Environmental time series were obtained by Principal Components Analysis of the PACEA data set, using only the longer-term observational series since 1950. Annual values were represented by the May observations (representing nominal time of ocean entry), and Nov-May differences (representing nominal shifts in conditions over the following half-year. The Okanagan run was forecast mainly by linear regression of AC1.3 on the previous year's AC1.2, and did not utilize environmental variates. The Wenatchee run is mostly AC1.2, and did not have an informative earlier appearance. I forecast AC1.2 with a stock-recruitment model from a log-linear multiple regression, including two of the environmental Principal Components. The Bonneville forecast was the sum of the the Okanagan and Wenatchee forecasts plus a small number set at the long term average of fish of indeterminate origin.

Data reduction of the environmental data by means of Principal Components Analysis was conducted in R. The remained of the calculations were done in Excel, including use of the multiple regression capability in the analysis toolkit.