4 The (mis) Match Between the Information Needs and Knowledge Availability
Political systems are caught in four to five year democratic cycles, while future climatic impacts are calculated for time scales that are much longer. In Table
1 it is shown that most studies focus on at least the year 2050. Policymakers are more interested in changes for the next couple of years, or what these changes mean for decisions they have to make on a short timescale. This is not true for all policymakers, as there are policymakers who are not chosen every four or five years and law and legislation are designed for longer term. Despite this, earlier studies showed that climate change is generally not seen as most important in the short term (Arnell and Delaney
2006; Ivey et al.
2004). Other political priorities dominate and it is easier to make decisions on issues that have a short time span. Furthermore, the short term socio-economic factors determining adaptive capacity are at least as important for vulnerability as climatic changes. Temporal mismatches occur when the short term temporal scale of policy makers and the long term temporal scale of the climate processes do not align (Cumming et al.
2006). Furthermore, Table
1 shows that the spatial resolution of RCMs of the studies has a maximum of 50 km. The spatial uncertainty of grid cells can be decisive for hydrological analysis of the river basin, making it difficult to make judgments on regional levels (ICPR
2009a). This also indicates that this low resolution does not always match the territorial boundaries of policymakers. The output of the hydrological model is usually a projected discharge for a specific location, like, for example, Lobith, the place where the Rhine enters the Netherlands. Local policymakers may need much more specific information. Temporal and spatial scaling complicate effective knowledge sharing between climate science and policy. This is further complicated by the fact that adding more spatial and temporal detail, often also adds more uncertainty (Alkhaled et al.
2007). Therefore, the choice of level and type of detail included in risk assessments should be driven by both scientific experts and policy makers, but this is often not the case.
Next to scaling and temporal issues, the representation of uncertainty for guiding decision-making faces a number of challenges. First, most studies quantify only a limited number of the types of uncertainties that have been mentioned in the previous section, often the total uncertainty is not clearly represented. Lack of transparency regarding the assumptions and uncertainties can lead to misunderstandings in the science-policy interface on the nature of the knowledge (Sluijs van der and von Krauss
2005). Second, the communication and representation of uncertainties is under a lot of debate. For example, the UK is the first country to present climate change projections for policy applications in a probabilistic framework (UKCP09) (Murphy et al.
2010). Some scientists are against this way of presenting uncertainties, as there are important limitations to our ability to project future climate conditions for adaptation decision-making (Hall
2007): uncertainties can only be quantified to a certain extent. Others find it is essential that GCM projections are accompanied by quantitative estimates of the associated probability (Giorgi
2005; Murphy et al.
2004; Wigley et al.
2003). Adding to this debate, Gawith et al. (
2009) explain that the experience with UKCP09 has taught that the provision of probabilistic climate scenarios must be accompanied by ongoing guidance and support. Another lesson from UKCP was that ongoing dialogue between those providing the scenarios and the communities using them is essential. Both lessons were motivated by the experiences from the UKCP02 program, which showed that users frequently chose the Medium-High climate change scenario, because it had the most detailed information and it was seen by some as presenting a ‘middle road’ or a ‘safe’ choice. It was also less resource intensive than having to apply four scenarios (Gawith et al.
2009). This experience and debate demonstrates that there is still much to be researched in communicating climate uncertainties and that interaction between scientists and policymakers is fundamental to constructively meet the challenges associated with climate change projections. Standard methodologies to include uncertainties in potential changes and assess their impact on projected estimates have yet to be developed (Prudhomme and Davies
2009). There remains a question as to whether it is possible to develop such a generic method that will fit all situations. Until then, the debate about how to present and how to manage uncertainties can be confusing and may make it more difficult for policymakers to formulate adaptation strategies on the basis of available scientific knowledge.
6 Dutch Case: Evolution of Design Discharge
Important policy variables in river basin management are politically agreed safety levels and design discharges derived from scientific analyses. Safety levels refer to the frequency of flood events that is considered to be acceptable. The amount of water per second that can be associated with these safety levels and which statistically has a certain probability to occur (‘design discharge’) is used to design adaptation or flood protection measures, e.g. to determine the necessary height of a river dike. Both safety level and design discharge differ between countries and vary over time as scientific insights and political priorities evolve.
