Advertisement

The role of uncertainty in climate change adaptation strategies—A Danish water management example

  • J. C. Refsgaard
  • K. Arnbjerg-Nielsen
  • M. Drews
  • K. Halsnæs
  • E. Jeppesen
  • H. Madsen
  • A. Markandya
  • J. E. Olesen
  • J. R. Porter
  • J. H. Christensen
Open Access
Original Article

Abstract

We propose a generic framework to characterize climate change adaptation uncertainty according to three dimensions: level, source and nature. Our framework is different, and in this respect more comprehensive, than the present UN Intergovernmental Panel on Climate Change (IPCC) approach and could be used to address concerns that the IPCC approach is oversimplified. We have studied the role of uncertainty in climate change adaptation planning using examples from four Danish water related sectors. The dominating sources of uncertainty differ greatly among issues; most uncertainties on impacts are epistemic (reducible) by nature but uncertainties on adaptation measures are complex, with ambiguity often being added to impact uncertainties. Strategies to deal with uncertainty in climate change adaptation should reflect the nature of the uncertainty sources and how they interact with risk level and decision making: (i) epistemic uncertainties can be reduced by gaining more knowledge; (ii) uncertainties related to ambiguity can be reduced by dialogue and knowledge sharing between the different stakeholders; and (iii) aleatory uncertainty is, by its nature, non-reducible. The uncertainty cascade includes many sources and their propagation through technical and socio-economic models may add substantially to prediction uncertainties, but they may also cancel each other. Thus, even large uncertainties may have small consequences for decision making, because multiple sources of information provide sufficient knowledge to justify action in climate change adaptation.

Keywords

Climate change Adaptation Uncertainty Risk Water sectors Multi-disciplinary 

1 Introduction

Climate change affects many aspects of human societies and the ecosystems on which they depend. Impacts on key sectors, such as agriculture, health, water supply, urban drainage, roads, buildings and the environment, can already be observed and are expected to increase in the future (IPCC 2007b; EU Commission 2009). The present climate projections exhibit large uncertainties arising among others from assumptions on greenhouse gas emissions, incomplete climate models and the downscaling of climate projections (IPCC 2007c). When assessing the physical impacts of climate change on water related sectors, traditional uncertainties in hydro-ecological models, such as data and parameter uncertainty and model structural uncertainty need to be addressed. For socio-economic impacts, additional uncertainties, involving aspects of costing and problem framing, need inclusion (van der Keur et al. 2008). The complete suite of uncertainties has been referred to as the uncertainty cascade (Hulme and Carter 1999; Katz 2002; Foley 2010).

Making climate change adaptation decisions is particularly difficult since they rely on uncertainties related to climate projections as well as to developments in natural systems and sectors that are affected by other uncertainties. Climate change impacts and adaptation also influence a wide range of stakeholders with different interests, making it difficult to distinguish uncertainties related to technical information stemming from different perceptions and understandings of issues that reflect stakeholder interests, perceived burdens and benefits. Decision-making in climate change adaptation deals with how, when and to what extent risks from climate events can and should be reduced, given private stakeholders’ interests and those of society at large. Uncertainties are seldom explicitly recognised and dealt with in practical climate adaptation planning (Preston et al. 2011).

Uncertainty has for many years been recognised by UN Intergovernmental Panel on Climate Change (IPCC) as crucial (IPCC 2007a), and it will receive even more attention in the forthcoming Fifth Assessment Report (AR5) (Yohe and Oppenheimer 2011). A goal of the AR5 is to apply “a common framework with associated calibrated uncertainty language that can be used to characterise findings of the assessment process” (Mastrandrea et al. 2011). According to an AR5 uncertainty guidance note, the degree of certainty of a key finding should be characterised qualitatively in terms of the confidence in the validity of a finding and the degree of agreement as well as in quantified measures of uncertainty (Mastrandrea et al. 2011). This approach has been criticised for being oversimplified and potentially leading to misleading overconfidence, because it “omits any systematic analysis of the types and levels of uncertainty and quality of the evidence, and more importantly dismisses indeterminacy and ignorance as important factors in assessing these confidence levels” (Curry 2011).

Our objectives are: (i) to outline a common uncertainty framework, including a terminology, that is generically applicable in climate change adaptation; (ii) to assess climate change related uncertainties in water related disciplines and sectors; and (iii) to evaluate strategies on how uncertainty affects climate change adaptation decision making. We have applied this framework to four water related sectors in Denmark. Given our focus we do not discuss all aspects related to adaptive management, such as resilience, adaptive capacity and social learning (Pahl-Wostl 2007; Lebel et al. 2010).

2 Uncertainty framework

2.1 Definition of uncertainty

We adopt the definition of Klauer and Brown (2003) that a person is uncertain if s/he lacks confidence about the specific outcomes of an event. This definition holds that for most technical and natural sciences, uncertainty is primarily an objective matter, whilst acknowledging that uncertainty includes subjective aspects.

2.2 Typology

Our typology, which is adapted from Walker et al. (2003), Refsgaard et al. (2007) and van der Keur et al. (2008), characterises all uncertainties according to three dimensions nature, level and source.

The nature of uncertainty can be epistemic, aleatory and ambiguity. Epistemic uncertainty is the uncertainty due to imperfect knowledge and is reducible by gaining more knowledge via research, data collection and modelling. Aleatory uncertainty, also termed ontological or stochastic uncertainty, is due to inherent variability. It can be quantified, but is stochastic and irreducible. Ambiguity results from the presence of multiple ways of understanding or interpreting a system. It can originate from differences in professional backgrounds, scientific disciplines, value systems and interests.

The level of uncertainty characterises how well the uncertainty can be described within the range from determinism to total ignorance (Fig. 1), where determinism is the ideal, non-achievable, situation where everything is known exactly and with absolute certainty. Within this range, statistical uncertainty can be described using well-known statistical terms. Scenario uncertainty, in general, cannot be described statistically but are used when possible outcomes are known but not all probabilities of such outcomes are present (Brown 2004). Qualitative uncertainty occurs when not even possible outcomes are known (Brown 2004). Recognised ignorance occurs when there is an awareness of lack of knowledge on a certain issue, but where it is not possible to categorise the uncertainty further. Total ignorance denotes a state of complete lack of awareness about imperfect knowledge.
Fig. 1

Levels of uncertainty. After Walker et al. (2003) modified by van der Keur et al. (2008)

The sources of uncertainty can be divided into uncertainties in input data, such as external driving forces and system data; model uncertainty, originating from the model structure including process equations, the model software and the model parameters (Refsgaard and Henriksen 2004); context uncertainty, such as the boundaries of the systems to be modelled, future climate and regulatory conditions when these aspects are not explicitly included in the modelling study; and uncertainty due to multiple knowledge frames, meaning that persons may have different perceptions of the main problems. The simultaneous presence of multiple frames of reference to understand a certain phenomenon may cause ambiguity.

Some sources of uncertainty are epistemic by nature, e.g. model structural uncertainty and model technical uncertainty; input data, e.g. future rainfall, contain aleatory uncertainty and these two sources typically dominate when dealing with technical and natural science aspects. Context and framing uncertainty are often of an ambiguity nature and are increasingly important, when political issues and stakeholder interests are included in a decision situation.

The above sources of uncertainty are composite groups, each of which may be decomposed into different sources. For instance there may be several types of input data with different uncertainty characteristics in terms of nature and level. Many uncertainties cannot be uniquely classified, but will often fall in different classes (Warmink et al. 2010). This situation may occur, when uncertainty is aggregated from various sources, and may be resolved by decomposing it into its several sources until the classification becomes unique.

