Monitoring and Evaluation of Smolt Migration in the Columbia River Basin : Volume III : Evaluation of the 1997 Predictions of the Run-Timing of Wild Migrant Yearling and Subyearling Chinook and Sockeye in the Snake River Basin Using Program RealTime.

Since the 1994 outmigration, program RealTime has been applied to provide in-season predictions of smolt outmigration timing for individual and aggregates of listed threatened and endangered Snake River salmon stocks. Results from the 1997 smolt outmigrations of wild Snake River yearling and subyearling chinook show prediction of run-timing can be accurately forecasted. The number of release sites meeting previous years criteria for RealTime forecasts dropped to five for the wild spring/summer chinook parr PIT-tagged in 1996: Catherine Creek, Imnaha, Lostine, Minam and South Fork Salmon Rivers. An experiment in lessening previous RealTime requirements for forecasting a outmigration in progress added three release sites of chinook: Lake Creek, Secesh and South Fork Wenaha Rivers; and one release of age 1+ sockeye at Redfish Lake. Passage indices provided by the Fish Passage Center for Lower Granite Dam were monitored for the wild subyearling chinook outmigration. Investigation continued into basing predictions on historical years with similar flows as a way to improve forecasting performance for the wild subyearling outmigration. Program RealTime's output is a series of estimated percentages of the status of the smolt outmigration throughout the season. To compare the performance the program from year to year, or to compare various assumptions used set up the forecasting, the mean absolute deviance (MAD) of the daily predicted outmigration-proportion from the actual outmigration-proportion is calculated post-season. Furthermore, these MAD's are considered for three periods of the season: the first 50% of the season, the second 50%, and the entire season.