1 Introduction

A range of medium- to long-term developments could influence the efficiency and cost-effectiveness of flood and drought risk management measures, for example in terms of the use of land. Some changes may lead to higher costs, e.g. building-over areas that could later have been more usefully deployed for adaptation measures, while other changes could lead to cost reductions, e.g. combining river widening with the replacement of sluices approaching the end of their life. Therefore, decisions on short-term measures should be taken in such a way as to avoid the unnecessary mounting of long-term costs, while agreements should be made on actions that could be linked efficiently (DP 2011).

Various planning approaches have been developed to take medium- to long-term developments into account in the assessment of short-term decisions. Relevant planning approaches (for flood and drought risk management) have been collated and reviewed by Walker et al. (2013). These include: robust decision-making (Lempert et al. 2003), adaptive policy making (Walker et al. 2001), adaptation tipping points (ATPs) (Kwadijk et al. 2010), adaptation pathways (Haasnoot et al. 2011), and dynamic adaptive policy pathways (Haasnoot et al. 2013). According to Walker et al. (2013), these approaches differ according to the level of uncertainty being addressed (from well-characterised to deep uncertainty) and the type of adaptation concept being used (from a static to a dynamic nature). Only few approaches (adaptive policy making, adaptation pathways, and dynamic adaptive policy pathways) explicitly consider dynamic adaptation of the long-term plan. That is, these approaches produce a long-term plan that is both robust (meaning that it performs satisfactorily under a wide variety of futures) and adaptive (meaning that it can be adapted to changing future conditions). Because of their dynamic nature, these approaches are particularly relevant for this paper. Their key characteristics are presented in Sect. 2, alongside the Adaptive Delta Management (ADM) approach (Isoard and Winograd 2013), being the focus of this paper. ADM has many similarities to adaptive policy making, adaptation pathways, and dynamic adaptive policy pathways, but it has been developed in a policy context. This policy concept (and associated planning approach) has been pioneered by the Thames Estuary 2100 project (EA 2009) and has been further elaborated by the Dutch Delta Programme (DP 2012). In the Delta Programme, various government authorities and social organisations collaborate on long-term plans for flood and drought risk management.

ADM is a transparent way of including and acting upon uncertainty around future changes in investment decision-making. It encourages an integrated approach to tasking (i.e. the tasks to be undertaken in the short, medium and long term) and minimises the regret with respect to maladaptation, which results from, e.g. over- or under-investment in flood and drought risk management. Key points of the ADM approach are (DP 2012):

  • Linking short-term decisions with long-term tasking around flood and drought risk management;

  • Incorporating flexibility in possible adaptation strategies (where effective);

  • Working with multiple adaptation strategies that can be used in an alternative pattern (i.e. adaptation pathways) (Haasnoot et al. 2011);

  • Linking different investment agendas.

Walker et al. (2013) observe that dynamic adaptation is increasingly being used in practice, including for flood risk management in the Thames Estuary in the UK (EA 2009) and flood and drought risk management in the Rhine–Meuse Delta in the Netherlands (DP 2014). However, they recommend that further work is done to bridge the gap between theory and practice. This paper contributes to bridging the theory-to-practice gap by presenting an in-depth account of how ADM has been applied for the Island of Dordrecht (the Netherlands). This is a dyke ring area in the Rhine Estuary–Drechtsteden area (Fig. 1). A dyke ring area is an area protected against floods by a single series of primary flood defences (dykes, dunes, hydraulic structures). The Island of Dordrecht contains various types of land use: residential, industrial and agricultural areas. Moreover, the governance structure is relatively simple: one regional water authority, one municipality and one safety authority are each responsible for one layer of multi-layered safety. These are, respectively: protection through dykes; prevention through spatial planning; and preparedness for emergency response. This makes it an attractive prototype example of a delta city threatened by floods. The case also has replication potential, because it is one of the (three) pilot studies of the Delta Programme regarding multi-layered safety. This concept has been introduced by the latest National Water Plan (VenW Verkeer en Waterstaat (VenW) 2009) to integrate spatial planning and emergency response into Dutch flood risk management, which traditionally has (and will maintain) a strong focus on protection. In this pilot study, the authorities (national, regional and local) are jointly exploring the opportunities and barriers for the implementation of multi-layered safety.

