A bivariate trend analysis to investigate the effect of increasing urbanisation on flood characteristics

Flood frequency analyses are usually based on the assumption of stationarity, which might be unrealistic if changes in climate, land uses or urbanisation impact the study catchment. Moreover, most non-stationarity studies only focus on peak flows, ignoring other flood characteristics. In this study, the potential effect of increasing urbanisation on the bivariate relationship of peak flows and volumes is investigated in a case study in the northwest of England, consisting of an increasingly urbanised catchment and a nearby hydrologically and climatologically similar unchanged rural (control) catchment. The study is performed via Kendall's tau and copulas. Temporal trends are studied visually and by formal tests, considering variables individually and jointly. Bivariate joint return period curves associated with consecutive time periods are compared to understand the joint implications of such bivariate trends. Although no significant bivariate trends were detected, hydrologically relevant trends were found in the urbanised catchment.