Table
3 shows different safety levels and corresponding design discharges for Germany and the Netherlands. The safety levels in the Netherlands are up to tenfold higher than in Germany. The Dutch norms are legally binding at the national level, while the German norm can differ between Länder, depending on historic water levels and local initiatives (Steenhuisen et al.
2006).
The estimation of the probability of an extreme event, that corresponds to a high safety level is far from trivial (Te Linde et al.
2010). Safety levels for the Rhine are relatively high and with only 110 years of observed discharge data available, statistical extrapolation leads to very high uncertainties (Klemes
2000). For recent applications, more sophisticated approaches have been developed that combine weather generators with hydrological models (Buishand and Brandsma
2001), to create such long discharge series that extrapolation is redundant. However, this approach is also under debate, as it requires hydrological modelling of extreme events, far beyond available time series of historic events (Te Linde et al.
2010).
Table
4 shows the history of design discharges over the previous century and the beginning of this century. The first design discharge as we define it today was set in 1956 after the major floods of 1953 in the Netherlands. After twenty years it became clear that a design discharge of 18,000 m3 s
−1 , with a safety level of 1/3000 would be too costly and the measures would have a huge impact on cultural, historical and nature values. The Becht Commission, assigned by the national government, determined that the safety level could be adjusted to 1/1250 and the design discharge could be decreased to 16,500 m
3 s
−1. Another twenty years later the design discharge was decreased further to 15,000 m
3 s
−1, because of a lot of public resistance against raising and broadening the dikes. This decrease in design discharge with the same safety level was consistent with a different statistical calculation method. The high waters of 1993 and 1995 placed safety back on the political agenda and the design discharge was raised again to 16,000 m
3 s
−1 in 2001 .
Table 4
Evolution of design discharges for the Dutch part of the Rhine basin (Kwadijk et al.
2008)
1926 | Level of 1926+1 m | – | Flooding 1926 |
1956 | 18,000 | 3000 | Flooding 1953 |
1976 | 16,500 | 1250 | Commission Becht |
1992 | 15,000 | 1250 | Public resistance—Commission Boertien |
2001 | 16,000 | 1250 | Flooding and evacuation 1995 |
2050a
| 18,000 | 1250 | Climate change—Second Delta Committee |
More extreme discharges are projected for the Rhine because of projected climate change, as explained in section
3, therefore the design discharge has been under discussion again. On the basis of a study of Middelkoop et al. (
2000) the Committee Water Management 21st century (WB21) has calculated an increase in design discharge of 5% per degree temperature rise. If a ‘middle’ scenario of the Royal Dutch Meteorological Institute (KNMI) is taken, this translates into a design discharge of 18,000 m
3 s
−1 for the Rhine. Spatial reservations are already made for the possibility of this discharge, although other measures taken at this moment are still based on a design discharge of 16,000 m
3 s
−1. If a more extreme scenario is taken, the maximum design discharge could in theory be up to 22,000 m
3 s
−1 for 2100. For this extreme scenario however, in practice the maximum discharge would be about 18,000 m
3 s
−1, because of flooding upstream the Rhine basin. This therefore means an upper limit of 18,000 m
3 s
−1 to the discharge that can reach the Netherlands (Kabat et al.