Fig. 1
figure 1

Location of the Island of Dordrecht (dyke ring 22 and unembanked areas) in the Rhine Estuary–Drechtsteden area

Full size image

ADM has been used to examine which measures and potential options are needed in Dordrecht and beyond (i.e. at the regional scale) for continued compliance with objectives for flood risk management. This paper provides an overview of the ADM approach and illustrates its application for the Island of Dordrecht. It also examines the added value and limitations of ADM, with a specific emphasis on the lessons learned from applying ADM in practice. This empirical evidence is expected to facilitate the use of ADM and similar planning approaches in practice and, ultimately, encourage sound decision-making on flood and drought risk management.

2 Adaptive Delta Management approach

The ADM approach aims to ensure that any short- to medium-term decision on flood and drought risk management is set within a framework that will not be maladaptive, if future developments (e.g. sea level rise) are different from what is currently predicted to be ‘the most probable’ (Reeder and Ranger 2011). It assumes a bandwidth of plausible future changes (i.e. scenarios) and subsequently systematically assesses the robustness and flexibility of proposed adaptation strategies. In this context, a key objective is to maximise robustness and/or flexibility to avoid later excess costs (e.g. increasing flood damages). Therefore, the ADM approach includes identifying ATPs, which are the points of reference (i.e. boundary conditions) where the objectives for flood and drought risk management are no longer met (Kwadijk et al. 2010). Combining the defined ATPs with climate and socio-economic scenarios provides information about the need for additional adaptation measures and the timing of these. The ATP analysis can, therefore, help to develop adaptation pathways. These refer to a sequence of measures and potential options which may be triggered before an ATP occurs. Adaptation pathways are commonly used in the context of long-term flood and drought risk management as they provide insight into the options, lock-ins as well as potential path dependencies and introduce the flexibility to adapt to a wide range of future changes.

Kwakkel et al. (2014) propose the use of Exploratory Modelling and Analysis (EMA) (Bankes 1993) and robust optimisation (Ben-Tal and Nemirovski 1998) for the design of adaptation pathways. Here, EMA is used to generate a large ensemble of plausible futures and to explore the performance of adaptation pathways over this ensemble. Robust optimisation facilitates the identification of a diverse set of promising adaptation pathways. The result of this approach is an adaptation map that shows the subset of promising adaptation pathways, together with the options for transferring from one Adaptation Pathway to another. In this approach, the decision-making process is focused on translating this adaptation map into a comprehensive plan of action. This includes deciding on which set of adaptation pathways to keep open, on its implications for short-term actions and on the establishment of a monitoring and evaluation system (Kwakkel et al. 2014). A potential drawback of the computational development of adaptation pathways is that decision-maker involvement is limited (to selected points in the process), as the policy analysts would carry out most of the steps. A further drawback is that it relies on fast hydrological models, since more complex models demand too much computing time for simulating the performance of adaptation pathways. This implies that some hydrological processes may be modelled less accurately and, as a result, uncertainties about the performance of options (e.g. estimated water stages) may be greater.

Another approach is to manually develop adaptation pathways, by exploring all the relevant policy options around an ATP, e.g. with a computational model. Each of these options can then be evaluated on its performance over changing future conditions. Here, more complex models can be used to obtain detailed information about the performance of options according to standards. However, when doing this some options may exclude others, and some sequences of options may be illogical (Haasnoot et al. 2013). In addition, other criteria for policy making, such as the severity of the impacts, the uncertainty involved and the desire to keep options open, could be used to develop a set of promising pathways. As such, the approach allows for more engagement with and input from decision-makers, for example on linking different investment agendas [i.e. project-level mainstreaming (Persson and Klein 2009)].

ADM sets out to manually develop adaptation pathways that will address the tasking for flood and drought risk management in a specific area. Starting with the current situation, a schematic overview charts out the short-term adaptation measures. It then looks at possible amendments of adaptive strategies in the medium to long term, including conditions under which it would be wise to shift to an alternative strategy. Finally, ADM focuses on possibilities for linking the implementation of strategies for flood risk management with other investment agendas (DP 2011).

