ent' with a surface temperature gradient from southwest to northeast 

 (Fig. 13) and an equatorward surface current with an offshore compo- 

 nent (Pattullo et al. 1969). At a depth of 30 m the vertical second deriv- 

 ative of temperature is -8. Ox 10 jO C/m 2 . The approximate scale forthe 

 vertical eddy diffusion coefficient is A H /A Z = 1.0x10* (Pond and 

 Pickard 1978); consequently /i 2 = l.Ox 10 J m 2 /s. Therefore, vertical 

 mixing, as parameterized in Equation (11), leads to a loss of heat at the 

 base of the water column at the rate of 0.2°C/mo. From the balance of 

 terms in Equation (11), we inferred mean vertical temperature advec- 

 tion (-wdT/dz) of -1.6°C/mo. In this region, the mean vertical tem- 

 perature gradient in the upper 30 m is approximately 0.05 °C/m. These 

 values imply a mean (upward) vertical velocity of 32 m/rao, which lies 

 within the range of estimates for coastal upwelling regions ("Wooster 

 and Reid 1963). Similar heat budget calculations for an upwelling 

 region off Cabo Bojador (Bowden 1977) yielded substantially larger 

 10-d mean vertical velocities (e.g., 4-10 m/d). The magnitudes of the 

 terms in Equation (11) imply that the primary balance in coastal 

 upwelling regions is between surface heat flux and horizontal and verti- 

 cal advection. 



The vertical velocity computed from the heat budget equation is 

 considerably larger than the mean open ocean vertical velocity 

 associated with horizontal divergence in the surface Ekman layer 

 (Yoshida and Mao 1957). We used the approximate relationship. 



k»( v*t) 

 pj 



(12) 



where k •( v • t) is the vertical component of wind stress curl, and/is 

 the Coriolis parameter, and distributions of wind stress curl (Nelson 

 1977) to independently compute a mean upwelling velocity of 5 m/mo. 

 The heat budget calculations would be consistent with this estimate of w 

 if the surface heat flux. Q m were distributed from the surface to a depth 

 of 50 m. 



The uncertainty in each of the terms in Equation (11) may be as 

 large as the error in Q s . which we estimated to be 15 to 70%. Roden 

 (1959) estimated an error of 50% in calculating the horizontal 

 advection term from observed velocity components and tempera- 

 ture gradients. In addition, the physical basis for parameterizing 

 vertical mixing as a function of a constant eddy coefficient. A 2 , and 

 the vertical second derivative of temperature, d 2 T/dz 2 As not well 

 founded. Therefore, the computed vertical eddy diffusion may 

 grossly underestimate the effects of mixing at the base of the mixed 

 layer (Niiler and Kraus 1977). In our heat budget calculations, an 

 order of magnitude increase in A z would have resulted in an approx- 

 imate balance between heat gain at the surface and heat loss by hori- 

 zontal advection and vertical mixing, without the requirement for 

 vertical advection. Because of the uncertainties associated with 

 heat budget calculations, the "balance" which we computed may 

 have been fortuitous. However, these calculations demonstrate that 

 our values of net surface heat exchange are not inconsistent with 

 independent estimates of temperature advection and diffusion. 



SUMMARY AND CONCLUSIONS 



Distributions of long-term composite monthly atmosphere- 

 ocean heat exchange processes have been presented for the Califor- 

 nia Current region over a 1 ° latitude-longitude grid. The monthly 

 fields represent summarizations of the radiative and turbulent heal 

 fluxes computed from individual surface marine weather reports 

 archived in the National Climatic Center's TDF-11 data file. This 



report complements the earlier work of Nelson ( 1977) who summa- 

 rized wind stress and wind stress curl over the California Current 

 region. 



The climatological fields of surface heat fluxes provide insight to the 

 important areas of heat gain and loss in the California Current. The 

 region off the west coast of the United States and Baja California is 

 characterized by net annual heat transfer from atmosphere to ocean. 

 Maximum net oceanic heat gain occurs in the coastal upwelling zone 

 off northern California as a consequence of relatively low cloud cover 

 compared with offshore distributions, suppression of latent heat flux, 

 and a reversal of the sensible heat transfer due to the presence of cold 

 water during summer. These same features (i.e., maximum net oceanic 

 heat gain, low cloudiness, and small turbulent fluxes in coastal upwell- 

 ing regions) are evident in all major eastern boundary current systems 

 (Hastenrath and Lamb 1978). Maximum heat loss occurs during winter 

 off the Pacific Northwest. Simplified heat budget calculations demon- 

 strate that our values of surface heat flux are not inconsistent with inde- 

 pendent estimates of horizontal and vertical temperature advection and 

 show the important contribution of advective processes in determining 

 the seasonal heat balance in coastal upwelling regions. 



Because of the heterogeneous character of the surface marine 

 weather observations, relatively large errors are associated with empiri- 

 cal estimates of the radiative and turbulent heat fluxes. Nonrandom dis- 

 tributions of reports, sampling errors, fair-weather bias, and methods of 

 computation introduce uncertainties which cast doubt on the validity of 

 heat exchange estimates derived from merchant vessel data. Reason- 

 able estimates of the error in each heat exchange component (e.g., 

 10%) lead to errors in the long-term mean net heat exchange, Q N , rang- 

 ing from 10 to 70% . Our analyses indicate that the principal spatial and 

 temporal features of surface heat flux over the California Current 

 region (e.g., maximum oceanic heat gain near the coast during sum- 

 mer) are independent of computational methods which use monthly 

 mean properties and variable exchange coefficients in the empirical 

 formulae as opposed to individual observations and constant coeffi- 

 cients. Spectrum analyses of representative time series of Q N show a 

 general absence of variance at low frequencies and demonstrate the 

 large signal-to-noise ratio of the annual cycle in midlatitude regions. 

 The lack of coherence between time series in adjacent 1 ° squares is 

 partially explained by the errors in monthly heat flux estimates, even in 

 densely sampled areas. 



The results of this study will be useful in modeling mean sea- 

 sonal variations in the thermal structure of the California Current. 

 Forthe purpose of indicating anomalies from the long-term means, 

 however, further research in marine boundary layer processes is 

 needed to place narrower confidence limits on the derived 

 atmosphere-ocean exchanges. Because the largest relative errors 

 are associated with heat exchange components which also have the 

 largest absolute magnitudes (i.e., Q s and Q E ), more extensive 

 experimental studies in the open ocean and in the coastal boundary 

 zone must be conducted 1) to improve the empirical formulae for 

 short-wave radiation by using more objectively measured parame- 

 ters (e.g., satellite observations of cloud cover), and 2) to evaluate 

 the moisture flux parameterization for wider ranges of atmospheric 

 stability and windspeed. With the increased application of air-sea 

 interaction research to global climatic and fisheries problems, there 

 is also an urgent requirement for consensus on the empirical formu- 

 lae and methods of computation to be used. Although the absolute 

 magnitudes of the long-term monthly mean values cannot be pre- 

 cisely fixed, the spatial and temporal consistency of independent 

 estimates of surface heat exchange indicate that the geographical 

 patterns (Appendix I) are realistic and significant. 



30 



