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.

Abstract

Listen

Translate article

Program RealTime provided tracking and forecasting of the 2000 in season outmigration via the internet for stocks of wild PIT-tagged spring/summer chinook salmon. These stocks were ESUs from nineteen release sites above Lower Granite dam, including Bear Valley Creek, Big Creek, Camas Creek (new), Cape Horn Creek, Catherine Creek, Elk Creek, Herd Creek, Imnaha River, Johnson Creek (new), Lake Creek, Loon Creek, Lostine River, Marsh Creek, Minam River, East Fork Salmon River (new), South Fork Salmon River, Secesh River, Sulfur Creek and Valley Creek. Forecasts were also provided for two stocks of hatchery-reared PIT-tagged summer-run sockeye salmon, from Redfish Lake and Alturas Lake (new); for a subpopulation of the PIT-tagged wild Snake River fall subyearling chinook salmon; for all wild Snake River PIT-tagged spring/summer yearling chinook salmon (new) and steelhead trout (new)detected at Lower Granite Dam during the 2000 outmigration. The 2000 RealTime project began making forecasts for combined wild- and hatchery-reared runs-at-large of subyearling and yearling chinook, coho, and sockeye salmon, and steelhead trout migrating to Rock Island and McNary Dams on the mid-Columbia River and the mainstem Columbia River. Due to the new (in 1999-2000) Snake River basin hatchery protocol of releasing unmarked hatchery-reared fish, the RealTime forecasting project no longer makes run-timing forecasts for wild Snake River runs-at-large using FPC passage indices, as it has done for the previous three years (1997-1999). The season-wide measure of Program RealTime performance, the mean absolute difference (MAD) between in-season predictions and true (observed) passage percentiles, improved relative to previous years for nearly all stocks. The average season-wide MAD of all (nineteen) spring/summer yearling chinook salmon ESUs dropped from 5.7% in 1999 to 4.5% in 2000. The 2000 MAD for the hatchery-reared Redfish Lake sockeye salmon ESU was the lowest recorded, at 6.0%, down from 6.7% in 1999. The MAD for the PIT-tagged ESU of wild Snake River fall sub-yearling chinook salmon, after its second season of run-timing forecasting, was 4.7% in 2000 compared to 5.5% in 1999. The high accuracy of season-wide performance in 2000 was largely due to exceptional Program RealTime performance in the last half of the season. Passage predictions from fifteen of the sixteen spring/summer yearling chinook salmon ESUs available for comparison improved in 2000 compared to 1999. The last-half average MAD over all the yearling chinook salmon ESUs was 4.3% in 2000, compared to 6.5% in 1999. Program RealTime 2000 first-half forecasting performance was slightly worse than that of 1999 (MAD = 4.5%), but still comparable to previous years with a MAD equal to 5.1%. Three yearling chinook ESUs showed moderately large (> 10%) MADs. These stocks had larger-than-average recapture percentages in 2000, producing over-predictions early in the season, in a dynamic reminiscent of migration year 1998 (Burgess et al., 1999). The passage distribution of the new stock of hatchery-reared sockeye salmon from Alturas Lake was well-predicted by Program RealTime, based on only two years of historical data (whole-season MAD = 4.3%). The two new run-of-the-river PIT-tagged stocks of wild yearling chinook salmon and steelhead trout were predicted with very good accuracy (whole-season MADs were 4.8% for steelhead trout and 1.7% for yearling chinook salmon), particularly during the last half of the outmigration. First-half steelhead predictions were among the season's worst (MAD = 10.8%), with over-predictions attributable to the largest passage on record of wild PIT-tagged steelhead trout to Lower Granite Dam. The results of RealTime predictions of passage percentiles of combined wild and hatchery-reared salmonids to Rock Island and McNary were mixed. Some of these passage-indexed runs-at-large were predicted with exceptional accuracy (whole-season MADs for coho salmon outmigrating to Rock Island Dam and McNary Dam were, respectively, 0.58% and 1.24%; for yearling chinook to McNary, 0.59%) while others were not forecast well at all (first-half MADs of sockeye salmon migrating to Rock Island and McNary Dams, respectively, were 19.25% and 12.78%). The worst performances for these mid- and mainstem-Columbia River runs-at-large were probably due to large hatchery release disturbing the smoothly accumulating percentages of normal fish passage. The RealTime project used a stock-specific method of upwardly adjusting PIT-tagged smolt counts at Lower Granite Dam. For chinook and sockeye salmon, the project continued using the 1999 formulation for spill-adjustment. For the new stock of wild PIT-tagged steelhead trout, a formula derived for steelhead trout only was used.