An investigation on the non-stationarity of flood frequency across the UK

Water infrastructure design and flood mitigation projects are currently based on a so-called stationary assumption. However, this assumption has been challenged in recent years due to increased extreme weather, potentially leading to the underestimation of flood quantiles and an increased risk of structural failures. In the present study, peak flow series from 158 gauging stations in the UK are analysed using both stationary and non-stationary models based on the generalised additive models for location, scale and shape (GAMLSS) framework. Nine candidate covariates are used for non-stationary models to obtain information on the significant drivers of non-stationarity in each flood series. Very few stations are thus found to show temporal trends, which suggests that time-varying non-stationary flood models may not always be suitable, and the use of physically-based covariates can provide better-fitting models. In particular, the non-stationary models using rainfall-related covariates display the best performance for the vast majority of stations, thus suggesting that the variability of the rainfall regime remains the dominant driver for changes in flooding. Moreover, two atmospheric circulation patterns, namely the North Atlantic Oscillation (NAO) and the East Atlantic Pattern, were found to be closely related to the non-stationarity of flooding at a number of stations. Unlike rainfall, however, the influence of these climate indices is not particularly dominant; hence, they are more suitably employed as components of a multiple-covariate non-stationary model. Besides, a direct linkage between the variability of flooding and climate changes has not been identified either regionally or globally. For the majority of stations (70%), the use of multiple covariates can provide a better non-stationary model than the use of a single covariate.