Communication about Water Management in the Australian
Capital Territory: A System Dynamics Modelling Approach
Sondoss El Sawah ', Alan McLucas ' and Jason Mazanov'
"University of New South Wales, Australian Defense Force Academy
Northcott Drive, Canberra ACT 2600 Australia
Telephone: +61 2 6268 8015, Facsimile: +61 2 6268 8666
s.elsawah@student.adfa..edu.au, a.mclucas@adfa.edu.au, j.mazanov(@adfa.edu.au
Abstract
The Australian Capital Territory (ACT) is increasingly threatened by risks to its water security driven
by climate change effects, growing population and water-intensive lifestyle. As an inland territory, the
ACT has few supply options which are economically and ecologically expensive. Demand management
strategies seek to deliver sustainable consumption patterns. Effective communication is an essential part
Jor achieving resilient reductions in consumption by raising public understanding of the problem in order
to inform decision making, stimulate public dialogue and ultimately promote behavioural changes.
Whereas System Dynamics is a promising approach for learning and communication about water
management, its potential for communicating systemic risks to the public has not been fully exploited yet.
This ongoing research aims to build a SD Interactive Learning Environment (ILE) to help residents in
the ACT to develop a systemic perspective about water management inherent complexity and uncertainty.
This paper describes a structured modelling process adopted to build the model through a series of
Ruowledge elicitation cycles, including interviews with stakeholders and electronic workbooks. A key
lesson can be taken from our experience through this modelling effort that the modelling process must be
flexible and adaptable with several research and real world trade-offs.
Keywords: Water management, knowledge elicitation, Australian Capital Territory region
1 Introduction
Recently, the Australian Capital Territory (ACT) has found its water supply increasingly
threatened by prolonged drought, bushfires and increasing consumer demand. In future,
risks will escalate as a consequence of climate change effects and continued population
growth. Strategies which seek to reduce the per capita demand for water, and ideally to
limit total consumption, appear to be obvious ways of achieving water resource
sustainability.
Programs which focus on improving the effectiveness in communicating to consumers
how water availability is threatened are considered to be an essential part managing
demand and ultimately consumption. Effective communication aims to increase public
understanding of the problem whilst better informing local decision making and public
acceptance of strategies which might be imposed in future to manage scarce water
resources. As this problem is inherently complex, effective risk communication is
problematic both for those developing management strategies and those who may have
such strategies imposed upon them.
Advances in computer modelling and simulation, supported by rapid progress in
Information and Communications Technologies (ICT) provide unprecedented
opportunities for demonstrating to the public the complex interrelationships between
supply and demand for water. System Dynamics (SD) modelling is a promising approach
for enabling learning and communication. While SD is often used to analyse the
dynamics of systemic problems and assess policy options, its potential for
communicating systemic risks to the public has not been fully exploited.
The ongoing research described in this paper aims to develop a SD-based Interactive
Learning Environment (ILE) to help ACT residents develop a systemic perspective of
the water management problem and to demonstrate plausible futures they might face.
Expected outcomes of this research are an understanding of options for facilitating
dialogue among stakeholders, effectively promoting water-wise attitudes and ultimately
influencing consumer behaviour.
This paper describes the methodology used to collect, analyse and metge views elicited
from stakeholders (i.e. users and managers) and to form them into a conceptual causal
feedback representation of the problem. This representation forms the basis for
subsequent stages in which quantitative models and computer simulations will be
developed. The methodology has five stages. Firstly, a preliminary conceptual model
was developed based on the systems thinking (ST) and SD literatures. Secondly, local
and expert knowledge was captured by eliciting the perceptions of water users (n=25)
and managers (n=10) in the ACT using semi-structured interviews. Cognitive Mapping
techniques were used to depict participants’ perceptions. Third, noting the potential
pitfalls of doing so, the various views were represented in a single conceptual model.
Fourth, an electronic workbook was also used to investigate stakeholders’ perceptions of
the causal relationships between selected strongly coupled variables and for obtaining
estimates of selected parameters needed for subsequent quantitative modelling activities.
Finally, a series of influence diagrams were developed to capture the variables and
relationships relevant to SD model building.