Some authors (Walker et al. 2003) do not distinguish ambiguity and epistemic uncertainty. This presupposes that more knowledge will reduce ambiguity. However, in decision making this is seldom the case, because more knowledge does not necessarily converge to a single truth as seen by stakeholders and hence to more certainty (Warmink et al. 2010). This is particularly true, when many different stakeholders’ interests, perspectives and perceptions are involved. In such cases interests and perceived uncertainty can mix together in complicated ways.

The distinction between aleatory and epistemic uncertainty is not always clear. One may argue that up to a certain level, random system behaviour is also lack of knowledge, and, therefore, that some of these stochastic uncertainties can be reduced, if we understand the system better. However in practice, the availability of resources such as time and money severely limit the possibility to reduce this kind of uncertainty. Warmink et al. (2010) therefore define aleatory uncertainty, denoted as natural uncertainty, as random system behaviour that cannot be explained adequately given the available resources.

2.3 Uncertainty cascade in climate change impact and adaptation for water systems

The uncertainty cascade found in climate modelling related to water systems includes uncertainties from emission scenarios and global and regional models (Foley 2010). We have included uncertainties due to statistical downscaling, water systems impacts and socio-economic impacts (Fig. 2). The main uncertainties in the different steps are characterized (Table 1) with respect to the sources and nature of uncertainty following our above typology. In our cascading process we see that:
Fig. 2

Structural elements in the assessment of climate change impacts and adaptation illustrating the uncertainty cascade

Table 1

Characterisation of key sources of uncertainty in the uncertainty cascade and their nature in relation to climate change adaptation. X, XX, XXX is a general guide on the relative importance level of the sources, although it must be emphasised that the importance of the individual sources of uncertainty is context specific

Steps in climate change adaptation analyses (chain in uncertainty cascade, Fig. 2)

Sources of uncertainty

Nature of uncertainty

Input data

Model

Context

Multiple knowledge frames

Ambiguity

Epistemic uncertainty (reducible)

Aleatory uncertainty (irreducible)

Parameter values

Model technical aspects

Model structure

Greenhouse gas emissions

    

XX

XXX

XXX

XX

 

Socio-economic scenarios

XX

  

XX

XX

XXX

XXX

XX

 

Future climate (Climate models)

GCMs

  

XX

XXX

   

XXX

 

RCMs

  

XX

XXX

   

XXX

 

Initial conditions/natural variability

XX

       

XXX

Downscaling/statistical correction

 

XXX

 

XX

   

XX

XX

Water system impacts (Hydro-ecological models)

X

XXX

X

XXX

XX

X

X

XXX

X

Socio-economic impacts (Socio-economic tools)

XX

  

XX

XX

XXX

XXX

XX

 

Adaptation measures

XX

XXX

X

XXX

XX

XXX

XXX

XXX

XX

Table 2

Examples of water related climate change adaptation issues related to agriculture in Denmark

Climate change impact

Adaptation

Type of problem

Consequence

Risk level

Dominating uncertainty

Option

Cost level

Intent

Action

Temporal scope

Spatial scope

Additional uncertainty

Source

Nature

Source

Nature

Increased summer droughts and higher water requirements cause by longer crop growth duration in warmer climate

Increase in irrigation requirements

Med

Projections of climate change, CO2 effects on evapotranspiration, market price for agricultural products

Epistemic+Aleatory

Adapt crop choice to existing water abstraction permissions (non-structural)

Low-Med

A (P)

R (C)

S

L

Same as for impact

Increase water abstraction permits (non-structural)

Low

P

A (C)

L

W

Multiple frames

Ambiguity

Increased winter rainfall

Increased inundation of agricultural fields during winter/spring

High

Projections of climate change, hydrological parameters of soils

Epistemic

More efficient tile drains (structural)

High

A

A (C)

L

L

Same as for impact

Conversion to perennial energy cropping (non-structural)

Low

A

A (C)

S (L)

L

Multiple frames

Ambiguity

Increased inundation of river valleys

High

Projections of climate change

Epistemic

Abandon agriculture in flood prone areas (non-structural)

High

A (P)

R (A)

S (L)

L (W)

Multiple frames

Ambiguity

Stream straightening and stream weed cutting (structural)

High

P

A (C)

L

W

Multiple frames

Ambiguity

Threshold conditions

Sudden changes in reliability in the production and quality of food and fibres

Med

Climate and impact models

Epistemic

Crop breeding to avoid impact (non-structural)

Med

A

R, A

S

L

Same as for impact

Change in diseases and pests

Sudden and large reduction in crop yields and quality

Med

Inherent uncertainty and impact models

Epistemic+Aleatory

Plant breeding, quarantine policies and pest/disease forecasting (non-structural)

Med-High

P

R

S

W

Same as for impact

Notes to characterisation of adaptation measures (see Fig. 3 and Section 3 for further details):

Intent: A: autonomous; P: planned

Action: R: reactive; C: concurrent; A: anticipatory

Temporal scope: S: short term; L: long term

Spatial Scope: L: localized; W: widespread

Future greenhouse gas (GHG) emissions cannot be known with certainty because they depend on future human decisions and are characterised by scenario uncertainty based on the IPCC scenarios (IPCC 2007c).

Uncertainties related to global (GCM) and regional climate models (RCM) are typically assessed by ensemble modelling, where multiple climate models with different process equations are used for making probabilistic projections (e.g. Tebaldi et al. 2005; Smith et al. 2009; Christensen et al. 2007a, 2010; Déqué and Somot 2010). A fundamental limitation here is lack of knowledge of all climate processes, such as natural feedbacks in the Earth’s carbon cycle (recognised ignorance). Another important source of uncertainty is lack of knowledge of initial ocean states, such as temperature and salinity. Different plausible initial conditions may result in significantly different climate projection pathways as climate systems have strong elements of natural variation (Hawkins and Sutton 2009).

For prediction of local water systems, climate variables are typically required at smaller spatial scales than those simulated by climate models. Furthermore, climate models often do not represent statistical properties of observed climate variables such as precipitation, which disenables impact modellers to use the outputs from climate models directly into impact models. Data from climate models are typically downscaled and statistically corrected (e.g. Wilby et al. 2002; Christensen et al. 2007b; Fowler et al. 2007; Kilsby et al. 2007) but different downscaling methods may give different input data, thus total errors can arise from both sources (e.g. Sunyer et al. 2011).

Identification and assessment of possible climate change adaptation measures uses basically the same tools as those for assessing climate change effects, i.e. combined hydro-ecological models and socio-economic tools, and hence all the uncertainties involved here are inherited. As adaptation decisions often involve considerable uncertainties, stakeholder and societal risk perception may, in some cases, become important, especially for situations where attitudes to risk may change over time. Furthermore, different stakeholders may have different perceptions of the consequences of alternative adaptation measures, reflecting their specific interests in relation to the affected case. This may contain more ambiguity than in other elements of the uncertainty chain.

3 Adaptation options

Options for adaptation can be classified as the intent, timing of the action, and their temporal and spatial scope(s) (Fig. 3). Other options of special interest include no regrets, and adaptation decisions can be characterised by their flexibility, reversibility and robustness. Uncertainty impinges differently on each class of actions in Fig. 3.
Fig. 3

Classification of adaptation measures. From Burton (2009)

Intent

In most sectors some adaptation actions will be autonomous and others will be planned. The former tend to be taken by individuals rather than public agencies, often in response to changes in the ambient environment or to changes in market conditions. They often consist of reactive measures of short-to-medium term duration and are frequently of local scope. Planned actions on the other hand are normally taken by public agencies or large private sector entities, often in anticipation of climate change and with a longer term perspective and wider spatial scope.

