INTRODUCTION 



This study is the second in a series concerned with the 

 analysis of sea-surface temperature observations. Tlie first 

 study" dealt with the effect of missing data in long time-series 

 sea-surface temperature measurements on certain regression and 

 autocorrelation analyses. 



For many decades observations of sea-surface temperature 

 have been taken and recorded by merchant and naval vessels. 

 Subsequently these observations have been catalogued and archived 

 by many agencies. In the United States these agencies include the 

 U, S. Naval Hydrographic Office (now the U. S. Naval Oceano- 

 graphic Office), the National Weather Records Center, and the 

 National Oceanographic Data Center. 



As the volume of data accumulated, it became the basis for 

 many generalized summaries. '''- All these summaries use 

 arbitrary temporal and spatial averaging. Kriimmel averaged the 

 data over all years and all months. His areas were 5 degrees of 

 latitude in the north-to-south dimension and extended east to west 

 across an entire ocean. Bohnecke, the U, S. Weather Bureau, and 

 the U. S. Naval Hydrographic Office used areas of 1-, 2-, or 

 5 -degree squares of latitude and longitude and averaged all years 

 together by month or season. 



In recent years a requirement by researchers in the fields 

 of fisheries oceanography, military oceanography, and meteor- 

 ological oceanography for more detailed descriptions of sea- 

 surface temperature distributions has developed. In response to 

 this requirement the Bureau of Commercial Fisheries and the 

 American Geographical Society have begun the preparation of 

 more detailed charts. The Bureau of Commercial Fisheries is 

 preparing detailed month-by-month charts of sea-surface 



11 



