The estimation of flood frequency curves by mapping from rainfall frequency curves

Abstract

Recent large flooding events have reinforced the need for prudent flood risk management. The July 2007 floods in Yorkshire and the Midlands and the November 2009 floods in the Lake District have highlighted the current vulnerability of key infrastructure and the built environment in the UK to flooding. This existing flood risk is coupled with concerns over the potential impacts of future climate change on flood regimes. Therefore, there is a need to develop tools and methodologies to assess the potential impact of likely climate change on flood risk. The link between large rainfall and flow events is first examined, as well as an assessment of the seasonality of these events. This reveals a distinct east-west split in the seasonal concentration of flooding. This work provides a basis for the development of a statistical modelling technique which estimates a catchment flood record on an event basis. The model uses estimates of the flood generating storm and the antecedent conditions to estimate a flow magnitude. The modelled flood record is then transformed into a flood frequency curve using an appropriate statistical method. Extensive testing of the model has assessed its robustness to the length of flood record used in fitting and its sensitivity to the input climate data. Several case studies using the UKCP weather generator show how the method works as well as providing an indication of how future climate changes may affect the flood frequency curve. The frequency curve mapping method developed here performs best on catchments whose flood regime is driven by rainfall. The use of a simple antecedent rainfall accounting method has been shown to perform as well as a quasi-physical soil moisture estimation method. The research undertaken offers several possibilities to develop understanding of flood frequency curves in catchments with short gauged records. This new methodology has the potential for further development and can be used to explore a wide range of future scenarios.