An advantage of the model cascade is that it calculates annual expected impacts based on flood hazard simulations for a large number of return-periods. The rapid nature of the cascade also allowed us to provide the first sensitivity assessment of a global flood risk assessment tool.
We found the model cascade to be relatively insensitive to the extreme value distribution used to estimate the low frequency flood volumes. However, the results are highly sensitive to the assumed flood protection standard; we highlight the development of a database of such standards at the global scale to be a research priority. Also, the results forced by the more sophisticated WATCH climate forcing data show large differences to those forced by ERA-40 data, which are subjected to a more simple bias correction only. We envisage several main potential applications of the model. One is the identification of risk hotspots, which is important for planning disaster risk reduction efforts (e.g. UNISDR 2011). Secondly, data on annual expected losses at the macro-scale are vital for the insurance and re-insurance industries. Similarly, large companies have an interest in examining potential losses and business interruption across the globe. We are further developing the model for estimating the benefits (avoided costs) of measures designed to reduce flood risk, which would provide important information for assessing the costs of (climate change) adaptation (e.g. Ward et al 2010, World Bank 2010). A planned integration of the model and its improvements in the IMAGE model suite (version 3.0) will allow the analysis of trends in flood risks within the context of integrated global assessments. Whilst the model cascade is now operational, and being
implemented in the context of several projects by different users, we are continually working on further improvements, and invite scientists and stakeholders to collaborate on aspects such as: validation; improving the model cascade; and inter-model comparison.