these "upwelling indices." Bakun (1973) extended the 

 concept of the upwelling index by computing estimates 

 of Ekman transport from routinely available analyzed 

 surface atmospheric pressure fields. Temporal fluc- 

 tuations are well described by these data. However, both 

 the derived quantities based on Hidaka's wind stress 

 data and Bakun's upwelling indices suffer from an in- 

 ability to characterize small-scale features of the Ekman 

 transport field, particularly near the coast. For example, 

 where these data may indicate surface divergence on the 

 large scale, smaller scale surface convergence might 

 exist, with associated effects on distributions of organ- 

 isms within the coastal upwelling zone. 



This report provides more detailed descriptions of the 

 surface wind stress distributions over the California Cur- 

 rent. The study differs from previous work by cal- 

 culating monthly mean values on a 1 -degree square area 

 basis. Surface marine wind observations have been 

 utilized in the computations. Roden (1974) evaluated the 

 surface wind stress on a 1-degree latitude-longitude grid. 

 However, Roden's distributions were derived from 

 monthly mean surface atmospheric pressure analyses 

 based on a 5-degree latitude-longitude grid. Thus, any 

 information concerning space scales smaller than 5 

 degrees is due to the particular interpolation scheme 

 used to refine the data to a 1-degree grid, rather than due 

 to observed data. 



The monthly mean data described in this report ade- 

 quately resolve the seasonal cycle, which is the dominant 

 time scale for coastal upwelling (Mooers et al. 1976). The 

 high resolution in space and time may provide the ob- 

 servational background for more detailed investigations 

 of the relationships among the local wind stress dis- 

 tributions, the equatorward surface current offshore, the 

 poleward surface and subsurface flow inshore, and dis- 

 tributions of certain species of fishes, such as the 

 northern anchovy, Engraulis mordax, and the Pacific 

 mackerel, Scomber japonicus. 



DATA REDUCTION 



The monthly mean distributions of surface wind stress 

 presented in Appendix I are based on summaries of data 

 contained in the National Climatic Center's file of sur- 

 face marine observations (Tape Data Family-11). The 

 total file contains approximately 40 million individual 

 ship reports dating from the mid- 19th century. Over 1 

 million of the reports are within the area of the Califor- 

 nia Current system. 



I have compiled long-term composite monthly fields of 

 surface wind stress on a 1-degree square area basis within 

 the geographical area outlined in Figure 1. The data grid 

 extends from lat. 20°N to 50°N and parallels the coast- 

 line configuration, extending 10 degrees of longitude in 

 the offshore direction. Each 1-degree quadrilateral is 

 centered on a whole degree of latitude and longitude. Ap- 

 proximately 25% of the total available reports contain 

 positions recorded to the nearest whole degree of latitude 

 and longitude. The grid orientation used in this study 



thus minimizes spatial bias which might be introduced 

 by summarizing the data according to the Marsden 

 square numbering system. 



The historical data contain errors in position, 

 measurement, and processing. A single pass editor was 

 used to remove gross errors in the data, including dup- 

 licate reports (0.5%), position errors (0.1%), and 

 measurement errors (0.5%). Erroneous wind directions 

 and wind speeds greater than 100 m s _1 were removed. 

 Reports of variable winds were treated as calms. 



An estimate of the surface wind stress was calculated 

 for each wind velocity report according to the bulk for- 

 mula: 



^<T>)=P a C D (\W w \U w , \WJV W ) (l) 



where r x and r, denote the eastward and northward 

 components of stress, p a is the density of air which was 

 considered to have a constant value of 0.00122 gem' 3 , 

 | W K | is the observed wind speed, and U w and V 10 are i 

 the eastward and northward components of the wind ve- 

 locity measured at a height of 10 m. The empirical drag 

 coefficient C D , referred to the 10-m level, was given a con- 

 stant value of 0.0013 (Kraus 1972). 



The resultant long-term monthly mean wind stress 

 vectors were computed as the arithmetic means by east 

 and north components of all available reports from 1850 

 to 1972 within a 1-degree square area. The appropriate 

 average is defined in Equation (2): 



(r. 



v- 



1 



N 



r y \ 



(2) 



where N is the total number of reports within a 1-degree 

 square area and month. The values (t, , r ) { were 

 evaluated according to Equation (1). A mean value for 

 each long-term month and square is therefore formed 

 from a data set which is independent of all other months 

 and squares. The monthly fields of surface wind stress are 

 displayed in Appendix I as vector quantities (Charts 1 to 

 12) and as east (Charts 13 to 24) and north (Charts 25 to 

 36) components. No attempt has been made to smooth 

 the fields, either by removing data which do not appear 

 to fit the distributions or by applying objective 

 smoothing procedures. The mean values were contoured 

 by computer and "bull's-eyes" in the contours, even 

 where they possibly reflect erroneous data, were left in 

 the charts as indications of the general degree of consis- 

 tency in the composite distributions. The bull's-eyes 

 may reflect either a paucity of ship observations, or ex- 

 treme variability associated with inadequate sampling of 

 strong winds. 



The spatial distribution of observations is biased in 

 that "ship of opportunity" reports are generally con- 

 fined to coastwise shipping lanes. Contours indicating 

 the total numbers of observations per 1-degree square 

 area are shown in Figure 2. The highest density of reports 

 is found within 300 km of the coast, exceeding 40,000 ob- 

 servations per 1-degree square in the area south of Point 

 Conception. The number of reports per 1-degree square 