The ST/SD modelling used in this research is distinguished by the following: as an action
research inquiry and that it seeks to demonstrate the recoverability criterion. That is,
whilst it is neither possible to comprehensively validate the research findings nor
precisely replicate the research elsewhere, every step in the research journey is traceable.
Hence understanding of problem and the manifestations of its complexity are enhanced
as are the insights into how to develop and implement effective risk communication
strategies.
The paper is organized as follows. Section (2) articulates the problem definition. The
adopted modelling process is described in section (3). The last section addresses the
conclusion and learning lessons.
2 Problem Definition
The Australian Capital Territory (ACT) lies entirely within the Upper Murrumbidgee
River Catchment. Being the largest urban centre in the Murray-Darling Basin, a seat of
the Commonwealth Government with a population of 360,000(Cooper, Tanner et al.
2007), the water security of the ACT is critically important. Over the last two decades the
ACT has experienced a rapid decline in its inflows (i.e. 25% below historic average). In
2003, the situation was exacerbated by bushfires unprecedented in recent history. These
bushfires burnt the vast majority of the ACT catchments. In year 2006, the ACT has
witnessed the lowest inflows in records which were worse than the worst case scenario
obtained from rainfall models. In order to meet demand, the region has significantly
drawn on the volume of water in storage. Figure 1 shows the decline in catchment
inflows (runoff) to drop below consumption in year 2006.
400000 2
A
ne
mw ol \ ok
ve ESS \ wey
Fa =
Figure 1: The decline in runoff relevant to consumption through years 1980-2008
In the ACT, water demand is mainly driven by residential use. As shown in Figure 2,
54% of water demand is used by households. About 44% of this is used outdoors as the
culture of English-life style lawns is dominant among many of the residents (Head and
Muir 2007). In addition to consumptive use, the ACT is obligated to releasing up to 269
GL as environmental flows necessary for the health of aquatic systems downstream
(Government 2004).
Queanbeyan & Other 10%
Units 6%
mnt 6% Kitchen
13% Laundry
18% Toilet
20% Bathroom
[| 39% Garden
4% Other Outdoor
Government 11%
Commerce 19%
Figure 2: The distribution of water demand in the ACT (Source: Government, 2004)
In future, the ACT will be faced by growing pressures on the water supply and demand
sides, including prolonged droughts, climate change effects and population growth.
Historically, rainfall in the Murrumbidgee catchment has shown considerable temporal
variability caused by the irregular El Nino events (Government 2004). Under climate
change scenario, the intensity and frequency of these dry-wet cycles will significantly
increase which may lead to a temporal shift in rainfall patterns. Similar shifts have already
been observed in other Australian cities, such as Perth (Marsden and Pickering 2006).
Annual rainfall is predicted to be in the range of a 9% decrease to a 2% increase while
annual evaporation is predicted to increase by between 1.4% and 9.1% (Government
2004). Whereas irrigation constitutes most of the ACT consumption, demand becomes
sensitive to increases in evaporation rate.
Management options may be broadly categorized into: increasing supply or reducing
demand. On the supply side, the ACT government continues to investigate a variety of
supply options. As an inland territory, options are limited for the ACT to initiate capital
and energy intensive projects, such as building a new dam and purchasing cross-borders
water. However, the long term sustainability of these solutions is seriously challenged by
the evolving climate change, likely ecological damage and economic uncertainty. On the
demand side, the government set targets of 12% reduction in per capita consumption by
year 2013 and 25% reductions by year 2023 (Government 2004). A combination of
demand management strategies, including price signals and water restrictions are being
used. Despite their perceived success is achieving immediate responses (Atwood,
Kreutzwiser et al. 2007), economic and regulatory instruments are not sufficient for
fostering resilient and voluntary behavioural changes. Communication strategies are
essential part for achieving long term reductions (Dietz and Stern 2002). Effective
communication programs aim to heighten public understanding of the problem in order
to inform decision making, stimulate dialogue among stakeholders and ultimately
promote behavioural changes (Renn 1998).
