From Data-Poor to Data-Rich
System Dynamics in the Era of Big Data
Erik Pruyt et al.
August, 2014
Abstract.
Although SD modeling is sometimes called theory-rich data-poor modeling, it does not
mean SD modeling should per definition be data-poor. SD software packages allow one to get
data from, and write simulation runs to, databases. Moreover, data is
in SD to calibrate parameters or bootstrap parameter ranges. But more could and should be
done, especially in the coming era of ‘Big Data’. Big data simply refers here to more data
than was until recently manageable. Big data often requires data science techniques to make
it manageable and useful. There are at least three ways in which big data and data science
may play a role in SD: (1) to obtain useful inputs and information from (big) data, (2) to
infer plausible theories and model structures from (big) data, and (3) to analyse and interpret
model-generated “brute force data”. Inte ly, data scien that are useful for
(1) may also be made useful for (3) and vice versa. There are many application domains in
which the combination of SD and big data would be beneficial. Examples, some of which are
elaborated here, include policy making with regard to crime fighting, infectious di
security, national safety and security, financial stress testing, market assessn
management.
also sometimes used
1 Introduction
Will ‘big data’ fundamentally change the world? Or is it just the latest hype? Is it of any interest
d? Could it affect the field of SD
modeling and simulation? And if so, how? These are some of the questions addressed in this paper.
But before starting, we need to shed some light on what we mean by big data. Big data simply
refers here to a situation in which more data is available than was until recently manageable. Big
data s chni and useful.
So far, the worlds of data science and SD modeling and simulation have hardly met. Although
SD modeling is sometimes called theory-rich data-poor modeling, it does not mean SD modeling
per definition is or ought to be data-poor. SD software packages allow one to get data from, and
write simulation runs to, databases. Morcover, data is also sometimes used in SD to calibrate
es could be labeled big data.
However, in an era of ‘big data’, there may be opportunities for SD to embrace data science
of big data. The! ays in which big data and data
science may play a role in SD: (1) to obtain useful inputs and information from (big) data, (2) to
infer plausible theories and model structures from (big) data, and (3) to analyses and interpret
model-generated “brute force data”. Interestingly, data science techniques that are useful for (1)
may also be made useful for (3) and vice versa.
More than just being an opportunity to be seized, adopting data science techniques may be
necessary, simply because some evolutions/innovations in the SD field result in ever bigger data
sets generated by simulation models. Examples of such evolutions/innovations include: spatially
[BenDor and Kazal especially
to the SD modeling and simulation field? Or should it be ignore
data often requires o es to make it bl
parameters or bootstrap parameter ranges. But none of these
are at least three
techniques and make us
if combined with GIS packages; individual agent-based SD modeling (Castillo and Saysal
[Osgood] |Feola et al] 2012) especially with ABM packages; hybrid ABM-SD modeling and
i 2
simulat [Ford] [1990); and multi-model multi-method SD modeling
and simul + deep uncertainty (Pruyt and Kwakkel] [Auping ef al]
Moorlag et al. F
There are also many application domains for which the combination of SD and big data or
data science would be very beneficial. Examples, some of which are elaborated here, include policy
making with regard to crime fighting, infectious
rity, financial stress testing, market assessment, asset management, and future-oriented technology
ases, cybersecurity, national safety and secu-
assessment.
The remainder of the paper is structured as follows: First, a picture of the future of SD and
big data is discussed in section] Approaches, methods, techniques and tools for SD and big data
are subsequently discussed in section] Then, some of the aforementioned examples are presented
in section] And we conclude this paper with a discussion and some conclusions.
2 Modeling and Simulation and (Model-Generated) Big Data
“Big” Data
Goal:
- Understanding’
- Experi i
- FOexploration
- Robust policie:
/
Réal dat Vander, \ Ly
(ul) _Z Unee! inty ranges <a
pee
L & Data tics 2)
\
A v
\
aa Inference of’ans. of models’
Sse
Figure 1: Picture of the near term state of science / long term state of the art
shows how real-world and model-based (big) data may in the future interact with modeling
and simulation.
L. In the near future, it will be possible for all system dynamicists to simultaneously use mul-
tiple hypotheses, ic. simulation models from the same or different traditions or hybrids,
for different goals including the search for deeper understanding and poli
mentation in a virtual laboratory, future-oriented exploration, and robust policy design and
robustness testing under deep uncertainty. Sets of simulation models may be used to rep-
resent different perspectives or plausible theories, to deal with methodologi tainty,
or to deal with a plethora of important characteristics (c.g. agent characteristics, feedback
and accumulation effects, spatial and network effects). This will most likely lead to more
model-generated data, even big data compared to sparse simulation with a single model.
ial
Some of these models may be connected to real-time or semi-real time (big) data streams,
and some models may even be inferred in part from (big) data sources.
I
a
Storing the outputs of these simulation models in database
niques may enhance our understanding, may generate polic
policy robustness across large multi-dimensional uncertainty
and applying data science tech-
insights, and may allow to test
Approaches, methods, techniques and tools that may enable system dynamics to deal with big
data are discussed below, followed by a few examples.
3 Methods, Techniques and Tools for SD & Big Data
There are many data science methods, techniques and tools for dealing with (big) data. Many
of them could be used to generate inputs for models or analyze model outputs. However, a
comprehensive discussion of these methods, techniques and tools is beyond the scope of this paper.
Below, we focus our attention on data science methods/techniques/tools for model-gencrated data,
for generating useful model inputs, and for inferring model structures.
The fundamental differences between data science for model-generated data and data
for real data is that (i) in modeling, cases (i.e. underlying cau
modeling, there is no missing output data, and (iii) in modeling, it is possible to generate more
model-based data if more data is needed.
‘ience
are —or can be~ known,
3.1 Approaches for dealing with big (model-based) data
3.1.1 Smartening to avoid big data
A first approach for dealing with big data —or rather not having to dee
“smarter” methods, techniques and tools, ic. methods, techniqu
1 with big data~ is to develop
and tools that provide more
insights and deeper understanding without having to create or analyze too big a da
A first example is the development of ‘micro-macro modeling’ (Fallah-Fini et al. which
allows to include agent properties in SD models, and hence, to avoid ABM which is computationally
more expensive and results in relative terms
make of.
Simil of formal modeling techniques that smartened the traditional SD
approach, are new methods, techniques and tools currently being developed to smarten “brute-
force” SD approaches. Instead of sampling from the inp space, adaptive ling approaches
are for example being developed to span the output s [Islam and Pruyt
ilustrated by [Miller!
in much larger data sets which are also harder to
ar to the de
ther example of this approach to big data,
is to use optimization techniques to perform directed search
rected “brute-force” sampling.
Kwakkel and)
to undi-
3.1.2 Data set reduction techniques
A second approach for dealing with big data is to use techniques to reduce the data set to man-
ageable proportions, using for example filtering, clustering, searching, and selection methods.
like feature selection (Kohavi and John|
could for example be used to limit the number of unimportant
) or random forests
generative structures and uncertainties,
adequately span the uncertainty space.
ation methods like similarity-based time series clustering based on (
BOT} or dynamic time warping (Petitjean) 2011] [Rakthanmanon]
oUt may for example be used to cluster similar behavior patterns, allowing the
selection of a small number of exemplars spanning all behavior patterns
Alternatively, conceptual clustering may be used. Conceptual
clustering is a technique developed to groupings » data, and to explain the
differences between the groupings. Conceptual clustering differs from standard clustering because
it gives clear, actionable rules which distinguish the different groups. While cluster analysis gives
the central tendency of a group, conceptual clustering describes the limits or boundaries to the
cluster.
and hence, the number of simulation runs required to
fachine learning techniques like PRIM (Friedman and Fisher Bryant and Lempert)
[Kwakkel and Pruyt] 2013bJa} [Kwakkel et al|[2014) and PCA-PRIM (Kwakkel ef al] 2013)
may be used to identify areas of interest in the multidimensional uncertainty space, for example
areas with high concentrations of simulation runs with very undesirable dynami
Other machine learning techniques lik ivan der Maaten and Hinton|
to cluster across multiple dimensions (see the example in subsection
may be used
3.1.3. Using methods and techniques for dealing with (model-generated) big data
A third approach for dealing with big data is to use methods and techniques that are appropriate
for dealing with (model-generated) big data.