There are four crucial steps when pursuing ADM (Bloemen 2013), and these are summarised below. Within the context of the Delta Programme, each step has been completed for the Island of Dordrecht. Three of the authors have been involved as participants in the Delta Programme, both at the regional/local level (Rhine Estuary–Drechtsteden and Island of Dordrecht) and at the national level (Staff Delta Commissioner). One author has been involved as an observer of the Delta Programme regarding the implementation of ADM.

2.1 Step 1: short term (current strategy)

  • Description of the current strategy, together with the reasoning process (ambitions, objectives and direction of action);

  • Translation of ambitions and objectives related to the current strategy into short-term tasking;

  • Identification of the measures that could contribute to meeting the short-term tasking;

2.2 Step 2: medium term (uncertainties, tipping points and alternative strategy)

  • Clarification of the key uncertainties from future changes influencing which tasks have to be undertaken in the medium term, such as climate change and socio-economic development;

  • Definition of the metrics (parameters) that describe the future developments, such as the magnitude of sea level rise in cm/year;

  • Modelling different future developments to provide insight into the occurrence of ATPs, where the current strategy will no longer be able to meet its objectives. Scenarios may be used to modify the boundary conditions (e.g. the magnitude of sea level rise) under which an ATP will occur into an estimate of when it is likely to occur. The time at which an ATP may occur defines the moment that a new strategy will be needed;

  • Development of the reasoning process (i.e. a vision on the preferred type of measures) related to the new strategy. This sub-step is optional and used to group measures into a coherent set that corresponds to a specific vision, philosophy or solution direction (e.g. protection from floods through dykes);

  • Identification of alternative measures and potential options that might alter the timing of ATPs and assessment of the efficiency (costs and effects, including the wider benefits) for the different scenarios;

  • Identification of the opportunities for mainstreaming adaptation with other planned investments in the area, such as the renewal of the urban fabric and infrastructure;

2.3 Step 3: long term (adaptation pathways)

  • Design of multiple adaptation pathways based on the efficiency of the individual measures and potential options (identified in the previous step). Adaptation pathways show the timing of implementation of measures and the points at which a decision should be made on the selection of potential options that can be called upon. These pathways also generate insight into the consequences of decisions in terms of lock-ins and possible options that remain open for the long term (Haasnoot et al. 2011). Opportunities for mainstreaming adaptation may be considered to adjust timing of implementation of measures (Gersonius et al. 2012b);

  • In addition, it is necessary to identify critical values (triggers) beyond which adjustments in the strategy will need to be made. Note that this is also relevant for the short and medium term;

2.4 Step 4: linking short term to long term (anticipatory actions)

  • Identification of anticipatory actions, which are necessary in the short term to enable the adaptation pathways. This comprises the initiation of:

    • Potential adjustments in legislation, rules and procedures;

    • Innovation and research;

    • Spatial reservations;

    • Securing of funds.

3 Case study: flood risk and resilience in Dordrecht

Surrounded by a series of rivers and canals—the Oude Maas, Beneden Merwede, Wantij, Nieuwe Merwede, Dordtse Kil—the city of Dordrecht is located on an island. Its population consists of around 120,000 inhabitants. Most residential, industrial and agricultural areas are located in a single polder area of about 7 ha, which is protected by a 37-km-long dyke ring (called dyke ring 22). Part of Dordrecht is situated outside the dykes: the unembanked areas. These areas are positioned at relatively high elevations, in addition to being protected by the Maeslant barrier and Hartel barrier. The historic port area is the lowest-lying unembanked area. The Island of Dordrecht lies in the transition zone between the tidal reach and the river regime reach, where the extreme water stages are influenced by both the high river run-off and storm surges from the sea. Together with the city of Rotterdam, a number of other municipalities and the surrounding industrial, agricultural and nature areas, Dordrecht is located in the Rhine Estuary–Drechtsteden area.

Using the ADM approach, a long-term plan has been developed for the management of flood risk and resilience in Dordrecht and beyond (i.e. at the regional scale). This plan comprises a set of measures that will be realised in the short term (up to and including 2028), but also looks ahead to the medium term (up to and beyond 2050).

