brs seminar series presents -...

41
BRS SEMINAR SERIES PRESENTS: Friday 22 October Earth System Science in the Early Anthropocene John Finnigan - CSIRO According to Nobel Prize winner Paul Cruetzen, the human race has entered the Anthropocene era when human influences on the earth system are as profound as those of natural forces. To understand the workings of the earth system, we must now understand the interactions between the biosphere and the social and economic systems of mankind. Fundamental to the behaviour of the earth system is the interaction between human learning and adaptation and the intrinsic dynamics of the biosphere. How can we combine models of the biosphere, rooted in the traditional domain of natural science, with the insights from social science and economics that describe human behaviour? The burgeoning field of ‘Complex System Science’ is beginning to provide the required methodology, as it demonstrates that many characteristics of the behaviour of societies can indeed be treated quantitatively, using methods developed to describe systems of all kinds from atoms to humans. This talk will show how new ‘agent-based’ approaches can be employed to represent agricultural systems, where human decisions play as large a role as climate, water and soil in the success of a farmer. It will then touch briefly on how more abstract techniques like ‘social network theory’ might allow questions about the long-term resilience and stability of such systems to be addressed. Finally, it will comment on the current big unknowns in these approaches. John Finnigan Holds a BSc from the University of Manchester and his PhD from the Australian National University. He has worked on the fluid dynamics of transonic wing designs for Hawker-Siddeley Aviation and on atmospheric dynamics at the Georgia Institute of Technology and at NOAA and NCAR in Boulder, Colorado. From 1989 to 1995 he was head of the CSIRO Centre for Environmental Mechanics, and variously Chief Research Scientist for the CSIRO Divisions of Land and Water and, later, Atmospheric Research. He was appointed founding Director of the CSIRO Centre for Complex Systems Science in 2001. He is heavily involved with the European and US programs on measuring the global carbon cycle and is a member of the Scientific Steering Committee for the IGBP-2 Program: Interactions between Land Ecosystems and Atmospheric Processes (iLEAPS)

Upload: others

Post on 07-Jun-2020

4 views

Category:

Documents


0 download

TRANSCRIPT

Page 1: BRS SEMINAR SERIES PRESENTS - data.daff.gov.audata.daff.gov.au/brs/data/warehouse/brsShop/data/finnigan22oct.pdf · BRS SEMINAR SERIES PRESENTS: Friday 22 October Earth System Science

BRS SEMINAR SERIES PRESENTS:

Friday 22 October

Earth System Science in the Early Anthropocene

John Finnigan - CSIRO

According to Nobel Prize winner Paul Cruetzen, the human race has entered theAnthropocene era when human influences on the earth system are as profound as thoseof natural forces. To understand the workings of the earth system, we must nowunderstand the interactions between the biosphere and the social and economic systemsof mankind. Fundamental to the behaviour of the earth system is the interaction betweenhuman learning and adaptation and the intrinsic dynamics of the biosphere.

How can we combine models of the biosphere, rooted in the traditional domain of naturalscience, with the insights from social science and economics that describe humanbehaviour? The burgeoning field of ‘Complex System Science’ is beginning to provide therequired methodology, as it demonstrates that many characteristics of the behaviour ofsocieties can indeed be treated quantitatively, using methods developed to describesystems of all kinds from atoms to humans.

This talk will show how new ‘agent-based’ approaches can be employed to representagricultural systems, where human decisions play as large a role as climate, water and soilin the success of a farmer. It will then touch briefly on how more abstract techniques like‘social network theory’ might allow questions about the long-term resilience and stability ofsuch systems to be addressed. Finally, it will comment on the current big unknowns inthese approaches.

John Finnigan• Holds a BSc from the University of Manchester and his PhD from the Australian

National University.• He has worked on the fluid dynamics of transonic wing designs for Hawker-Siddeley

Aviation and on atmospheric dynamics at the Georgia Institute of Technology and atNOAA and NCAR in Boulder, Colorado.