The current communication strategies are “hints and tips” based programs, at which the
key message is focused on the daily measures of water saving (e.g. shorter showers
campaign). In parallel, users need and have the right to develop a systemic understanding
about the problem causes, potential effects and effectiveness of different management
options (Hjorth and Bagheri 2006). Communication messages should help users to
address the following questions:
1. What factors are responsible for deriving water resource behaviour in the ACT?
2. What are the plausible futures for the ACT’ water resource?
3. How effective are different supply and demand management options?
3 The adopted Modelling Process
This research follows a structured and transparent modelling process augmented by
semantically rich “real world” interviews and cognitive mapping (Eden and Ackermann
1998), analysis of causal structures through an integrated approach using qualitative
modelling and quantitative SD modelling and simulation (Mclucas 2001; Mclucas 2003;
Mclucas 2005). The methodological roots for this process are grounded in soft
operations research and SD literatures with particular emphasis on SODA (Strategic
Options Development and Analysis), Cognitive Mapping and SD (Coyle 1996; Sterman
2000). Designed as an action research, we strive for a recoverable research process
through which the methodological details and potential outcomes are well-declared to
audiences (Checkland 1998).
The modelling process started by designing a preliminary model and cascaded through a
series of knowledge elicitation tasks in order to reach a conceptual representation of the
problem. This work was done over one year period. The overall adopted modelling
process and outcomes are depicted in Figure 3, with chronological order from [Step 1]
through [Step 13].
31 Preliminary Model Design
As a departure point, a preliminary model was created to articulate the problem based on
relevant literature [Step 1]. This model was the basis for structuring the questions used
for subsequent data collection [Step 2]. Figure 4 represents the preliminary model.
Literature Review —pocument anaysis—»| Preliminary ie)
Official Reports Model
a an
Nw anagers?
script Interview script
iG)
fs) i)
$8 Plnverviews with users a
Users’ Interview
Dee
a besiu)
) feeviews with
waite +) Usets* cognitive By-prodict insights for managers
mags risk communication 4 |
a
strategies in the AC’
BI Manager’ — | [8]
Validation interviews cognitive maps
with users »)
9)
” von El,
template for _//
Conceptual managers
Model
[3]
1:
Decomposed into sub-moduels in [12]
Influence SD Model
[11] Eworkbook >} Diagram
Figure 3: Overview of the modelling process
Sources of Uncertainiry
Natural climate variability
srowth
Climate change effects
ta
SUPPLY oe
DEMAND
Ecological
Requirements
Buildi
Purchasing
Investing in recycling plant
Water price
cation/co:
Technologic:
Policy Levers
Figure 4: The preliminary model developed at the outset of the research project
32 Eliciting Local Knowledge
Local knowledge or public perceptions constitute not only a rich and but a legitimate
problem representation (Garvin 2001). If users' perceptions provide the basis for their
behaviours then their perceptions are critical for water management. In risk
communication literature, effective communication interventions are preceded by a deep
investigation of the audiences' existing knowledge and beliefs (Morgan, Fischhoff et al.
2002). Therefore, the purpose of this knowledge elicitation task was to capture users’
perceptions about the problem causes, effects and potential mitigation strategies.
A semi-structured interview probed around a set of anchor topics was used to gain an
understanding of the extent of participants’ knowledge. The purpose of the in-depth
interviews is ideas rather than data collection (Oppenheim 1992). Much of the value of
the open ended interviews can be obtained by a relatively small sample, where a sample
of 20-30 individuals should reveal most beliefs held by any substantial frequency of the
population from which they were selected (Morgan, Fischhoff et al. 2002).
The interviewing process was conducted as two sessions. The main session (45-60
minutes) was used to data collection [Step 3]. Interviews were transcribed and organized
into cognitive maps [Step 4]. Figure 5 illustrates an example of a user's cognitive map. A
second session (20-30 minutes) was organized to validate the developed maps, refine
language ambiguities and ensure consistent terms. Users were invited to give feedback
about their cognitive maps which were updated accordingly [Step 5]. A detailed
description of this step can be found in (El Sawah et al, 2008). Findings were used to
generate more questions in the managers' interviews script [Step 6].
