fields during the 7-yr period were missing from the FNWC 

 archives. Certain of the missing fields had been rejected as 

 erroneous. Some had been lost due to various equipment 

 failures. As a result, some 113 of the daily averages reflect 

 only three of the four daily synoptic periods, 15 contain two 

 of four, and 2 contain only one. The days for which the daily 

 averages represent less than all four synoptic samplings are 

 listed in Table 2. 



Distortion of Spatial Distributions 



Bakun (1973) listed two sources of distortion in the 

 spatial distributions of upwelling intensity indicated by the 

 monthly indices presented therein. In that case monthly 

 mean pressure data were used for the purpose of extending 

 the series back through 1946 in a reasonably consistent man- 

 ner. A major distorting effect was attributed to the use of 

 mean data in conjunction with the nonlinear stress law 

 (Equation (1)), resulting in an underestimate which depend- 

 ed on the variability of the wind field during the month. To 

 partially compensate for underestimating the mean stress, a 

 high value (0.0026) of the drag coefficent was used. Since the 

 normal variability differs among different geographical 

 areas, indicated upwelling at certain locations was amplified 

 relative to other locations. The series presented here cover 7 

 yr, for which there are available a quite complete and 

 internally consistent set of 6-hourly synoptic pressure 

 analyses. Thus, there is no need to resort to mean data and 

 that particular source of distortion is not encountered. 



The second source of spatial distortion discussed by 

 Bakun (1973) applies also in the present case. This involves a 



discontinuity in the onshore-offshore sea-level atmospheric 

 pressure gradient caused by the coastal mountain ranges. In 

 the interior valley of southern California, in particular, an 

 intense localized thermal low develops during the summer. 

 Because of the 3-degree mesh length of the computation 

 grid, the effect of this low on the computed indices may be 

 amplified relative to its actual effect on the wind stress field 

 on the coastal side of the mountain range. 



This effect can be illustrated by comparing the distribu- 

 tion of means by month and location of the 6-hourly 

 upwelling indices to a corresponding summary of offshore 

 Ekman transports produced from actual wind observations 

 made by ships at sea. Figure 3(a) displays the annual cycle 

 obtained by Bakun et al. (1974). The figure summarizes over 

 75,000 individual wind reports from the National Climatic 

 Center's file of marine surface observations over the 20-yr 

 period, 1948 through 1967, within the 1-degree squares 

 (Figure 3(c)). The Ekman transport was computed for each 

 wind observation and the offshore components, grouped by 

 long-term month and by 1-degree square, were averaged. 

 Figure 3(b) displays time-series isograms of the 7-yr 

 (1967-73) mean monthly values of the 6-hourly upwelling 

 indices. Some of the detail apparent in Figure 3(a) is 

 smoothed out in Figure 3(b) because the upwelling indices 

 are computed at 3-degree intervals using derivatives 

 computed across a 6-degree length scale whereas the 

 averages making up Figure 3(a) were formed at 1-degree 

 intervals and contain data only from the surrounding 

 1-deg^ee square. However, comparison of the two figures 

 indicates a definite southward shift of the region of 

 maximum magnitude of the upwelling indices. The indica- 



J F M S M J 



I50W I40W 13PW I20W now HOW 



Figure 3. — Long-term annual cycle of offshore Ekman transport. I'nits are cubic meters per seond per 100 m of coastline, (al Cycle formed by averaging 

 transports computed from individual wind observations taken at sea within the one degree coastal squares indicated in Figure 3lc). (b) Cycle formed from 

 6-hourly upwelling indices averaged by month at the locations indicated in Figure 3(c). (c) Coastline drawing. Coastal squares for Figure 3(a) are shaded. 

 Locations for Figure 3(bl are indicated by dots. 