• From 1989 to 1995 he was head of the CSIRO Centre for EnvironmentalMechanics, and variously Chief Research Scientist for the CSIRO Divisions of Landand Water and, later, Atmospheric Research.

• He was appointed founding Director of the CSIRO Centre for Complex SystemsScience in 2001.

• He is heavily involved with the European and US programs on measuring the globalcarbon cycle and is a member of the Scientific Steering Committee for the IGBP-2Program: Interactions between Land Ecosystems and Atmospheric Processes(iLEAPS)

Page 2: BRS SEMINAR SERIES PRESENTS - data.daff.gov.audata.daff.gov.au/brs/data/warehouse/brsShop/data/finnigan22oct.pdf · BRS SEMINAR SERIES PRESENTS: Friday 22 October Earth System Science

Earth System Science in theEarly Anthropocene

John Finnigan

CSIRO Atmospheric Research

Canberra, Australia

Page 3: BRS SEMINAR SERIES PRESENTS - data.daff.gov.audata.daff.gov.au/brs/data/warehouse/brsShop/data/finnigan22oct.pdf · BRS SEMINAR SERIES PRESENTS: Friday 22 October Earth System Science

The Anthropocene Era

Human influence on the earth’s systems is as great as thatof natural forces

Page 4: BRS SEMINAR SERIES PRESENTS - data.daff.gov.audata.daff.gov.au/brs/data/warehouse/brsShop/data/finnigan22oct.pdf · BRS SEMINAR SERIES PRESENTS: Friday 22 October Earth System Science

Contents• Are we really in the Anthropocene?• Complex System Science and Earth System

Science• Systems and Complexity• emergence and self-organisation

• Two Approaches to modelling Complex Systems• Dynamical Systems Theory and its limitations• Agent Based Modelling

• What Lies Beneath-Dynamics on Networks• Interactions and network topology• Growth, form and function of networks• Dynamics on networks-evolution and change

• Summary and Conclusion

Page 5: BRS SEMINAR SERIES PRESENTS - data.daff.gov.audata.daff.gov.au/brs/data/warehouse/brsShop/data/finnigan22oct.pdf · BRS SEMINAR SERIES PRESENTS: Friday 22 October Earth System Science

From: Steffen et al. 2003

A wide variety ofsocial and economicindicators show asharp change ofslope within thedecade centred on1950

Page 6: BRS SEMINAR SERIES PRESENTS - data.daff.gov.audata.daff.gov.au/brs/data/warehouse/brsShop/data/finnigan22oct.pdf · BRS SEMINAR SERIES PRESENTS: Friday 22 October Earth System Science

From: Steffen et al. 2004

A similar range ofbio-geophysicalindicators start tochange rapidly at thesame time. Whathappened?

Page 7: BRS SEMINAR SERIES PRESENTS - data.daff.gov.audata.daff.gov.au/brs/data/warehouse/brsShop/data/finnigan22oct.pdf · BRS SEMINAR SERIES PRESENTS: Friday 22 October Earth System Science

To Understand and predict the behaviour of earthSystems in the Anthropocene we need to treathuman and natural systems at the same level.

Definitions

• Earth Systems Science is the integrated study ofphysical, chemical, biological, and social processes thatare important to the functioning of the Earth

• It employs the tools of Complex Systems Science tounderstand the fundamentally connected character ofnatural and human dominated systems.

Page 8: BRS SEMINAR SERIES PRESENTS - data.daff.gov.audata.daff.gov.au/brs/data/warehouse/brsShop/data/finnigan22oct.pdf · BRS SEMINAR SERIES PRESENTS: Friday 22 October Earth System Science

What is Complex System Science?