Within planned actions we have some that consist of ‘hard’ or ‘structural’ measures (e.g. investment in physical assets) and others of ‘soft’ or ‘non-structural’ measures (e.g. increases in the prices of water services). In the context of high uncertainty (especially of the epistemic kind) one could argue that non-structural planned measures are preferred to structural measures, on the grounds that as new knowledge is gained it will be easier and less costly to modify the response than it would be with the structural planned actions. One could argue that if there is a real risk of system collapse, then physical planned actions are appropriate. But in such cases it may be better to require relocation (or at least contingent plans for relocation), which would constitute a non-structural measure. In addition, one has to take account of the uncertainty related to behavioural actions (ambiguity) on non-structural incentives (e.g. taxation), which constitute a major part of uncertainty in non-structural measures (Arnell et al. 2004).

Timing of action

In general, adaptation measures combine reactive and anticipatory forms: increased frequency of storms mandates the need to develop early warning systems and protective physical structures, but it also increases the need for more effective action once storms occur. With greater uncertainty it would be preferable to rely on reactive measures, because they are taken with more information at hand, but that is not necessarily the case. A high level of uncertainty may include the possibility of an outcome that would make reactive measures ineffective or very costly in terms of loss of life and property. In such cases it may be desirable to go for the anticipatory action, although ideally it should be in a form that is flexible and reversible.

Temporal and Spatial Scope

Similar considerations prevail with respect to temporal and spatial scope. Other things being equal, the greater the spatial scope of an action and the longer its duration, the greater will be the uncertainty associated with its consequences. In adaptation, actions often need to be taken with long time periods in mind. Investments in road design or the dimensions of sewerage systems are influenced by conditions over the lifetime of the capital, which is measured in decades. Since conditions at the end of the period are highly uncertain, political and economic concerns can exclude adaptation options. However, non-climate proofing of investments with a long lifetime can also be very costly. The compromise can be to plan on the basis of a ‘likely’ scenario, but then leave open the possibility that more costly changes can be made later, when more precise information is available.

No-regrets, flexibility, reversibility and robustness

In all adaptation decisions the consequences of uncertainty are reduced when actions fall into the category of no-regrets, or that have properties of flexibility, reversibility and robustness. No-regrets actions are those that one would wish to take for other reasons and that also have benefits in terms of reducing the impacts of possible climate change. If these actions are truly justified on other grounds than their climate benefits, then the uncertainty of their climate benefits is of no consequence as it does not affect the decision. Such cases, however, are very rare; in most circumstances, the action can be modified to increase the climate adaptation benefits and there is a choice to be made of how much such modification is justified.

4 Water sector examples—a Danish context

4.1 Climate change projections for Denmark

The current national strategy for adaptation (Danish Energy Agency 2008) to climate change in Denmark is based on climate change projections for the period 2071–2100 for the IPCC A2 and B2 SRES scenarios, using climate scenario data from the HIRHAM4 model (Christensen et al. 1996), as utilized in the European climate downscaling project PRUDENCE (Christensen et al. 2007a). According to these projections, Denmark will probably warm by approximately 1–2°C, relative to the pre-industrial level, in the near future and by approximately 2–3°C towards the end of the century. The largest temperature increase is expected during winter. Equivalent projections for the mean annual precipitation show a small increase in the short term, becoming more pronounced in the long term. The strongest positive precipitation response is seen during winter, whereas little or no change is predicted during summer. Extreme precipitation events are generally expected to be stronger. The general picture is of warmer and wetter winters; drier summers with enhanced extreme precipitation events and longer dry periods, although the above climate projections are subject to very considerable uncertainties (Boberg 2010).

4.2 Climate change impacts and possible adaptation measures for water related sectors in Denmark

The key uncertainties related to climate change adaptation in a Danish context are illustrated by examples from the water-related sectors agriculture, freshwater ecology, water infrastructures in rural areas and urban water infrastructures (Tables 2, 3, 4 and 5). The examples have been selected to illustrate the variety of the climate change adaptation issues and uncertainties with a focus on issues requiring a planning horizon of more than 20 years, when a climate change signal could begin to dominate the natural climate variability, and where a relatively long economic lifetime of adaptation projects makes it important to give careful consideration to uncertainty. In Tables 2, 3, 4 and 5, climate change impacts are characterised with respect to risk level and dominating uncertainty following the uncertainty terminology outlined in Section 2, while the adaptation options are characterised according to cost level (High/Medium Low) and the classification outlined in Section 3. The three cost levels apply for internal comparisons within each of the four sectors, but may not be applicable for comparisons from one table to another.
Table 3

Examples of climate change adaptation issues related freshwater ecology and quality in Denmark. See notes on adaptation measures below Table 2

Climate change impact

Adaptation

Type of problem

Consequence

Risk level

Dominating uncertainty

Option

Cost level

Intent

Action

Temporal scope

Spatial scope

Additional uncertainty

Source

Nature

Source

Nature

Change in precipitation → Higher variability in flow; flooding of riparian zones and streams

Reduced biodiversity, lower chan-ce of fulfilling WFD

High

Climate models+downscaling+Hydrological models

Epistemic+Aleatory

(Re)-establishment of wetlands and buffer zones, (structural)

Med

P,A

A,C

S

L (W)

Multiple frames

Ambiguity

Accepting flooding (non-structural)

Low

A

A,R

S

L,W

Multiple frames

Ambiguity

Increase in nutrient loading to lakes → Higher eutrophication

Lower chance of fulfilling WFD

High

Context (land use and agricultural practice)+Climate model structure+hydro-ecological model parameters and structure

Epistemic+Aleatory

Reduce external loading from sewage and agriculture (structural)

High

P

A;C

S

L,W

Multiple frames

Ambiguity

Re-establishment of wetlands and buffer zones (structural)

Med

P,A

A,C

S

L (W)

Multiple frames

Ambiguity

Re-meandering of streams (structural)

Med

P

A

S

L

Multiple frames

Ambiguity

Reduced weed cutting (structural)

Med

P

A,C

S

L,W

Multiple frames

Ambiguity

Accepting flooding (non-structural)

Low

A

A,R

S

L,W

Multiple frames

Ambiguity

Reduce internal loading in lakes by biomanipulation and/or sediment capping (structural)

Med

P

A,C

S

L

Same as for impact

Higher temperatures → Changes in stream and lake ecosystems; higher meta-bolism in streams; higher risk of oxygen depletion

Reduced biodiversity; lower chan-ce of fulfilling WFD

High

Climate models+Hydro-ecological model parameters+structure

Epistemic+Aleatory

More trees in the riparian zone and establishment of overhanging bank vegetation (structural)

Low

A,P

A,C

S

L

Multiple frames

Ambiguity

Higher temperatures → Shift in biological structure leading to higher eutrophication and higher frequency of toxic algae in lakes

Lower chance of fulfilling WFD

Med

Climate models+Hydro-ecological model parameters+structure

Epistemic+Aleatory

Reduce external organic and nutrient loading by better treatment of sewage, more sustainable agriculture(structural)

High

P

A,C

S

L,W

Multiple frames

Ambiguity

(Re)-establishment of wetlands and buffer zones (structural)

Med

P,A

A,C

S

L (W)

Multiple frames

Ambiguity

Re-meandering of streams (structural)

High

P

A

S

L

Multiple frames

Ambiguity

Reduced weed cutting. (structural)

Med

P,A

A,C

S

L,W

Multiple frames

Ambiguity

Table 4

Examples of change adaptation issues related to water infrastructure in rural areas in Denmark. See notes on adaptation measures below Table 2

Climate change impact

Adaptation

Type of problem

Consequence

Risk level

Dominating uncertainty

Option

Cost level

Intent

Action

Temporal scope

Spatial scope

Additional uncertainty

Source

Nature

Source

Nature

Water supply. Changes in groundwater recharge or acceptable influence on streamflow in critical low flow periods

Change in how much groundwater can be abstracted in a sustainable manner due to either problems in aquifer or low flow conditions in stream.

