The linear cloud factor in Equation (3) is appropriate for low strato- 

 cumulus clouds. Reed (1976) proposed an alternate linear correction 

 (1-0.7C) which is more suitable for the higher clouds typical of the 

 tropics. However, we could not discern any significant difference in 

 seasonal cloud cover, by cloud type, which would allow us to use the 

 alternate cloud factor in the lower latitudes of the study region. 



The bulk aerodynamic formulae for turbulent fluxes of latent and 

 sensible heat across the air-sea interface in a neutrally stable atmo- 

 spheric boundary layer (Kraus 1972) can be expressed in the forms: 



and 



Q E = p a LC E (q -q m )U w 



Qc=P lJ c l ,C^T,-TJU t0 . 



(4) 



(5) 



In these equations, p„ is the density of air, L is the latent heat of 

 vaporization, c p is the specific heat of air, C E and C H are experimen- 

 tally determined exchange coefficients for water vapor and sensible 

 heat, q and q w are the specific humidities of the air in contact with 

 the sea surface and at 10 m or deck level, T s - T a is the sea-air tem- 

 perature difference, and U l0 is the windspeed at 10 m or ship ane- 

 mometer height. The equations are based on the assumption of a 

 "constant flux" layer in the first few tens of meters above the sea 

 surface, which allows the use of routine ship observations rather 

 than measurements at the wave-perturbed, air-sea interface. 



The specific humidities required in Equation (4) are not available 

 in surface marine weather reports, but can be approximated from 

 the relationship between specific humidity, q. and vapor pressure. 

 e. expressed in Equation (6): 



<? = e 



(6) 



where € is a known constant ratio of the molecular weight of moist 

 air to dry air, equal to a value of 0.622. We assumed constant values 

 for density p„=1.22 kg/m'. latent heat of vaporization, 

 L = 2.45x 10" J/kg, and specific heat of air, c p = l.OOx 10 3 J/kg per 

 °C, and expressed the empirical exchange coefficients C E and C H , 

 referred to the 10 m level, as constants equal to 0.0013 (Kraus 

 1972; Coantic 1974"'; Anderson and Smith 1981). Substitution of 

 these values and Equation (6), with P= 1.013.25 mbar, into Equa- 

 tion (4) yields the following formulae: 



Q E = 2.38(0.98e„-e a )U ll 



e f =i.59(r,-rj{/,„. 



(7) 



(8) 



The saturation vapor pressure over pure water at the sea surface 

 temperature. e„(mbar). was computed from a formula given by 

 Murray (1967) and multiplied by 0.98 to account for the effect of 

 salinity (Miyake 1952). 



Long-term monthly mean heat exchange components were com- 

 puted from all available reports between 1921 and 1972 within each 

 1 ° square area. The appropriate average is defined in Equation (9): 



1 



{Qs.Qb.Qe,QcQx) = -W E (Qs.Qb-Qe,QcQs), (9) 



1=1 



where n is the total number of reports within a particular 1 ° square 



-KToantic. M. 1974. Formules empiriques d'evaporaiion. Note de ta Con- 

 vention C.N EX. O. 1 M.S. T. No. 74/951. 24 p.. Xerox. 



area and month. The values (Q s , Q B , Q E , Q c , Q N ), were evaluated 

 according to Equations (2), (3), (7), (8), and (1), respectively. Each 

 individual report was weighted equally. Therefore, the mean values 

 of the heat exchange components for each long-term month and 1° 

 square are formed from a data set which is independent of all other 

 months and squares. 



Monthly mean heat exchange fields are presented in Appendix I 

 and include incident solar radiation corrected for cloud cover and 

 reflection, Q s (Charts 1 to 12); effective back radiation, Q B (Charts 

 13 to 24); latent heat flux, Q E (Charts 25 to 36); sensible heat flux, 

 Q c (Charts 37 to 48); and net heat exchange across the air-sea inter- 

 face, Q N (Charts 49 to 60). No attempt has been made to smooth the 

 fields, either by removing data which do not appear to fit the distri- 

 butions or by applying objective smoothing procedures. The mean 

 heat exchange fields were contoured by computer, and "bull's- 

 eyes" in the contours, even where they may reflect erroneous data, 

 were left in the charts as indications of the general degree of con- 

 sistency in the composite distributions. 



Error Analysis 



Errors associated with the computations of large-scale air-sea 

 interaction transfers described in this report reflect 1) nonconform- 

 ities in the spatial and temporal distributions of the surface marine 

 observations, 2) inadequacies in data quality, and 3) uncertainties 

 in the empirical heat exchange formulae. Random and systematic 

 errors in the marine surface data primarily result from improper 

 instrument calibration and measurement techniques, and inaccura- 

 cies introduced by data reduction at sea or in the process of archiv- 

 ing different data sets in the common TDF-11 format. These 

 sources of error may introduce large variance in the monthly distri- 

 butions of surface heat flux which might otherwise be interpreted as 

 actual seasonal or nonseasonal variability related to geophysical 

 processes. 



The spatial distribution of the TDF-11 reports is known to be 

 biased to coastwise and transoceanic shipping lanes between major 

 seaports. Figure 2 shows the distribution of total numbers of obser- 

 vations per 1° square area used in this study for the calculation of 

 heat exchange processes. The actual numbers of observations per 

 month at each grid point are tabulated in Appendix II and shown as 

 auxiliary data in Figures 3 and 4 for the months of July and Decem- 

 ber, respectively. The highest density of reports exists in a zone 

 along the coast approximately 300 km wide. The largest numbers 

 of reports are in the area of the Southern California Bight. The 

 numbers of observations per 1 ° square per month ranged from 7 off 

 the coast of Washington in December to more than 2,400 off Los 

 Angeles in March. August, and October. The influence of mer- 

 chant vessel traffic between San Francisco and Hawaii is evident in 

 a zone approximately 200 km wide extending offshore, with more 

 than 200 observations/mo per 1° square. Approximately 29% of 

 the 1 ° squares contained fewer than 50 observations/mo, primarily 

 reflecting the sparse distribution of weather reports between lat. 

 20°N and 30°N at distances > 400 km from the coast of Baja Cali- 

 fornia. Nearly 40% of the long-term composite means were based 

 on between 100 and 500 observations/mo. More than 500 

 observations/mo were available in 5% of the 1° squares. 



Nonuniform distributions in time introduce additional bias in the 

 sample statistics for the composite monthly mean heat exchange 

 processes. Although all acceptable surface marine weather reports 

 between 1921 and 1972 were used in this study, approximately 70 

 to 80% of the observations were collected after 1950. Therefore, 

 the long-term mean values more properly represent estimates for 