Employing
over Government bein, inconveniently i ‘The ACT
Goreroment Sees reonveniently_, Residents ate —p» The ACT being
failure to read the “~ late in building a Prdesigned water Yeung their garden brown-green as it
future dam restrictions die used to be
Governméft Underground :
avoiding cost of - water users py ne
large investments - building new Business growth in
—, subuxbs the ACT
ACT Wateg Security F
Releasing y Metering units for thei
environmental lows water usage..not metering
Population growth
unit
s for thei water usage
Our responsibility
of keeping water Water available in the __ Urban water 4 ___Units consumption being,
streams health ACT dams consumption abig eae demand in
the A
Growing Ea Public garden and
sports field water Waier et
consuming plants
consumption
8 such as ric velit
crops such astice ack of rafAfall N house development
and cottan Governntent being Policy
> the biggest water
demand in the
Government wasting
Natural climdte water g, automatic
variabil irrigation)
Households water
consumption
ACT
People developing,
Residents pefeciving ig don't worry/care
government wasting water attitudes towards es, av
while they are asked to save water saving — Dae.
3 water
Government G Sine perception of
paving suicene__ Government Soddpe oe
F the tank's thcouraging people
Rebate OF Fe ete tae the storage levels
price aa People affording
Government —_____» money for
pushing prices whatever they want
higher to have
Figure 5: An illustrative example of a user's cognitive map.
33 FEliciting Expert Knowledge
Expert knowledge has been increasingly recognized as an important input for informing
and guiding environmentally related decisions (Fazey, Proust et al. 2006). Through their
experience, experts have acquired extensive knowledge about the dynamic complexity of
water management and adaptation policies (Fazey, Fazey et al. 2005). At this step, we
aim to capture this wealth of knowledge using a semi-structured interviewing process
(45-75 minutes) [Step 7]. Ten highly experienced managers were recommended by the
water management authority in the ACT for participation in the study. Their expertise
covered the main business sectors including: supply, demand, and quality management.
Six participants were distinguished for their cross functional knowledge, compared to
others whose knowledge was focused on a specific area of expertise. Interviews were
transcribed and organized into cognitive maps [Step 8].
Because of the managers’ tight schedule, a second validation session could not be
organized. Alternatively, an electronic validation template was prepared to summarize the
key causal assertions extracted from their maps. Managers were asked to accept/reject
relationships and justify their choices. Cognitive maps were updated according to the
results of the validation template [Step 9]. Figure 6 illustrates an example of a manager's
cognitive map.
The fact that Corer Investment in eakging
catchments the most the ‘a én Pressure to purmore ACT being at higher
protein comparison to severe water restition level of resticons
(Queanbeyan and Gucgneby Water security inthe measures Units consumption being
Waters be ACT dam fm ee ai
es acity bein i
The decision to arg aia’ Te at iat
— Poof the ACT
get Googong dam
si \ intheg man frome consumption s
7 ; Benin | attbuted to
j te being detached houses
ACTEW aa ing the Supply passed — |
negotiating fo management vant the dovnstedl Relea Resideats earning
purchase water ——_stratepy.epending on a geservgirs users e j chews how
aes ae an aaa cavironmentl i a
licenses from ke npr mange owe advertisement how eflcenty
igators vey | canpsigns ging government and
Invesnein bia ee IE
vstenaee eesentin Rainfall sdaking into Dea -
w
nompotable , iigaton being the
eyeing pion ene thou cumin Ven uci ig largest consumption "The percepion dat
wate eyeing offto thy resewoir threatened, veetation/ Aquatic
option
ACT accesing water
from the
sine ‘Murrunbidgee River
Investment in
construtinga new
network for iigaton
/ supply
Pressuf€ on the___—
government to
Hea anil Ah residents ae the only one
targted.she perception
i thatthe whole community
Research progtans being angel
carted out for finding | /
effective management |
The catchment being strategies Comminity being
on board
see |
|
Water qualgy Sediments loving Lidents doing high \
ie into the reser “
Pershanent impuc wate saving
: : detegrated
ard iv a es “ conservation bchaviou dow impact
ye ae nena ¥ measures water saving behaviour
‘etl fareenle _
sia destruction Ae
Stromlo Treatment Bushes Grin Residents
Changes in Plant Xen as using ge
rainfall pattems 07° 7S 82003 shies 7 water for
/ _ ¥ ACTEW communicating. \ hie garden
Z s ca, to residents the large
Naat Sa Clit Change es xpi Reems wsingiper SUNS pea wing
vanity GIRO ystems spake shower buckets
Figure 6: An illustrative example of a user's cognitive map.
