Event Detail

Event Type: 
Applied Mathematics and Computation Seminar
Friday, October 26, 2012 - 05:00
GLK 113

Speaker Info

OSU Statistics

Part of The Oregon Plan for Salmon and Watersheds requires the Oregon Department of Fish and Wildlife to monitor survival and migration of salmonid fishes in coastal basins. The work described in this talk supports that effort by examining estimates of smolt (juvenile migrating fish) survival under several different migration (run) distributions, survival probability schemes and sampling schemes. We simulate migration numbers using Poisson distributions, and we add random noise to several systematic survival probability models to account for daily fluctuations in survival probability. We consider three sampling schemes to compare the effectiveness of our survival estimates. All estimates are evaluated by examining estimation error (the difference between the estimate and the known survival probability) and confidence interval coverage—we expect this coverage to be 95% if the estimation procedure is valid.  A systematic sampling scheme appears to perform best, though there is enough uncertainty in both the shape and timing of the run distributions and in the survival probabilities to make estimates from all sampling schemes less than optimal.

This is joint work with Jeremy Romer and Carl Schreck.