2009). The design discharge has been reason for a lot of discussion. The example of Table
4 illustrates the high impact of extreme events on the formulation and implementation of adaptation strategies. The determination of design discharges from statistical analyses of the measured peak discharges faces various problems. The estimation of the 1250 year discharge event from statistical information in a discharge record of about 100 years involves a strong extrapolation, which is quite uncertain. Recent developments like the development of GRADE (Generator of Rainfall And Discharge Extremes) (de Wit and Buishand
2007) have improved these extrapolations, but do not eliminate all uncertainty. The design discharge of 16,000 m
3 s
−1 was included in water safety legislation in the Netherlands in 2001, before research was done on flood safety in Germany in 2004. Without additional flood-protection measures in Germany an amount of 16,000 m
3 s
−1 would not reach the Netherlands, as the Niederrhein would flood in Germany when the discharge is between 11,000 m
3 s
−1 and 16,000 m
3 s
−1, transboundary floods would occur at 14,000 m
3 s
−1. This means that in case of large-scale flooding, the peak discharge at Lobith is reduced (Kroekenstoel and Lammersen
2005). The cooperation and communication between the Netherlands and Germany definitely could have been better, for example, it could be unnecessary for the Netherlands to take measures for extreme discharges, if Germany is not doing this.
This case is a typical example of a ‘predict-then-act’ approach. Science and projections are taken as a starting point and the strategy is based on these projections. The strategy is vulnerable to uncertainty and surprises, as it relies on the scientific accuracy of the projection. If the projections are not accurate and the design discharge would be estimated wrongly, the damage could be huge. This example also shows that transboundary cooperation is essential for effective river basin management. The measures taken in the Netherlands should be adapted to measures in the other riparian countries, especially Germany and vice versa.
In the Netherlands the ‘assess-risk-of-policy’ approach has been applied for the area of water management using the concept of “adaptation tipping points”. These “tipping points” are reached if the current management strategy can no longer meet its objectives (Kwadijk et al.
2010). Only beyond the tipping points an additional adaptation strategy would be needed. The focus of this approach is on the resilience of the water system. The results of this study also have been input to the authoritative study on future adaptation options by the 2nd Delta Committee (see section
5). A number of case studies on sea level rise in the Netherlands which have explored this approach suggest that it may better match the way policy makers address questions than the ‘predict-then-act’ approach. The results have shown, for example, that for dikes along the tidal river area no major technical and financial adaptation tipping points will be reached any time soon, but that potential tipping points might arise on the social- and political level. Social acceptability, for example, of living behind giant dikes may decline (Kwadijk et al.
2010).These experiences suggest that a ‘assess-risk-of-policy’ approach might be useful or at least complementary to the more commonly used ‘predict-then-act’ approach.
7 Discussion, Conclusions and Recommendations
In this paper we have identified factors that facilitate or constrain effective risk management with respect to climate adaptation in transnational river basins. The Rhine river basin was taken as a case study area, as it is a large international river basin with a history of droughts and floods. Three questions were addressed in particular: ‘How are climate change uncertainties dealt with?, ‘How does a (mis)match between information needs and knowledge availability across different geographic and administrative scales stimulate or constrain effective adaptation policy development?’, and ‘What is the effect of (lack of) transboundary cooperation on adaptation management?’ A number of findings emerge:
7.1 Scientific Uncertainties Provide Opportunities for Politically Strategic Water Safety Choices
A view on history shows that design discharges that have been established by water managers were at least informed by statistical analyses from scientific and technical advisors (see section
6). So, the demand of knowledge by policymakers appears to be matched by the supply by scientists. However, the degree to which statistical calculations determine the design discharge can be debated, as over the last century a number of times the design discharge in the Netherlands changed not only as a result of new scientific insights or statistical methods, but also as result of extreme events, financial considerations or public opposition. Extreme events increase the level of public attention and sense of urgency and design discharges were increased to ease these public concerns. After some time remembrance of extreme events seem to fade away in the minds of people and the design discharges were lowered, requiring less costly measures. The political and societal discussion that follows extreme events offers a particular window of opportunity for scientists and scientific information to play a role in policy making (Arnell and Delaney
2006). This is confirmed in a comparative study by Krysanova et al. (
2010) where it was found that experts in different large river basins perceived a climate-related disaster amongst the most important drivers for development of adaptation strategies. But in turn, once the disaster is over, there is a tendency to return to the original situation instead of developing long-term policies (Christoplos
2006). While after an extreme event re-active measures are taken, climate adaptation strategies, targeting future extreme events, ought to be pro-active. This proves to be very challenging as it is more difficult to create a sense of urgency for events that have not happened yet.