34 Building a Conceptual Model
The purpose of this step [10] was to create a conceptual model for the problem in order
to: (1) model the knowledge and arguments discovered so far, merging the various views
so that "synergy" and creativity become possible; and (2) sharpen the authors!
understanding about the dynamics of the problem and the appropriate level of details for
quantitative model building. In the literature, these representations (known as “cause
maps”) are often built in group settings at which different groups can contribute directly
to map building by capturing views, negotiating and reaching a consensus (Howick, Eden
et al. 2008). Whereas this step was planned in our original methodology outline, these
focus groups were not run because of time constraints for the water managers. A process
of comparison, aggregation and merging was undertaken by the modellers to create a
shared representing without suppressing the inherent diversity *. The conceptual model
highlights the perceived gaps and overlays in the perceptions of managers and users. For
example, while users believed that investment in building a new dam will automatically
lead to additional inflows to the reservoirs, managers challenged this assumption
considering other rainfall-independent supply sources as the best strategy to cope with
climate change effects. Figure 7 shows the developed conceptual model.
35 Developing Influence Diagrams
This stage focused scoping the key variables and causal relationships which were
candidates for quantitative modelling. Initially, an electronic workbook was used to
identify those elements in the conceptual model on which participants did not agree
[Step 11]. The electronic workbook contained 4 sub-models, centred on four decisions
(dependent variables) in the conceptual model: water supply, water demand, water quality
and total costs. Causal loop diagrams were used to model interrelationships for their ease
of communication. Participants were invited to accept/reject or add variables to each
sub-module. The analysis of the collected data helped to determine the
endogenous/exogenous variables and feedback loops which are candidates for
quantitative modelling. Afterwards, a series of influence diagrams, with varying levels of
details (Coyle 1996), were developed to depict the problem dynamics. Figure 8 shows an
aggregate view influence diagram for the problem.
3.6 SD Model Building
The SD model building is still under progress. The proposed model can be divided
intol0 sub-systems that have relatively sparse interactions with the remainder of the
model. Figure 9 illustrates the sector boundaries and internal interactions.
* The term conceptual model rather than cause map is used to distinguish the developed representations
from maps built in group settings.
Ne 2
Attractiveness of
using water
water, ——
NK
ining:
Balance between
Auractiveness of environment and,
living in the ACT humnad water
= needs
Auster:
driest continent
Personal gardens
Total costs
Rice/cotton
Environmental ~gcagticultare water
ACT at higher restrictions
Dripper
systems
Shower buckets
Grey water
the ACT dams
‘Negotiation to 2
flows ‘Consumption :
Wat eee” population
assed to seonth
‘Water security in the Base
ceuei Beer SK
Households
‘water available ine Pan coraumption =
SS Rainfall water tanks
purchase water a
recycling option,
Legend Common concept/relationship Managers’ view
enlarging the Cotter
dam
Investment in
building a new dam.
‘Users? view
Geese toes ——
Teeyatore soon eee ‘Treated me tag 4
Dam overflows ~ eden
Hipseseeiet connumpion
potable water Investment in non- \ bac
recycling option yotable water 4
in Investéaent in i wares
Inflows to the
ue reservoirs
consumption Water quality
for regrowth
Sediments
UGranosstig-wdler Runoff Destruction of lowing into the
one he vegetation reservoirs
Murrumbidgee River Government's
political and
economic agenda i iulenars
Carbon emissions —_ 8 Cm
‘Natural climate eats Evaporation
bushfires rates
variability se ami
S
Rainfall
patterms: ‘Climate Change éffects
Water wise
gardens
Smart meters
Permanent
Rainfall water
tanks
connected
internally
Figure 7: A. concepbual framework depicting the problem as perceived by water users and managers.
