exchange coefficients. C E and C H . were used in Equations (4) and (5) to 

 estimate the turbulent fluxes of latent and sensible heat. Different 

 results might have been obtained if the radiative and turbulent fluxes 

 had been calculated from monthly mean atmospheric properties or if 

 the transfer coefficients for the turbulent fluxes had varied with wind- 

 speed or atmospheric stability. These aspects are discussed below. 



Exchange Processes Computed from 

 Monthly Mean Atmospheric Properties 



Previous summaries of large-scale heat exchange processes have 

 used mean monthly values of atmospheric and oceanic properties in 

 the empirical formulae, primarily due to the lack of sufficient data 

 to perform the calculations on shorter time scales. Malkus (1962) 

 has shown that the use of mean monthly values can lead to underes- 

 timates of the turbulent heat fluxes. For the sake of comparability. 

 we have assembled long-term monthly mean atmospheric proper- 

 ties (Appendix ni) and used these values to recompute the heat 

 exchange components. 



Figure 5 shows a scatter diagram of all monthly gridpoint values of 

 net heat exchange computed using mean monthly properties. Q X {M), 

 versus the mean monthly net heat exchange computed from the individ- 

 ual synoptic weather reports, QM)- The slope of the least-squares 

 regression of Q X (M ) on Q s (l ) is 1 .02 and the coefficient of determina- 

 tion (Br) between the two data sets is 0.99. Because the correlation for 

 this regression may be artificially inflated by the seasonal cycle in both 

 data sets, linear regressions were computed by month for each of the 

 heat exchange components. Table 1 lists the coefficients of determina- 

 tion, significant above the 99% level, and slopes of the regression 

 lines, each of which is based on 300 pairs of observations. 



The mean heat transfers computed from the individual surface 

 weather reports show a high degree of correlation with those computed 

 from mean monthly atmospheric and oceanic properties. The highest 

 correlations are found for Q B and Q s (7?- >0.99). The regressions for 

 Q c show the lowest correlations and the greatest degree of seasonal var- 

 iability (0.88 <,ff <0.96). 



The slopes of the regression lines provide an estimate of the average 

 differences between values of the radiative and turbulent fluxes com- 

 puted by the two methods. The largest deviations occur for incident 

 solar radiation. The monthly values, Q S (M), are from 1 to 18% larger 

 than corresponding estimates, Q s (f). based on synoptic reports. A simi- 

 lar relationship between Q S (M) and Q s ([) has been demonstrated by 

 Seckel ( 1 970) and Husby and Seckel ( 1 975 ) . The correlations for latent 

 and sensible heat exchange show that computations which use monthly 

 mean properties consistently underestimate the turbulent fluxes by 1 to 

 8% for Q E and 3 to 13% for Q c , although in August, September, and 

 October, the Q E (M) values are approximately 1% larger than Q B (I). 



R : 



0.997 



m 



/ 



R' = 



0.994 







Y = 



A + BX 







A = 



B = 



2.598 

 1.023 



~~. 30B " 

 o 



2BB- 



d }< 







iea- 





-jra 





-18B iJT 



IBB 2KJ 323 IM 







..;* ' "103 - 



QNi I 1 







-2Z3 - 









-3BB ■ 









-iCtt - 





Figure 5.— Linear regression of monthly net heat exchange values computed 

 from mean monthly meteorological properties, Q N (M), versus monthly net 

 heat exchange computed from individual reports. Q N (I). Units are \\7m-. The 

 correlation coefficient is R = 0.997 and the coefficient of determination is R 2 = 

 0.994. The dashed diagonal line represents the regression equation 

 Q N (M) = 2.598-H.023Q N (I). 



These differences are comparable with values of about 7% discussed 

 by Malkus (1962) and are within the uncertainties of the determina- 

 tions discussed earlier. 



Estimates of net heat exchange computed by the two methods are 

 well correlated (0.95 <R 2 < 0.98). Slopes > 1 .0 indicate that the values 

 based on monthly parameters. Q S {M). are consistently more positive 

 than the monthly averages of individual estimates, Q X (I), although this 

 generalization is not valid for all areas in all months. Distributions of 

 the ratio Q X (M)/Q X (T) are shown in Figures 6 and 7 for July and 

 December, respectively. In July the values for Q X {M) are approximately 

 3 % greater than those for Q x (l) , except near the coast of southern Baja 

 California, between Point Conception and Cape Blanco, and off Van- 

 couver Island. The use of long-term monthly mean atmospheric prop- 

 erties results in 5 to 10% higher oceanic heat gain in these regions 

 compared with the monthly means of synoptic estimates. During 

 December, the ocean loses heat over the entire California Current 

 region (Chart 60). Values of Q X (M)/Q X (I)<\.Q occur over approxi- 



Tahle 1.— Monthly values of coefficient of determination (R-) and slope (B) of regression line for linear regres- 

 sions between heat exchange components (Q s , Q„, Q E , Q c , Q N ) estimated from monthly mean atmospheric 

 properties, Q(M) and those computed from individual reports, Q([). 



Feb. 



Mar. 



Apr. May June July Aug. Sept. 



Oct. 



Nov. 



Dec. 



ft* 



0.998 



0.997 



0.993 



0.991 



0.991 



0.991 



0.994 



0.989 



0.990 



0.995 



0.998 



0.999 



1.016 



1.028 



1.051 



1.079 



1.164 



1.181 



1.156 



1.096 



1.056 



1.036 



1.021 



1.009 



e°s 



.997 



.998 



.998 



.997 



.998 



.998 



.999 



.998 



.996 



.995 



.997 



.997 



1.003 



1.000 



1.003 



1.003 



.995 



.993 



.992 



.989 



.987 



1.010 



1.004 



.996 



< 



.954 



.968 



.975 



.986 



.986 



.980 



.974 



.962 



.971 



.972 



.970 



.963 



.929 



.923 



.961 



.988 



.987 



.985 



.988 



1.022 



1.007 



1.012 



.961 



.926 



c B 



.934 



.881 



.927 



.924 



.945 



.955 



.963 



.955 



.948 



.931 



.941 



.909 



.953 



.865 



.939 



.946 



.966 



.953 



.944 



.966 



.926 



.943 



.960 



.926 





.970 



.957 



.958 



.952 



.956 



.963 



.979 



.958 



.962 



.974 



.980 



.965 



1.018 



1.029 



1.048 



1.086 



1.049 



1 .085 



1.121 



1.091 



1.067 



1.022 



1.025 



1.017 



11 



