(1958) and Johnson et al. (1965) used 4.7 for 

 wind measurements at 20 m. height. The coef- 

 ficient of Laevastu (1960) varies with wind speed 

 but at 10 m. sec."' the evaporation would be 

 about the same as that computed by Wyrtki. Be- 

 low this wind speed, values would be higher and 

 above this speed they would be lower than Wyrt- 

 ki's values. For a wind speed of 8 m. sec."' 

 measured at 10 m. above sea level, Roden's 



(1959) coefficient is also 6. 



In this paper the formula suggested by Malkus 

 has been used. In the trade wind zone it Is rea- 

 sonable to assume neutral stability at the usual 

 height of marine wind measurements. The un- 

 certainties in this height are therefore not crit- 

 ical, and the drag coefficient for neutral stabil- 

 ity (Malkus, 1962: fig. 6, p. 110) can be used. 

 Inspection of the wind summaries in table A and 

 the smoothed values in table B show that the spa- 

 tial and temporal variations in the wind speed 

 generally lie in the range of 4 to 12 m. sec.~' 

 Within this range the nondimensional drag coef- 

 ficient changes rapidly. In the calculations of 

 this paper the variable nature of the coefficient 

 has therefore been taken into account. 



The equation 



C - 

 D 



arctan (W - 8) 

 1.96 



+ 1.6 



10 



-3 



approximates the curve for the drag coefficient, 

 CjD, as a function of wind speed, W (inm. sec."'), 

 under conditions of neutral stability given by 

 Malkus, and facilitates the computation of the 

 evaporation. 



In the evaporation equation of this paper the 

 saturation vapor pressure over pure water has 

 been multiplied by 0.98 to account for the salt 

 effect (Miyake, 1952). Also the computer pro- 

 gram was written so that for negative sea-air 

 vapor pressure differences, the computed heat 

 of evaporation would be 0, This part of the com- 

 puter program does not affect the results of ta- 

 ble B since monthly mean values of the sea-air 

 vapor pressure difference are always positive 

 (see table A). The program, however, affects 

 results based on evaporation rates computed for 

 each set of meteorological observations. 



Since fog is rarely reported except, possibly, 

 in the northern region of the area considered in 

 this paper (McDonald, 1938), minimum sea-air 

 vapor pressure differences that are negative in 

 table A are probably due to erroneous observa- 

 tions. The computer program, therefore, re- 

 duces the error if evaporation rates of individ- 



ual sets of observations are summed. Where a 

 negative sea-air temperature difference may 

 actually have occurred, the computer program 

 reflects the belief that the heat of condensation 

 gained by the ocean is small (Roden, 1959). 



Q(C), Conduction of Sensible Heat 



The exchange of sensible heat across the sea 

 surface is the smallest term in the net heat ex- 

 change across the sea surface. It is generally 

 computed from the ratio of Q(C) to Q(E), the 

 Bowen ratio, and depends upon the temperature 

 difference between water and air and the wind 

 speed. Wyrtki (1966) used the form suggested 

 by Malkus (1962) with a conduction factor of 

 3.86. Johnson et al. (1965) used a factor of 3 

 and Roden (1959), 3.96. Laevastu's (1965) fac- 

 tor again varies with the wind speed but at 10 m. 

 sec."' it is essentially the same as that used by 

 Wyrtki or Roden. When the sea-air tempera- 

 ture difference is negative, Laevastu used the 

 same form as other workers with a factor of 3. 



Here a factor of 4 is used and no change is 

 made when the sea-air temperature difference 

 becomes negative. 



EVALUATION OF RESULTS 



Uncertainties in the estimation of large-scale, 

 sea-air interaction processes calculated in this 

 paper are due to (1) inadequacies in the distri- 

 bution and quality of the data, (2) the methods of 

 processing which must be used because of inad- 

 equate spatial and temporal distribution of ob- 

 servations, and (3) uncertainties in the empiri- 

 cal formulas. The empirical formulas which 

 are the subject of intensive research, and their 

 uncertainties have been thoroughly treated in 

 the literature cited here, and elsewhere. The 

 evaluation is, therefore, concerned with the data 

 inadequacies, effects of these inadequacies on 

 the heat exchange results, and the effects on the 

 results of the processing methods which must 

 be used because of data inadequacies. Finally, 

 comparisons with other results in the North Pa- 

 cific and interseason and interyear compari- 

 sons are made. 



Inadequacies in the Distribution 

 and Quality of Data 



Table A shows that south of lat. 15° N. the 

 number of observations per month is small and 

 that some 5° squares have no observations. For 

 these latitudes the results presented in table B 