Examples include optimization iques like stochastic optimization (in Vensim), SOPS (in
Powersim), or -even better— (multi-objective) robust optimization
e.g. to identify policy levers and define policy triggers across large multidimensional uncertainty
spaces: these methods actually require many simulation runs.
The same is true for methods for testing policy robustness across a wide ranges of uncertain-
ties These outcomes are usually hard to visualize, unless they ar
unambiguous. Thi -ase with other ‘big data approa od are new ways
to visualize th 1s of multi-dimensional outcomes i
Machine learning techniques may possibly also be used to infer Foce of models and
plausible models from data. A linear dynamical system
machine learning technique which both models data in the presence of unc Srainty, and reproduces
a range of dynamical patterns consistent with real-world observations. The technique i
to older approaches in dynamic modeling and control, such as Kalman filtering. The particular
appeal of the technique is the potential automated production of an SD model straight from
data. The model is i
to translate the evidenc
into a full s
em dynamic model. The modeling approach might als
be useful bec: ures and parameterizes uncertainty in the data, showing a range of SD
parameters which are consistent given the data.
Methods and techniques that could easily be made useful for dealing with model-based big
a include Formal Model Analysis tect s (Kampmann and Olival 2009} [Saleh e¢ al.|
mathematical methods (Kuipers) atistical screening techniques (Ford and Flynn|
, statistical pattern testing techniques like the ones in/BTS-IT
e it stru
3.2 Tools for dealing with models and (big) data
Database connectivity and data management procedures of some of the traditional software pack-
ages have been improved. For example, Vensim DSS allows to read from and write to databases
via ODB'
1See the|Vensim DSS Supplement] (Ventana Systems| 2010) 2011) and|this video
More powerful though is the use of scripting languages like Python
to control traditional SD software packages, manage data and perform advan
analysis on stored data. TU Delft’s EMA workbench software is an open source tool in python
that integrates multi-method simulation with data visualization and analy:
Modeling and simulation across platforms is also likely to become reality in a few years time.
The XMILE (eXtensible Model Interchange LanguagE) project
aims at facilitating the storage, sharing, and combination of simulation
models across software packages and z some) modeling schools. It may also allow for easier
connections with databases, statistical and analytical software packages, and ICT infrastructure.
Or:
‘XMILE will enable System Dynamics models to be reused with Big Data sets to show
icies produce different outcomes in complex environments. Models
ibraries, shared within and between organizations, and
with common vocabulary. XMILE will also
support the integration of system dynamics models and simulations into mainstream
analytics software. ]
used to communicate different outcom«
Note however that this is already possible to a large extent by using scripting languages or
software packages with scripting capabilities such as the aforementioned EMA Workbench.
4 Examples
4.1 Crime Fighting under Deep Uncertainty
This example relates to crime fighting (s
. An exploratory SD model and related tools
were developed for the police in view of increasing the effectiveness of their fight against high
impact crime (HIG). Although HIC comprises robbery, mugging, burglary, intimidation, assault
and grievous bodily harm, the pilot reported on in this paper focussed on fighting robbery «
burglary. HIC require a systemic perspective and approach: they are characterized by important
systemic effects in time and space, such as learning and specialization effects, ‘waterbed effects’
between different HICs and precincts, accumulations (prison time) and delays (in policing and
jurisdiction), preventive effects, and other 1 effects (ex post preventive me
also characterized by deep uncertainty: ion of these perpetrators is unknown
it is known that these crimes are mainly committed by opportui
and mobile criminal teams= and even. though ‘their archetypal ctinie-related habits are. known,
accurate time and geographically specific predictions cannot be made. At the same time is part
of the HIC system well-known and is near-real time information related to these crimes available.
The main goals of this pilot project were to support strategic policymaking under deep un-
s of policies to fight HIC. The SD model was used as an
engine behind the interface, to explore effects under deep uncertainty, i.e. to generate and explore
model-based big data in view of generating policy insights, and to identify real-world pilots that
c/would increase the understanding about the
nally to compare the rol of interventions across bles of tl Is of plausible futures.
Real world information and insights from the real-world pilots would then be used to improve and
update the model. Today, geo-spatial real-world crime related data is available in near real-time
and could be updated automatically. An evolved version of this model could thus automatically
update the information, and, by doing so, increase the model’s value for the strategic decision
makers. An even more advanced version may use hybrid models or a multi-method approach,
with more attention paid to individuals, individual obj
caus
a large fré
certainty and to monitor the effectivene
em and effectiveness of interventions, and fi-
spatial characteristics, and networks.
The latter will further expand the amount of model-generated data.
2The EMA
vorkbench can be downloaded for free from, |attp://simulation.tbm.tudelft .nl/ema-workbench/|
//waa.oasis- open. org/committees/tc_home .php?wg_abbrev=xmile| accessed on 18 March 2014.
(V) monitoring of RW data
from pilots and HIC system
(Il) Interface
for policy-makers
(I ESDMA model
See 2
(II)
Analytical un
module
a = Real world
aah (IV) ace SS
erveey
Figure 2: HIC model, interface, analyses under deep uncertainty, real-world pilots, and monitoring
of real-world data
4.2 Integrated Risk-Capability Analysis under Deep Uncertainty
The third example deals with integrated risk-capability anal
tional Safety and Security as discussed by [Pruyt ef al]
multi-model approach results in three instances of big data. Large ensembles of plausibl
are generated using multiple simulation models for each of a few dozen risk types. Time series
s under deep uncertainty for Na-
‘utures
clustering and machine learning techniques are used to identify and select a dozen of exemplars
~ exemplary in terms of outputs and origin. These exemplars are used as inputs for a capability
analysis under deep uncertainty, resulting for each exemplar in hundreds to thousands of simula-
tion runs. Again, exemplars are selected from these ensembles. Either they are used to assess the
effectiveness of alternative sets of capabilities, or they are used as starting point of an automated
s, settings
of some of the capabilities, and exogenous uncertainties may in the future be set with real-world
data.
But the real big data problem in this case is one of model-generated big data. Smart sampling
techniques and time-series classification methods that together allow to identify the largest variety
of behavior patterns with the minimal amount of simulations are desirable for this computational
approach, for performing an d multi-h ability analysis over many risks is —
due to the Multi-Objective Robust Optimization method used— computationally very expensive.