It has two elements:

• Systems– collections of interacting things

• Complexity– the essence of which is the property of self-

organisation or emergence of structure fromthe interaction between the constituent partsof the system

Page 9: BRS SEMINAR SERIES PRESENTS - data.daff.gov.audata.daff.gov.au/brs/data/warehouse/brsShop/data/finnigan22oct.pdf · BRS SEMINAR SERIES PRESENTS: Friday 22 October Earth System Science

Modelling the Earth System

When we model the Earth Systemwe are effectively describing a setof agents that have more or lesscomplex attributes and more orless complex interactions

Human decisions andreactions

Page 10: BRS SEMINAR SERIES PRESENTS - data.daff.gov.audata.daff.gov.au/brs/data/warehouse/brsShop/data/finnigan22oct.pdf · BRS SEMINAR SERIES PRESENTS: Friday 22 October Earth System Science

Some complex systems can be reduced to low-dimensionalmodels and studied as ‘dynamical systems’. This is the

traditional approach to modelling

1x

3xxa

13a

32a

a

In low dimensional models, interactions between manysimpler ‘agents’ are reduced to interactions between a fewcomplex aggregates

Page 11: BRS SEMINAR SERIES PRESENTS - data.daff.gov.audata.daff.gov.au/brs/data/warehouse/brsShop/data/finnigan22oct.pdf · BRS SEMINAR SERIES PRESENTS: Friday 22 October Earth System Science

• State variables: B(t) = biomassH(t) = human population

• Dynamical equations: dB/dt = N - kB - E dH/dt = g (E - mH)

• Model for resource production: E = cBH

– more humans extract more biospheric resource

– each human extracts better as the biomass increases (B is a surrogate forquality of life)

An extreme example: Human-biosphere interactionas a two-equation dynamical system: a model

Net PrimaryProduction of

biomass

Resourceproductionby humans

Respirationof biomass

Surplus inresource

production

Populationgrowth rate

H BTwo interacting agents

Page 12: BRS SEMINAR SERIES PRESENTS - data.daff.gov.audata.daff.gov.au/brs/data/warehouse/brsShop/data/finnigan22oct.pdf · BRS SEMINAR SERIES PRESENTS: Friday 22 October Earth System Science

Human-biosphere interaction as a dynamical systemTrajectories on a (B,H) plane for 6 scenarios lead to simple, fixed-pointattractors

Eden Occupy Disaster

Grow Subsist Exploit

Occupy

0

0.2

0.4

0.6

0.8

1

1.2

1.4

1.6

1.8

2

0 10 20 30 40 50B = biomass

H=

hu

ma

ns

Grow

0

1

2

3

4

5

6

7

0 10 20 30 40 50B = biomass

H=

hu

ma

ns

Subsist

0

2

4

6

8

10

12

14

16

18

20

0 10 20 30 40 50B = biomass

H=

hu

ma

ns

Exploit

0

1

2

3

4

5

6

0 10 20 30 40 50B = biomass

H=

hu

ma

ns

Disaster

0

0.2

0.4

0.6

0.8

1

1.2

1.4

0 5 10 15 20 25 30

B = biomass

H=

hu

ma

ns

Eden

0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1

0 10 20 30 40 50

B = biomass

H=

hu

ma

ns

Page 13: BRS SEMINAR SERIES PRESENTS - data.daff.gov.audata.daff.gov.au/brs/data/warehouse/brsShop/data/finnigan22oct.pdf · BRS SEMINAR SERIES PRESENTS: Friday 22 October Earth System Science

10102883dxxydtdy

But even a slight increase in complexity leads to behaviour thatis much more difficult to interpret or even to calculate

x

y

The Lorentz Attractor

Page 14: BRS SEMINAR SERIES PRESENTS - data.daff.gov.audata.daff.gov.au/brs/data/warehouse/brsShop/data/finnigan22oct.pdf · BRS SEMINAR SERIES PRESENTS: Friday 22 October Earth System Science

In ‘Agent-Based Modelling’, we adopt a quitedifferent strategy. We disaggregate the system

into many simple parts

We aim to capture the complexity and non-linearity of thesystem through the pattern of the links between manyagents with simple attributes. This is a robust strategycopied from nature

Page 15: BRS SEMINAR SERIES PRESENTS - data.daff.gov.audata.daff.gov.au/brs/data/warehouse/brsShop/data/finnigan22oct.pdf · BRS SEMINAR SERIES PRESENTS: Friday 22 October Earth System Science

Reynold’s Boids (1986) Simple local interactionslead to complex group behaviour

AlignementAlignement CohesionCohesion SegregationSegregation

Page 16: BRS SEMINAR SERIES PRESENTS - data.daff.gov.audata.daff.gov.au/brs/data/warehouse/brsShop/data/finnigan22oct.pdf · BRS SEMINAR SERIES PRESENTS: Friday 22 October Earth System Science

Crowd Behaviour-More complex but stilllocal interactions

Agent Based Modelsimulation of a crowdescaping from a fire.

