Executive Summary Regulating the timing and volume of water released from storage reservoirs (often referred to as flow augmentation) has become a central mitigation strategy for improving downstream migration conditions for juvenile salmonids in the Snake River. The success of the flow augmentation, in turn, depends on releasing reservoir waters when and where wild smolt will benefit the most. This requires the ability to predict in real time the status and trend in the outmigration of listed

more »

... hreatened and endangered stocks. This study evaluated the feasibility and the performance of two alternative statistical algorithms to predict outmigration status of Snake River wild spring chinook. Using historical trends in PIT-tag detections of wild chinook smolt at Lower Granite Dam, pattern recognition techniques were developed to predict the percent of the run-to-date and days to a specific percent of the run. The statistical methods are based on algorithms that smooth historical trends in PIT-tag arrivals and a generalized least squares decision criterion. The methods were evaluated for 16 different river runs of chinook, as well as composites over various river basins. A bootstrapping approach across historical years provided the means to measure the accuracy and precision of predictions and construct approximate interval estimates. The recommended predictors have an average error rate across stocks of fish and seasons of &9.6% about the true percent of the run-to-date. The best performance was for Catherine Creek with an average error of +4.2%. The worst performance occurred at Big Creek with an average error of +19.0%. An interactive graphical analysis program written in C-language for an X-Window@ environment has been developed to analyze outmigration data for select stocks of Snake River spring chinook.