High

Climate models+hydro-ecological model parameters+structure (geology)

Epistemic

Relocation of groundwater abstraction—influencing also the protection zones (item below) (structural)

Med

P

P

L

L

Same as for impacts

Changes in objectives and risk willingness (non-structural)

Low

P

C

S

L

Multiple frames

Ambiguity

Water supply. Changes in wellfield capture zones

The selected areas for groundwater proctection will be the wrong area.

Med

CHG emissions+climate models+hydro-ecological model parameters+structure (geology)

Epistemic

Increase protection areas to account for worst case (structural)

High

P

P

L

L

Same as for impacts

Changes in strategy, increased risk to protect wrong area (non-structural)

Low

P

C

S

L

Multiple frames

Ambiguity

Inundations of roads

Road traffic interrupted

Med

CHG emissions+climate model structure

Epistemic+Aleatory

New design to avoid inundation (structural)

High

P

P

L

W

Same as for impacts

Close roads+warning in critical periods (non-structural)

Low

P

C

S

L

Multiple frames

Ambiguity

Undermining of road foundation due to increasedgroundwater table

Roads deteriorate

Med

Climate models+hydro-ecological model parameters and structure (geology)

Epistemic

New designs to accept high groundwater table (structural)

High

P

P

L

W

Same as for impacts

New designs to avoid high groundwater table (structural)

High

P

P

L

W

Same as for impacts

Drainage or pumping scheme to keep groundwater table low (structural)

Low

P

C

S

L

Same as for impacts

Table 5

Examples of water related climate change adaptation issues related to urban water infrastructure in Denmark. See notes on adaptation measures below Table 2

Climate change impact

Adaptation

Type of problem

Consequence

Risk level

Dominating uncertainty

Option

Cost level

Intent

Action

Temporal scope

Spatial scope

Additional uncertainty

Source

Nature

Source

Nature

Increasing precipitation extremes

Increased flood risk in flat areas and local depressions

High

Climate models+downscaling+hydrologic and hydraulic models

Epistemic+Aleatory

Larger pipes (structural)

Med

P

C

L

W

Multiple frames

Ambiguity

Local retention (structural)

High

P

C

S

W

Multiple frames

Ambiguity

Changing land use practices (relocation) (structural)

Med –High

P

A

M

L

Multiple frames

Ambiguity

Contingency planning,incl. forecasting (non-structural)

Low

P

C

S

L

Multiple frames

Ambiguity

Increasing sea surge levels

Increased flood risk in coastal regions

High

Climate models+downscaling+hydrologic and hydraulic models+areal planning

Epistemic+Aleatory+Ambiguity

Dikes (structural)

High

P

A (R)

L

L

Multiple frames

Ambiguity

Changing land use practices (non-structural)

Low-High

P

R (A)

S—L

L

Multiple frames

Ambiguity

Changing precipitation patterns

Increased risk of unstable foundations of buildings and infrastructure

Med

Hydrology+context

Epistemic+Aleatory

Changing building design standards, higher use on tile pipes (structural)

Med

P

C

M

L

Same as for impact

Increasing risk of heat waves

Water supply standards are not met

Med

Context (design and operation practices)

Aleatory+Ambiguity

Use of intermediate cooling systems on water supply (non-structural)

Low

P

C

L

L

Same as for impact

4.2.1 Agriculture

Potential impacts

Projected climate change will lead to warmer season crops, in particular grain maize, taking over part of the cereal area (Olesen et al. 2007). Agricultural land use will also be determined by market and European Union (EU) Common Agricultural Policy factors and technological changes leading to changes in economic competiveness of different crops and cropping systems. Higher frequencies of summer droughts will lead to higher demands for irrigation leading to increased water demand in late summer. Changes in crop productivity and crop type will lead to higher fertiliser demand. Higher temperatures lead to increased soil organic matter turnover causing higher soil contents of mineral nitrogen. This, in combination with higher winter rainfall and larger rainfall intensities, will increase the risk of nitrogen and phosphorus losses to the aquatic environment. Higher winter rainfall will also increase the chance of inundation of agricultural fields during autumn, winter and spring. This increases the need for tile draining of soils. There are indications of thresholds in both crop yield responses to climatic changes leading to increased variability in crops yield (Porter and Semenov 2005; Kristensen et al. 2011) and to increased risk of nitrate leaching (Fig. 4). It is also likely that the increasing temperatures, in particular winter temperatures, will lead to increased risks of attack by pests and diseases (Olesen et al. 2011).
Fig. 4

Probability of mean N leaching from winter wheat (Triticum aestivum L.) exceeding a threshold level of 70 kg N ha−1 for sandy soils. Results were calculated for projected climate conditions for the 21st century for the A1B emission scenario (Børgesen and Olesen 2011)

Possible adaptation measures

Adaptation options to increasing wet conditions during autumn, winter and spring include improved drainage of agricultural fields (renewing tile drains) and conversion of agricultural land with risk of inundation to perennial energy crops (e.g. willow), or, in severe cases, land abandonment. Increased risks of summer drought may be countered by enhancing irrigation capacity, but also by having crop mixtures with variation in timing of irrigation needs. In case of severe water restrictions, there may be a need for improved forecasting of water availability for irrigation. The risks of increased erosion and nutrient losses during high intensity rainfall events may be reduced by improved soil cover, in particular by growing catch crops during autumn and winter (Olesen et al. 2011). The risk of changes in low yield can to some extent be avoided through targeted plant breeding, and there is also some scope for breeding against plant diseases, although other options for integrated pest and disease control should also be considered and adjusted to the changing environmental conditions. Breeding should probably also address the possibilities for improved water use efficiency under higher CO2 concentrations, although the genetic basis for this is poorly known and also unlikely to have major impact as reduced water loss from leaves is likely to lead to lower CO2 uptake.

Dominating uncertainties

Partial analyses of the uncertainty chain (Olesen et al. 2007) show that there is large uncertainty associated with future land use, which will have considerable interactions with and dependencies on future climate, but this nexus has not been sufficiently investigated (Schröter et al. 2005). Uncertainties associated with impact models have so far been little explored (Rosenzweig and Wilbanks 2010; Rötter et al. 2011).

4.2.2 Freshwater environment

Potential impacts

Climate change will most probably lead to more variation in discharge in streams with higher flow (and flooding) in winter and less in summer, possibly with periodic drought in some streams. This may impoverish flora and fauna in streams (Table 3). Higher precipitation will enhance leaching of nutrient from soils and bank erosion and, thereby, the external loading of nitrogen, phosphorus and sediment to lakes. This, in turn, will lead to a poorer ecological status of lakes, with higher levels of potentially toxic cyanobacteria (blue-green algae) in eutrophic lakes and lower water clarity (Adrian et al. 2009). The projected increase in temperature will result in higher biological metabolic rates and oxygen consumption and thus lower oxygen concentrations in streams. This can lead to lower self-purification capacity and consequently lower stream water quality, ecological status and a reduced biodiversity. Higher temperatures will most likely reinforce eutrophication of lakes due to changes in the fish community (smaller fish, faster reproduction, higher predation on algal grazers) that via a set of positive feedback mechanisms leads to reduced water clarity (Jeppesen et al. 2010). Moreover, toxic cyanobacteria thrive well at high temperatures (Paerl and Huisman 2008). Consequently, the critical nutrient loading to obtain good ecological status according to the EU Water Framework Directive (WFD) will likely be lower, and a further reduction of the external nutrient loading will be required to counteract this deterioration (Jeppesen et al 2009, 2011).