Compared to traditional SD, it could be labeled ‘big computing’.
search for robust capability sets (using robust optimization techniques). Some risk class
Exploratory Scenario ‘Scenario Discovery Single-hazard Integration of single-hazard | Future: Automated
Modeling Generation and Set Selection Capability Analyses cas overall representative ual danadck
CApolicies Scenarios ofall risks over all risks
xuncertainties toNRAscores set to CAUDU
—
¢
>
Radicalization
8
a 8?
all-hazard design of capability policies
—
e
Automated search for most robust capa
|
|
Iterative
Figure 3: Model-Based Integrated Risk-Capability Analysis (IRCA)
4.3 Future-Oriented Technology Assessment
Machine learning techniques could also be used to inform model/theory building. In this exam-
ple, based on (Cunningham and Kwakkel] , a machine learning technique called t-NSE
der Maaten and Hinton] 2008) is used to better und d recent technological de in
Electric Vehicles, Hybrid Electric Vehicles, and Plug-in Hybrid Electric Vehicles. T-NSE is a tech-
nique for dimensionality reduction that is particularly well suited for visualizing high-dimensional
datasets. It is primarily a visualization technique. Noncthel
ing trend in the analysis of data as well, known as topological data analysis. Topological data
analys enables researchers to capture the broad features of a data set without
requiring 5 ses which may have generated
the data. This is in sharp contrast to parametric statistics which may require strong assumptions
about the data, and may demand an extensive base of prior hypotheses about the data. Topolog-
ical data analysis approaches may be particularly suitable as a means of structuring SD outputs.
These structured outputs may enhance understanding of the d pability of the s
may allow comparison between seemingly very different dynam! and may enable the
ss it is representative of an emerg-
‘ametric knowledge of the underlying proces
identification of interesting or under-explored regimes of behavior.
is are: type, weight, output power, battery capacity, accel-
ions, gas mileage and electric range of the vehicles. These dimensions are first
ef] t-SNE then reduces our 7
jectories: HEVs have
on HEV
Dimensions included in the anal:
eration, CO emi
rescaled so that similarities in design performance can be asses
dimensions to 2 dimensions + time. Figure [I[a) shows the individual traj
radically shifted their evolution since 2006. And Figure[i[b) shows the convergence betwe
and PHEV on the one hand and EV on the other. Note that these plots are dimensionle:
for time).
A slightly more detailed analysis of performance gradients shows that PHEV, HEV and EVs
are converging: HEVs and PHEVs are on the one hand hybridizing with pure electric vehicles
“The nominal variable ‘type’ is converted to a logical variable (EV or not). Gas mileage and electric range are
converted to a ‘functional range’ variable. Data is ranked sorted by column, and the data replaced with its ranks.
Repeated values are replaced with an average rank.
ate 8 Many vehicles at pn
is | 2 both frontiers on
}2008 * }2008
}2006 2 }2006
al erfomance
}2004 ape j2004
2
aoc aoc
4
2200 zc00
‘
sc00 so08
*
a a a a
(a) Trajectories of HEV, PHEV and EV (b) Frontiers, trade-offs, and performance gradient
Figure 4: t-SNE applied to data regarding the multi-dimensional development of Electric Vehicles
Hybrid Electric Vehicles, and Plug-in Hybrid Electric Vehicles
slowly shedding reliance on their IC engines. EVs on the other hand display eroding technological
goals: they are growing more powerful and lighter in weight, able to cruise for greater ranges
on a charge, but their acceleration rates are declining. Finally, CO» reduction is the most rapid
indicator of change. This information could be used to inform model building or to compare model
outcomes with real data. Note also that t-SNE can be applied to multi-dimensional outcomes of
simulation models, including their trajectories.
4.4 Monitoring Infectious Diseases
eS, like the 2009 A(H1N1)n flu. This case is described
lodeling and Analysis in (Pruyt and Hamarat
ed to illustrate by [Pruyt et that more can be done with SD models.
‘tainty is taken into account scriously, then model-based analyses ought to show
le futures. Over time, information will narrow down the ensemble of plausible futures.
Near real-time geo-spatial data (twitter, medical records, et cetera) could be used in combination
with SD models to reduce the uncertainty space. More real data then results in a reduction of the
model-gencrated data.
The last c
s about (new) infectious dis
d with Exploratory System Dy:
5 Discussion and Conclusions
In this paper, we addressed the combination of SD modeling and simulation and data science to
deal with so-called ‘big data’.
Although their combination is promising, it will also require investments by the SD community.
To name a few: Multi-method and hybrid modeling approaches should be further developed in
order to make existing modeling and simulation approaches appropriate for dealing with agent-
system character atial and network aspects, deep uncertainty, and possibly other aspec
Data science and machine learning techniques should be further developed into techniques that
can provide useful inputs for simulation models; Data science and mathematical techniques should
be further developed into techniques that can be used to infer parts of models from data;
connectors to databases and other programs need to be provided by software developers; Machine
learning algorithms and formal model analysis methods should be further developed into tools
that help to generate deeper understanding and generate useful policy insights; Methods and
tools should berdeveloped to: turn intuitiverpolieymaling intoxmodel:based policy design and
automated robustness testin:
be developed to make more
There are multiple approaches for dealing with big data: three of these approaches were dis-
cussed in the paper. Some machine learning techniques have been introduced and their application
demonstrated on a few cases.
; And new analytical approaches and visualization techniques should
se out of models and data.
References
Auping WL, Pruyt E, Kwakkel J. 2012. Analysing the uncertain future of copper with three
exploratory system dynamics models. In Proceedings of the 30th International Conference of
the System Dynamics Society. St.-Gallen, CH, System Dynamics Society.
Bankes SC. 1993. Exploratory modeling for policy analysis. Operations Research 41(3): 435-449.
Bankes $C. 2002. Tools and Techniques for Developing Policies for Complex and Uncertain Sys-
tems. In Proceedings of the National Academy of Sciences of the United States of America 99(3):
7263-7266.
BenDor TK, Kaza N. 2012. A theory of spatial system archetypes. System Dynamics Review 28():
109-130.
Breiman L. 2001. Random Forests. Machine Learning 45(1): 5-3:
Bryant BP, Lempert RJ. 2010. Thinking inside the box: A participatory, computer:
ssisted ap-
proach to scenario discovery. Technological Forecasting & Social Change 77(1): 34-49.
G. 2009. Topology and Data. Bulletin (New Series) of the American Mathematical
, 46(2): 255-303.
Castillo D, Saysel AK. 2005. Simulation of common pool resource field experiments:
a behavioral model of collective action. Ecological Economics 55(3): 420-436,
http://dx.doi.org/10.1016/j.ecolecon.2004.12.014.
Clemson B, Tang Y, Pyne J, Unal R. 1995. Efficient methods for sensitivity analysis. System
Dynamics Review 11(1): 31-49.
Cunningham SW, Kwakkel JH. 2014. Technological Frontiers and Embeddings: A Visualization
Approach. Portland International Conference on Management of Engineering and Technology
(PICMET), Kanazawa, Japan: July 27-31, 2014.
Diker VG, Allen RB. 2005. XMILE: towards an XML interchange language for system dynamics
models. System Dynamics Review 21(4): 351-359.
Dogan G. 2007. Bootstrapping for confidence interval estimation and hypoth
rameters of system dynamics models. System Dynamics Review 23(4): 415-436.
testing for pa-
Eberlein RL, Chichakly KJ. 2013. XMILE: a new standard for system dynamics. System Dynamics
Review 29(): 188-195.
Feola G, Gallati JA, Binder CR, 2012. Exploring behavioural change through an agent-oriented
system dynamics model: the use of personal protect
in Colombia. System Dynamics Review 28(): 69-93.
among
Fallah-Fini S, Rahmandad R, Chen HJ, Wang Y. 2013. Connecting micro dynamics and population
distributions in system dynamics models. System Dynamics Review 29(4):197-215
Fiddaman TS. 2002. Exploring policy options with a behavioral climate-economy model. System
Dynamics Review 18(2): 243-267.
