Sunteți pe pagina 1din 8

12

nd
International Conference on Urban Drainage, Porto Alegre/Brazil, 11-16 September
Page 1 of 8
Accounting for climate change in urban drainage and flooding:
contrasting alternative approaches to devising adaptive strategies
B. Gersonius
1
*, R. Ashley
1
, A. Jeuken
2
, A. Pathirana
1
and C. Zevenbergen
1



1
UNESCO-IHE Institute for Water Education, Westvest 7, Delft, 2611 AX, The Netherlands.
2
Deltares, Daltonlaan 400, Utrecht, 3584 BK, The Netherlands.
*Corresponding author, e-mail b.gersonius@unesco-ihe.org

ABSTRACT
Two frameworks are presented that can be used to account for climate change uncertainty in
investment decision making related to urban drainage and flooding systems, either cause-
based or effect-based. In the former, Real-In-Options (RIO) is proposed as an approach to
identify the optimal set of adaptive strategies in response to advances in knowledge about
future climate change. Other approaches align with the effect-based framework. A relatively
simple approach to implementing this framework is to use Adaptation Tipping Points (ATPs),
which have been defined as the points where the magnitude of climate change is such that the
current strategy can no longer meet the pre-set objectives. Responses for this approach aim to
extend the location and timing of these ATPs to some acceptable future point. This paper
compares the RIO approach with the ATP approach. The emphasis is on the procedural steps,
benefits and limitations. The paper concludes with a summary of the key characteristics to
assist in choosing the most appropriate approach.


KEYWORDS
Adaptation Tipping Points; climate change; flood risk management; Real-In-Options; urban
drainage


1. INTRODUCTION
Decision making for investments in urban drainage and flooding systems needs to take
account of climate change uncertainty. This is because of two features associated with such
systems. First, the consequences of investment decisions for these systems have to be lived
with for a long time, which means that the associated uncertainties can grow larger. Second,
potential irreversibilities in choices can lead to a need for larger construction initially,
particularly in 'hard' structural measures; which allows for headroom for later adjustment.
New approaches to devising adaptive strategies are needed to address these uncertainties
(Kundzewicz et al., 2008). Otherwise, such strategies can be maladaptive, resulting in
unnecessary costs of potentially irreversible measures (Barnett and O'Neill, 2010). Two
frameworks are presented that can be used to account for climate change uncertainty in
investment decision making related to urban drainage and flooding systems (Jones and
Preston, 2010), framed around a Driver-Pressure-State-Impact-Response (DPSIR) continuum.
The cause-based framework begins by considering the climate system (drivers) and moves
through the pressures, state process to predict the impacts. Responses are then formulated to
cope with these in a way that maintains expected performance levels. The alternative, effect-
based framework starts with quantifying the outcome in the form of impact thresholds that
define possible system states, and then the likelihood of attaining or exceeding this outcome is
assessed.
12
nd
International Conference on Urban Drainage, Porto Alegre/Brazil, 11-16 September
Page 2 of 8
The cause-based framework is most widely used in practice. For instance, the conventional
Net Present Value (NPV) approach uses climate scenarios to predict the impacts and to
develop adaptive strategies based on these impacts. The limitation of the conventional NPV
approach is the reliance on best estimate climate scenarios, which are expected to be precise
forecasts of future climate change. However, such precise forecasts cannot be produced by
climate modelling (Dessai et al., 2009). Where irreducible uncertainties exist, adaptive
strategies can still be developed in the face of these uncertainties. Real-In-Options (RIO) is a
relatively new approach that can be used for managing future flood risk in the cause-based
framework, which explicitly acknowledges and allows for the lack of certainty about climate
change by building in flexibility into infrastructure systems (Gersonius et al., submitted).
Other approaches align with the effect-based framework, and they can be carried out almost
independently of climate scenarios. For example, Kwadijk et al. (in press) have developed a
relatively simple approach based on the concept of Adaptation Tipping Points (ATPs) to
assess whether, and for how long, the current strategy or alternative, adaptive strategies will
continue to be effective under different climate conditions. This paper compares the RIO
approach with the ATP approach, based on literature and our implementation experience for
two case studies. The emphasis is on the procedural steps, benefits and limitations. The paper
concludes with a summary of the key characteristics to assist in choosing the most appropriate
approach.


