The Salish Sea is home to Canada's largest port, the cities of Vancouver and Victoria border it and it hosts shellfish farms, whale watching, commercial, first nations and recreational fisheries. Predicting the 'weather' of the ocean provides oil spill drift tracks, storm surge warnings, low pH warnings for shellfish farmers and high current maps for marine pilots. SalishSeaCast is a real-time ocean prediction system for the Salish Sea including a fluid dynamics model, plankton biological model, carbon model and it drives a near-coast and wave models. SalishSeaCast gives "nowcasts" and predictions out to 36 hours in the future multiple times per day. In this presentation, we will discuss what is added to a fluid dynamics model to make a real-time prediction system, how we automate it and how we distribute the information. Two physical oceanography processes of the Salish Sea are particularly difficult to model. First, the strong mixing region through the Gulf and San Juan Islands which controls the estuarine exchange flow with the ocean and second, the Fraser River buoyant plume which defines the stratification of the Southern and Central Strait of Georgia. The first process is now well simulated through a combination of changes to code choices and boundary conditions. The second process is still a work in progress. We will present the sensitivity of the plume to turbulence parameters and bathymetry. Future plans for the project, including oil spill tracking and oil spill sensitivity will finish the presentation.