Resilient by design: a modelling approach to support scenario and policy analysis in
the Olifants River Basin, South Africa.
Jai K. Clifford Holmes’, Sharon Pollard’, Harry Biggs’, Kundai Chihambakwe’,
Willem Jonker’, Theo York’, Fabio Andres Diaz** & Jill H. Slinger’
* Corresponding Author, email: jai.clifford.holmes@gmail.com;
! Institute for Water Research, Rhodes University, P.O. Box 94, Grahamstown, 6140, South Africa;
? Association for Water and Rural Development (AWARD), P.O. Box 1919, Hoedspruit, 1380, South Africa;
3 Independent consultants, Stellenbosch, South Africa;
* Institute of Social Studies, Erasmus University Rotterdam, P.O. Box 29776, 2502 LT, The Hague, the Netherlands
ae Department of Politics and International Studies, Rhodes University, P.O. Box 94, Grahamstown, 6140, South Africa;
Faculty of Technology, Policy and Management and the Faculty of Civil Engineering and Technical Geosciences, Delft
University of Technology, P.O. Box 5015, 2600 BX Delft, the Netherlands
Summary of paper to the 34" International Ci of the System Dynamics Society, 17 — 21 Jul
2016,
in Delft, the Netherlands.
DRAFT v.22 (for conference proceedings)
19" August 2016
Abstract:
In this paper, we report on a concerted modelling effort in the South African water resources
sector in which system dynamics provides the paradigmatic framing for both a stakeholder
engagement process and for developing an underpinning, integrative simulation model. We
describe the design of the parallel modelling approach and examine the progress to date. We
highlight the crafting of the engagement sessions to fit the different characteristics and needs
of the participating stakeholders and reflect upon outcomes so as to set the project path into
the future. In this, we choose to build resilience by a progression from collecting and
exploring individual experiences of stakeholders’ problems; to connecting individuals’
experiences to their broader sectors’ problems; and finally to cross-connecting the sectors’
impacts and requirements to one another under different scenarios of change.
Keywords: climate change; water management; public policy; stakeholder engagement; resilience;
simulation modelling
Please note: This is a summary document reporting on research funded by the United States Agency for
International Development, under USAID Southern Africa - RFA-674-12-000016 RESilience in the
LIMpopo Basin Program (RESILIM). The RESILIM-O part of the programme is implemented by the
Association for Water and Rural Development (AWARD), in association with project partners. The full
copy of the DRAFT paper has been excluded from the conference proceedings in order to be revised
ahead of submission to the Journal of Systemic Practice and Action Research (between December 2016 —
January 2017).
ueries about this research can be directed to the corresponding author
jai.clifford.holmes@ gmail.com).
Page | of 9
Fisher DH. 1987. Knowledge acquisition via incremental conceptual clustering. Machine Learning
2 (2): 139-172.
Islam, T. and Pruyt, E. 2014. An adaptive sampling method for examining the behavioral spectrum
of long-term metal scarcity. In: Proceedings of the 30th International Conference of the System
Dynamics Society. Delft, NL, System Dynamics Society.
Michalski RS. 1980. Knowledge acquisition through conceptual clustering: A theoretical framework
and an algorithm for partitioning data into conjunctive concepts. International Journal of Policy
Analysis and Information Systems 4: 219-244.
Roweis S, Ghahramani Z. 1999. A Unifying Review of Linear Gaussian Models. Neural Computa-
tion 11, pp.305-345.
Ford A. 1990. Estimating the impact of efficiency standards on the uncertainty of the northwest
tem. Operations Research 38(4):580-597
Ford, A. and Flynn, H. (2005), Statistical screening of system dynamics models. System Dynamics
Review 21(1): 273-303
Forrester JW. 2007. System dynamics — the next fifty years. System Dynamics Review 23(2-3):
359-370.
Friedman JH, Fisher NI. 1999. Bump hunting in high-dimensional data. Statistics and Computing
9(2): 123-143.
Groves DG, Lempert RJ. 2007. A new analytic method for finding policy-relevant scenarios. Global
Environmental Change 17(1): 73-85.
Hamarat C, Kwakkel JH, Pruyt E. 2013. Adaptive robust design under deep uncertainty. Techno-
logical Forecasting & Social Change 80(3): 408-418.
Hamarat C, Kwakkel JH, Pruyt E, Loonen ET. 2014. An exploratory approach for adaptive
policymaking by using multi-objective robust opti s Modelling Practice and
Theory. Accepted for publication.
Kampmann CE, Oliva R. 2008. Structural dominance analysis and theory building in system
dynamics. Systems Research and Behavioral Science, 25(4): 505-519.
Kampmann CE, Oliva R. 2009. Analytical methods for structural dominai
dynamics. In: R. Meyers (Ed.), Encyclopedia of Complexity and Systems Scie
York, p8948-8967
Kohavi R, John GH. 1997. Wrappers for feature subset selection. Artificial Intelligence 97(1-2),
273-324
Kuipers J. 2014. Formal Behaviour Classification under Deep Uncertainty: Applying Formal Anal-
ysis to System Dynamics. In: Proceedings of the 30th International Conference of the System
Dynamics Society. Delft, NL, m Dynamics Society.
Kwakkel, J.H., Auping, W.L., Pruyt, E. 2014. Comparing Behavioral Dynamics Across Mod-
cls: the Case of Copper. In: Proceedings of the 30th International Conference of the System
Dynamics Society. Delft, NL, System Dynamics Society.
Kwakkel JH, Auping WL, Pruyt E. 2013. Dynamic scenario discovery under deep uncertainty:
The future of copper. Technological Forecasting € Social Change 80(4): 789-800.
Kwakkel JH, Pruyt E. 2013a. Using system dynamics for grand challenges: The ESDMA approach.
Systems Research € Behavioral Science. doi: 10.1002/sres.2225
Kwakkel JH, Pruyt B. 2013. Exploratory modeling and analysis, an approach for model-based
foresight under deep uncertainty. Té | Forecasting & Social Change 80(3): 419-431.
Lempert RJ, Groves DG, Popper SW, Bankes $C. 2006. A general, analytic method for generating
robust strategies and narrative scenarios. Management Science 52(4): 514-528.
Lempert RJ, Popper SW, Bankes SC. 2003. Shaping the next one hundred years: New methods
for quantitative, long-term policy analysis. RAND report MR-1626, The RAND Pardee Center,
Santa Monica, CA.
Lempert RJ, Schlesinger ME. 2000. Robust strategies for abating climate change. Climatic Change
45(3-4): 387-401.
McKay MD, Beckman RJ, Conover WJ. 2000. A comparison of three methods for selecting values
of input variables in the analysis of output from a computer code. Technometrics 42(1): 55-61.
McKinney W. 2012. Python for Data Analysis: Data Wrangling with Pandas, NumPy, and
IPython. O'Reilly Media, p470
Miller J. 1998. Active nonlinear tests (ANTs) of complex simulation models. Management Science
44: 820-830.
Ruben Moorlag, Willem Auping, Erik Pruyt. 201. Biloting the effects of shale gas development
on natural gas markets: a multi-method a In: dings of the 32nd International
Conference of the System Dynamics Society, Delft, NL, System Dynamics Society.
Moxnes E. 2005. Policy sensitivity analy:
namics Review 21(2): 123-145.
imple versus complex fishery models. System Dy-
Oliva R. 2003. Model calibration as a testing strategy for system dynamics models. European
Journal of Operational Research 151(3): p552-568.
