolette and Chabot (1969) to remove 

 transitory clouds artifically from the HEIR data involves a highly 

 selective composite of several days' data. An example of this 

 compositing is shown in Figure 8. 



The region shown in the figure — the Gulf of California and a 

 small portion of the eastern Pacific — has relatively simple sea 

 surface temperature structure with distinctive land masses easily 

 defined by infrared sensors. 



*A comparatively minor difficulty in the Nimbus 2 HRIR data was the 

 presence of systematic and erratic electronic "noise" generated on 

 board the spacecraft and by the data reduction system on the ground. 

 This has been partially surmounted by the use of filter programs 

 devised by NASA's Goddard Space Flight Center. 



13 



