Setting design inflows to hydrodynamic flood models using a dependence model

In setting design inflows to hydrodynamic models of flood flow along rivers, there can be a conflict between site-specific hydrological estimates of flow for a given return period and what the river model calculates as it routes flood hydrographs. This paper describes research carried out as part of the Flood Studies Update programme in Ireland, aimed at developing guidance on how to divide up river models and set the magnitude and timing of their inflows so that conditions in the model reach correspond to the expected design flood return period. A model for the joint distribution of flood peaks at pairs of catchments has been developed. The relationship between flood return periods is linked to physical differences between catchments. The model thus allows estimation of the statistical distribution of the flood return period expected at one site during a flood of specified return period elsewhere. A separate regression model predicts the relative timings of flood peaks on pairs of rivers. A summary of the resulting practitioner guidance is given, along with an overview of the testing of the method. The paper concludes with a discussion of the potential for application of the spatial dependence model to other problems in hydrology.