Oreskes N, Shrade
Numerical Models in the Earth Sciences. $i
Frechette K, Belitz K. 1994. Verification, Validation, and Confirmation of
nce. 263(5147): 641-647.
Osgood N. 2009. Lightening the performance burden of individual-based models through dimen-
sional analysis and scale modeling. System Dynamics Review 25(): 101-134.
Petitjean FO, Ketterlin A, Gancarski P. 2011. A global averaging method for dynamic
time warping, with applications to clustering. Pattern Recognition 44(3): 678-693
doi:10.1016/j.patcog.2010.09.013
Pruyt E. 2013. Small system dynamics models for big issues: Triple jump towards real-world
complexity. TU Delft Library, Delft. Available at http: //simulation.tbm.tudelft .n1/|
Pruyt E, Hamarat C. 2010. The Influenza A(HIN1)v Pandemic: An Exploratory System Dynamics
Approach. In Proceedings of the 28th International Conference of the System Dynamics Society.
Seoul, Korea, System Dynamics Society.
Pruyt E, Hamarat C, Kwakkel JH. 2012. Integrated risk-capability analysis under deep uncer-
tainty: an integrated ESDMA approach. In Proceedings of the 30th International Conference of
the System Dynamics Society. St.-Gallen, CH, System Dynamics Society.
Pruyt, E. and Islam, T. 2014. The Future of Modeling and Simulation: Beyond Dynamic Com-
plexity and the Current State of Science. In: Proceedings of the 30th International Conference
of the System Dynamics Society. Delft, NL, System Dynamics Society.
Pruyt E, Kwakkel JH. 2014. Radicalization under Deep Uncertainty: A Multi-Model Exploration
of Activism, Extremism, and Terrorism. System Dynamics Review 30(1-2):1-28.
Pruyt E, Kwakkel JH. 2012. A bright future for system dynamics:
science and more. In Proc
From art to computational
ings of the 30th International Conference of the System Dynamics
Society. St.-Gallen, CH, System Dynamics Society.
Pruyt E, Kwakkel JH, Hamarat C. 2013. Doing more with models: Illustration of a system dy-
namics approach for exploring deeply uncertain issues, analyzing models, and designing adaptive
robust policies, In Proceedings of the #1st International Conference of the System Dymamics So-
ciety. Cambridge, MA, System Dynamics Society.
Rakthanmanon T. 2013. Addressing Big Data Time Series: Mining Trillions of Time Series Subse-
quences Under Dynamic Time Warping. ACM Transactions on Knowledge Discovery from Data
7(3): 10:1-10:31. doi:10.1145/2510000/2500489
Ruth M, Pieper F. 1994. Modeling spatial dynamics of sea-level rise in a coastal area. System
Dynamics Review 10(): 375-389.
Saleh M, Oliva R, Kampmann CE, Davidsen PI. 2010. A comprehensive analytical approach for
policy analysis of system dynamics models. European Journal of Operational Research 203(3):
673-683
Struben J. 2005. Space matters too! Mutualistic dynamics between hydrogen fuel cell vehicle
demand and fueling infrastructure. In Proceedings of the 2005 International Conference of the
System Dynamics Society. System Dynamics Society.
Taylor TRB, Ford DN, Ford A. 2010. Improving model understanding using statistical screening.
m Dynamics Review 26: 73-87
Van Rossum G. 1995. Python Reference Manual. CWI, Amsterdam.
van der Maaten LJP, Hinton GE. 2008. Visualizing High-Dimensional Data Using t-SNE. Journal
of Machine Learning Research 9: 2579-2605
Ventana Systems Inc. 2010. Vensim DSS Reference Supplement. Ventana System Inc.
Ventana Systems Inc. 2011. Vensim Reference Manual. Ventana System Inc.
Weaver CP, Lempert RJ, Brown C, Hall JA, Revell D, Sarewitz D. 2013. Improving the contribu-
tion of climate model information to decision making: the value and demands of robust decision
frameworks. Wiley Interdisciplinary Reviews: Climate Change 4(1): 39-60.
Yiicel G. 2012. A novel way to measure (dis)similarity between model behaviors based on dynamic
pattern features. In: Proceedings of the 30th International Conference of the System Dynamics
Society, St. Gallen, Switzerland, 22 July-26 July 2012.
Yiicel G, Barlas Y. 2011. Automated parameter speci
on behavior pattern features. System Dynamics Review 27(2): 195-215.
edback models based
Ezekiel, M., 1938. The Cobweb Theorem. The Quarterly Journal of Economics, 52(2), pp.255—
280. Available at: http://www jstor.org/stable/1881734.
FAO, 2013. FAOSTAT: Food and Agricultural Commodities Production; Commodity per
Country. Available at: http://faostat3.fao.org/home.
Femandes, R.A.S., Braga, M.J. & Lima, J.E. de, 2010. Elasticidade de transmissao e formacao
espacial de precos de leite ao produtor nos maiores estados produtores. Revista de
Economia e Administracao, 9(3), pp.368-385.
Ferreira, A.H. & Gomes, A.P., 2004. Eficiéncia técnica no curto e no longo prazo em sistemas de
producao de leite com gado holandés, mestigo e zebu. In 42° Congresso da Sociedade
Brasileira de Economia, Administragao e Sociologia Rural - SOBERCongresso Brasileiro
de Economia e Sociologia Rural. Brasilia: SOBER, p. 16.
Figueiredo, A.M. & Gomes, A.L., 2009. Consideracées sobre estratificagao de produtores de
leite em Minas Gerais. In XLVII Congresso da Sociedade Brasileira de Economia,
Administragao e Sociologia Rural - SOBER. Porto Alegre-RS, pp. 1-11.
Frigon, M., Doyon, M. & Romain, R., 1999. Asymmetry in Farm-Retail Price Transmission in
the Northeastern Fluid Milk Market, Connecticut.
Ghaffarzadegan, N. & Tajrishi, A.T., 2010. Economic transition management in a commodity
market: the case of the Iranian cement industry. System Dynamics Review, 26(2), pp.139-161.
Gomes, A.L., ALVES, E.R.A. & GOMES, A.T., 2003. O comportamento do preco do leite
recebido pelo produtor: uma abordagem de curto e longo prazo. In XLI Congresso
Brasileiro de Economia e Sociologia Rural. Juiz de Fora-MG.
Gomes, S.T., 2006. Diagndstico da pecuaria leiteira do estado de Minas Gerais em 2005 P. M.
G. Lins & P. S. Vilela, eds., Belo Horizonte: SEBRAE-MG, OCEMG e SENAR-AR/MG.
Gongalves, R.M.L. et al., 2008. Analysis of technical efficiency of milk-producing farms in
Minas Gerais. Economia Aplicada, 12(2), pp.321-335.
Hoffmann, R., 2007. Elasticidades-renda das despesas e do consumo de alimentos no Brasil em
2002-2003. In F. G. . SILVEIRA et al., eds. Gasto e consumo das familias brasileiras
contemporaneas. Brasilia: Ipea, p. 463-483.
Hommes, C.H., 1994. Dynamics of the cobweb model with adaptive expectations and nonlinear
supply and demand. Journal of Economic Behavior & Organization, 24(3), pp.315-335.
Kaldor, N., 1934. A Classificatory Note on the Determinateness of Equilibrium. The Review of
Economic Studies, 1(2), pp.122-136.
Liehr, M. et al., 2001. Cycles in the sky: understanding and managing business cycles in the
airline market. System Dynamics Review, 17(4), pp.311-332.