Placing a pillar nearthe exit counter-intuitively increases theexit rate and reducesinjuries.

Conventional treatmentof crowd movementusing diffusionequations would predictthe opposite

Diagram from Bonabeau (2002)

Page 17: BRS SEMINAR SERIES PRESENTS - data.daff.gov.audata.daff.gov.au/brs/data/warehouse/brsShop/data/finnigan22oct.pdf · BRS SEMINAR SERIES PRESENTS: Friday 22 October Earth System Science

Bounded rationality or Limited Cognition is acharacteristic of all sentient creatures interacting

with each other and their environment

Whether we are modelling socialinsects or human societies (orsystems of inanimate objects),individual agents have limited‘knowledge’. These limits aredefined by their neighbourhood, aconcept that can be interpreted bothliterally and more abstractly

Bounded rationality is incorporatednaturally in Agent-based Models

Page 18: BRS SEMINAR SERIES PRESENTS - data.daff.gov.audata.daff.gov.au/brs/data/warehouse/brsShop/data/finnigan22oct.pdf · BRS SEMINAR SERIES PRESENTS: Friday 22 October Earth System Science

How can we use these concepts of disaggregatedsimple agents and bounded rationality to modelhuman interaction with the biosphere?

We will look at a model of resource use conflict inan irrigated rice farming system in Thailand

The issue being addressed was

Does Upstream Water Management affectdownstream farm sustainability ?

Page 19: BRS SEMINAR SERIES PRESENTS - data.daff.gov.audata.daff.gov.au/brs/data/warehouse/brsShop/data/finnigan22oct.pdf · BRS SEMINAR SERIES PRESENTS: Friday 22 October Earth System Science

Example of Modelling a Human-Biophysical system withan Agent Based Model: CatchScape

Location:Location:Mae Uam Catchment

45 km2, Chiang Mai Province

Characteristics:Characteristics:2 irrigation schemes.

Rice/Soybean/Onion cropping system.

Issue:Issue:Does Upstream Water Management affectthe downstream farm sustainability ?

Acknowledgements to Pascal Perez (ANU) and the CIRAD Cormas Group(Becu et al., 2001)

Page 20: BRS SEMINAR SERIES PRESENTS - data.daff.gov.audata.daff.gov.au/brs/data/warehouse/brsShop/data/finnigan22oct.pdf · BRS SEMINAR SERIES PRESENTS: Friday 22 October Earth System Science

RR

SR

RO

DD

PLOT

CRCROP

AET

ID

IR Canal

VOL

Structure of CatchScape: Sentient and inanimateagents

River

Farmer

Village

Page 21: BRS SEMINAR SERIES PRESENTS - data.daff.gov.audata.daff.gov.au/brs/data/warehouse/brsShop/data/finnigan22oct.pdf · BRS SEMINAR SERIES PRESENTS: Friday 22 October Earth System Science

Social Interactions: Task based and local learning

Collaboration

Negotiation

level 1

Negotiation

level 2

Water ?

Water ?

OK

Water ?Water ?

OK

Water ?