7.2 Scientific Support to Water Management Strategies Currently Inadequately Addresses Uncertainties
Even if communication between scientists and policymakers in the area of water safety appears to have been quite satisfactory, particularly in The Netherlands, some questions can be asked. First of all, the question of selection of long-term climate scenarios is interesting. While initially a “best guess” middle scenario was used, and even incorporated in legislation, later a more “worst case” scenario was applied, although not in all cases. It is not completely clear if this was a decision by the relevant policymakers or by the scientific experts and what arguments were behind such decisions. At the same time, model calculations generally not only used one scenario, but also the output of only one global climate model, ignoring differences between model outcomes. It might be that for the coming decades the differences in terms of runoff projections between scenarios and climate models are relatively small and multiple model runs would be too costly, but this is not systematically discussed in the various papers and reports underpinning Dutch water policy.
In general, research on the human dimensions of climate change suggests that available information on climate change is often not perceived to be useful for policymakers, or is misused and contributes to undesired outcomes (Sarewitz and Pielke
2007). In national and regional Dutch and German adaptation strategies uncertainties are mentioned in rather general terms, but it is not explicitly explained how governments could deal with these uncertainties. As a consequence, policy makers can use uncertainties strategically, as illustrated by the evolving choices on design discharges. At the same time, scientific output in the area of water management often does not provide the policy makers with clear information about the uncertainties and how to manage them. Three mismatches between the supply of knowledge and the demand of policy makers relate to spatial and time scaling, and to the scope and form of information provided. Most climate change information is available at long-term temporal scales and large spatial scales, but most management plans or adaptation strategies, from the Water Framework Directive to national plans, have their goals set for at the latest 2015, and usually focus on smaller scales (municipalities, regions, water basins). As to scope and form: often the information provided is too complex, and not expressed in terms directly relevant for the policy question that is supposed to be addressed. Policy makers mostly need information that is simple, and relevant for short-term local decisions. Of course, this is not easy and will not solve all the climate related policy challenges, as for example, environmental policy decision making tends to be highly politicized (Castree and MacMillan
2001). Juntti et al (
2009) discuss some of the challenges in the science policy interface. Firstly, they argue that the notion of validity of evidence would benefit from a more transparent treatment of the division into lay and expert knowledge in evidence generation. Secondly, the range of involved interests adds to the political struggle and finally it is argued that knowledge is only turned into ‘evidence’ when the political climate is ripe for a problem to be identified. Turnpenny et al. (
2009) add to this discussion that technical uncertainties are often invoked as a reason for policy direction. These findings underline the arguments of this paper, the exchange of knowledge between science and policy is not straightforward and there are many factors that influence this process. For both scientists and policy makers it is important to be aware of these influences and to be clear about the choices and underlying assumptions that are made.
7.3 Early Experiences with ‘Assess-Risk-Of-Policy’ Analysis of Options (Looking at the Climate Resilience of Development Plans Rather Than Linking Adaptation Options to Projected Impacts) Suggest That This Method May Be Applied More Widely
Because climate change is framed as a global problem, ‘predict-then-act’ scenario approaches are most commonly used in developing climate adaptation strategies and measures. This approach is strong in coping with statistical uncertainties and can profit from the large amount of available impact assessments. However, projections of future climate change also have uncertainties that cannot be quantified. Too much focus on climate change scenarios alone may lead to ineffective risk management. In the Netherlands, for example, the ‘predict-then-act’ approach may not lead to optimal decision making in the water sector in terms of robustness, flexibility and costs, if only one scenario and one model is chosen as a best or worst case estimate (Kwadijk et al.
2010). The approach ignores governance questions. The ‘assess-risk-of-policy’ approach recognizes local interests and conditions, and offers possibilities to deal with uncertainties that cannot be quantified, by focusing on the resilience of the system. Research on this approach has only recently started, e.g. with the concept of adaptation tipping points. First results of this method show that it can offer policy makers a new, complementary tool for evaluating adaptation strategies that also addresses their non-climate priorities and maybe a different view on the urgency of adaptation to climate change. Therefore it would be interesting to do more research on ‘assess-risk-of-policy’ approaches and test these approaches more widely.
