A time Series of mean weekly sea surface temperature (SST) images was used
to investigate the relationship between fluctuations in the marine survival
of hatchery-reared coho salmon and coastal ocean dynamics off the north-we
stern United States (51 degrees to 37 degrees N) between 1985 and 1996, usi
ng univariate and nonlinear bivariate regression analysis. Ocean conditions
were matched against survival for a number of different annual time frames
according to the sum of negative or positive weekly SST anomalies. From th
e univariate analyses, the sum of negative anomalies from April to June, wh
en the juvenile salmon first enter the ocean, was found to have an R-2 of 0
.88 against survival with 1991 excluded as an outlier. The bivariate multip
le regressions used the sum of negative anomalies from April to lune as the
first independent variable. When the sums of positive anomalies from the f
ollowing periods during the fishes' second calendar year in the ocean were
each used as the second independent variable, the R-2 Values were all great
er than or equal to 0.92 (with no data points excluded): January to June, F
ebruary to lune, April to June, March to lune. These results are discussed
within the context of coastal ocean processes. It is concluded that the ana
lysis of SST image time series might allow management to make reasonable fo
recasts of hatchery-reared coho salmon survival.