As far as historical marine surface observations are 

 concerned, little can be done about the subjective nature of 

 the cloud observations. However, uncertainty in the 

 radiation estimate due to the second cause has been 

 reduced by the inclusion of Laevastu's (1960) and Tabata's 

 (1964b) cloud correction formulae in Equation (2). Tabata's 

 formula is based on extensive OWS-P observations. 

 Laevastu's formula is based on less extensive observations 

 made on the U.S. S. Rehoboth. Ta.ha.tsL states that when 

 monthly mean cloudiness is used, about 70% of the 

 estimated values fall within 5% of the observed values. 

 Laevastu estimates, when leaving out days with a 

 cloudiness of more than eight tenths of sky covered, that 

 his radiation values are within 5% of measured values 

 during about 42% of the days and within 10% of the 

 measured values during 51% of the days when measure- 

 ments were made. We estimate that our radiation values in 

 Appendix II are better than the underestimates reported 

 by Quinn and Burt (1968) and possibly lie within 10% of the 

 true values. 



Next in importance in the net heat exchange across the 

 sea surface, Q(N), is the heat used for evaporation, Q(E). 

 We have discussed the uncertainties in the drag coefficient 

 under neutral conditions. Values of the neutral drag 

 coefficient referred to the 10-m level in recent field 

 experiments range from 0.0010 to 0.0016. Variations in 

 stability and wave spectra, and the assumption that the 

 transfer coefficients of heat, water vapor, and momentum 

 are equal, increase the uncertainty in the magnitudes of the 

 derived turbulent exchange processes. 



Verification of the derived evaporation rate and 

 determination of its accuracy cannot be made at this time 

 because direct measurements have not been possible. 

 However, gross water vapor budget estimates such as 

 those by Riehl et al. (1951) and measurement of vertical 

 eddy fluxes during BOMEX (Holland 1972) indicate that the 

 derived evaporation is of the correct order of magnitude. 



Third in importance is the effective back radiation, Q(B). 

 Budyko (1974) states that formulae for Q(B) have been 

 checked by many observations obtained during the 

 International Geophysical Year at actinometric stations 

 in the USSR. He states that Berliand's formula (our 

 Equation (3)) is well corroborated for observations made at 

 average and high humidities. However, verifications at sea 

 are few. Measurements of Q(B) during the Trade Wind 

 Zone Oceanography investigation reported by CharneU 

 (1967) ranged from 58 to 173 cal cm'^ day"'. The mean 

 monthly Q(B) computed by Seckel (1970) for the months 

 and area of those observations fell within the above range. 

 Charnell's (1967) observations indicate that the upward 

 long-wave radiation followed the Stefan-Boltzmann law 

 with an average emissivity of 0.99 and with values ranging 

 from 0.96 to 1.1. The downward sky radiation, dependent 

 on the water vapor content of the atmosphere as well as 

 the type, amount, density, and height of clouds, is more 

 difficult to verify without extensive observations. For 

 example, 10 24-h observations made off the Oregon coast 

 (Reed and Halpern 1975) gave average Q(B) values only 

 50% of that calculated with Equation (3). 



The primary cause for the differences between the 

 observed and calculated values is the cloud correction 

 factor. The coefficient in the cloud factor was determined 

 for the average type and height of cloudiness occurring in a 

 given latitude band (presumably over the USSR). The 

 example g^ven above illustrates that empirical formulae 



derived for average conditions do not necessarily hold for a 

 short duration such as 10 days or a month or for a specific 

 location within the latitude band. 



Although the accuracy of the Q(B) calculated for OWS-V 

 cannot be given, interseason and interannual comparisons 

 of Q(N) are not expected to be significantly affected. The 

 average Q(B) calculated for OWS-V (Appendix II) shows an 

 annual range of 39 cal cm"^ day '' compared to ranges of 288 

 and 595 cal cm"^ day'^ for the calculated Q(E) and Q(N), 

 respectively. 



The conduction of sensible heat, Q(C), is subject to the 

 same limitations as the Q(E) but is of relatively small 

 magnitude. Errors in Q(N) due to uncertainties in Q(C) are 

 expected to be smaller than those contributed by the other 

 heat exchange processes. 



Again, the wind stress on the sea surface is subject to 

 the same limitations as the turbulent transfers of water 

 vapor and sensible heat. Thus, we are unable to determine 

 the accuracy of any of the turbulent transfer processes. 



Q(N) is the difference of large numbers. The relative 

 error in Q(N) is therefore potentially much larger than that 

 for the individual exchange processes. For example, if 

 Q(S) is in error by 10% during July when Q(S) averages 

 473 cal cm"^ day "', then Q(N), with an average value of 278 

 cal cm"^ day', will be in error by about 17%. 



The values of the exchange processes listed in Appendix 

 II must therefore be regarded as indices whose absolute 

 magnitude is in doubt. Nevertheless, these indices are 

 useful in climatic scale applications when interseason and 

 interannual comparisons are to be made. 



DISCUSSION 



In this section we wiU take the results of Appendix II at 

 face value and draw attention to the air-sea interaction 

 processes that are of climatic significance at OWS-V and in 

 the net annual heat loss area of the north Pacific Ocean. 



First, consider the relative magnitudes of the heat 

 exchange processes at OWS-V in terms of their modification 

 of Q(S), using the 1956-70 average values (Fig. 3). The figure 

 shows that Q(E) is the most important process by which heat 

 is lost from the sea surface. Of the heat lost annually, Q(E) 

 contributes 63%, Q(B) 26%, and Q(C) 11%. 



Figure 3 also shows that the annual cycle is divided into 

 a warming portion lasting from April through September 

 and a cooling portion lasting from October through March. 

 There is a net annual heat loss of 32 cal cm"^ day"' at 

 OWS-V which agrees with Wyrtki's (1965. fig. 1) chart 

 value. 



Monthly values of Q(N) and Q(E) 



Monthly values for Q(N) and Q(E) and their anomalies 

 are shown in Figures 4 to 6. Values prior to April 1955 

 were not corrected to reflect the change in location of 

 OWS-V. Anomalies are calculated from the April 1955 to 

 March 1971 monthly mean values of Q(N) and Q(E). Note 

 that, particularly during the heat loss portion of the annual 

 cycle, the pattern of the Q(N) and Q(E) curves are similar. 

 This similarity is pronounced during the fall 1967 to winter 

 1968. The high net heat loss in November 1967 followed by 

 low heat loss in December 1967 and then high heat loss in 

 February 1968 was primarily caused by the heat used for 

 evaporation. Similarities in the Q(N) and Q(E) anomaly 

 patterns are also apparent. 



