Implementation issues for total risk calculation for groups of sites

For flood events in a large territory, different locations will have their own distinct probability distribution for flooding. However, if these locations are linked through shared meteorological and/or hydrological processes, then these probability distributions are likely to be correlated. This implies that simple summation of calculation results for flood risks or expected flood damage for the separate locations will give a distorted view of the risk or expected damage for the region as a whole. The question examined in this paper is the extension of the concept of a link between a once in N years event and the resulting damage from one dimension to multiple dimensions. Copulas are shown to represent this link.