Marques, P.R. et al., 2015. Competitiveness levels in cattle herd farms. Ciéncia Rural, 45(3),
pp.480-484.
Meadows, D.L., 1970. Dynamics of Commodity Production Cycles, Waltham, MA: Pegasus
Communications.
Munshi, K.D. & Parikh, K.S., 1994. Milk supply behavior in India: Data integration, estimation and
implications for dairy development. J ournal of Development Economics, 45(2), pp.201-223.
13
Nicholson, C.F. & Stephenson, M.W., 2014. Dynamic Market Impacts of the Dairy Margin
Protection Program. Journal of Agribusiness, 32(2), pp.165-192.
Nicholson, C.F. & Stephenson, M.W., 2015. Milk Price Cycles in the U.S. Dairy Supply Chain
and Their Management Implications. Agribusiness: An International Journal, 31(4),
pp.507-520.
Novo, A.M. et al., 2013. Feasibility and competitiveness of intensive smallholder dairy farming
in Brazil in comparison with soya and sugarcane: Case study of the Balde Cheio
Programme. Agricultural Systems, 121(0), pp.63-72.
Olaya, C., 2015. Cows, agency, and the significance of operational thinking. System Dynamics
Review, 31(4), pp.183-219.
Pagel, EJ., 2005. Dynamic patterns of change in structure under different support policy
regimes: an examination of U.S. dairy farming. MS Thesis, Cornell University.
Rooney, M., Nuttall, WJ. & Kazantzis, N., 2015. A dynamic model of the global uranium
market and the nuclear fuel cycle. Resources Policy, 43, pp.50-60.
Scalco, P.R. & Braga, M.J., 2014. Measuring the Degree of Oligoposony Power in the Brazilian
Raw Milk Market. International Food and Agribusiness Management Review, 17(2), p.20.
Simoes, A.R.P., Silva Junior, V.P. & Protil, R.M., 2015. Perfil tecnologico da pecuaria leiteira
em Minas Gerais. In 53° Congresso da Sociedade Brasileira de Economia Administragao e
Sociologia Rural - SOBER. Joao Pessoa - PB, p. 15.
Smith, P.C. & van Ackere, A., 2002. A note on the integration of system dynamics and economic
models. J ournal of Economic Dynamics and Control, 26(1), pp.1-10.
Souza, D.P.H., 2000. Andlise da estrutura de custo e preco de sobrevivéncia dos principais
sistemas de producao de leite. Vicosa-MG: Universidade Federal de Vicosa.
Sterman, J.D., 2000. Business Dynamics: Systems Thinking and Modeling for a Complex World
1st ed., Bostom: Mc Graw-Hill.
Tauer, L.W., 1998. Estimates of individual dairy farm supply elasticity, Cornell.
Zoccal, R. et al., 2007. Distribuicdo espacial da pecuaria leiteira no Brasil. In Reunion latino
americana de produccion animal (ALPA). Cuzco, Peru: ALPA, p. 4.
ISDC 2016 paper (RESILIM-O) 19/08/2016
1. Introduction
The resilience of ecosystems and people to changing climatic conditions, rapid urbanization,
ongoing commercialization of agriculture, industrialization and degradation of the natural
environment, is recognized as problematic in sub-Saharan Africa (Turton, 2013). The central role
played by water security in coping with both the developmental challenges as well as building
resilience, is also recognized (Muller et al., 2009; V6rdsmarty et al., 2010). The RESILIM-O
research programme, funded by the United States Agency for International Development:
Southern Africa, focuses on water security as a unifying thread in building resilience. RESILIM-
O’s overarching objective is to improve the trans-boundary management of the Limpopo River
Basin to enhance the resilience of people and ecosystems. The RESILIM-O research programme
is implemented by the Association for Water and Rural Development (AWARD). AWARD has a
specific focus on the Olifants River Basin (hence RESILIM-O), which is a major tributary of the
Limpopo River and is an international watercourse shared by South Africa and Mozambique (see
Figure 1).
=r
Hydrological boundaries in | USAID | SOUTHERN AFRICA
the Olifants Catchment niatatehentiabanabiainsts
Middle Olifants “Agric
JT Loskop (633
Legend
© Town
River
Moons
Transfrontier Park
International border
Figure 1: Hydrological boundaries in the Olifants Catchment of the Limpopo basin. Source: AWARD.!
In the first phase of RESILIM-O, a grounded understanding of water resource management and
the ecosystems in the Olifants catchment was developed. A Resilience Assessment Process
(RAP), designed to garner understanding of risks and vulnerabilities and build more resilient
" http://award .org.za/resilim-o/olifants-river-basin/
Page 2 of 9
ISDC 2016 paper (RESILIM-O) 19/08/2016
practices across stakeholder groups, was pioneered in this phase, according with Objective 1 of
RESILIM-O, namely:
To institutionalise systemic, collaborative planning and action for resilience of
ecosystems and associated livelihoods through enhancing the capacity of stakeholders
to sustainably manage natural resources of the Olifants River Basin under different
scenarios.
In 2015, the RESILIM-O project management team decided to focus attention on the Selati River
quaternary catchment (B72 in Figure 1), identifying the Selati to be an exemplar of land uses and
key social dynamics within the Olifants catchment. The dominant land-uses in the Selati
catchment are game ranching; dryland agriculture and rangelands; mining; urban settlements in
villages and towns; and conservation, both within public parks (specifically the Kruger National
Park) and private game reserves, such as the Selati Game Reserve (Pollard & Laporte, 2015).
Central problems in the Selati pertain to ephemeral river flow, fluctuating water quality, mining
effluent, inadequate access to drinking water, and inadequate sanitation and the poor management
of wastewater. The governance and management dimensions of these challenges differ from the
upper to the lower reaches of the Selati River, yet the cumulative impact of these diverse
activities calls for integrated, catchment-level management and strategic planning. The longer-
term projections of the declining yield, associated with increased evaporation and decreased
rainfall in the upper Selati (Climate System Analysis Group (CSAG), 2016), provides a further
reason why scenario thinking and planning is required in the region. However, many of the
disbenefits associated with Selati River being in a poor state are externalised, with the
biodiversity impacts of elevated phosphate and sulphate concentrations experienced primarily by
Kruger National Park, which is located outside of the Selati River catchment but which is reliant
on environmental compliance within the Selati. Hence, there is a disjuncture between the low set
of incentives for stakeholders within the Selati to improve the state of the river, and the high
requirement for a functioning river system by stakeholders outside of the catchment. Indeed,
payment for ecosystem services have been considered in the region (Chapman, 2006) and offsets
are being debated.
In this paper we report on the Selati sub-catchment case study. In 2015, RESILIM-O began using
system dynamics modelling (SDM) in an integrative fashion, drawing on the earlier stakeholder
engagements and collaborative research efforts in Phase | (including the dedicated modelling on
water quality, hydrology, and land use patterns that had been undertaken within the broader
Olifants catchment). We examine how the System Dynamics modelling (SDM) process acted to
support understanding, planning, and action systemically, and we seek to develop scientific
insights on the concerted modelling effort in which system dynamics provided the paradigmatic
framing for both a stakeholder engagement process and for developing an underpinning,
integrative simulation model.