Page 22: BRS SEMINAR SERIES PRESENTS - data.daff.gov.audata.daff.gov.au/brs/data/warehouse/brsShop/data/finnigan22oct.pdf · BRS SEMINAR SERIES PRESENTS: Friday 22 October Earth System Science

farmer

Off-farmdecision

Negativecash

Season

Farmingdecision

Dryincome

>

wages

Off-farmdecision

no

wet

dry

no

yes

yes

Individual decisions: Rule based but allowing multiplestrategies. Aggregated to village level this leads to adaptive

behaviour

Page 23: BRS SEMINAR SERIES PRESENTS - data.daff.gov.audata.daff.gov.au/brs/data/warehouse/brsShop/data/finnigan22oct.pdf · BRS SEMINAR SERIES PRESENTS: Friday 22 October Earth System Science

An appropriateviewpoint to testthe assumption

Dry season cropping pattern

Exploring the System with Viewpoints

amLegendAmount of cashFig. 7. Spatial distributionHousehold ’s cash

Assumption:Farmer’s income dichotomycomes from the plot positionwithin the irrigated scheme

Importance offarmer’s waterexpectation too

Page 24: BRS SEMINAR SERIES PRESENTS - data.daff.gov.audata.daff.gov.au/brs/data/warehouse/brsShop/data/finnigan22oct.pdf · BRS SEMINAR SERIES PRESENTS: Friday 22 October Earth System Science

Living in theLiving in the AnthropoceneAnthropocene::Virtual experiment with water thievesVirtual experiment with water thieves

nn The modelled result of <5% of the villageThe modelled result of <5% of the villagepopulation breaking the social rulespopulation breaking the social rulesgoverning water allocation had a greatergoverning water allocation had a greaterimpact on the village economy than twoimpact on the village economy than twosuccessive drought years.successive drought years.

02040608012345678910 Thives scenar

Page 25: BRS SEMINAR SERIES PRESENTS - data.daff.gov.audata.daff.gov.au/brs/data/warehouse/brsShop/data/finnigan22oct.pdf · BRS SEMINAR SERIES PRESENTS: Friday 22 October Earth System Science

Evidently, we can build models of human-biosphere interaction using the techniques

of ‘Agent based Modelling’

• But we want to be able to make general deductions orpredictions about the behaviour of such systems byunderstanding them at a deeper level.

• We want to go beyond merely running many simulations withdifferent assumptions and parameterisations.

• We would like to be able to relate properties of humanecosystems like resilience, adaptation or prosperity to thesystem structure.

• Disaggregating these systems into many interactingelements allows us to employ the tools of networktheory to do this.

Page 26: BRS SEMINAR SERIES PRESENTS - data.daff.gov.audata.daff.gov.au/brs/data/warehouse/brsShop/data/finnigan22oct.pdf · BRS SEMINAR SERIES PRESENTS: Friday 22 October Earth System Science

Examples of Networks

Food webof LittleRock Lake,Wisconsin

New YorkStateelectricpowergrid

A portion of themolecularinteraction map forthe regulatorynetwork thatcontrols themammalian cellcycle

The Global carbon Cycle

Page 27: BRS SEMINAR SERIES PRESENTS - data.daff.gov.audata.daff.gov.au/brs/data/warehouse/brsShop/data/finnigan22oct.pdf · BRS SEMINAR SERIES PRESENTS: Friday 22 October Earth System Science

Dynamics on Networks

• Complex interactions onsimple networks

• Simple interactions oncomplex networks

• Evolving network structure

• Co-evolution of structure andinteractions

Page 28: BRS SEMINAR SERIES PRESENTS - data.daff.gov.audata.daff.gov.au/brs/data/warehouse/brsShop/data/finnigan22oct.pdf · BRS SEMINAR SERIES PRESENTS: Friday 22 October Earth System Science

The opposite of a regular lattice is a randomnetwork-what can we say in this case?

Start with a set of nodes and add connections randomlybetween them. The first ‘structures that emerge are ‘trees’.