Extracts from the
Climate change Humans : a
fate Ebvircamenia) finvmament Murrumbidgee Nea. Business
: flows > Growth
i ; ome demand Rate
M ach « “BL 4
Rainfall Rate Runoff Rate
» ; Population
\ ‘Be
:
» Infiltration yew Fidld
nee < Capacty: changein Qverflows* . V
B3) 4 } rypoff 4 stotage levels
- i v Z -
Evapo- + 4
vapo- - sou. orstu Tafliation. yy . Residential
transpiration» SOIL MOISTURE | © Capacity i -
Rate -apacity Storage capacity™ ra demand a
: “ / : ~~ Delay.
Drought Index : Perceived a Per capita eameation effects
: ’ storage levels Awater use
Frequency + Potential Impact cod “, Soke fay
eee Fire Forest Bo of fires on the i Delay BZ comumprion \ Dilceetiacts
Danger Index 4 catchments . F F, A
i H eee Ce “ / \
/ | re / Attractiveness ‘ / Outdoor Delay
i ' ‘a P Pes OF Indoor water water use
THegetng Fire ++ NUMBEROF™ consumption i. 4 SS Water price
Events "occurrence" FIRES / Delay
Indeor _Qutdeer Restrictions
Efficienc} ;
Efficiency iency Effects
measures
measures
Figure 8: An aggregate influence diagram describing the key variables and relationships deriving the problem behaviour.
10
E
Variables
Precipitation rate]
ogenous
Bs
Rain fallindependent
inflows
Extracts from the
Murrumbidgee River
Economic/Energy
costs
Carbon emission
‘Total monetary costs
Storage capacity
Rainfall-dependent
inflows
Temperature Soil moisture
Infiltration rate
Runoff rate
——»
% Change in runoff
Residential demand
Catchment,
yr [
RESERVOIRS
Water in reservoirs
Storage levels
Ecological Water
Requirements
Environmental
Storage level
Overflows
Perceived
leyel
Non residential demand
Environmental flows rate
ACT Environmental
flows Guidelines
Policy
Soil Moisture Deficit
Fires
Drought Index
Frequency of fires
—
Severity of fires impact
Forest danger Fire Index
Fires Triggering
Population size
—
Residential Demand
Per capita use
Indoors water use
Outdoors water use
Attractiveness of
consumption,
Non-tesidential
Levers
Business
Growth
Demand
ff
Efficiency measures
Price effects
Education effects
Regula
ions effects
Figure 9: A sector boundary diagram of the proposed model
11
4 Conclusion and Learning Lessons
The goal of this ongoing research is to communicate a "big picture" understanding about
the evolving risks of water scarcity in the ACT using a SD based ILE. We follow a
transparent modelling process which cascades from tich individual views and cognitive
maps to qualitative models to a formal quantitative model. Through this process,
stakeholders' knowledge is regarded as a crucial input for effective model building.
This paper reports the knowledge elicitation process used to collect different, understand
and merge different views into a conceptual model. First, a preliminary model was
developed based on content analysis of relevant literature. This model represented the
initial dynamic hypothesis about the problem behaviour. A series of semi-structured
interviews were conducted to elicit the perceptions of users and managers. Cognitive
mapping technique was used to map the elicited data in terms of causal assertions about
the problem. Whereas a second interviewing session was organized to validate users!
maps, an electronic template was designed to validate managers' maps because of their
tight schedule. The collected data was used to create a conceptual model. Finally, an
electronic workbook is used to scope the key variables and interrelationships which are
candidates for quantitative modelling. Data collected from the workbooks are used to
develop a series of influence diagrams, which will be directly mapped into the SD model.
Early indications are that the research methodology is proving to be highly effective from
an analytical viewpoint. Critical factors in achieving the research aims are:
1. Having the resources to engage with a sufficiently large set of stakeholders to
elicit their mental models and capture them in cognitive mapping format. These
activities are time consuming and labour intensive.
2. Engaging sufficiently with stakeholder groups, both managers and consumer to
ensure the knowledge capture processes are comprehensive.
3. As far as it is possible, validating the SD models.
4, Designing the simulations in ways that engage players and realistically test their
decision making skills.