[paper excluded from here onwards]
Page 3 of 9
ISDC 2016 paper (RESILIM-O) 19/08/2016
Acknowledgements:
The research reported on in this paper was funded by the USAID: RESILIM-O program. The
USAID: RESILIM-O program is funded by the United States Agency for International
Development, under USAID Southern Africa — RFA-674-12-000016 RESilience in the LIMpopo
Basin Program (RESILIM). The program is implemented by the Association for Water and Rural
Development (AWARD), in association with project partners. A full list of the 50+ deliverables
that have been developed in the System Dynamics Modelling (SDM) component of the
RESILIM-O programme can be accessed via the link provided in Appendix A.
ResiMod version | was primarily developed by Willem Jonker and Theo York (independent
consultants) with oversight and input from Jai Clifford-Holmes (an AWARD Research
Associate). Additional technical input and modelling was provided by Fabio Diaz (an
independent consultant also contracted by AWARD).
The Social Ecological Systems (SES) working group of RESILIM-O, led by Sharon Pollard,
provided crucial modelling input and the original project framework within which the modelling
could be undertaken. Key inputs from the following AWARD researchers and research associates
are gratefully acknowledged: Derick du Toit, Eureta Rosenberg, Dirk Versveld, Stephen Holness,
Jan Graf, Hugo Retief, Taryn Kong, Charles Chikunda, Tebogo Mathebula, Fredrick Govere,
Reuben Thifhulufwelwi, Manuel Mangombeyi, Farnaz Farhang, and Wehncke van der Merwe.
The input from Doreen Robinson, Regional Environment Chief, USAID: Southern Africa, is also
gratefully acknowledged, as is the SDM training facilitated at AWARD by Josephine Musango.
Initial modelling efforts were presented for peer-review as two presentations at the 3“ Eskom
System Dynamics Conference (in collaboration with the South African System Dynamics
Chapter), held in Johannesburg, South Africa, on the 10" November 2015. The prototype
dashboard design produced by Sovtech for AWARD was presented as a draft only and remains
the intellectual property of Sovtech, pending the outcome of the contractual negotiations between
AWARD and Sovtech between April — May 2016.
Finally, the authors acknowledge the valuable input from the following stakeholders: managers
and technicians from the Ba-Phalaborwa Local Municipality; Safety, Health, Environmental
Quality (SHEQ) engineers and officers from the Palabora Mining Company (PMC) and Foskor
Mining; Commercial and emerging farmers and irrigation managers from the Selati Irrigation
Board (SRIB); water resource managers from Kruger National Park and bio-monitoring river
managers from the South African Earth Observation Network (SAEON).
Page 4 of 9
ISDC 2016 paper (RESILIM-O) 19/08/2016
2. References
Andersen, D. F., Vennix, J. a M., Richardson, G. P., & Rouwette, E. a J. a. (2007). Group model
building: problem structuring, policy simulation and decision support. Journal of the
Operational Research Society, 58(5), 691-694. doi:10.1057/palgrave .jors.2602339
Association for Water and Rural Development (AWARD). (2016). Report-back to the Ba-
Phalaborwa Local Municipality: Systems modelling, wastewater management, and
scenarios for building resilience. P.O. Box 1919, Hoedspruit, Limpopo, 1380, South Africa.
Chapman, A. (2006). Hydrology and land use in the Ga-Selati catchment. CSIR, Pretoria, South
Africa and International Institute for Environment and Development, London, UK.
Clifford-Holmes, J. K. (2015a). Fire and Water: A Transdisciplinary Investigation of Water
Governance in the Lower Sundays River, South Africa. PhD thesis, Rhodes University,
South Africa.
Clifford-Holmes, J. K. (2015b). SYSTEM DYNAMICS-BASED MODELLING PROCESS FOR
THE RESILIM-O PROJECT: Process Plan. Draft V.2 (20th July 2015), consulting report
produced for the Association for Water and Rural Development (AWARD) for the USAID
Southern Africa RESILIM-O programme (RFA-674-12-000016).
Clifford-Holmes, J. K., Slinger, J. H., de Wet, C., & Palmer, C. G. (n.d.). Modeling in the
“Muddled Middle”: A Case Study of Water Service Delivery in Post-Apartheid South
Africa. In C. Garcia-Diaz & C. Olaya (Eds.), Social Systems Engineering: The Design of
Complexity. Wiley (in preparation).
Clifford-Holmes, J. K., Slinger, J. H., Mbulawa, P., & Palmer, C. G. (2015). Modes of Failure of
South African Local Government in the Water Services Sector. In Conference Proceedings
of the 33rd International Conference of the System Dynamics Society, 19-23 July 2015.
Cambridge, Massachusetts, USA.
Climate System Analysis Group (CSAG). (2016). Advisory Report: Climate Scenarios for
modelling the hydrological flow for the Ga-Selati Catchment. Advisory report submitted to
the Association for Water and Rural Development (AWARD) for the USAID Southern
Africa RESILIM-O programme (RFA-674-12-000016).
Cockerill, K., Passell, H., & Tidwell, V. (2006). Cooperative Modeling: Building Bridges
between Science and the Public. Journal of the American Water Resources Association,
42(2), 457-471. doi:10.1111/j.1752- 1688 .2006.tb03850.x
Etienne, M. (2014). Companion Modelling: A Participatory Approach to Support Sustainable
Development. Dordrecht: Springer. doi: 10.1007/978-94-017-8557-0
Ford, A. (2009). Modeling the Environment (Second ed.). Washington D.C.: Island Press.
Hovmand, P. S., Andersen, D. F., Rouwette, E. A. J. A., Richardson, G. P., Rux, K., & Calhoun,
A. (2012). Group Model-Building “ Scripts ” as a Collaborative Planning Tool. Systems
Research and Behavioural Science, 29, 179-193. doi:10.1002/sres
Hovmand, P. S., Rouwette, E. A. J. A., Andersen, D. F., Richardson, G. P., & Kraus, A. (2013).
Scriptapedia 406. Retrieved from
http://tools.systemdynamics .org/scrpda/scriptapedia_4.0.6.pdf
Jonker, W., York, T., & Clifford-Holmes, J. K. (2015a). ResiMod — Problem Conceptualisation:
Draft 2 of ResiMod deliverable 1, 30/09/2015. Internal report produced for the Association
for Water and Rural Development (AWARD) for the USAID Southern Africa RESILIM-O
programme (RFA-674-12-000016).
Page 5 of 9
ISDC 2016 paper (RESILIM-O) 19/08/2016
Jonker, W., York, T., & Clifford-Holmes, J. K. (2015b). ResiMod v.1. Modelling Process and
Structure Report. Draft 1 of ResiMod Deliverable 6. Internal report produced for the
Association for Water and Rural Development (AWARD) for the USAID Southern Africa
RESILIM-O programme (RFA-674-12-000016).
Lane, D. C. (2010). Participative modelling and big issues: Defining features of system
dynamics? Systems Research and_ Behavioral Science, 27(4), 461-465.
doi:10.1002/sres.1048
Langsdale, S. M. (2007). Participatory model building for exploring water management and
climate change futures in the Okanagan Basin, British Columbia, Canada. PhD dissertation,
The University of British Columbia.
Mangombeyi, M., Pollard, S., & Hansen, K. (2015). Overview concept map: Challenges faced in
poor performing wastewater treatment plants (WWTP) from field visits experiences of 9
WWTP in the Olifants catchment. Internal report produced by the Association for Water and
Rural Development (AWARD) under the for the USAID RESILIM-O project. P.O. Box
1919, Hoedspruit, 1380, South Africa.
Muller, M., Schreiner, B., Smith, L., Koppen, B. Van, Sally, H., Aliber, M., ... Pietersen, K.
(2009). Water security in South Africa (No. 12). Development Planning Division Working
Paper Series. No. 12.
