James 



TABLE III HIMBER OF DAILY OBSERVATIOHS PER FIVE DEGREE SQUARE 



Complex Area Smooth Area 

 Chart Reliability H5 90 S%~ 65 75 85 95^ 



Data !iype 



Present Ship 8 I6 40 12 72 * * 



Injection Data 



Random NSRT Reports 7 12 35 5 6 8 20 



Plus Some AEOT, Buoy- 

 Mostly ART, Buoy Data 7 8 30 5 5 6 8 



Pltis Some KSRT 



*Not attainable with present type data 



The restilts of these tests are considered valid on the basis 

 of the niunber of tests conducted including computer and manual, tests 

 for randomness, for specified data distribution and for error func- 

 tions. One proced\jre that was not treated in these tests was the use 

 of continuity analyses. These charts obviously help in the prepar- 

 ation of a sjTioptic chart when correct. If wrong they simple mislead 

 the analyst. On the ass\imption that their popular use by all anal- 

 ysis groups shows a degree of corrections it is concluded that all 

 scores would be improved slightly by their use in the tests. The 

 slopes of the curves, however, would not be materially altered. 



The najor conclusions are as follows: 



(1) Planning for synoptic network should emphasize high 

 concentrations of data with average accuracies to +1.0*^ for corxplex 

 areas. In smooth areas the emphasis should be oxi a few highly ac- 



^ 



curate (less than +1T') fixed observations. 



(2) Data should be increased about four times in com^ 

 plex areas and improved in accuracy. For smooth areas the quantity 

 of data need only be improved. 



(3) ART surveys should be confined generally to the 

 highly coa^lex areas with an occasional flight for continuity in the 

 snraoth areas. Buoy arrays should be utilized in smooth areas where 

 their high accuracy and fixed position contribute greatly to anal- 

 yses reliabilitj'-. 



150 



