Computation and storage of all these descriptive terms on the SERC 

 deck were not necessary. However, it was envisioned that many statisti- 

 cal analyses using the SERC deck woiJ.d be carried out on a computer with 

 relatively slow input and output rates and a small memory. Thus consider- 

 able computer time and memory woxild be saved if the descriptive terms were 

 stored on the punchcards. 



Harvey (l) states that temperatures recorded at standard depths on 

 the National Oceanographic Data Center geographical BT file do not repro- 

 duce a BT trace completely. As the SERC deck was originally intended for 

 special research projects requiring a high degree of accuracy, it contains 

 a finer gradation of the BT data. Although small depth- temperature inter- 

 vals reproduce a BT trace more accurately, decrease of the intervals con- 

 sumes more space on a punchcard. Intervals of 20 feet between the surface 

 and 360 feet and intervals of 40 feet at depths below 36O feet were used as 

 a compromise between the two extremes. Computation errors introduced by 

 this choice of intervals as opposed to either a 5- or a 10-foot interval 

 are a subject for a separate report, 



ANALYSIS OP STATION ECHO DATA 



Aneilysis of the Historical Data 



In order to obtain a quant it ive description of temporal and spatial 

 variations in the thermal structure parameters, the ECHO historical deck 

 was sorted into 1-degree quadrangles by months and years. The following 

 parameters were chosen to adequately describe the thermal structure: 



(1) 



surface temperature 





(2) 



layer depth 





(3) 



maximum gradient 



(a) average gradient 







(b) temperature of the 



upper bound 





(c) depth of the upper 



bound 





(d) thickness 





ih) 



mean gradient 



(a) average gradient 





(5) 



first gradient 



(a) average gradient 







(b) depth of the upper 



bound 





(c) thickness 





A Royal-McBee LGP-30 computer was programmed to compute the mean, 

 standard error of the mean, variance, and standard deviation. These 

 statistics were computed for the above-mentioned parameters for each 

 1-degree quadrangle by months for all years. Computations were not made 

 for groups of less than 10 BT observations. Locations of observations 

 were assumed to be accurate. Observations taken at 350N,ii-8°W are arbi- 

 trarily assigned to the 1-degree quadrangle to the northwest. 



