Evaluation of the 2008 Predictions of Run-Timing and Survival of Wild Migrant Yearling Chinook and Steelhead on the Columbia and Snake Rivers.
Columbia Basin Research uses the COMPASS model on a daily basis during the outmigration of Snake River Chinook and steelhead smolts to predict downstream passage and survival. Fish arrival predictions and observations from program RealTime along with predicted and observed environmental conditions are used to make in-season predictions of arrival and survival to various dams in the Columbia and Snake Rivers. For 2008, calibrations of travel and survival parameters for two stocks of fish-Snake River yearling PIT-tagged wild chinook salmon (chin1pit) and Snake River PIT-tagged steelhead (lgrStlhd)-were used to model travel and survival of steelhead and chinook stocks from Lower Granite Dam (LWG) or McNary Dam (MCN) to Bonneville Dam (BON). This report summarizes the success of the COMPASS/RealTime process to model these migrations as they occur. We compared model results on timing and survival to data from two sources: stock specific counts at dams and end-of-season control survival estimates (Jim Faulkner, NOAA, pers. comm. Dec. 16, 2008). The difference between the predicted and observed day of median passage and the Mean Absolute Deviation (MAD) between predicted and observed arrival cumulative distributions are measures of timing accuracy. MAD is essentially the average percentage error over the season. The difference between the predicted and observed survivals is a measure of survival accuracy. Model results and timing data were in good agreement from LWG to John Day Dam (JDA). Predictions of median passage days for the chin1pit and lgrStlhd stocks were 0 and 2 days (respectively) later than observed. MAD for chin1pit and lgrStlhd stocks at JDA were 2.3% and 5.9% (respectively). Between JDA and BON modeling and timing data were not as well matched. At BON, median passage predictions were 6 and 10 days later than observed and MAD values were 7.8% and 16.0% respectively. Model results and survival data were in good agreement from LWG to MCN. COMPASS predicted survivals of 0.77 and 0.69 for chin1pit and lgrStlhd, while the data control's survivals were 0.79 and 0.68. The differences are 0.02 and 0.01 (respectively), nearly identical. However, from MCN to BON, COMPASS predicted survivals of 0.74 and 0.69 while the data controls survivals were 0.47 and 0.53 respectively. Differences of 0.27 and 0.16. In summary: Travel and survival of chin1pit and lgrStlhd stocks were well modeled in the upper reaches. Fish in the lower reaches down through BON suffered unmodeled mortality, and/or passed BON undetected. A drop in bypass fraction and unmodeled mortality during the run could produce such patterns by shifting the observed median passage day to appear artificially early.
- Research Organization:
- Bonneville Power Administration (BPA), Portland, OR (United States)
- Sponsoring Organization:
- USDOE
- DOE Contract Number:
- 39016
- OSTI ID:
- 947611
- Report Number(s):
- P109799; R&D Project: 198910800; TRN: US200906%%86
- Country of Publication:
- United States
- Language:
- English
Similar Records
Monitoring and Evaluation of Smolt Migration in the Columbia Basin : Volume IX : Evaluation of the 2001 Predictions of the Run-Timing of Wild and Hatchery-Reared Migrant Salmon and Steelhead Trout Migrating to Lower Granite, Rock Island, McNary, and John Day Dams using Program RealTime.
Monitoring and Evaluation of Smolt Migration in the Columbia River Basin : Volume VI : Evaluation of the 2000 Predictions of the Run-Timing of Wild Migrant Chinook Salmon and Steelhead Trout, and Hatchery Sockeye Salmon in the Snake River Basin, and Combined Wild Hatchery Salminids Migrating to Rock Island and McNary Dams using Program RealTime.