Modelling extreme rainfall and tidal anomaly

Abstract

Tides consist of two major components, astronomical tide and tidal anomaly. Astronomical tide is well understood and can be predicted, quite precisely, well in advance. However tidal anomaly cannot be predicted with such certainty for even short lead times. Tidal anomaly is largely dependent on prevailing atmospheric conditions, and as such can be greatly affected by local storms. A consequence of the dependence between tidal anomaly and nearby extreme storms is that any assumption of independence between tidal water levels and rainfall is incorrect. This is of particular importance in coastal catchments and estuarine systems, where high tail-water levels can increase the severity of any fluvial flooding. Ignoring the relationship between extreme tidal levels and extreme rainfall can lead to serious underestimation of the flood risk in coastal catchments. This paper examines the relationship between extreme rainfall and tidal anomaly, which is not necessarily characterised by simple correlation, and presents a method for modelling the joint distribution. A method to simulate stochastically similar data-series, suitable for use either in a continuous simulation model or in an event-based model, is developed.