Changes in seasonal compound floods in Vietnam revealed by a time-varying dependence structure of extreme rainfall and high surge

Compound floods due to intense rainfall and storm surges in coastal areas have shown an increasing trend in some parts of the world, and many studies suggested a strong link with climate change. Yet, such link has not been fully explored and quantitively assessed. In this paper, we demonstrate the development and application of a nonstationary framework to determining different compound scenarios, where individual drivers and their interactions have altered under climate change. The framework has been applied to one of the most flood-prone areas: the Ho Chi Minh City of Vietnam, to help analyze the present and future compound flood risks in both the dry and wet seasons driven by the joint effect from heavy inland rainfall and high skew surge. Over the period of 1980–2017, the two drivers are found to be significantly correlated in March and April, corresponding to the transition from dry-to-wet seasons. We also find that the commonly-used traditional multivariate statistical models underestimate the flood magnitudes for both the current (represented by 2020) and future (represented by 2050) scenarios, when compared with the results produced by the nonstationary methods. In addition, the results reveal that the dry season is expected to receive more floods triggered by the increased intensity and frequency of rainfall extremes, with the magnitude reaching a similar level to that of the wet season. This is in line with the climate projections under RCP4.5 and 8.5 scenarios although the duration of dry spells is expected to increase and the total annual rainfall to decrease in Vietnam. The simulated flood inundations indicate remarkable increases in flood magnitude and extension, especially at the locations identified as low risk by the stationary models.