Document Type


Publication Date



Columbia River (Or. And Wash.) -- Tides -- Analysis, Plumes (Fluid dynamics) -- Mathematical models, Marine ecology


Semi-operational daily forecasts of circulation from an observatory for the Columbia River estuary-plume-shelf system routinely support oceanographic cruises, by providing 24h-ahead estimates of plume location and structure for planning purposes and for near real-time interpretation of observations. This paper analyzes forecast skill during spring/summer cruises in 2004-2007. Assessment addresses both qualitative descriptions of major plume trends and features and quantitative representation of data from (primarily) vessel-based flow-through, cast and towed systems. Forecasts emerge as robust predictors of plume location and variability, with skill that has grown over time, at least in part due to improvements in model algorithms. When the same version of SELFE is used as the common computational engine, forecast skill is very comparable to the skill of multi-year simulation databases, which are computed retrospectively. As a measure of forecast skill, 55% of the predictions of surface salinities came within 2 km of observations, and directional shifts in response to coastal winds were well predicted. Quantitative skills for other aspects of the plume structure vary. Skill is highest for flow-through (and near-surface TRIAXUS) salinities. Forecasts also capture aspects of the vertical structure of the plume as represented by CTD casts (undulating TRIAXUS). Sub-tidal velocities, as compared against fixed station data, are least well described among examined variables. Overall, circulation forecasts help level the playing field among chief scientists with diverse disciplinary expertise. The chief scientist?s expertise on plume physics determines whether forecasts should be used for training and land-assisted planning, or as a sophisticated en route planning and interpretation tool. Effective interpretation of vessel observations in context of forecasts requires understanding of physical processes and modeling limitations.


This is the Author's Original Manuscript of a work that was submitted to Journal of Geophysical Research on July 11, 2008.

Persistent Identifier