Possible adaptation measures

Adaptations include a shift to lower input and more perennial agriculture, improved nutrient and soil management with less loss of nutrients to surface waters, reduced loading from point sources and, where appropriate, re-establishment of lost wetlands, establishment of riparian buffer zones, re-meandering of channelized streams and less drastic weed-cutting in streams. Potential measures to reduce local temperature increases are the establishment of trees in the riparian zone and more overhanging bank vegetation. Relevant restoration measures in lakes include methods to reduce internal phosphorus loading (Cooke et al. 2005) and methods to reduce the amount of coarse fish (Hansson et al. 1998).

Dominating uncertainties

We are not aware of studies that have been made to compare the effects of the various sources of uncertainties listed in Table 1. Our expectation is that the dominating sources of uncertainty are related to climate models, precipitation and runoff of water and nutrients, change in land use and management induced by climate and environmental change, and how biota and ecosystem functions change with climate.

4.2.3 Infrastructure in rural areas—water supply and roads

Potential impacts

Water supply, which is almost 100% groundwater derived in Denmark, will be affected by climate change, because both groundwater recharge and low flows and ecological conditions in streams and wetlands may be affected (van Roosmalen et al. 2007). This will influence the amount of water available for water abstraction, the effect of groundwater abstraction on water quality and the ecological state of freshwater bodies. Furthermore, there is a risk that the groundwater protection measures will be applied to the wrong areas, because the groundwater abstraction catchments are climate dependent. For roads in rural areas the two main potential problems are inundations resulting in traffic interruption, and undermining of road foundations due to increased groundwater table.

Possible adaptation measures

We envisage that the main options related to water supply would include relocation of well fields/water works and increase of groundwater protection areas to account for uncertainty of the location of the groundwater abstraction catchments under a future climate. Non-structural measures might include changes in objectives and in our risk willingness to accept a lower reliability for supply of clean drinking water without causing adverse environmental effects. These measures could have major impacts on environment and agriculture. Similarly, the envisaged adaptation with respect to roads include both structural and non-structural measures such as new designs, real-time flood forecasting and warning of closing of roads, and pumping of groundwater to keep groundwater table low.

Dominating uncertainties

The combined effects of the various sources of uncertainties (Table 1) show that for water supply we expect that the dominating source for the structural measures are related partly to the climate models and partly to the hydro-ecological models with a particular emphasis on the geological data and interpretation of geological structures. The dominating uncertainty for the non-structural measure would be ambiguity on risk willingness aspects. For road infrastructure adaptation measures we expect that the main uncertainties are climate models and hydro-ecological models.

4.2.4 Urban water infrastructure

Potential impacts

Climate change is expected to lead to increases in extreme precipitation. With the current estimates of increases in design level frequencies of 30–40% for precipitation (Arnbjerg-Nielsen 2012), case studies have suggested that the present hydraulic capacity of the urban drainage system will result in a 10-fold increase in cost of urban flooding compared to current annual losses (Arnbjerg-Nielsen and Fleischer 2009). Increases in mean sea level and storm surges will affect coastal areas even more, with changes in the hazard frequencies of more than one order of magnitude. Changing precipitation patterns are expected to lead to changes in the high ground water levels which may in some cases lead to unstable buildings because of high groundwater levels. Water quality in the water supply sector may be jeopardized by heat waves leading to high temperatures in the water supply network and thus potentially higher microbial and chemical activity.

Possible adaptation measures

Adaptation measures related to increased flood risks from precipitation can be divided into the following four categories: increasing drainage capacity away from depressions/flat areas, local source control and infiltration from areas upstream depressions/flat areas, changing land use practices in depressions/flat areas and improved contingency planning. While some of these measures rely to some degree on an autonomous and reactive adaptation scheme, the increase in flood risk due to sea surges must rely on both planned and anticipatory adaptation methods. The other adaptation issues (Table 5) are less important economically, and the adaptation options are easier to identify.

Dominating uncertainties

Climate change impacts are all related to hydrological extremes, which are subject to substantial epistemic uncertainty. However, when considering the increases in flood risk due to precipitation extremes, the uncertainty of the climate change impact is small compared to the uncertainty of the adaptation measures. All four adaptation options involve planning from public authorities. They are ranked in Table 5 according to the degree of technical/public planning relative to the public involvement. The choice of optimum adaptation is to some extent based on this ranking, which is why ambiguity is an important additional uncertainty. Equity issues between the few percent of property owners with very high increases in risk and the vast majority of property owners with a very low and negligible increase in risk may also influence the adaptation method. Using solely economic evaluation methods, the increasing drainage capacity option can clearly be recommended. However, using other objectives the other options may turn out favourably.

5 Discussion of climate change adaptation strategies under uncertainty

Our analyses of uncertainties, risks and adaptation characteristics have revealed a number of key messages that are discussed below.

5.1 Climate change adaption decisions needed in spite of large uncertainties

It is unquestionable that the uncertainty cascade includes many sources and that their propagation through technical and socio-economic models further adds to the overall prediction uncertainties of climate change impacts. It is therefore sometimes argued that decision making about climate change adaptation actions has to be postponed until more knowledge becomes available and less uncertain decisions can be made. We argue that, in spite of all these uncertainties, we often have sufficient knowledge to make decisions on climate change adaptations, and that such adaptation decisions could be taken now.

Example 1—Urban drainage

The design rainfall criteria for small drainage systems are expected to increase by 30% over the next 100 years. The confidence interval for this estimate is 5–75% (Arnbjerg-Nielsen 2012). This large uncertainty on design criteria has little influence, when drainage systems are constructed in new urban areas, because the marginal cost of implementing extra drainage capacity is very low. Hence, often an increase of 50% drainage capacity can be achieved for 10% of the overall cost. Therefore it will usually always be a good idea to increase the design level for newly built drainage systems.

Example 2—Agriculture/freshwater ecosystems

Several studies have shown that nitrate leaching from agriculture is likely to increase due to climate change, and that these projections are subject to considerable prediction uncertainties originating from all sources in the uncertainty cascade (Olesen et al. 2007; Børgesen and Olesen 2011). The increased nutrient load will in turn affect the freshwater ecosystems, and a relevant question is to which extent the uncertainties on future N-leaching should affect decisions on adaptation. Figure 4 shows results from a probabilistic analysis of N-leaching for sandy soils in Denmark, where uncertainties in climate projections are inferred from the ENSEMBLES (van der Linden and Mitchell 2009) results of multiple climate models for assessing N-leaching from climate data. In spite of considerable uncertainties on N-leaching (not shown) the probability of N-leaching exceeding a critical threshold (70 kg N/ha/year) is shown in Fig. 4 to follow a clear increasing trend during the coming decades. Thus, in this case, uncertainties on the underlying nitrate leaching turn out to become less important for long-term decisions, because the critical threshold here will be passed with a high probability.

5.2 Assess adaptation now as a basis for optimal timing

Despite uncertainties, there is enough information about the future climate change impacts to realise that without any adaptation the consequences will generally be negative. It is therefore necessary to consider the adaptation options now. This does not necessarily imply that actions with large investments need to be implemented right away, but that long term planning efforts should be initiated already now.

Example—Future water supply

As drinking water pumped from aquifers typically is 50–100 years old the protection measures decided through the ongoing groundwater protection programme have a long-term perspective. However, the delineated groundwater abstraction catchments for which groundwater protection action plans are being prepared may turn out to comprise the wrong areas, because these catchments are climate dependent. Although the outcome of climate change uncertainty analyses may show that the locations of the catchments are more uncertain than without considering climate change aspects, it is still important to include climate change uncertainties in the assessments now. An implementation strategy could be to design flexible action plans that can be modified, as more precise climate change predictions become available during the coming years.

