Table 1. — C^astluie orientations chosen to characterize the dominant coastline trend 

 at each location on a scale of approximately 200 miles. The directions of vectors tan- 

 gent to the coast with the ocean on the right are given in degrees from true north. 



LOCATION COASTLINE LOCATION COASTLINE LOCATION COASTLINE 



Lat. Long. ANGLE Lat. Long. ANGLE Lat. Long. ANGLE 



waters to replace those driven offshore by the stress of the 

 wind. The characteristic coastline orientations used to 

 resolve the offshore component are listed in Table 1. 



Constants Employed in the Stress Calculation 



Following the procedure of Fofonoff (1960), ^ the den- 

 sity of air used in Equation (1) was considered to have a con- 

 stant value of 0.00122 g cm' . This is very near to the U.S. 

 standard atmosphere sea level air density value (0.001226 

 g cm' ) chosen by the U.S. Weather Service as representing 

 average conditions over the United States at lat 40° N. 



The drag coefficient, Cd , was considered to be a constant 

 equal to 0.0013. Recent research (Davidson 1974) has 

 demonstrated dependence of the drag coefficient both on 

 atmospheric stability and on spectral properties of the ocean 

 waves. However, a generally agreed-upon formulation does 

 not appear well enough established at this time to warrant 

 attempting to incorporate these effects in the computations 

 of upweUing indices. The value 0.0013 is within the range of 

 recent estimates for constant drag coefficients. DeLeonibus 

 (1971) gives 0.0012 ± 0.0004 as characteristic of constant 

 drag coefficients measured in neutrally stable conditions 

 from the Argus Island tower off Bermuda. Denman and 

 Miyake (1973) arrive at 0.00163 ± 0.00028, with no observed 

 dependence on stability, from measurements at ocean 

 station Papa. For our purposes, the absolute value is less 

 important than intercomparability among various data sets. 

 The value 0.0013 was used by Bakun, McLain, and Mayo 

 (1974) in a climatological study, and is compatible with 

 Bakun's (1973) discussion of the effect of monthly mean data 

 on such computations. 



Atmospheric Pressm-e Data 



The choice of time scale (6-hr) and space scale (3-degTee 

 grid mesh) used in the calculations is largely dictated by the 

 time and space distributions of the basic observational data. 

 Meteorological observations are commonly made and collect- 

 ed via the international data network at 6-hr intervals. By 

 convention, the times of observation are at 0000, 0600, 1200, 

 and 1800 Greenwich mean time. The derived distributions 

 are properly viewed as simulataueous instantaneous sam- 

 plings rather than as averages over any time interval. Any 

 averaging which may take place as part of a measurement 

 procedure is intended merely as a means of filtering the very 

 small-scale turbulent variations (having periods less than 



several minutes) in order to reveal the underlying larger 

 scale patterns which are of interest. 



A typical distribution of pressure observations over the 

 northeast Pacific Ocean for a particular 6-hr sampling (Fig. 

 2) suggests futility in attempting to resolve details in the 

 distribution on a scale much smaller than three degrees. On 

 any scale much larger than this, however, important 

 features of coastline trend are lost. 



The method of atmospheric pressure analysis presently 

 in use at FNWC is described by HoU and Mendenhall (1972). 

 The method not only incorporates pressure reports but also 

 accompanying wind reports in the form of equivalent 

 pressure g^radients. The result thus reflects a greater 

 amount of observational data than would a separate analysis 

 of either pressure or wind. Objective forecast procedures 

 are applied to the previous analysis (6 h earlier) to supply 

 additional information. 



About 1% of the 6-hourly synoptic sea-level pressure 



^Fofonoff, N.P. 1960. Transport Computations for the North 

 Pacific Ocean 1958. Fish. Res. Board Can.. Manuscr. Rep. (Oceanogr. 

 Limnol.) 80, 81 p. 



Figure 2.— Distribution of observations made at 1800 GMT, 5 January 

 1974, which arrived at the Fleet Numerical Weather Central in time for 

 incorporation in the operational surface pressure analysis. 



