We developed methods to forecast survival and cumulative percent passage for subyearling chinook salmon Oncorhynchus tshawytscha at a dam to help managers effectively time the release of reservoir water to mitigate for passage delays and reduced survival. We tagged Snake River subyearling chinook salmon upstream of a dam from 1993 to 1998 and determined when a subsample of the tagged fish passed the dam. We randomly selected data (1993, 1994, 1996, and 1998) to develop a quadratic discriminant function for predicting which fish would survive to the dam and to develop a multiple-regression equation to predict the date survivors would pass the dam. We used the predicted passage dates within a year to calculate a daily cumulative percent passage forecast and then calculated a 90% forecast interval that varied by year, depending on the number of predicted survivors. We validated the forecast method using data for 1995 and 1997. The 1995 forecasted passage curve differed from the observed passage curve for 25 d of the 168-d emigration season. The 1997 forecasted and observed passage curves were similar for the entire 168-d emigration season. The 90% forecast interval was ±18.8% in 1995 and ±22.4% in 1997. We conclude that our method is a valid tool for in-season water management but acknowledge the potential for interannual variability in forecast performance.