Ostrom, E. (2010a). Beyond Markets and States: Polycentric Governance of Complex Economic
Systems. American Economic Review, 100(June), 641-672.
Ostrom, E. (2010b). Polycentric systems for coping with collective action and global
environmental change. Global Environmental Change, 20, 550-557.
doi:10.1016/j.gloenvcha.2010.07.004
Pollard, S., Biggs, H., Rydannykh, A., du Toit, D., & Laporte, A. (2013). The Shared River
Initiative - Phase II - Part 3 - Development of a participatory framework for understanding
Water-related ecosystem services within the context of Classification and the Reserve. Water
Research Commission (WRC) Project No. KS5/1920). Retrieved from
http://www.wrc.org.za/Knowledge Hub Documents/Research Reports/TT 574 -13.pdf
Pollard, S., & Laporte, A. (2015). Living in Phalaborwa: a collaborative view of the system
through the eyes of people. Association for Water and Rural Development (AWARD). P.O.
Box 1919, Hoedspruit, 1380, Limpopo, South Africa.
Richardson, G. P. (2013). Concept models in group model building. System Dynamics Review,
29(1), 42-55. doi:10.1002/sdr.1487
Rouwette, E. A. J. A., & Vennix, J. A. M. (2006). System Dynamics and Organizational
Interventions. Systems Research and Behavioural Science, 23(4), 451-466. doi:10.1002/sres
Star, S. L., & Griesemer, J. R. (1989). Institutional Ecology, “Translations” and Boundary
Objects: Amateurs and Professionals in Berkeley’s Museum of Vertebrate Zoology, 1907-
39. Social Studies of Science, 19(3), 387-420.
Stave, K. A. (2010). Participatory system dynamics modeling for sustainable environmental
management: Observations from Four Cases. Sustainability, 2(9), 2762-2784.
doi:10.3390/su2092762
Sterman, J., Franck, T., Fiddaman, T., Jones, A., McCauley, S., Rice, P., ... Rooney-Varga, J. N.
(2014). WORLD CLIMATE: A Role-Play Simulation of Climate Negotiations. Simulation
& Gaming, 10468781 13514935-. doi:10.1177/1046878113514935
Stirzaker, R., Biggs, H., Roux, D., & Cilliers, P. (2010). Requisite Simplicities to Help Negotiate
Page 6 of 9
ISDC 2016 paper (RESILIM-O) 19/08/2016
Complex Problems. AMBIO, 39, 600-607. doi:10.1007/s13280-010-0075-7
Terreblanche, S. (2002). A history of inequality in South Africa, 1652-2002. Pietermaritzburg:
University of Natal Press.
Tidwell, V. C., Passell, H. D., Conrad, S. H., & Thomas, R. P. (2004). System dynamics
modeling for community-based water planning: Application to the Middle Rio Grande.
Aquatic Sciences, 66(4), 357-372. doi:10.1007/s00027-004-0722-9
Turton, A. (2013). WATER: Overview of the Water Situation in South Africa. In D. Parry-
Davies (Ed.), The Enviropaedia: Eco-Logical Edition (Thought Leadership for Sustainable
Development) (pp. 336-339). Cape Town: Eco-Logic Publishing.
van den Belt, M. (2004). Mediated Modeling: A System Dynamics Approach to Environmental
C Building. Washington, DC: Island Press.
Vennix, J. A. M. (1999). Group model-building: tackling messy problems. System Dynamics
Review, 15(4), 379-401.
Videira, N., Antunes, P., Santos, R., & Lopes, R. (2010). A participatory modelling approach to
support integrated sustainability assessment processes. Systems Research and Behavioral
Science, 27, 446-460. doi:10.1002/sres
Vordsmarty, C. J., McIntyre, P. B., Gessner, M. O., Dudgeon, D., Prusevich, A., Green, P., ...
Davies, P. M. (2010). Global threats to human water security and river biodiversity. Nature,
467(7315), 555-561. doi: 10.1038/nature09549
Vriens, D., & Achterbergh, J. (2006). The Social Dimension of System Dynamics-Based
Modelling. Systems Research and Behavioural Science, 23(4), 553-563. doi:10.1002/sres
Warren, K. (2014). Agile SD: fast, effective, reliable. Conference proceedings of the 32nd
International Conference of the System Dynamics Society, 21-24 July 2014, Delft, the
Netherlands. Retrieved from
http://www.systemdynamics.org/conferences/20 1 4/proceed/papers/P1126.pdf
END OF DRAFT PAPER
Page 7 of 9
ISDC 2016 paper (RESILIM-O) 19/08/2016
Appendix A Summary of the supplementary material
1. Association for Water and Rural Development (AWARD). (2016). Report-back to the Ba-
Phalaborwa Local Municipality: Systems modelling, wastewater , and scenarios for
building resilience. P.O. Box 1919, Hoedspruit, Limpopo, 1380, South Africa.
To download:
https://www.dropbox .com/s/iebmt0db04nhgao/ 1 .%20Report%20for%20BPLM_Feb2016%20upd
ate_reduced.pdf?dl=0
2. Powerpoint slides from the mining workshop ran with Palabora Mining Company (PMC):
To download:
https://www.dropbox.com/s/6ttikpz8dk9iivy/2.%20PRESENTATION%20PMC%2012%20NOV.
pdf?dl=0
3. Example of blank baseline survey used in the System Dynamics Modelling (SDM) workshops
To download:
://www dropbox.com/s/rexgek1675zuqb8/3.%20SDM%20 Worksho
rmers%20Baseline%20Questions_HB%2BJCH_10March_final.pdf?dl=0
%20Commercial %20Fa
4. Association for Water and Rural Development (AWARD). (2016). Key to SDM [System
Dynamics Modelling] deliverables to date (as of 06/04/2016). AWARD internal reporting. P.O.
Box 1919, Hoedspruit, Limpopo, 1380, South Africa.
To download:
https://www.dropbox.com/s/okljh1ii1z82tj0/6.%20Key %20to%20SDM %20deliverables_6April2
016.pdf?dl=0
Page 8 of 9
ISDC 2016 paper (RESILIM-O) 19/08/2016
Appendix B Summary of desirable features of a core modelling platform
RESILIM-O
Table B.1 : List of desirable features for the core modelling platform in the RESILIM-O project. Source:
drawn from Pollard et al. (2013: 50).
Feature
Description
Intuitive & diagrammatic
Facilitates model construction, model transparency, and
stakeholder involvement in the modelling process.
Expressive
Allows various classes of model to be implemented (e.g. system
dynamics, agent-based, stochastic, spatial).
Computationally efficient
Allows for complex models to be run quickly.
Allows for rapid experimentation with simple models.
Scalable
Enables the handling of very large models, both in terms of
number of equations and degree of disaggregation.
Conceptual closeness
Conceptually close to other peri-modelling activities
(e.g. concept mapping, causal loop analysis, ecosystem services
concepts).
Reduces the “conceptual gap” between the per-modelling
activities and a corresponding model.
Integration and conversion
friendly
Allows for software integration with other modelling platforms.
Allows for automatic conversion between the platform’s native
model-representation format and the formats used by other
modelling systems.
Obtains maximum leverage from available resources.
Enables the model to be linked to external data sets.
Modularity Supports modular modelling.
Allows for the re-use of sub-models.
Supports nario planning | Enables scenarios to be modelled through alternative model
settings, linked to external data sets and/or other models.
Customisable
Allows for the customisation of the run-time user interface: to
make it possible for people with no modelling experience to
experiment with the models.
Page 9 of 9