5.3 Risk willingness differs among individuals and stakeholders

The willingness to accept specific risks may differ greatly between individuals and policy makers as well as among different groups of stakeholders. Stakeholders are affected by risks in different ways, e.g. there is a difference between being a house owner who could be directly affected by extreme weather, because s/he lives near the coast, and another home owner who is not directly exposed to the risks. Furthermore, the temporal and spatial scope for many stakeholders is likely to be short-term and/or localized and dependent on the perceived risks they are subject to personally, while policy makers are inclined to focus on the expected risks across the perspective of all stakeholders in society and on long-term and/or widespread adaptation measures.

Example—Agriculture and freshwater ecology

Decision making on adaptation measures to ensure good ecological status will involve stakeholders (agriculture, environment) with different interests and perceptions of what constitutes the most important problem. This may result in considerable ambiguity, which needs to be reduced to achieve a political decision. In such situations it is important to separate the different types (natures) of uncertainty, because epistemic uncertainty can be reduced by obtaining more knowledge, while ambiguity requires dialogue to achieve better mutual understanding.

5.4 Risk strategies should not be based on status quo attitudes to risk acceptance

It is necessary to discard the “status quo” as the base for risk acceptance. As the present risk attitude is a function of present societies and climate, so will the future risk strategies be a function of future conditions. However, it is very difficult to predict risk attitudes of future generations. Neglecting to address the uncertainties and risks properly is likely to lead to short term solutions, which, at best, could prove insufficient on the long term.

Example—inundations of roads

The standard for road designs today is that roads should not be inundated. On the other hand it is accepted that under some extreme weather conditions roads may be closed, e.g. due to snow or due to heavy wind for some of the large bridges. Future climate projections show very large uncertainties for extreme precipitation events. The key uncertainty here is aleatory by nature implying that we somehow have to live with it. If future roads were to be designed to have a very high certainty of completely avoiding inundations, it would require very costly road constructions that most likely will turn out to be designed very much with a conservative view of risk. Hence, it is likely that the construction costs in some cases can be reduced significantly by allowing roads, on rare occasions, to be inundated, and implement an online forecasting system to warn the traffic about closing of the road—equivalent to the situation for heavy snow or heavy wind at some bridges. Such change in design philosophy will require a change in the risk acceptance among policy makers and hence also among the general public.

6 Discussion of novelty in relation to IPCC

Although the definitions of uncertainty in the IPCC AR4 Synthesis Report (IPCC 2007a) and in the AR5 guidance note (Mastrandrea et al. 2011) are not conflicting with our terminology, the focus in the IPCC uncertainty work differs from our uncertainty framework. The IPCC uncertainty guidance focuses on development of a language suitable for communication of uncertainty based on evidence and agreement. While our framework cannot replace IPCC guidance due to its lack of focus on communication, it offers a more fundamental characterisation of uncertainty with focus on the nature of uncertainty (whether reducible or non-reducible), the level of uncertainty (whether it can be described statistically, as scenarios, qualitatively or is due to ignorance) and the source (origin) of uncertainty. Our framework with its more comprehensive uncertainty characterisation addresses the concerns raised by Curry (2011) who argues that the present IPCC approach is oversimplified and can lead to misleading overconfidence.

Another particular strength of our framework is that it is suitable for adaptation planning. Our examples have illustrated how uncertainty may or may not matter very much, depending on the context and how the characterisation of uncertainty into different sources and nature of uncertainty can help in decisions on adaptation strategies.

7 Summary and conclusions

The uncertainty framework has been used to study climate change impacts and adaptation options for four water related sectors in Denmark: agriculture, freshwater ecology, water infrastructure in rural areas and urban infrastructures. We find that (i) there are considerable uncertainties on climate change impacts; (ii) the dominating sources of uncertainty differ greatly among the various problems; (iii) most uncertainties on impacts are epistemic by nature implying that they are reducible; and (iv) the uncertainties on adaptation measures are complex with ambiguity often being added on top of the impact uncertainties.

The impacts of different sources of uncertainties on the risk assessment and evaluation of adaptation options are context specific. Depending on the physical system and the economic lifetime of adaptation measures being considered, different uncertainty sources in the uncertainty cascade may dominate. The relative importance of the different uncertainty sources also depends on the temporal and spatial scales of the physical impact assessments. For instance, evaluation of urban drainage designs depends on extreme precipitation characteristics with spatial and temporal resolutions in the order of, respectively, 1–10 km2 and 1 hour or less, and hence uncertainties related to regional climate model projections and downscaling are dominating. On the other hand, for assessing the impacts on water resources on national and regional levels, the uncertainties in the GCM are often more important than those related to the downscaling. For some types of problems, uncertainties in the physical impact model may dominate compared to those in climate forcing.

For adaptation characteristics we find that (i) the two infrastructure sectors are dominated by planned adaptation measures, while the agricultural and freshwater ecology sectors include both planned and autonomous adaptation options; (ii) adaptation options include both reactive, concurrent and anticipatory actions in all four sectors; (iii) temporal scope varies from short term for all options in the freshwater ecology sector to a mixture of short-term and long-term in the other sectors; (iv) spatial scope varies between localized and widespread in all four sectors; and (v) non-structural measures are typically characterised by being reactive actions with short-term temporal and localized spatial scopes, with relatively low costs, but also with a relatively high level of ambiguity. This implies that non-structural measures, irrespective of whether they are planned or autonomous, are flexible tools, because they often can be decided reactively with short-term perspectives and localised effects, and they are relatively cheap. However, non-structural measures cannot be used to address all types of problems, and adaptation strategies will often include combinations of infrastructure investments and non-structural measures. Furthermore as non-structural measures often assumes a change of behaviour, such as accepting to live with climate effects, there often is an extra source of uncertainty related to obtaining some degree of societal consensus about a changed attitude to risk.

The uncertainty cascade includes many sources of uncertainty and their propagation through technical and socio-economic models may add substantially to prediction uncertainties. It is therefore sometimes argued that decision making should be postponed until more knowledge becomes available. However, it should be recognised that even large uncertainties may in some contexts imply small consequences for decision making. Therefore, in spite of large uncertainties, there is often sufficient knowledge (certainty) to justify action in climate adaptation (Arnbjerg-Nielsen and Fleischer 2009; Jeppesen et al. 2011; Olesen et al. 2011).

Notes

Acknowledgements

The present study was funded by a grant from the Danish Strategic Research Council for the Centre for Regional Change in the Earth System (CRES—www.cres-centre.dk) under contract no: DSF-EnMi 09–066868. CRES is a multidisciplinary climate research platform, including key Danish stakeholders and practitioners with a need for improved climate information. Erik Jeppesen was also supported by CLEAR2 and the EU project REFRESH. The contribution of Anil Markandya was in part supported by the LIFE+ programme. Constructive comments to the manuscript by Hans Jørgen Henriksen, GEUS, is acknowledged. We appreciate the constructive suggestions from two anonymous reviewers.

Open Access

This article is distributed under the terms of the Creative Commons Attribution License which permits any use, distribution, and reproduction in any medium, provided the original author(s) and the source are credited.

