Enhancing flood risk system robustness in practice

Decisions about flood risk management are usually based on the reduction in flood risk compared to the implementation costs of the strategy. It is common practice to express flood risk (the combination of flood probabilities and potential flood damages) into a single number. The downside of this approach is that explicit information about how the system responds to the whole range of possible discharges is lacking. This type of information is relevant when a robust system is desired. We consider robust (fluvial) flood risk systems to have the ability to remain functioning under a range of river discharges. To understand system robustness, it is thus useful to analyze how the flood impact varies over a range of possible events. We applied a method to quantify the robustness of flood risk systems on a fluvial flood risk system in the Netherlands: the IJssel River, a branch of the Rhine River, and its flood-prone area. We compared four alternative system configurations of the IJssel River Valley, and analyzed their robustness, applying the following criteria: resistance threshold, response severity, proportionality and recovery threshold. The robustness analysis provides us with clues about how to enhance a flood risk system’s robustness. We found that a system with unbreachable embankments scores best on overall robustness, because it has the least uncertain resistance threshold, the most proportional damage increase with increasing discharge, and the least uncertainty about where and when floods will occur. This shows that a robustness perspective helps to develop strategies that reduce the flood risks without increasing the consequences of beyond-design floods. 1998). In addition, some authors propose taking into account worst-case scenarios (Merz et al. 2010) or start so-called ‘possibilistic thinking’ instead of ‘probabilistic thinking’ (Clark 2005). Thus, in addition to the traditional comparison of flood risk and costs, it is advocated to analyse ‘what if’ design conditions are exceeded. In literature on socio-ecological systems, the proposed way to deal with uncertainties is to aim for a robust or resilient system, instead of trying to control external disturbances. Among other things, control means that the variability of the system is reduced to make its behaviour better predictable: floods hardly happen. However, the downside of too much control is that unanticipated events may cause surprise and crisis (Holling 1996): when a flood does happen it will be a disaster. The idea of steering on system persistence (thereby allowing disturbances) instead of system stability was first introduced by Holling (1973) for ecosystem management, and later extended to the management of socio-ecological systems (Carpenter et al. 2001, Walker & Salt 2006, www.resalliance.org). This type of management is called ‘resilience’. In the field of flood risk management, however, the term resilience is associated with the ability to recover from the response to a disturbance (De Bruijn 2005), which is a narrower interpretation than that of the ecological and socioecological literature. To avoid confusion, we use the term system robustness for the ability to remain functioning under a range of possible disturbance magnitudes (see also Mens et al. 2011). In this paper, we analyse robustness of a socioeconomic system to river flood waves, by providing insight into the system response to a range of river flood waves, including the extreme ones. From such an analysis, not only the level of protection but also the potential consequences of all possible discharge waves, and the balance between them, will become clear. This ‘robustness perspective’ fits well in the recent policy developments on flood risk management in the Netherlands. In 2009, the Dutch government introduced what is referred to as a multi-layered approach to sustainable flood risk management, which states that three layers are required to manage flood risk in the long term: 1) flood defenses to protect against flooding, 2) spatial planning to limit the flood consequences, and 3) well-organized emergency management (National Government 2009). However, this multi-layered approach does not yet have any official status. The idea of using system robustness as decision criterion in addition to damage risk, fatality risk and costs, was already tried out in De Bruijn et al. (2008) and Klijn et al. (2012), but only in a qualitative manner. In this paper, we compare alternative system configurations for the IJssel River Valley (a branch of the Rhine River) in the Netherlands, based on quantified robustness criteria. We apply the method as introduced in Mens et al. (2011) and adjusted in Mens & Klijn (in prep.). The following criteria together provide an indication of system robustness (see Figure 1): Resistance threshold, or the smallest river discharge that will cause substantial economic