7.4 Development and Implementation of Adaptation Options Derived from Integrated Analysis at the Full River Basin Level Rather Than Within the Boundaries of the Riparian Countries can Offer New Opportunities, But Will Also Meet With Many Practical Challenges
The history of water management in the Rhine basin has shown that international cooperation can be successful. Agreements on water pollution of the Rhine have led to a successful improvement of water quality. A comparative study of Ma et al. (
2008) showed that the 1998 Rhine Convention is the best transboundary water treaty for enforcement, capability and treaty implementation. This can be an example for other transboundary cooperation, e.g. to address climate change adaptation in the most cost effective manner. Taking a closer look at regional policy practices along member states’ borders, however, suggests that cooperation is often still viewed as problematic. So, while ‘Europe’ is striving for a borderless river basin management, harsh realities reflected in regional practices do not always meet these expectations (Wiering et al.
2010). International cooperation in river basins with respect to climate change adaptation is very important, as measures in one country could have negative effects in another or country-by-country measures could be less effective or more expensive than measures optimized over the full river basin. In the case of the Rhine, the latter can be illustrated by the current understanding that the design discharge of 16,000 m
3 s
−1 was included in Dutch legislation before research was done on the impacts of floods on high water in Germany. Results of this research showed for example that an extreme discharge of 18,700 m
3 s
−1 at Lobith would be reduced to 15,500 m
3 s
−1 at Lobith because of flooding in Germany (Lammersen
2004). Of course, this may change as the climate changes and further protective measures are taken throughout the river basin. This example shows the potential importance of enhanced cooperation, especially since the projection of climate change impacts suggests that more adaptation measures will be necessary in the futureIf the difficulties caused by different institutional arrangements and cultural differences were to be explicitly recognized and systematically addressed, more effective transnational collaboration would be possible. However, to reach this goal, political will from the riparian countries is essential. Until now this will and the means to put this will into action is not clearly expressed in the governmental documents on climate adaptation that we have analyzed.
7.5 Knowledge Gaps
We identified a number of knowledge gaps that require research attention. While much is known about technical aspects of measures, institutional barriers for pro-active adaptation are less well understood. Research has addressed the problem of climate change uncertainties in climate and impacts models separately, but the consequences of the propagation through the various analytical steps for risk management is poorly understood. The discussion on climate-related uncertainties is mainly science-driven, and more attention is required on how policymakers deal with them: the communication of uncertainties should be fit for purpose. The implementation of adaptation measures depends on interactions of different governance levels—more research is required to understand how this affects the formulation and actual implementation of adaptation strategies. So far, the most common approach to impacts and adaptation assessment is the projected climate impacts-driven ‘predict-then-act’ approach—more attention is required to alternative, or complementary ‘assess-risk-of-policy’ approaches in support of the enhancement of climate resilience. Different countries in transnational river basins use different methods and climate impact information. Research to better understand the constraints and opportunities of transboundary cooperation with respect to climate change impacts and adaptation assessment in international river basins would be useful. This paper is based on literature review and informal contacts, for a better understanding of the details of how past decisions were made, more systematic research supported by well-structured interviews would be a useful complement to the literature review. While some of these suggestions are likely to be addressed in new national research programmes, such as Knowledge for Climate in the Netherlands and Klimzug in Germany, stronger and sustained international research collaboration would strengthen the scientific quality and policy-relevance of the projects.
Acknowledgements
We would like to thank the following persons for taking part in meetings and interviews: Gert Bekker (IVM, VU University), Hendrik Buitenveld (RIZA), Suraje Dessai (Exeter University), Jaap Kwadijk (Deltares) and Tom Raadgever (TU Delft), The two anonymous reviewers are also acknowledged for their valuable comments, references and suggestions.