264 



Fishery Bulletin 88(2), 1990 



these stocks of chinook salmon. However, because the 

 correlations are with three oceanographic variables 

 rather than consistently with one, they are somewhat 

 difficult to interpret in terms of a causal mechanism. 

 To attempt to solve this problem, we sought a single 

 indicator of variability in the coastal ocean. We com- 

 puted the principal components (Harris 1975) of 

 variability in sea surface temperature, sea level height, 

 and upwelling index, in the hope that they would have 

 a meaningful physical interpretation. The resulting 

 first principal component is larger than the second and 



third by about a factor of two, depending on season, 

 and its loading is approximately equal for all three 

 variables (positive for sea level height and ocean 

 temperature and negative for upwelling index). 

 Positive values of sea level height and surface tem- 

 perature are consistent with the higher latitude effects 

 of ENSO events as would be propagated by coastal 

 trapped waves, but the negative value of upwelling 

 index would require atmospheric teleconnection (recall 

 the upwelling index is not a measure of actual upwell- 

 ing, but an estimate based on atmospheric pressure 



