which were converted to equivalent 10 m windspeeds in knots and 

 meters per second according to the international scale of 1946 (see 

 List 1949: 1 19). The standard deviation of the overall error of an 

 estimated windspeed amounts to 0.58/, where / is the Beaufort 

 interval (Verploegh 1967), and corresponds to +0.7 to ±2.4 m/s 

 for equivalent windspeeds of 0.9 m/s to 30.5 m/s. The remaining 

 53% of the observations were estimated windspeeds which did not 

 agree numerically with the Beaufort wind force scale but typically 

 were originally reported in 5-kn intervals for windspeeds >20 kn. 

 Therefore, the possible error associated with these estimates is 

 approximately ± 1.3 m/s. An error of 1 m/s in estimating wind- 

 speed results in equivalent changes in the rates of evaporation and 

 sensible heat transfer of 10 to 20% for the mean windspeeds of 5 to 



10 m/s most commonly observed in the California Current region. 

 In using the historical surface marine weather reports, we have 



assumed that the wind observations were taken at a standard height 

 of 10 m, or that the Beaufort force estimates were equivalent to 

 windspeeds measured at this height. For surface winds which have 

 been measured by shipboard anemometers, this assumption can be 

 grossly inaccurate, since known anemometer heights vary from 7 to 

 37 m. the mean height being 21 m (Cardone 1969). If anemometer 

 heights were known for the vessel reports comprising the 12% of 

 the total observations which were measured as opposed to esti- 

 mated quantities, a neutrally stable logarithmic wind profile could 

 have been used to correct each observation from the measurement 

 height to the 10 m reference level. Except for the relatively few 

 anemometer heights from military and research vessels listed in 

 Cardone s report, the anemometer heights for reports in the TDF- 



1 1 file cannot be readily determined. By assuming that windspeeds 

 have been measured at 10 m. we possibly overestimate the turbu- 

 lent fluxes by 12% for winds actually measured at 37 m and under- 

 estimate the fluxes by 3% for measurements taken at 7 m. 



Conclusive verification of the derived rates of sensible and latent 

 heat transfer and the associated accuracies of the bulk formulae has not 

 been possible owing to the lack of extensive direct, over-water mea- 

 surements of the temperature and moisture fluxes. On the basis of 30 

 eddy flux measurements of water vapor, Friehe and Schmitt (1976) 

 concluded that latent heat flux is adequately described by the bulk for- 

 mula (Equation (4)) with an uncertainty of 15 W/m 2 (standard devia- 

 tion). The more numerous determinations of sensible heat flux also 

 conform to the empirical equation with an uncertainty of approxi- 

 mately 3.8 to 8.4 W/m : (standard deviation). However, the direct mea- 

 surements of sensible heat flux show a dependence on atmospheric 

 stability and indicate a small positive flux of approximately 0.15 W/ 

 m 2 , even when the value predicted from the term, (T r T a )U m , is zero. 

 Although Laevastu "s (1960) equation for the turbulent transfer of sensi- 

 ble heat predicts a nonzero flux for U m = m/s (but not for (2^- 

 r„)=0°C), the expected values from the more commonly used 

 formula (Equation (5)) may include a 1 to 2% bias in addition to ran- 

 dom measurement errors. The results of Anderson and Smith (1981) 

 generally confirm the bulk formulae with uncertainties of ± 10 to 

 ±25%, and demonstrate the applicability of Equation (4) to negative 

 fluxes of water vapor (i.e., condensation). 



