Characteristics and mechanisms of flood evolution under climate change

Abstract

Flood is among the most frequent natural disasters and is of great concern worldwide. However, it is unclear how flood would evolve under climate change and whether flood changes would follow the path of extreme precipitation. Therefore, the characteristics of flood evolution under climate change and the mechanisms behind deserve more exploration.

Chapter 2 of the thesis investigated whether the opposing conclusions in existing studies about flood and extreme precipitation sensitivity to warming climate are caused by anthropogenic activities, mainly urbanization. The catchments with different imperviousness in the United States were considered. The results show that although higher catchment imperviousness can increase the connection between flood and extreme precipitation, flood sensitivity to temperature and the sensitivity difference between flood and extreme precipitation did not change significantly with imperviousness. This indicates that the influences of anthropogenic activities on flood scaling are limited compared to local hydrometeorological features.

Chapter 3 focused on the hydrometeorological issues in determining the sensitivity of flood to warming climate. A large number of natural catchments (imperviousness below 5%) across the Continental United States (CONUS) were studied. Contrasting sensitivity patterns were found. This indicates that the sensitivity of flood to warming climate can be either similar, higher or lower than that of extreme precipitation. Generally, similar sensitivity was found in regions where the timing of extreme precipitation and flood is correspondent, and with similar temporal evolution in extreme event timing and magnitude. In these regions, floods are dominantly caused by extreme precipitation. Therefore, the contrasting sensitivity of flood and extreme precipitation is determined by different local flood generating mechanisms.

Chapter 4 quantified the relative contributions of different flood generating mechanisms to flood magnitudes across CONUS. Four hydrometeorological flood drivers were considered comprehensively, depending on whether soil moisture condition or snowmelt influences flood generation. Nonlinear machine learning models were used to quantify the relative contributions of these drivers to floods. Unlike previous studies, it was found that snowmelt/rain on snow upon saturated soil was the largest contributor to the floods in the midwest and northeastern CONUS. This finding highlights that both snowmelt and soil moisture conditions are key issues in the midwest and northeastern CONUS, instead of only snowmelt as has been reported previously.

Chapter 5 further investigated flood generating mechanisms in CONUS in terms of the seasonality of flood and the four hydrometeorological flood drivers. A clustering approach based on the seasonality of flood and hydrometeorological extremes was developed to classify the catchments in CONUS. The clustering approach clearly revealed the regional differences in flood relationships with the four hydrometeorological flood drivers, which contributed to the understanding of flood generation across CONUS.

Overall, this thesis systematically explored the characteristics and mechanisms behind flood evolution under climate change. It reveals the potential reasons for the debate in existing studies about the sensitivity of flood and extreme precipitation to climate warming. The approaches to flood generating mechanism studies in this thesis are also applicable to flood exploration in other regions, so as to unveil flood generating mechanisms worldwide.