3.1 Step 1: short term (current strategy)

Under the current strategy, flood risks are primarily managed through dyke improvements and by replacing the existing storm surge barriers. Without these barriers, the dykes in Dordrecht would need to be 0.5 m higher. Since 1961, protection standards for the dyke ring areas in the Netherlands have been established by national law as the average exceedance frequency of the design water level that the dykes must withstand (VenW 2010). In Dordrecht, the maximum acceptable overtopping probability has been set at 1/2000 per year (which is legally binding).

The short-term tasking (up to and including 2028) has been inventoried in the second and third Statutory Assessment of primary flood defences. These were based on the defined protection standard and the corresponding design water levels. Findings of the Statutory Assessment (Fig. 2) indicated that 28 % of the dykes in Dordrecht were below standards (due to, among other, changes in hydraulic peak conditions) and required improvement (PZH 2011). These dyke improvements are incorporated, or will be incorporated, into the implementation programme for flood protection measures (IenM 2011).

Fig. 2
figure 2

Short-term tasking (i.e. the required improvement tasks) in Dordrecht; ‘no judgement’ means that a more complex assessment of the performance of dykes according to standards is needed

Full size image

There are no legal protection standards for the unembanked areas. According to the National Water Plan (VenW 2009), the residents and users of these areas are responsible for taking consequence-reducing measures (where there is an unacceptable risk). This could include using elevated ground floor levels, dry proofing and wet proofing ground floors. Dry proofing may involve shielding, where the floodwater is kept out of the building by installing temporary barriers. Wet proofing, on the other hand, is based on the acceptance of water entering the building and involves using materials that will minimise the impact of floodwater on fabric and fixtures.

3.2 Step 2: medium term (uncertainties, tipping points and alternative strategy)

Uncertainty arises from a range of external factors, and these include future changes in sea level, river discharges and soil subsidence. These future changes increase the hydraulic load conditions and/or reduce the resistance and, therefore, generate additional tasks to be undertaken in the medium term (up to 2050). These tasks occur particularly in the transitional areas of the Rhine Estuary–Drechtsteden area. The Delta Scenarios ‘Steam’ and ‘Rest’ were used in analysing the tasking following from these changes. These scenarios combine the climate scenarios from the Royal Netherlands Meteorological Institute (KNMI-2006 scenarios)  (van den Hurk 2007) and the socio-economic scenarios from the collaborating planning bureaus (WLO-2006 scenarios). In the Steam scenario, rapid climate change (4 °C temperature rise in 2100) is combined with socio-economic growth (25 million inhabitants in 2100), whereas in the Rest scenario, moderate climate change (2 °C temperature rise in 2100) is combined with socio-economic squeeze (12 million inhabitants in 2100).

In addition to these external factors, there are also internal influences on the tasking. At the launch of the national Delta Programme, it was found that population and economic values in the area have increased considerably since the first Delta Committee (which was set up after the floods of 1953) issued its recommendations in 1961, on which the current protection standards are largely based. To this end, the Delta Programme sub-programme ‘Safety’ has examined whether the current levels of protection (protection standards) are still sufficient, using a risk-based approach. This analysis has shown that the Island of Dordrecht, among other parts of the Rhine Estuary–Drechtsteden area, has to increase its level of protection (these are ‘areas of attention’). The Delta Programme 2015 (DP 2014) has presented a proposal for updating the protection standards, as part of the Delta Decisions. The new protection standards are based on the following goals: everyone living behind a dyke may expect a tolerable risk level of at least 1 in a 100,000 per year. This means that the probability of dying as a result of a flood for any individual should be no greater than 0.001 % per year. Moreover, additional protection is offered in areas where there may be: potentially large groups of victims; major economic damage; and/or major damage as a result of the failure of vital and vulnerable infrastructure of national importance. In the proposal of the sub-programme Safety, a flood probability of 1/10,000 per year has been established for the northern dyke segment of Dordrecht and 1/3000 per year for the southern dyke segment of Dordrecht.

Figure 3 shows the future tasking following from both external factors (sea level rise, shifts in river discharges and soil subsidence up to 2050) and internal influences (updating of the protection standards).