References

  1. Adrian R, O’reilly CM, Zagarese H, Baines SB, Hessen DO, Keller W, Livingstone DM, Sommaruga R, Straile D, van Donk E, Weyhenmeyer GA, Winder M (2009) Lakes as sentinels of climate change, Limnol. Oceanogr 54:2283–2297Google Scholar
  2. Arnbjerg-Nielsen K (2012) Quantification of climate change effects on extreme precipitation used for high resolution hydrologic design. Urban Water Journal, in press, doi: 10.1080/1573062X.2011630091
  3. Arnbjerg-Nielsen K, Fleischer HS (2009) Feasible adaptation strategies for increased risk of flooding in cities due to climate change. Water Sci Technol 60(2):273–281CrossRefGoogle Scholar
  4. Arnell NW, Livermore MJL, Kovats S, Levy PE, Nicholls R, Parry ML et al (2004) Climate and socio-economic scenarios for global-scale climate change impacts assessments: characterising the SRES storylines. Glob Environ Chang 14:3–20CrossRefGoogle Scholar
  5. Boberg F (2010) Weighted scenario temperature and precipitation changes for Denmark using probability density functions for ENSEMBLES regional climate models. Danish Climate Centre Report, http://www.dmi.dk/dmi/dkc10-03.pdf
  6. Børgesen CD, Olesen JE (2011) A probabilistic assessment of climate change impacts on yield and nitrogen leaching from winter wheat in Denmark. Nat. Haz. Earth Syst. Sci 11(9):2541–2553. doi: 10.5194/nhess-11-2541-2011
  7. Brown JD (2004) Knowledge, uncertainty and physical geography: towards the development of methodologies for questioning belief. Trans Inst Br Geogr 29(3):367–381CrossRefGoogle Scholar
  8. Burton I (2009) Climate Change and the Adaptation Deficit. In: Schipper ELF, Burton I (eds) The earthscan reader on adaptation to climate change. Earthscan, LondonGoogle Scholar
  9. Christensen JH, Christensen OB, Lopez P, van Meijgaard E, Botzet M (1996) The HIRHAM4 Regional Atmospheric Climate Model; DMI Scientific Report 96–4. [Available from DMI, Lyngbyvej 100, Copenhagen Ø, DENMARK]Google Scholar
  10. Christensen JH, Carter TR, Rummukainen M, Amanatidis G (2007a) Evaluating the performance and utility of regional climate models: the PRUDENCE project. Clim Chang 81(1):1–6. doi: 10.1007/s10584-006-9211-6 CrossRefGoogle Scholar
  11. Christensen JH, Hewitson B, Busuioc A, Chen A, Gao X, Held I, Jones R, Kolli RK, Kwon W-T, Laprise R, Magaña Rueda V, Mearns L, Menéndez CG, Räisänen J, Rinke A, Sarr A, Whetton P (2007b) Regional Climate Projections, Chapter 11 in: Climate Change 2007: The Physical Science Basis. Contribution of Working Group I to the Fourth Assessment Report of the ntergovernmental Panel on Climate Change [Solomon, S., D. Qin, M. Manning, Z. Chen, M. Marquis, K.B. Averyt, M. Tignor and H.L. Miller (eds.)]. Cambridge University Press, Cambridge, United Kingdom and New York, NY, USAGoogle Scholar
  12. Christensen JH, Kjellström E, Giorgi F, Lenderink G, Rummukainen M (2010) Weight assignment in regional climate models. Clim Res 44:179–194. doi: 10.3354/cr00916 CrossRefGoogle Scholar
  13. Cooke GD, Welch EB, Peterson SA, Nicholas SA (2005) Restoration and management of lakes and reservoirs. CRC Press, Boca Raton, FloridaCrossRefGoogle Scholar
  14. Curry J (2011) Reasoning about climate uncertainty. Clim Chang 108:723–732CrossRefGoogle Scholar
  15. Danish Energy Agency (2008) Strategi for tilpasning til klimaændringer I Danmark (In Danish—Stragy for climate change adaptation in Denmark). Danish Energy Agency, CopenhagenGoogle Scholar
  16. Déqué M, Somot S (2010) Weighted frequency distributions express modelling uncertainties in the ENSEMBLES regional climate experiments. Clim Res 44:195–209CrossRefGoogle Scholar
  17. EU Commission (2009) White Paper. Adapting to climate change: Towards a European framework for action. Downloaded January 2011 from http://eur-lex.europa.eu/LexUriServ/LexUriServ.do?uri=COM:2009:0147:FIN:EN:PDF
  18. Foley AM (2010) Uncertainty in regional climate modelling: A review. Prog Phys Geogr 34(5):647–670CrossRefGoogle Scholar
  19. Fowler HJ, Blenkinsop S, Tebaldi C (2007) Review Linking climate change modelling to impacts studies: recent advances in downscaling techniques for hydrological modelling. Int J Climatol 27:1547–1578CrossRefGoogle Scholar
  20. Hansson L-A, Annadotter H, Bergman E, Hamrin SF, Jeppesen E, Kairesalo T, Luokkanen E, Nilsson P-Å, Søndergaard M, Strand J (1998) Biomanipulation as an application of food-chain theory: constraints, synthesis, and recommendations for temperate lakes. Ecosystems 1:558–574CrossRefGoogle Scholar
  21. Hawkins E, Sutton R (2009) The potential to narrow uncertainty in regional climate predictions. Billetin of the American Meteorological Society 90:1095–1107CrossRefGoogle Scholar
  22. Hulme M, Carter TR (1999) Representing uncertainty in climate change scenarios and impact studies. In: Carter TR, Hulme M, Viner D (eds) Representing uncertainty in climate change scenarios and impact studies. Proceedings of the ECLAT-2 Helsinki Workshop, 14–16 April 1999, Climatic Research Unit, Norwich, p 11–37Google Scholar
  23. IPCC IPCC (2007a) Climate Change 2007: Synthesis Report. Contribution of Working Groups I, II and III to the Fourth Assessment Report of the Intergovernmental Panel on Climate Change [Core Writing Team, Pachauri, R.K and Reisinger, A. (eds.)]. IPCC, Geneva, Switzerland, 104ppGoogle Scholar
  24. IPCC (2007b) Climate Change 2007: Impacts, Adaptation and Vulnerability. Contribution of Working Group II to the Fourth Assessment Report of the Intergovernmental Panel on Climate Change, M.L. Parry, O.F. Canziani, J.P. Palutikof, P.J. van der Linden and C.E. Hanson, Eds., Cambridge University Press, Cambridge, UK, 976ppGoogle Scholar
  25. IPCC (2007c) Climate Change 2007: The Physical Science Basis. Contribution of Working Group I to the Fourth Assessment Report of the Intergovernmental Panel on Climate Change [Solomon, S., D. Qin, M. Manning, Z. Chen, M. Marquis, K.B. Averyt, M. Tignor and H.L. Miller (eds.)]. Cambridge University Press, Cambridge, United Kingdom and New York, NY, USA, 996ppGoogle Scholar
  26. Jeppesen E, Kronvang B, Meerhoff M, Søndergaard M, Hansen KM, Andersen HE, Lauridsen TL, Liboriussen L, Beklioglu M, Özen A, Olesen JE (2009) Climate change effects on runoff; catchment phosphorous load and lake ecological state, and potential adaptations. Journal of Environmental Qualitu 38:1930–1941CrossRefGoogle Scholar
  27. Jeppesen E, Meerhoff M, Holmgren K, González-Bergonzoni I, Teixeira-de Mello F, Declerck SAJ, De Meester L, Søndergaard M, Lauridsen TL, Bjerring R, Conde-Porcuna JM, Mazzeo N, Iglesias C, Reizenstein M, Malmquist HJ, Liu ZW, Balayla D, Lazzaro X (2010) Impacts of climate warming on lake fish community structure and potential ecosystem effects. Hydrobiologia 646:73–90CrossRefGoogle Scholar