Page 29: BRS SEMINAR SERIES PRESENTS - data.daff.gov.audata.daff.gov.au/brs/data/warehouse/brsShop/data/finnigan22oct.pdf · BRS SEMINAR SERIES PRESENTS: Friday 22 October Earth System Science

Continue adding connections and feedback loopsor cycles form

Page 30: BRS SEMINAR SERIES PRESENTS - data.daff.gov.audata.daff.gov.au/brs/data/warehouse/brsShop/data/finnigan22oct.pdf · BRS SEMINAR SERIES PRESENTS: Friday 22 October Earth System Science

When the number of connections is ~1/2 thenumber of nodes, a ‘giant’ structure appears and

most nodes are connected

Page 31: BRS SEMINAR SERIES PRESENTS - data.daff.gov.audata.daff.gov.au/brs/data/warehouse/brsShop/data/finnigan22oct.pdf · BRS SEMINAR SERIES PRESENTS: Friday 22 October Earth System Science

The Connectivity Avalanche

Number ofconnections, n

OverallConnectivity

Doublejump

PhaseTransition

fully connected

Double jump and emergence of a “giant” appears when the number of connectionsn ~N/2, with N the number of nodes. Network is practically fully connected whenn~N/2 Log(N). (Erdos and Renyi, 1960)

treescycles

Giantgroupemerges

Page 32: BRS SEMINAR SERIES PRESENTS - data.daff.gov.audata.daff.gov.au/brs/data/warehouse/brsShop/data/finnigan22oct.pdf · BRS SEMINAR SERIES PRESENTS: Friday 22 October Earth System Science

Consequences of the ‘ConnectivityAvalanche’

• Gridlock can emerge in networks processing fluctuating inputs. Ifelements are operated near 100% capacity they shed ‘load’ to otherelements when their inputs exceed their capacity. The shedding isequivalent to setting up links between elements and, if the system isnear the phase transition, blockage can ‘avalanche’ through thesystem.

• Examples are traffic jams, power blackouts, hospital bed overloads?LAN lock-ups, Internet (IP/TCP), inefficient businesses, CSIRO?

• Foot and Mouth Disease: movement of herds between abattoirs,markets and to take advantage of consumer preferences and bizarreEU subsidies took the herds past a critical level of connectivity andallowed F&M to spread faster than controls could detect infectiousanimals. Another element at work here was the "small world”phenomenon.

Page 33: BRS SEMINAR SERIES PRESENTS - data.daff.gov.audata.daff.gov.au/brs/data/warehouse/brsShop/data/finnigan22oct.pdf · BRS SEMINAR SERIES PRESENTS: Friday 22 October Earth System Science

Foot and Mouth Disease in Britain A Consequence of the Connectivity Avalanche?

Economic rationalization of abattoirsand bizarre EU subsidies increasedthe connections between herds to acritical point.

Changes to F&M reporting rules mayhave delayed the isolation ofinfectious animals.

The relationship between theseactions and the epidemiology of F&Mwas not appreciated in advance (atleast where it mattered) because thelivestock industry was not viewed asan integrated system.

Page 34: BRS SEMINAR SERIES PRESENTS - data.daff.gov.audata.daff.gov.au/brs/data/warehouse/brsShop/data/finnigan22oct.pdf · BRS SEMINAR SERIES PRESENTS: Friday 22 October Earth System Science

Growth, Form and Function ofNetworks

Regular Network: each node has the samenumber of connections

Homogeneous network: Number of connectionsper node varies but there is a clear average value(follows e.g. a Poisson distribution). Networkslike this result from randomly connecting nodes.Near the phase transition they are vulnerable torandom removal of links

Heterogeneous or ‘scale free’ network: There is noaverage number of connections per node (followsa power law): Living networks that grow byaccretion often have this dendritic form. They areresilient to random removal of links but vulnerableto a targeted attack that removes a key node)

Page 35: BRS SEMINAR SERIES PRESENTS - data.daff.gov.audata.daff.gov.au/brs/data/warehouse/brsShop/data/finnigan22oct.pdf · BRS SEMINAR SERIES PRESENTS: Friday 22 October Earth System Science

Protein-protein interactions in the yeastproteome-Growth, form and function

Map of protein–proteininteractions. The largestcluster, which contains 78%of all proteins, is shown. Thecolour of a node signifies thephenotypic effect of removingthe corresponding protein(red, lethal; green, non-lethal;orange, slow growth; yellow,unknown

Page 36: BRS SEMINAR SERIES PRESENTS - data.daff.gov.audata.daff.gov.au/brs/data/warehouse/brsShop/data/finnigan22oct.pdf · BRS SEMINAR SERIES PRESENTS: Friday 22 October Earth System Science