2. REAL-IN-OPTIONS APPROACH
Real Options (RO) is a recognised procedure to handle uncertainties in infrastructure
investments through valuing flexibility. Originally developed as a means to predict the value
of financial options, RO analysis integrates expected changes in future levels of uncertainty
into economic analysis. It therefore offers a major development on the conventional Net
Present Value (NPV) approach. RO can be categorised as those that are either on or in
systems (De Neufville, 2003). RO on systems are options applied to the investment decision
making process related to infrastructure systems, treating the infrastructure design as a black
box. RO in systems (or: RIO) however, are created by changing the infrastructure design as
uncertainty is resolved. In the RIO concept, the option is thus based on a technical
characteristic of the infrastructure system that is not apparent if the system is treated as a
black box. For RIO analysis, a number of characteristics are required that traditional RO
approaches do not deal with, such as technical details and interdependency/path-dependency
among options. Wang and de Neufville (2004) have proposed an optimisation-based RIO
analysis procedure that is able to manage such interdependency/path-dependency features.
Their procedure is based on the scenarios established by a binomial (path-dependent) tree.

Framework used
RIO is a relatively new concept to account for uncertainty in the cause-based framework. This
approach uses climate scenarios, as climate is the main driver of the impacts, to devise an
optimal set of adaptive strategies. The development of RIO analysis provides a procedure to
find out which flexibilities, that permit the infrastructure design to be adjusted over time, are
worth their cost. The value of flexibility is determined based on a probability distribution for
possible system states in future epochs. This approach assumes that the uncertainty cannot be
completely resolved over time, but rather that due to advances in knowledge the probability
distribution will be adjusted in the future. The adjustment of probability distributions for
climate change plays a significant role in informing the size and timing of investments in
adaptation. Such distributions can either be based on climate modelling data or subjectively
formulated by experts.
12
nd
International Conference on Urban Drainage, Porto Alegre/Brazil, 11-16 September
Page 3 of 8
Procedural steps
The procedure for RIO analysis has been developed by Wang and De Neufville (2004), and
later modified by Gersonius et al. (2010) for the context of adapting urban drainage and
flooding systems to climate change. It comprises the following steps:
1. Define system and objectives
2. Quantify drivers and possible system states
3. Identify potential options for adapting
4. Simulation and options analysis
5. Optimisation by GAs

Gersonius et al. (submitted) provide an example of the application of RIO analysis to the
modification an urban drainage system in West Garforth, England. The system consists of 85
sewer conduits, 9 possible storage facilities and 15 sub catchments. Associated with these
components are the design variables, which define possible system configurations. The design
variables can be changed over time, reflecting an incremental process of adaptation to climate
change with a flexible design. There are different ways to build in the flexibility into the
infrastructure design. In the example used, flexibility arises from the possibility of replacing
sewer conduits, upsizing storage facilities, and disconnecting back roofs in the sub
catchments. Due to the large number of design variables and their interactions, coupled with
the considerations of technical and real option constraints, cost-effective system design is
complex. In this regard, optimisation by GAs has been applied to identify the optimal set of
adaptive strategies (consisting of the first-epoch configuration and subsequent configurations)
in response to advances in knowledge about future climate change.

