Traditionally, flood management has focused on providing protection against floods up to a given return-period, but recently there has been a shift towards a more risk based approach, whereby flood risk is defined as the probability of flooding multiplied by the consequences. However, both the probability of flooding and the consequences are expected to change in the future as a result of changes in climate and socioeconomic factors.
To date, there is relatively little known internationally about how these will influence flood risk. Most studies examining this issue have considered a limited number of discrete climate change scenarios, representing single future pathways. One of the reasons that only a limited number of climate change scenarios are used in flood risk assessment is the large amount of computational time required to convert discharge into inundation maps using hydraulic models. Moreover, for the same reason, flood risk is usually estimated based on the damage resulting from a very limited number of flood return-periods (for example European states are only obliged to map flood extents for three return-periods).
Some of these issues will be addressed in the project Attention to Safety 2. Several aims of the project are to:
(b) develop a rapid inundation model to map inundation depths for large numbers of scenarios and return-periods;
(c) assess how flood risk estimates are influenced by using the probabilistic climate change scenarios.
In brief, the method consists of developing long time-series (3000-years) of climate parameters (temperature, precipitation), by statistical resampling from 30-year series, for simulations derived from seven Regional Climate Models (RCMs) and twelve General Circulation Models (GCMs). Each of these climate time-series (runs) will then be used as input to a hydrological model (HBV) to simulate daily discharge time-series of 3000-years length; for each run a statistical relationships will be derived between discharge and return-period. We will also develop a rapid inundation model to estimate flood extent and depth, based on stage-discharge relationships, and use this to produce flood maps for large numbers of return-periods (>50) for each run. These inundation maps will be used as input in a flood damage model (Damagescanner), together with land use maps and estimates of land value, to estimate damage for each return-period and for each run. Finally, risk curves will be developed for each run, in order to assess the difference in risk (annual average damage) between the different runs.
The research shows the effects of estimating risk based on damage estimates for large numbers of return-periods compared to the more common approach of using just a few return periods. It also demonstrates how the estimation of flood risk is affected by using climate data from a large number climate models.
Contact information: Dr Philip Ward (project leader)