Fig. 3
figure 3

Medium-term tasking (i.e. the required improvements tasks) in Dordrecht following from both external factors and internal influences

Full size image

The Island of Dordrecht is faced with a significant task up to 2050 to improve the dykes that will likely lead to problematic spatial claims (e.g. when the required space for dyke improvements is not available in urban areas). This underscores the importance of a comprehensive vision on and strategy for flood risk management. The question then arises as to whether the current strategy (of dyke improvements) is adequate to meet the future tasking in an efficient and cost-effective manner, or whether there needs to be a shift in strategy. Two aspects of the tasking potentially call for a shift in strategy:

  • Further improvement of the dyke along the Voorstraat is complex, because this dyke is covered by a large number of historic buildings of monumental importance. Traditional heightening will lead to very high costs (about 550 million Euros) and has major spatial consequences (e.g. potential construction damages to historic buildings).

  • A desire to minimise investments in the improvement of the southern dyke segment that might subsequently be found to be not the best, if a decision is made in the future to make more room for the river in the south.

As such, an alternative strategy has been developed for Dordrecht that addresses all three layers of multi-layered safety. In this strategy, prevention and preparedness measures (layers 2 and 3) are combined with protection measures (layer 1) in order to achieve the required level of protection (Fig. 4). The reasoning process and measures related to the alternative strategy are summarised below.

Fig. 4
figure 4

Multi-layered safety for the Island of Dordrecht (courtesy of DeUrbanisten)

Full size image

By transforming the northern dyke segment into a strong dyke (i.e. a flood probability of 1/10,000 or lower), Dordrecht can be safer than with an economically optimal standard for the entire dyke ring, for about the same cost. This targeted measure in layer 1 is sufficient to meet the tolerable risk level (chance of fatalities is no more than 1/100,000) and reduces the risk of large groups of victims to virtually zero. Economic damages and casualties due to a dyke breach can be further reduced by using regional defences to compartmentalise the Island of Dordrecht into safety zones (layer 2). These safety zones may be used for preventive evacuation on the island itself and/or to speed up recovery (layer 3). This also calls for thorough preparation, e.g. by robust design of critical infrastructure networks (layer 2) and improved risk and crisis communication (layer 3).

As part of the strategy development, it has been necessary to consider how the implementation of the measures for the short and medium term can be combined, or linked with other spatial and/or economic developments and ambitions in the area, a process termed ‘mainstreaming’. The following opportunities for mainstreaming adaptation have been identified:

  • The Regional Water Authority (Hollandse Delta) has made an extra investment in safety with the regular improvement of the dyke section Kop van ‘t Land (north-eastern dyke segment), as part of the Second Flood Protection Programme. This excess safety (over and above the required safety level by the current protection standards) is sufficient to meet the updated protection standards, not only now, but also up to 2050. In this way, it has been possible to limit the medium-term tasking.

  • A possible alternative to the traditional heightening of the dyke along the Voorstraat is to build a new flood defence along the River Oude Maas to replace the existing dyke (Hinborch 2010). This new flood defence will consist of temporary (moveable) flood barriers in the quays. Because this barrier will not be visible under normal circumstances, it will have little impact on the historical town centre. The barrier will provide considerable synergistic benefits (i.e. reduced flood damages) by protecting the unembanked part of the historical town centre against flooding, requiring only slightly higher investment costs. As such, it provides the opportunity to link the tasking for the unembanked areas with the required dyke improvements for the Voorstraat (up to 2050).

  • The Ecoshape consortium, which is a collaborative body comprising the business community, knowledge institutions and government bodies, has conducted a study into the potential of green adaptation measures (building with nature) in Dordrecht. These measures could be implemented as alternative measures to address the safety tasking (as opposed to traditional dyke heightening), such as for the maintenance and management of the foreland (Koch et al. 2009). A willow tree forest in front of the dyke at Kop van ‘t Land can provide safety through wave reduction. This means that the dyke can be lower than in a traditional design to prevent overtopping. Furthermore, green adaptation measures can deliver added value through a combination of nature development and safety (e.g. Fig. 5).

