| Abstract/Description or Keywords |
Mature salmon returning to spawn in their natal streams are sensitive to both river flows and temperatures. Enhancements to existing forecast models result in significant reductions in the root mean square (RMS) forecast errors for both flow and temperature in the semi-weekly 10-day forecasts made during the salmon migration season. The Fraser watershed model is replaced by a statistical model that projects future flows using the latest observation as an initial condition and a slope consistent with the historic rate of change. This new method reduced RMS errors by as much as 38%. When the model flows were adjusted iteratively by feeding the flow error back into the system, the average RMS forecast flow error was reduced from 18.7% to 6.4% in 2000 and from 16.8% to 7.4% in 2002. The coupled temperature model combines atmospheric forcing with transport and velocities from the flow models. When the temperature model was run using the iterative feedback scheme, the RMS forecast error was reduced from 0.85 °C to 0.59 °C in 2000 and from 1.18 °C to 0.94 °C in 2002. Key words: Fraser River, temperature model, flow model, data assimilation, salmon. |