The optimisation model has been implemented in a computer program written in C++ which
has then been used for the identification and analysis of possible strategies. The objective
function was to minimise the expected life cycle cost, subject to the condition that there is
sufficient hydraulic capacity in place in each epoch and each climate state to always meet the
protection standard in that state. In West Garforth, the protection standard is: no surface
flooding for a design storm with a 1 in 30 year recurrence interval. However, due to climate
change, there is uncertainty about the intensity of the design storm for future epochs. To
represent this uncertainty it is assumed that the change in intensity follows a geometric
Brownian motion (GBM). This assumption has the property that the variance of the uncertain
parameter increases over time. GBM is one of the most important basic notions of stochastic
processes, and in particular, is the basis of options theory. The input parameters for the GBM
model used were: a drift rate of 0.018% per year and a volatility of 1.573% per year for
the change in intensity of the design storm. This was obtained from the climate modelling
data available from the UKCP09 probabilistic projections (Murphy et al., 2009). The outputs
from the RIO analysis model give the optimal set of adaptive strategies over the three defined
epochs (1990s-2020s, 2020s-2050s, and 2050s-2080s) (Table 1). The resulting strategy is to
build configuration A1 in the first epoch, to build configuration A2 if the intensity goes up by
13%, and to build configuration A3 only if the intensity increases by 28%.
12
nd
International Conference on Urban Drainage, Porto Alegre/Brazil, 11-16 September
Page 4 of 8
Table 1. Optimal set of adaptive strategies
Start of epoch 1 (1990s-2020s) Start of epoch 2 (2020s-2050s) Start of epoch 3 (2050s-2080s)

Climate change factor = 1.28
Configuration A3 build
Climate change factor = 1.13
Configuration A2 build
Climate change factor = 1.13
No build
Climate change factor = 1.00
No build
Climate change factor = 1.13
Configuration A2 build
Climate change factor = 1.00
Configuration A1 build
Climate change factor = 1.00
No build
Climate change factor = 1.00
No build
Climate change factor = 0.89
No build
Climate change factor = 1.00
No build
Climate change factor = 0.89
No build
Climate change factor = 0.89
No build
Climate change factor = 0.78
No build

Benefits and limitations
RIO analysis has considerable potential to support investment decision making for urban
drainage and flooding systems, particularly if the required adaptation is particularly sensitive
to the magnitude of climate change. The benefit of the approach is that it identifies the
optimal set of adaptive strategies in reaction to changes in knowledge about future climate
change. This will allow the size and timing of investments to be informed by new information
regarding climate change, and so improve the economic efficiency of such investments. The
total life cycle costs will therefore likely be lower, as demonstrated by some recent
applications (Gersonius et al., submitted; Woodward et al., 2010). Using RIO also reduces
initial capital costs and spreads the costs over the lifetime of the infrastructure system which
is important, especially in a time of economic stringency.

The main limitation of the application of RIO is that it requires accurate probability
distributions for climate change. The accuracy of climate projections is, however, limited by
fundamental, irreducible uncertainties (Dessai et al., 2008). The uncertainties associated with
climate projections arise from model errors, internal variability, and emissions scenario
uncertainty (Cox and Stephenson, 2007). Whilst some of these uncertainties can potentially be
reduced by more research (e.g., model errors), other uncertainties simply cannot (e.g.,
emissions scenario uncertainty). This means that there will always be some level of
irreducible uncertainty associated with predictions of climate change. Other limitations of the
RIO approach are that it is complicated and time-intensive to develop and run (Water Utility
Climate Alliance, 2010). This is due to the potentially large number of design variables,
epochs and boundary conditions. In addition, there is no procedure for monitoring information
in order to adjust or update/reassess the resulting adaptive strategy over time (Rahman,
Walker and Marchau, 2008).


3. TIPPING POINT APPROACH
The ATP approach is aimed at assessing whether, and for how long, the current strategy or
alternative, adaptive strategies will continue to be effective under future climate conditions. It
uses the concept of ATPs, reached if the magnitude of climate change is such that the current
strategy can no longer meet the pre-set objectives (Kwadijk et al., in press).

12
nd
International Conference on Urban Drainage, Porto Alegre/Brazil, 11-16 September
Page 5 of 8
Framework used
The ATP approach is effect-based. In this, an impact threshold is defined according to the
imposed standards or decided upon by the stakeholders and the likelihood of attaining or
exceeding this threshold is assessed. The approach starts the analysis with the policy
objectives, which determine the maximum allowable impact. The current strategy to achieve
these objectives is also defined. This is followed by a sensitivity analysis of the strategy in
dealing with future climate conditions to determine the system boundary conditions (state)
where the impact thresholds are exceeded. Up to this point the analysis is independent of
climate scenarios. Next, the state of the system is related to pressures in terms of the climate
scenarios. The climate scenarios are used to define the moment in time when an alternative,
adaptive strategy will be needed. Analysing the potential options for adapting and the ATPs
will result in the definition of a number of adaptive strategies. Once a realistic and acceptable
adaptive strategy has been selected, this strategy will then need to be continually monitored
and reviewed.

