On analysis and forecasting of surges on the west coast of Great Britain

Summary. Surges are analysed by the regression technique. On the west coast of Britain surges are strongly correlated with meteorological variables such as barometric pressure and winds. The north—south component of wind is more effective than the east‐west component in the generation of surges at Fish‐guard and Portpatrick. The surge elevations at these ports are linearly proportional to the wind velocity. This pattern changes near Liverpool where the east—west component of wind dominates the development of surges. Here the surges are related to the square of the wind velocity. Moreover, the north‐south component of wind over St George's Channel is marginally more effective than the north—south component of wind in the local area of the port, which suggests that a component of the surge near Liverpool is transported through St George's Channel. Spatial correlations of Fishguard, Holyhead and Liverpool surges also support this result. However, it is found that the forrnulation of a regression equation with simultaneous input of meteorological variables and spatial surges is ill‐conditioned. Non‐linearity resulting from surge‐tide interaction is negligible at Fishguard and Portpatrick. The development of non‐linearity starts when surges progress from Fishguard to Holy‐head. The non‐linear component is small near Holyhead but becomes a significant part of surges observed at Liverpool. The non‐linear component may be represented as a modulation of the tide by a slow time‐varying component of the surge. The component of surge variance which can be estimated by the regression model is over 70 per cent for Liverpool, 60 per cent for Holyhead and 50 per cent for Fishguard.