Influence of wind field and open boundary input upon computed negative surges in the Irish Sea

Abstract

A hydrodynamic model of the west coast of Britain (extending from about 57°N to 48°N and 10°W to −4°E), employing a range of open boundary conditions, is used to simulate the wind forced response during the period 25 March to 5 April 1994 during which there are positive and negative surges. Strong uniform winds over the region during 30–31 March have the potential to generate a positive surge, and the passage of a depression on 1 April has the potential to produce a negative surge. A simulation using local wind forcing (namely, that in the region covered by the model) with an elevation-specified boundary condition does give a negative surge on 1 April, although this is not actually observed due to “far-field” effects which prevent a flow out of the Irish Sea, and hence the associated negative surge. Calculations with an elevation-specified open boundary condition that could include “far-field” effects (primarily produced by shelf wide winds, and introduced into the region by specifying the observed surge elevation along the open boundary) show that the observed positive surge on 30–31 March is accurately reproduced but that the model still gives a negative surge on 1 April. The application of a radiation boundary condition is equally successful in reproducing the positive surge, but in addition it provides a “far-field” input on 1 April which prevents the outflow from the region and gives elevations in good agreement with observations. It is concluded that the use of either an elevation-specified or radiation boundary condition in a west coast model is equally successful at times of positive surges (which supports previous work). However, new conclusions are being made concerning negative surges. These are in essence, that under conditions in which the local wind field would produce a negative surge, but “far-field” inflow into the west coast region prevents this, an elevation-specified open boundary is not appropriate. In this situation, a radiation condition with input from a shelf wide model is most appropriate.