    Fig. 5
    figure 5

    Impression of green adaptation measures for the north/south compartmentalisation dyke, left under normal circumstances and right in case of a dyke breach in the south. In this example, the nature development along the dyke has been designed to break the flood waves, in addition to enriching the landscape (courtesy of D.EFAC.TO)

    Full size image

3.3 Step 3: Long term (adaptation pathways)

In accordance with the ADM approach, an exploration of the adaptation pathways has been carried out for flood risk and resilience in Dordrecht. This involved specifying what is required to keep options open for the long term (up to 2100), so as to be able to shift from the current strategy to another in the future. The options that need to be kept open followed from the alternative strategy determined for the future (i.e. the multi-layered safety strategy). Based on this exploration, it appears to be possible to take no-regret decisions for the short term. An example of a no-regret decision is using the regional defences to compartmentalise the dyke ring area. In this way, it is possible to lower the level of protection for the southern dyke segment by a factor of between 3 and 10 as a result of the reduced flood consequences with compartmentalisation. This short-term decision allows postponement of the decision point for improving the southern dyke segment until after 2050. In this way, the decision to shift from dyke improvement to making more room for the river (as potentially required to manage upstream flooding) could be kept open for the long term.

The various adaptation pathways for the management of flood risk in Dordrecht are given in Fig. 6. These paths provide insight into which adaptation measures need to be taken when and how long-term tasking impacts upon short-term decisions.

Fig. 6
figure 6

Adaptation measures and pathways for flood risk; short term is the period up to and including 2028, medium term is the period from 2028 up to 2050, and long term is the period from 2050 up to 2100

Full size image

3.4 Step 4: linking short term to long term (anticipative actions)

It is not yet known how agreements on measures in layers 2 and 3 of multi-layered safety can be laid down. If the protection levels are going to be based in part on measures in layers 2 and 3, then it is essential to guarantee that they will actually provide the necessary protection (DP 2014). To this end, the central government and the regional partners (regional water authority, province, municipality and safety region) have initiated a Multi-year Programme for Infrastructure, Spatial Planning and Transport (MIRT) study into the added value of and the possibilities for (safeguarding) measures in layer 2 and 3. The MIRT study will also look for a better connection between flood risk management and spatial planning and further develop the opportunities for mainstreaming adaptation and for spatial or economic development.

4 Discussion

4.1 Added value of ADM

The ADM approach has been applied in Dordrecht to develop robust and flexible flood risk management strategies. Using the ATP approach, it was evaluated as to what extent the proposed measures within the various strategies are able to perform their objectives in the short, medium and longer term. The performance of all measures was assessed for the most extreme scenarios that were deemed plausible (i.e. the ‘Rest’ and ‘Steam’ scenarios). In addition, flexibility was ensured by the development of adaptation pathways that can be accelerated, slowed down or adjusted through exploiting options that were previously kept open (as shown in Fig. 6). It should be noted that not only the technical aspects (i.e. efficiency and cost-effectiveness) of the measures were assessed during the process of strategy development, but also their legitimacy and social feasibility. This led, for example, to the development of alternatives to dyke strengthening, such as temporary flood barriers and green adaptation measures. It can, therefore, be argued that as much as possible is being done to avoid investments in measures that may be regretted on the long term. As such, the approach of ADM has provided a means for adjusting short-term decisions in the context of long-term tasks for flood risk management.

Moreover, the approach of ADM was instrumental in tailoring the strategy to the local context of the Island of Dordrecht. Particularly during the second step of ADM, the spatial and economic characteristics of the area were explicitly considered in the development of the flood risk management strategy. To this end, an integral approach was taken to include the perspectives of all key stakeholder groups and to identify and take advantage of possibilities for linking different investment agendas. In addition, spatial–economic perspectives have been developed in order to create an optimal fit of the flood risk management strategy with the spatial and economic characteristics of the area. These are: strong, urbanised dykes in the north and robust sea clay islands in the south. As a result, it is expected that ADM will in the long term provide more benefits at lower cost and thus greater value for money, than a more traditional approach.

4.2 Limitations of ADM

The application of ADM in Dordrecht showed that it involves a time-consuming process that requires detailed knowledge about the flood risk management system. Moreover, it should be realised that success is not guaranteed, because the effectiveness of proposed strategies can only be measured after the future has unfolded.