These factors are being address as the research proceeds. So far, we obtained two
pragmatic lessons from our experience through this modelling effort. Lesson 1: The modelling
process must be flexible and adaptable with several research and real world trade-offs. Although
managets' input is considered a critical element to the research process, it was very hard
to intensively engage them in the research activities (e.g. focus groups). We altered some
of the research activities to balance between the results' validity and maintaining a good
relationship with out client, such as using e-workbooks and validation templates. Lesson 2:
The modelling process may have many by-product outcomes. Knowledge accumulated through the
different elicitation cycles illuminated many useful insights for guiding the design of
public policies about the attitudes of users towards water conservation policies and the
gap between local and experts! perceptions.
Finally, simulations will be comprehensively tested and before being demonstrated and
made available for public evaluation in mid-2009. Internet- accessible simulations will be
used to gather data about how players adapt to possible future scenarios. Subsequent
stages of the research will test the extent to which player learning influences their
behaviour as water consumers.
12
REFERENCES
Atwood, C., R. Kreutzwiser, et al. (2007). "Residents' Assessment of an Urban Outdoor
Water Conservation Program in Guelph, Ontario." Journal of the American
Water Resources Association 43(2): 13.
Checkland, P. a. H., Sue (1998). "Action research: its nature and validity." Systemic
Practice and Action Research 11(1): 13.
Cooper, M., J. Tanner, et al. (2007). Next steps to ensure water security for the ACT’
region. ACT, Water Security Taskforce Chief Ministet’s Department.
Coyle, R. G. (1996). System dynamics modelling: a practical approach, Chapman & Hall.
Dietz, T. and P. C. Stern (2002). New Tools for Environmental Protection: Education.
Information, and Voluntary Measures, National Academy Press.
Eden, C. and F. Ackermann (1998). Making Strategy: The Journey of Strategic
Management. London, Sage.
Fazey, L., J. A. Fazey, et al. (2005). "Learning more effectively from experience." Ecology
and Society 10(2): 22.
Fazey, L, K. Proust, et al. (2006). "Eliciting the implicit knowledge and perceptions of
on-ground conservation managers of the Macquarie Marshes." Ecology and
Society 11(1): 28.
Garvin, T. (2001). "Analytical paradigms: The epistemological distances between
scientists, policy makers, and the public." Risk Analysis 21(3): 13.
Government, A. (2004). Think Water, Act Water Vol 1: Strategy for sustainable water
resource management in the ACT.
Government, A. (2004). Think Water, Act Water Vol 2: Explanatory Document.
Head, L. and P. Muir (2007). "Changing cultures of water in eastern Australian backyard
gardens." Social and cultural Geography 8(6): 18.
Hjorth, P. and A. Bagheri (2006). "Navigating towards sustainable development: A
system dynamics approach." Futures 38(1): 19.
Howick, S., C. Eden, et al. (2008). "Building confidence in models for multiple audiences:
The modelling cascade." European Journal of Operational Research 186(3): 16.
Marsden, J. and P. Pickering (2006). Securing Australia's urban water supply:
Oppottunities and impediments. Victoria, Marsden Jacob Associates.
Mclucas, A. C. (2001). An investigation into the integration of qualitative and quantitative
techniques for addressing systemic complexity in the context of organizational
Strategic Decision Making. School of Civil Engineering. Canberra, University of
New South Wales.
Meclucas, A. C. (2003). Decision making: risk management, systems thinking and situation
awareness. Canberra, Argos Press.
Mclucas, A. C. (2005). System dynamics applications: a modular approach to modelling
complex world behaviour. Canberra, Argos Press.
Morgan, M. G., B. Fischhoff, et al. (2002). Risk Communication: A Mental Models
Approach, Cambridge University Press.
Oppenheim, A. N. (1992). Questionnaire Design, Interviewing, and__ Attitude
Measurement, Continuum
Renn, O, (1998). "The role of risk communication and public dialogue for improving risk
management." Risk Decision and Policy 3(1): 26.
Sterman, J. (2000). Business Dynamics: Systems Thinking and Modeling for a Complex
World. Boston, Irwin McGraw-Hill.
13
