Figure 16. — Discretization grid used in calculating the wind stress 

 curl. 



per 100 km is equivalent to 1 X 10~' dyne cm" . Negative 

 values of wind stress curl are shaded. 



Small scale features evident in these distributions 

 should be viewed with caution. Detail within a single 1- 

 degree square area which is not supported by similar 

 values in surrounding squares probably reflects "noise" 

 in the monthly means of surface wind stress. Objective 

 smoothing procedures applied to the monthly mean wind 

 stress fields would result in more homogeneous dis- 

 tributions of wind stress curl. A particular method has 

 been described by Evenson and Veronis (1975). 



Characteristic absolute magnitudes of the spatially 

 averaged wind stress curl are 1 X 10 -8 dyne cm -3 . This 

 value is approximately a factor of two less than the sum- 

 mertime mean reported by Halpern (1976) for an up- 

 welling region near the Oregon coast. Considering the 

 time and space averages used in the present study, this 

 difference appears to be reasonable. However, the values 

 are consistent with wind stress curl calculations reported 

 by Saunders (1976). The probable errors associated with 

 these estimates of wind stress curl may be calculated 

 from a knowledge of the spatial distributions of standard 

 errors of the (wind stress) component means (see Appen- 

 dix II). On average, the expected error is 1 X 10 ~ 8 dyne 

 cm -3 , with maximum and minimum errors of 4 X 10 ~ 8 

 dyne cm -3 and 1 X 10" 10 dyne cm" 3 , respectively. Of 

 course, these values apply to particular individual 1- 

 degree squares. Large values generally correspond to 

 "holes" in the distributions, which are easily seen in the 

 monthly charts. For larger space scales, errors as- 

 sociated with the gradients of the mean wind stress com- 

 ponents would tend to cancel. Thus, greater confidence 

 may be expected for the patterns of wind stress curl 

 which are consistent over several degrees of latitude and 

 longitude. 



The large-scale features of positive and negative curl 

 along the coast are significant. Important details to note 

 are the sign of the wind stress curl at the coast, and the 

 position of the line of zero wind stress curl. A general 

 feature common to all months is the occurrence, on 

 average, of positive wind stress curl near the coast, and 

 negative curl at some distance offshore. This feature is 

 well developed from May to September. Greater spatial 

 variability is evident in winter distributions. 



Existence of an offshore wind stress maximum results 

 in a line of zero wind stress curl approximately parallel to 

 the coast. Positive curl occurs inshore of the maximum 

 wind stress. Negative curl in the offshore region is as- 

 sociated with the anticyclonic atmospheric circulation 

 over the interior ocean. The positive curl near the coast is 

 related to topography and to local features in the surface 

 wind stress distributions. 



The distributions from December to March are charac- 

 terized by positive wind stress curl near the coast from 

 San Francisco to northern Baja California, and south of 

 Punta Eugenia. Large areas of associated Ekman diver- 

 gence extend several hundred kilometers off the coast. 

 The patterns of wind stress curl are less well behaved 

 near the coasts of Oregon and Washington. However, 

 negative wind stress curl near this coastal region appears 

 to be typical of the distributions during the winter. 



During spring and summer upwelling seasons, the 

 dominant patterns of surface wind stress curl are easily 

 recognized. The line of zero wind stress curl parallels the 

 coast approximately 200 to 300 km offshore, along the en- 

 tire boundary from northern Baja California to Van- 

 couver Island, and from June through September. 

 Yoshida and Mao (1957) placed this line at ap- 

 proximately 500 km from the coast. Considering the 

 coarse resolution (5-degree squares of latitude and 

 longitude) of their data, this disparity is not surprising. 



The months of April-May and October-November ap- 

 pear as transitional periods. During transition from 

 spring to summer, the negative curl along the coasts of 

 Oregon and Washington shifts to positive curl. The off- 

 shore distribution takes on a more uniform character. 

 Scattered regions of positive and negative curl are replac- 

 ed by a large area of negative curl. The late fall transition 

 is marked by a total breakdown in the curl distributions 

 within the northern sector of the grid. 



Several local (positive) curl maxima are associated 

 with major topographic changes in the coastline con- 

 figuration. Large values of positive wind stress curl are 

 found near Cape Blanco, Cape Mendocino, San Fran- 

 cisco, and Point Conception. These features may be real, 

 or they may be artifices of the finite difference cal- 

 culations or the data distributions. However, one does 

 note that near Point Conception, the values of positive 

 wind stress curl are consistent with a decrease in the 

 magnitude of the surface wind stress in the lee of the 

 point. Where areas of positive wind stress curl extend off- 

 shore of capes, there would tend to be a continued, al- 

 though much reduced, level of upwelling outside of the 

 primary coastal upwelling regime. 



21 