It appears that keeping options open may often be logical in theory, but it can be difficult in the political reality of decision-makers to establish and maintain broad political support for doing so. Following economic theories about real options and adaptive policy making, thinking in options can result in optimised cost-effectiveness over the long term (Gersonius et al. 2012a; Walker et al. 2001). However, previous research has shown that a focus of decision-makers on the status quo, combined with risk and uncertainty aversion can be a barrier for the development and appraisal of adaptive approaches as this can negatively influence the willingness to invest in keeping options open (Defra 2009, Sayers et al. 2012). According to insiders in various sub-programmes of the Delta Programme, there has been political resistance against keeping options open and taking measures that are needed for keeping options open. Several explanations were suggested: (1) it is often unclear to decision-makers whether the options are realistic and whether they will ever be realised; (2) the spatial claims of keeping options open put potential constraints on regional developments; and (3) most decision-makers have the desire and habit to take decisions for visible short-term action rather than for inaction in the short term directed to facilitating future actions. These explanations suggest that clarity about the applicability, added value and constraints is desired as early as possible. From this perspective, it would be worthwhile to consider options as provisional decisions for taking action which can be postponed or cancelled under specific conditions.

In addition, it was identified that the ATP Approach, which is one of the key elements of ADM, has its limitations. In particular, it appeared difficult to apply this approach to river flooding. Based on the experiences of project managers within the Delta Programme, a recent evaluation of the implementation of ADM in the Delta Programme provided several explanations that limit the identification of ATPs (Rijke 2014): (1) tipping points do not occur if strategies are inherently flexible (e.g. in case of sand nourishment—this strategy could theoretically be extended indefinitely because of the abundant availability of sand and ongoing natural processes of sediment transport along the coast line); (2) tipping points cannot be projected if climate signals are weak in comparison with the variability of pressures on the system (e.g. in the Netherlands this is the case for river discharges); and (3) in the absence of binding safety norms, it is not straightforward to determine threshold values (e.g. in the case of unembanked areas). Not surprisingly, the same study identified that responsible project managers in the Delta Programme considered the identification of tipping points as one of the most challenging aspects of ADM (Rijke 2014).

5 Conclusions

It is increasingly being recognised that contemporary flood risk management calls for strategic approaches that take the dynamic nature of the flooding system into account as future conditions may change from those that currently exist. These innovative approaches seek to embed resilience and adaptive capacity within the choices made, requiring a continuous process of adaptation planning requiring monitoring and adjustment. They aim to ensure that adaptation decisions made today are robust under a changing and uncertain future and that the right decisions are taken at the right time. Based on these principles, ADM has been developed to provide a basic framework for the Dutch Delta Programme, in which various authorities and other organisations collaborate to develop a national strategy for flood and drought risk management. This paper deals with its application in the pilot island of Dordrecht, where it has been applied in a regional context. As methods and tools to value the effectiveness of ADM are still lacking, this pilot is considered as a pioneering example to systematically analyse the performance of current and proposed strategies in Dordrecht in the short, medium and long term and the implication of long-term changes for short- and medium-term decisions.

For the context of Dordrecht, ADM has been demonstrated to be instrumental for decision-makers to help to understand the suite of options open to them and how they can be combined into ‘decision pathways’ that create a portfolio of measures for the coming century. Moreover, ADM has provided guidance for the Dordrecht pilot to explore and identify investment decisions that need to be taken in the short term. It has become clear that this innovative approach requires a continuous effort of monitoring, evaluation and adjustment which in turn calls for collaboration between different stakeholders and joint decision-making and thus contributes to the development of new governance and institutional arrangements in the domain of flood risk management. For example, this pilot has evidenced that a profound change in the way long-term investments are identified and planned will be required. It has also shown its potential to further strengthen the conceptual relevance, legitimacy and feasibility of the resulting flood risk strategy at a regional/local and national level.

The application of ADM, however, needs further exploration to better understand how best it could be applied in practice and to assess the effectiveness of measures. This holds true both for non-structural measures such as planning and emergency management as well as for structural measures such as hard infrastructure where engineered solutions are likely to play a role in the approach of ADM. Related to the latter, it still needs to be seen whether such an adaptive approach can be effectively applied into practice as within the time scales of decision-making detection of climate change signals is a difficult task. For tipping points related to slowly changing conditions, it may work, but for those conditions that are affected by extreme events such an approach runs the risk of being too late. The process in Dordrecht also indicates that ADM has the potential to help to merge various investment agendas and look for opportunities for mainstreaming with planned investments. It needs to be seen whether the prospects of including multiple functions via, e.g. mainstreaming will be feasible in the longer term.