  28. Jeppesen E, Kronvang B, Olesen JE, Søndergaard M, Hoffmann CC, Andersen HE, Lauridsen TL, Liboriussen L, Meerhoff M, Beklioglu M, Özen A (2011) Climate change effect on nitrogen loading from catchment in Europe: implications for nitrogen retention and ecological state of lakes and adaptations. Hydrobiologia 663:1–21CrossRefGoogle Scholar
  29. Katz RW (2002) Techniques for estimating uncertainty in climate change scenarios and impact studies. Clim Res 20:167–185CrossRefGoogle Scholar
  30. Kilsby CG, Jones PD, Burton A, Ford AC, Fowler HJ, Harpham C, James P, Smith A, Wilby RL (2007) A daily weather generator for use in climate change studies. Environ Model Softw 22:1705–1719CrossRefGoogle Scholar
  31. Klauer B, Brown JD (2003) Conceptualising imperfect knowledge in public decision making: ignorance, uncertainty, error and ‘risk situations’. Environmental Research, Engineering and Management 27(1):124–128Google Scholar
  32. Kristensen K, Schelde K, Olesen JE (2011) Winter wheat yield response to climate variability in Denmark. J Agric Sci 149:33–47CrossRefGoogle Scholar
  33. Lebel L, Grothmann T, Siebenhüner B (2010) The role of social learning in adaptiveness: insights from water management. International Environmental Agreements 10:333–353CrossRefGoogle Scholar
  34. Mastrandrea MD, Mach KJ, Plattner G-K, Edenhofer O, Stocker TF, Field CB, Ebi KL, Matschoss PR (2011) The IPCC AR5 guidance note on consistent treatment of uncertainties: a common approach across the working groups. Clim Chang 108:675–691CrossRefGoogle Scholar
  35. Olesen JE, Carter TR, Diaz-Ambrona CH, Frontzek S, Heidmann T, Hickler T, Holt T, Minguez MI, Morales P, Paalutikof JP, Quemada M, Ruiz-Ramos M, Rubæk GH, Sau F, Smith B, Sykes MT (2007) Uncertainties in projected impacts of climate change on European agriculture and terrestrial ecosystems base don scenarios from regional climate models. Clim Chang 81:123–143CrossRefGoogle Scholar
  36. Olesen JE, Trnka M, Kersebaum KC, Skjelvåg AO, Seguin B, Peltonen-Saino P, Rossi F, Kozyra J, Micale F (2011) Impacts and adaptation of European crop production systems to climate change. Eur J Agron 34:96–112CrossRefGoogle Scholar
  37. Paerl HW, Huisman J (2008) Blooms like it hot. Science 320:57–58CrossRefGoogle Scholar
  38. Pahl-Wostl C (2007) Transistions towards adaptive management of water facing climate and global change. Water Resources Management 21(1):49–62CrossRefGoogle Scholar
  39. Porter JR, Semenov MA (2005) Crop responses to climatic variability. Philosophical Transactions of the Royal Society B 360:2021–2035CrossRefGoogle Scholar
  40. Preston BL, Westaway RM, Yuen EJ (2011) Climate adaptation planning in practice: an evaluation of adaptation plans from three developed nations. Mitig Adapt Strateg Glob Chang 16:407–438CrossRefGoogle Scholar
  41. Refsgaard JC, Henriksen HJ (2004) Modelling guidelines—terminology and guiding principles. Advances in Water Resources 27(1):71–82CrossRefGoogle Scholar
  42. Refsgaard JC, van der Sluijs JP, Højberg AL, Vanrolleghem PA (2007) Uncertainty in the environmental modelling process—A framework and guidance. Environ Model Softw 22:1543–1556CrossRefGoogle Scholar
  43. Rosenzweig C, Wilbanks TJ (2010) The state of climate change vulnerability, impacts, and adaptation research: strengthening knowledge base and community. Clim Chang 100:103–106CrossRefGoogle Scholar
  44. Rötter RP, Carter TR, Olesen JE, Porter JR (2011) Crop–climate models need an overhaul. Nature Climate Change 1:175–177CrossRefGoogle Scholar
  45. Schröter D, Cramer W, Leemans R, Prentice IC, Araújo MB, Arnell NW, Bondeau A, Bugmann H, Carter TR, Gracia CA, de la Vega-Leinert AC, Erhard M, Ewert F, Glendining M, House JI, Kankaapää S, Klein RJT, Lavorel S, Lindner M, Metzger MJ, Meyer J, Mitchell TD, Reginster I, Rounsevell M, Sabaté S, Sitch S, Smith B, Smith J, Smith P, Sykes MT, Thonicke K, Thuiller W, Tuck G, Zaehle S, Zierl B (2005) Ecosystem service supply and vulnerability to global change in Europe. Science 310:1333–1337CrossRefGoogle Scholar
  46. Smith RL, Tebaldi C, Nychka D, Mearns LO (2009) Bayesian modelling of uncertainty in ensembles of climate models. J Am Stat Ass 104:97–116CrossRefGoogle Scholar
  47. Sunyer MA, Madsen H, Ang PH (2011) A comparison of different regional climate models and statistical downscaling methods for extreme rainfall estimation under climate change. Atm Res. doi: 10.1016/j.atmosres.2011.06.011
  48. Tebaldi C, Smith R, Nychka D, Mearns L (2005) Quantifying uncertainty in projections of regional climate change: a Bayesian approach to the analysis of multi-model ensembles. J Climate 18:1524–1540CrossRefGoogle Scholar
  49. van der Keur P, Henriksen HJ, Refsgaard JC, Brugnach M, Pahl-Wostl C, DeWulf A, Buiteveld H (2008) Identification of major sources of uncertainty in current IWRM practice. Illustrated for the Rhine basin. Water Resources Management 22:1677–1708CrossRefGoogle Scholar
  50. van der Linden P, Mitchell JFB (eds) (2009) Summary of research and results from the ENSEMBLES project. ENSEMBLES: climate change and its impacts. Met Office Hadley Centre, ExeterGoogle Scholar
  51. van Roosmalen L, Christensen BSB, Sonnenborg TO (2007) Regional differences in climate change impacts on groundwater and stream discharge in Denmark. Vadose Zone Journal 6:554–571CrossRefGoogle Scholar
  52. Walker WE, Harremoës P, Rotmans J, Van der Sluijs JP, Van Asselt MBA, Janssen P, Krayer von Krauss MP (2003) Defining uncertainty a conceptual basis for uncertainty management in model-based decision support. Integr Assess 4(1):5–17CrossRefGoogle Scholar
  53. Warmink JJ, Janssen JAEB, Booij MJ, Krol MS (2010) Identification and classification of uncertainties in the application of environmental models. Environ Model Softw 25:1518–1527CrossRefGoogle Scholar
  54. Wilby RL, Dawson CW, Barrow EM (2002) SDSM—A decision support tool for the assessment of regional climate change impacts. Environ Model Softw 17:147–159CrossRefGoogle Scholar
  55. Yohe G, Oppenheimer M (2011) Evaluation, characterization, and communication of uncertainty by the intergovernmental panel on climate change—an introductory essay. Clim Chang 108:629–639CrossRefGoogle Scholar

Copyright information

© The Author(s) 2012

Authors and Affiliations

  • J. C. Refsgaard
    • 1
  • K. Arnbjerg-Nielsen
    • 2
  • M. Drews
    • 3
  • K. Halsnæs
    • 4
  • E. Jeppesen
    • 5
  • H. Madsen
    • 6
  • A. Markandya
    • 7
  • J. E. Olesen
    • 8
  • J. R. Porter
    • 9
  • J. H. Christensen
    • 3
  1. 1.Geological Survey of Denmark and Greenland (GEUS)København KDenmark
  2. 2.Department of Environmental EngineeringTechnical University of DenmarkLyngbyDenmark
  3. 3.Danish Meteorological InstituteKøbenhavn ØDenmark
  4. 4.Technical University of DenmarkRoskildeDenmark
  5. 5.Aarhus UniversitySilkeborgDenmark
  6. 6.DHIHørsholmDenmark
  7. 7.BC3 Basque Centre for Climate ChangeBilbaoSpain
  8. 8.Aarhus UniversityTjeleDenmark
  9. 9.University of CopenhagenTaastrupDenmark

Personalised recommendations