The first problem was to determine hov; closely the computed 

 values for incoming radiation represented actual conditions dur- 

 ing the forecast period. To do this pyrheliometer records were 

 used. For each day during the daylight hours the average rate 

 of incoming radiation was computed from the curve of pyrhelio- 

 meter record number 3j by using finite differences at l5-minute 

 intervals,. These observed values are given in the third column 

 of table IV, A similar process was carried out with curves of 

 pyrheliometer record numbers 1 and 2, Adding the two gave the 

 average daily rate of reflected radiation. These values for 

 Q^q£. are shown in the fourth coliaran of table IV, The similari- 

 ties between the computed and observed rates of incoming radi- 

 ation, as shown in the first and third colii^s of table IV, are 

 remarkable 



The next step in the computation was to get the total 

 incoming and back radiation for the eight-day period. Adding 

 the values for Q^^^ in 'the first column of table IV and multi- 

 plying by 13 hours or 780 minutes give a total incoming rad- 

 iation of 1^607 g. cal. cmr2. Adding the values of % in the 

 second column of table IV and multiplying by 2k hours or ikkO 

 minutes give the total of 1,26? g. cal. cm."^. Subtracting Q^ 

 from Qab gives a total of 3,339 g. cal. cml^ absorbed by the 

 water column during the period. 



It is now necessary to convert this amount of absorbed 

 heat to temperature changes vrith depth. A representative 

 transparency station, occupied on 23 April, v:as chosen. These 

 data were taken by the use of colored filters, thus giving a 

 separate curve for each spectral band. Figure 5 presents the 



24 J 