Equations (7) and (8) are generally used to parameterize the loss 

 of heat from ocean to atmosphere due to evaporative and convec- 

 tive processes. The reverse processes of condensation and sensible 

 heat transfer to the sea are also predicted by the formulae. Negative 

 values of (T r Tj are commonly measured in eastern boundary cur- 

 rent regions during periods when coastal upwelling establishes a 

 stable atmospheric boundary layer. Comparable negative sea-air 

 vapor pressure differences (e„-ej are not observed as often. Over 

 large regions of the oceans, vapor pressure decreases with height, 

 although the opposite behavior can occur over cold water areas 



(Roll 1965). Laevastu and Harding (1974) demonstrated a rela- 

 tively rapid 5-h response of the properties of the surface air to 

 changes in the properties of the sea surface. Therefore, in midlati- 

 tude and tropical regions away from coastal boundaries, the maxi- 

 mum atmospheric vapor pressure would be expected to correspond 

 to the saturation vapor pressure at the sea-surface temperature, and 

 evaporation would occur. 



During spring and summer, modification of the atmospheric 

 boundary layer by coastal upwelling processes produces stable stra- 

 tification which is favorable for the formation of advection fog. 

 particularly along the coasts of central and northern California and 

 Oregon. When the contrast between warm, relatively moist air 

 overlying a cooler sea surface is large, fog may persist even when 

 the wind is strong. The resulting downward flux of water vapor 

 transfers latent heat from atmosphere to ocean. A downward flux of 

 moisture may also exist in a stable boundary layer in the absence of 

 fog, as the data of Laevastu and Harding (1974) and Anderson and 

 Smith (1981) demonstrate. 



In a study of the heat exhange processes over the North Pacific 

 trade wind zone, Seckel (1970) assumed that negative sea-air vapor 

 pressure differences were due to erroneous observations of sea sur- 

 face temperature, vapor pressure of the air, or both, and in such sit- 

 uations set the computed latent heat flux equal to 0. Over the 

 California Current region, we found that 10% of the usable surface 

 weather reports resulted in negative fluxes of latent heat. If these 

 values were due to bad observations, then an approximately equal 

 number of positive values should also be affected by poor quality 

 observations, but we had no a priori reason to reject 20% of the data 

 or possibly alias the long-term mean distributions by neglecting the 

 rate of heat change through condensation, as Roden (1959) did in a 

 previous study of the heat balance of the California Current region. 

 However, additional quality control tests were implemented to 

 remove unreasonably large negative and positive fluxes of both 

 latent and sensible heat. 



Editing procedures consisted of comparing observed surface proper- 

 ties with the average joint probability density functions of air-sea tem- 

 perature difference versus windspeed and sea-air vapor pressure 

 difference versus windspeed computed from the entire data set. Pairs of 

 data values which fell outside a 1 % bound of the joint probability den- 

 sity estimates were rejected. This method was relatively conservative, 

 and <0.5% of the data was removed. The distributions could have 

 been trimmed more severely. For example, in an earlier version of the 

 flux calculations, we used Seckel's (1970) method and set the latent 

 heat flux equal to zero for negative sea-air vapor pressure differences. 

 Comparisons of the two sets of values showed that the ratio of the dif- 

 ference between the old and new values to the standard error of the 

 mean of the recomputed data was consistently < 1.0, except in regions 

 within 200 km of the coast during months when coastal upwelling 

 occurs. We concluded that our results would not be severely distorted 

 by using Equation (8) for computing both positive and negative latent 

 heat fluxes, and might be more representative of actual surface layer 

 conditions near the coast. 



Our computations indicate an upward flux of latent heat (Appen- 

 dix I, Charts 25-36) over the entire California Current region 

 except at lat. 40°N, long. 124°W in May, July, August, and Sep- 

 tember when the mean fluxes are -0.6, -11.0. -12.1. and -8.7 W/ 

 m-\ at lat. 4 1 °N, long. 124 °W in July and August ( - 6.0 and - 0.4 

 W/m 2 ), and at lat. 43°N, long. 124°W in July (- 18.2 W/m 2 ). 

 Negative (condensation) values may be expected for periods of sev- 

 eral hours to a few days; however, it seems unlikely that latent heat 

 transfer from atmosphere to ocean would be characteristic in a 

 long-term monthly mean sense. 