WWW Showing major ISPs

Page 37: BRS SEMINAR SERIES PRESENTS - data.daff.gov.audata.daff.gov.au/brs/data/warehouse/brsShop/data/finnigan22oct.pdf · BRS SEMINAR SERIES PRESENTS: Friday 22 October Earth System Science

Effect of network structure on the Modelling ofEcosystem State

()dxabxrfxdt()()() 1pppphxfx

ABM where each animal has the averagechance of contacting an infected animal

DE model based on averageprobabilities of contact Discrete contact with a random

fraction of the total herd

Effect of realisticclustering of contacts

x is the proportion of a herd infected by a disease

(Bonebeau, 2002)

Page 38: BRS SEMINAR SERIES PRESENTS - data.daff.gov.audata.daff.gov.au/brs/data/warehouse/brsShop/data/finnigan22oct.pdf · BRS SEMINAR SERIES PRESENTS: Friday 22 October Earth System Science

Evolving Network structure: Why do scale –freenetworks occur so often in natural and artificial

systems?

Scale-free networks are:

• A natural outcome of systems thatgrow by linking new nodespreferentially to the the bestconnected existing nodes(Barabasi and Albert, 1999)

• Optimal for conserving energywhen travelling between nodes

• Reflect stable dynamics in theunderlying system (Brede andFinnigan, 2004)

• Resistant to random attack onlinks or nodes (but vulnerable totargetted attack)

Page 39: BRS SEMINAR SERIES PRESENTS - data.daff.gov.audata.daff.gov.au/brs/data/warehouse/brsShop/data/finnigan22oct.pdf · BRS SEMINAR SERIES PRESENTS: Friday 22 October Earth System Science

Where are we now in applying these techniques topractical systems?

• We are developing a variety of agent-based models of systems asdiverse as arid rangelands (N Queensland), irrigated farming(Burdekin), dryland agriculture (WA wheatbelt), managed fisheries (NWShelf, GB Reef), the privatised electricity market (NEMCAS), Socialsystems (management of the Swan River, Melbourne drug scene).

• We are simultaneously analysing these systems as networks andbuilding dynamic models to analyse system stability, capturingproperties like rapid, irreversible changes of state in ecosystems withcompeting organisms (foliage types, herbivores, predators) or theformation of social networks through transmission of ideas orinformation

• Some results are already being applied but there are many openquestions such as optimum ways to capture human decision making inthese models.

Page 40: BRS SEMINAR SERIES PRESENTS - data.daff.gov.audata.daff.gov.au/brs/data/warehouse/brsShop/data/finnigan22oct.pdf · BRS SEMINAR SERIES PRESENTS: Friday 22 October Earth System Science

Summary and Conclusions

• Dynamical systems theory is limited in its ability to capture human-ecosystem interaction because even if we can reduce complexsystems to just a few, non-linearly interacting elements, ourmathematical tools are usually inadequate to analyse them.

• Agent-Based Modelling in contrast provides a set of tools to simulatethem in a convincing way

• Meta analysis using network theory shows that network topologyputs strong constraints on system behaviour independent of thedetails of the interactions (and vice-versa?)

Page 41: BRS SEMINAR SERIES PRESENTS - data.daff.gov.audata.daff.gov.au/brs/data/warehouse/brsShop/data/finnigan22oct.pdf · BRS SEMINAR SERIES PRESENTS: Friday 22 October Earth System Science

Summary and Conclusions

• Many naturally evolved or growing systems exhibit scale-freearchitecture

• We can now show that this architecture is a signature of a stable,resilient dynamical system

• We can postulate that systems are more robust to disturbance iftheir complexity is embodied in their network topology rather than inthe interactions between the agents: after all we see that natureseems to have adopted this strategy in many evolved systems

• This gives us some measures with which to test human ecosystemsfor resilience-if we can bridge the gap between simple abstractmodels and complex facsimiles of real systems

• More Information: http://www.dar.csiro.au/css/index.htm