Procedural steps
The procedure for the ATP approach, following Kwadijk et al. (in press), is:
1. Define system and objectives
2. Quantify the maximum allowable Impacts
3. Assess under which boundary conditions the impact thresholds are exceeded
4. Identify potential options for adapting
5. Repeat step 2-3 for the identified set of options

An example of the ATP approach is provided by Gersonius et al. (in preparation), who have
applied it to a minor drainage system in the neighbourhood Wielwijk in Dordrecht, the
Netherlands, comprised of a combined sewer network. The relevant objective here was to
provide adequate flood protection. According to the protection standard, the minor drainage
system is expected to be large enough to convey the full flow for a 1 in 2 year recurrence
interval event without surface flooding. Hydrological and hydraulic simulation of the sewer
network was used to assess the system boundary conditions, where the system fails to meet
the protection standard. The location of the ATP for the current strategy is shown in Figure 1
(indicated by the end of the horizontal bar). It can be seen from the location of the ATP that,
in the current situation, the minor drainage system fails to meet the protection standard
determined by the municipality.

KNMI G scenario
KNMI W scenario
Rainfall change (%)
Current situation
Redevelopment without disconnection measures
Redevelopment with disconnection measures
2050
20 % 40 % 60 % 80 % 100 %
2050

Figure 1. ATP analysis for the minor drainage system

The next step has been to devise an incremental process of adaptation, starting with those
options that achieve quick gains for minimal cost. Therefore, the options identification aimed
to make use of opportunities for bringing together flood risk management with urban
regeneration and renewal. Taking the case where opportunities occur within the building and
development programme, Table 2 shows the planned adaptation options for the short and
medium-term horizon. These options involve disconnecting 1.16 ha of closed paved area,
10.24 ha of open paved area and 4.78 ha of roofed area, and diverting the runoff water to the
open water system. This has been proposed in a series of workshops attended by urban
12
nd
International Conference on Urban Drainage, Porto Alegre/Brazil, 11-16 September
Page 6 of 8
planners, sewerage managers, water managers (including the waterboard), project managers,
scientists, and inhabitants.

Table 2. Planned adaptation options for the short to medium-term horizon
Roof
disconnection
Disconnection of
open paved areas
Disconnection of
closed paved areas
[ha] [ha] [ha]
Reddersbuurt/Westervoeg
Tromptuinen
Wielwijkpark
Total 4.78 10.24 1.16

Assessing the ATPs for the adaptive strategy showed that the conveyance capacity of the
minor drainage system will remain effective up to a change in intensity of rainfall of 30%
after the planned adaptation options have been implemented. The KMNI06 G and W climate
scenarios (Hurk, 2007) were used to assess how long it takes until this ATP is reached.
According to the W climate scenario, this will occur around 2070. This ATP would lead to a
reconsideration of the adaptive strategy.

Benefits and limitations
The benefit of the ATP approach is that it is almost independent of climate scenarios, and in
particular of climate change probabilities. Rather, it requires a range of plausible climate
scenarios that can be used to assess the durability of the current strategy and alternative,
adaptive strategies. The ATP approach examines the effects of different possible magnitudes
of climate change on the protection standard, without considering probability distributions for
climate change. Climate change becomes relevant for investment decision making only if it
would lead to the crossing of an impact threshold. In this sense, the approach is more
dependent on stakeholder engagement to define the impact thresholds and the adaptive
strategy that is realistic and acceptable. Kwadijk et al. (in press) point out that the application
of ATPs answers the basic question of decision makers and other stakeholders: How much
climate change can the current strategy cope with? They found that expressing uncertainty in
terms of the period that the current strategy is effective (i.e. when an impact threshold will be
reached), appears more understandable for decision makers, than defining the likelihood of a
specific outcome in a specific epoch. Based on these findings, they conclude that the ATP
approach is useful to reduce the complexity of effect-based approaches. Another benefit
identified by Kwadijk et al. (in press) is that the ATP approach allows easier updating, when
new climate scenarios become available. In addition, Gersonius et al. (in preparation)
recognise that the ATP approach allows easier integration with adaptive capacity approaches.
The integration with adaptive capacity approaches includes the recognition of the wider
contexts (physical, social, political and economicl) in which adaptation has to take place. A
focus on connecting adaptation with urban dynamics could help to reduce climate adaptation
costs and to make use of synergistic benefits.

