Edinburgh Napier University

  Despite a tendency for the complexity of physical–biological models to increase, simple coupled models remain useful for some applications and can provide insights into crucial links between physical and biological processes. This argument is illustrated with an account of a simple 3-box model intended to help assess the capacity of fjords to assimilate nutrients from fish farms. The model, a dynamic version of the UK " Comprehensive Studies Task Team " (CSTT) steady-state model for eutrophication, was applied to Loch Creran (Scottish Western Highlands) and was implemented using Stella 8 and tested using historical data from 1975 (before the installation of a salmon farm) and field data collected in 2003, during the period of operation of the farm. The model's biological state variables are chlorophyll, dissolved inorganic nitrogen (DIN) and dissolved inorganic phosphorus (DIP), and it includes a simple runoff model to convert rainfall into river discharge. The physical processes involved in exchange between the loch and the adjacent waters of the Firth of Lorne were parameterised as a constant daily exchange rate. Between 1975 and 2003, local inputs of nutrient increased but, despite this, there was little apparent increase in nutrient concentrations in the loch, and observed chlorophyll concentrations decreased substantially. Model simulations of chlorophyll and DIN agreed well with observations in 1975, as did DIN simulations in 2003. However, simulated chlorophyll was overestimated in 2003. Some of the agreement between observations and simulations come from the use of observed boundary conditions to force the model. However, even when boundary conditions are subtracted from simulations and observations, the simulations in most cases retain a significant correlation with observations, demonstrating that the model's 'interior' processes do add to its ability to replicate conditions in the loch.