Causes of dependence between extreme floods

Abstract

Compound events, like compound floods, have rapidly aroused interest due to the strong impacts associated with them. The spatial dependence has a fundamental role in the dynamics of these events, and causative investigations of their origins could contribute to elucidate their dynamics. Here, we addressed the pairwise spatial dependence between annual maximum (instantaneous) discharges occurring in river stations located in the United Kingdom. First, we tested the hypothesis that the dependence comes from the co-occurrence of annual maxima using Kendall's tau measure of association and its conditional version, calculated from the non-co-occurrent values. This hypothesis, commonly accepted in literature, would attribute to the co-occurrence of the origin of the spatial dependence between extreme floods. The analysis showed how there is also dependence between annual maxima pertaining to catchments located very far from one another, and where the co-occurrence of annual maxima is small, if not zero. We formulated a general hypothesis to explain the spatial dependence between annual maxima: dependence is the compound result of co-occurrences, and climatological and hydrological similarities. The origin of dependence is more complex than what is presently stated in the literature. Thus, not only is synchronization a cause, but similarities in climate and hydrological response may also play a role. We introduced three dissimilarity indices and dependence-dissimilarity maps to illustrate this general hypothesis.