Although the outputs from the ATP approach are easier to understand for decision makers, it
can still be difficult to make a decision based on the outputs. This is because the approach
forces the decision makers to explicitly decide, through their choice of strategy, those future
climate conditions under which the system is likely to lack resilience (Lempert et al., 2004).
In addition, expressing uncertainties in time with bar charts leads to a pseudo-certainty effect.
The bar charts depict a definite limit, which is a simplification of the actual system reaction.
The approach does not represent uncertainties associated with, for example, model
simplification and the choice of the standard. The treatment of uncertainty is therefore
12
nd
International Conference on Urban Drainage, Porto Alegre/Brazil, 11-16 September
Page 7 of 8
incomplete, and there is an argument here to use other uncertainty approaches, such as Monte-
Carlo simulation, to quantify the uncertainty about the system reaction. Finally, the sensitivity
analysis over the range of future climate conditions, as needed to determine the system
boundary conditions where the impact thresholds are exceeded, can be a time-intensive
process due to the potentially lengthy run times of hydrological and hydraulic models.


4. CHOOSING AN APPROACH
Table 3 summarizes each approach, in terms of the following the key characteristics.
1. Framework: identifies the direction in which the cause and effect chain (e.g. from pressure
to state to impact) is followed in the reasoning.
2. Aim: gives the main aim of using the approach.
3. Focus: determines whether scenarios and/or stakeholders are driving the approach.
4. Scenario requirements: considers the type of scenarios needed to apply the approach.
5. Ease of development: considers the capacities and capabilities needed to develop the
approach.
6. Ease of running: considers the level of effort needed to run the approach.
7. Ease of output use: defines the ease in interpreting and making decisions based on the
results obtained.
8. Ease of updating: defines the ease in updating/reassessing the strategy, when new climate
scenarios become available.
9. Ease of integration with adaptation: defines the ease in integrating the approach with
adaptive capacity approaches.

Table 3. Key characteristics of each approach
RIO approach ATP approach
Framework Cause Effect Effect Cause
Aim Optimised strategy Socially acceptable strategy
Focus Scenario-driven Scenario-/stakeholder-driven
Scenario requirements Probabilistic scenarios Plausible scenarios
Ease of development Low High
Ease of running Low Medium
Ease of output use High Medium
Ease of updating Low High
Ease of integration with adaptation Medium High

The characteristics summarized in Table 3 can be used as a starting point for identifying
which approach to use under what circumstances. The selection of an approach will depend
on a number of factors, including: (1) the knowledge about the probabilities of climate
change; (2) the knowledge about the likely success of options; and (3) the capacities and
capabilities available. If the probabilities of the drivers from climate change are available and
there is agreement on the potential options for adapting, then the RIO approach may be most
appropriate as an attempt to identify the optimal set of adaptive strategies. This will however,
be dependent on the capacities and capabilities available to effectively use the probabilistic
climate data in optimisation procedures. If probabilities of climate change are not available
and/or there is not yet agreement on the potential options for adapting, then the ATP approach
may be most appropriate in an attempt to identify a realistic and acceptable adaptive strategy.
In the same way, this will be dependent on the capacities and capabilities available to
meaningful engage all the relevant stakeholders in the decision making process. Finally, the
type of adaptation is also important. For mainstreamed adaptation, the ATP approach is
recommended, because it allows easier integration with adaptive capacity approaches. For
12
nd
International Conference on Urban Drainage, Porto Alegre/Brazil, 11-16 September
Page 8 of 8
stand-alone adaptation, however, integration with adaptive capacity approaches is typically
less important and either approach may be used.


5. CONCLUSIONS
This paper has compared the RIO approach with the ATP approach. These approaches differ
in fundamental ways, as shown in the discussions above. It is concluded that each approach
has merits under particular circumstances. In addition, there are some cases where both
approaches could be used in conjunction. For example, at the early stages of the adaptation
planning process the ATP approach might be adopted in an attempt to short-list one or more
acceptable strategies. At a later stage, these strategies may then be analysed with the help of
RIO in order to optimise the size and timing of the investments. Lastly (where required), the
economic efficiency of the alternative, adaptive strategies can be compared in order to aid the
final decision making process.


ACKNOWLEDGEMENT
This work has been supported by the EU's Interreg IVB project MARE.


REFERENCES
Barnett J. and ONeill S. (2010). Maladaptation. Global Environmental Change, 20, 211-213.
Cox P. and Stephenson D. (2007). A changing climate for prediction. Science, 317, 207.
Dessai S., Hulme M., Lempert R. and Pielke Jr R. (2008). Climate prediction: a limit to adaptation. Living with
climate change: are there limits to adaptation, 49-57.
Gersonius B., Ashley R., Pathirana A. and Zevenbergen C. (2010) Managing the flooding system's resiliency to
climate change. Proceedings of the ICE-Engineering Sustainability, 163, 15-23.
Gersonius B., Ashley R., Pathirana A. and Zevenbergen C. (submitted) Climate change, real or not; building
flexibility into water infrastructures. Submitted to Climatic Change.
Gersonius B., Ashley R., Nasruddin F., Pathirana A. and Zevenbergen C. (in preparation). A hybrid approach to
devising climate adaptation: application to a stormwater system in Dordrecht, the Netherlands.
Hurk van den B. (2007). New climate change scenarios for the Netherlands. Water science and technology, 56,
27-33.
Jones R.N. and Preston B.L. (2010) Adaptation and risk management. Climate Change Working Paper No. 15.
Centre for Strategic Economic Studies, Victoria University, Melbourne, Australia.
Kundzewicz Z., Mata L., Arnell N., Doll P., Jimenez B., Miller K., Oki T., Sen Z. and Shiklomanov I. (2008).
The implications of projected climate change for freshwater resources and their management.
Hydrological Sciences Journal/Journal des Sciences Hydrologiques, 53, 3-10.
Kwadijk J., Haasnoot M., Mulder J., Hoogvliet M., Jeuken A., Krogt R., Oostrom N., Schelfhout H., Velzen E.,
Waveren H. and Wit M. (in press) Adapting to sea-level rise in the Netherlands. Submitted to Wiley
Interdisciplinary Reviews: Climate Change.
Lempert R., Nakicenovic N., Sarewitz D. and Schlesinger M. (2004). Characterizing Climate-Change
Uncertainties for Decision-Makers. An Editorial Essay. Climatic Change, 65, 1-9.
Murphy J., Sexton D., Jenkins G., Booth B., Brown C., Clark R., Collins M., Harris G., Kendon E. and Betts R.
(2009). UK Climate Projections Science Report: Climate Change Projections. Met Office Hadley
Centre, Exeter, UK.
Neufville R. (2003). Real Options: Dealing With Uncertainty in Systems Planning and Design. Integrated
Assessment, 4, 26-34.
Rahman S., Walker W. and Marchau V. (2008). Coping with Uncertainties about Climate Change in
Infrastructure Planing An Adaptive Policymaking Approach. Final Report.
Water Utility Climate Alliance. (2010). Decision Support Planning Methods: Incorporating Climate Change
Uncertainties into Water Planning.
Wang T. and Neufville de R. (2004). Building Real Options into Physical Systems with Stochastic Mixed-
Integer Programming, pp. 1719.
Woodward M., Gouldby B., Kapelan Z., Khu S.T. and Townend I. (2010). The use of real options in optimum
flood risk management decision making.

S-ar putea să vă placă și