- 1. USING INSTITUTIONAL ARRANGEMENTS AND HYBRID AUTOMATA FOR
REGIONAL SCALE AGENT-BASED MODELING OF LAND CHANGE Srgio Souza
Costa Orientadores: Dr. Gilberto Cmara Dra. Ana Paula Dutra de
Aguiar
2. Using institutional arrangements and hybrid automata for
regional scale agent-based modeling of land change 2Land Change
Introduction 3. Using institutional arrangements and hybrid
automata for regional scale agent-based modeling of land change
3Land Change: Amazonia Introduction 4. Using institutional
arrangements and hybrid automata for regional scale agent-based
modeling of land change 4Level of Analysis Micro (Social Science)
Macro (Natural Science) Introduction 5. Using institutional
arrangements and hybrid automata for regional scale agent-based
modeling of land change 5Level of Analysis Micro (Social Science)
Macro (Natural Science) Introduction 6. Using institutional
arrangements and hybrid automata for regional scale agent-based
modeling of land change 6Pattern-based models ! How changes in the
regional (or national) scale affect the smaller units of the model,
usually pixels or cells. Introduction Relation between land changes
and spatial variables 7. Using institutional arrangements and
hybrid automata for regional scale agent-based modeling of land
change 7Pattern-based models ! Introduction Relation between land
changes and spatial variables Analysing a deforestation hotspot 8.
Using institutional arrangements and hybrid automata for regional
scale agent-based modeling of land change 8Level of Analysis Micro
(Social Science) Macro (Natural Science) Introduction 9. Using
institutional arrangements and hybrid automata for regional scale
agent-based modeling of land change 9Agent-based models
Introduction 1. Agents, environmental and interactions. 2.
Emergence concepts (Local actions lead to global patterns) 3. No
central authority (autonomos) 4. Heterogeneity Macro Pattern
Interactions between agents and the environment. 10. Using
institutional arrangements and hybrid automata for regional scale
agent-based modeling of land change 10 Agent-Based Modeling of Land
change Introduction TheoreticalModels Empiricalmodels Agent-based
models (ABM) range from theoretical to empirical. Theoretical
models use simple generalizable ideas, whereas empirical models
require more complexity and case-specific data.. 11. Using
institutional arrangements and hybrid automata for regional scale
agent-based modeling of land change 11 Agent-Based Modeling of Land
change Introduction TheoreticalModels Empiricalmodels Agent-based
models (ABM) range from theoretical to empirical. Theoretical
models use simple generalizable ideas, whereas empirical models
require more complexity and case-specific data.. 12. Using
institutional arrangements and hybrid automata for regional scale
agent-based modeling of land change 12Problem definition
Introduction The modellers rarely have access to individual data to
represent and locate households in land change agent-based model.
13. Using institutional arrangements and hybrid automata for
regional scale agent-based modeling of land change 13Problem
definition Introduction The lack of data is an even greater problem
in large frontier area, like Brazilian Amazonia. The modellers
rarely have access to individual data to represent and locate
households in land change agent-based model. 14. Using
institutional arrangements and hybrid automata for regional scale
agent-based modeling of land change 14Problem definition
Introduction Large extension of municipalities The lack of data is
an even greater problem in large frontier area, like Brazilian
Amazonia. The modellers rarely have access to individual data to
represent and locate households in land change agent-based model.
15. Using institutional arrangements and hybrid automata for
regional scale agent-based modeling of land change 15Problem
definition Introduction Intense land changes and land market. Large
extension of municipalities The lack of data is an even greater
problem in large frontier area, like Brazilian Amazonia. The
modellers rarely have access to individual data to represent and
locate households in land change agent-based model. 16. Using
institutional arrangements and hybrid automata for regional scale
agent-based modeling of land change 16Intense changes 1985 1995
2006 2010 Area of farms (km) 2394 9601 17917 * Number of farms 1375
5893 7148 * Deforestation (km) 426 4070 16762 19271 Head of cattle
24494 230875 1912033 2290538 Population 14016 69117 85751 125030
Evolution of deforestation, population and agrarian structure in
the study area. Sources: (IBGE, 2006; INPE, 2010b) Problem
definition 17. Using institutional arrangements and hybrid automata
for regional scale agent-based modeling of land change 17Intense
changes Problem definition Evolution of deforestation patterns in
the study area. 18. Using institutional arrangements and hybrid
automata for regional scale agent-based modeling of land change
18Intense changes Introduction Concentrationoflandownership
inTucum. In 1996, farms with individual areas greater than 1000 ha
accounted for 8% of the total area of farms. In 2006, they
accounted for greater than 60% of the total. In the same period,
the number of farms decreased from 2518 to 1039 19. Using
institutional arrangements and hybrid automata for regional scale
agent-based modeling of land change 19Problem definition
Introduction Different actors and strategies. Intense land changes
and land market. Large extension of municipalities The lack of data
is an even greater problem in large frontier area, like Brazilian
Amazonia. The modellers rarely have access to individual data to
represent and locate households in land change agent-based model.
20. Using institutional arrangements and hybrid automata for
regional scale agent-based modeling of land change 20Different
actors and strategies Problem definition 21. Using institutional
arrangements and hybrid automata for regional scale agent-based
modeling of land change 21 Problem definition 61,000 ha 30,000 ha
50-200 ha 60,000 ha 30,000 ha 50 ha 20,000 ha 20,000 ha 200 ha
Different actors and strategies 22. Using institutional
arrangements and hybrid automata for regional scale agent-based
modeling of land change 22Problem definition Introduction The
modellers rarely have access to individual data to represent and
locate households in land change agent-based model. The lack of
data is an even greater problem in large frontier area, like
Brazilian Amazonia. The behaviour of actors changes in response to
internal and external conditions. Intense land changes and land
market. Large extension of municipalities Different actor and
strategies. 23. Using institutional arrangements and hybrid
automata for regional scale agent-based modeling of land change 23
Lagoa do Triunfo Farm: 19962 HA Problem definition Changing of
behaviour 24. Using institutional arrangements and hybrid automata
for regional scale agent-based modeling of land change 24 Problem
definition 0 200 400 600 800 1000 1200 1400 1600 1800 2000 APA
Triunfo do Xingu - 2007 Lagoa do Triunfo Farm: 19962 HA Evolution
of deforestation in the Lagoa do Triunfo Farm: ~653 ha/year.
Changing of behaviour 25. Using institutional arrangements and
hybrid automata for regional scale agent-based modeling of land
change 25Changing of behaviour Problem definition 0 20 40 60 80 100
120 140 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011
Prot. Integral 2005 0 50 100 150 200 250 300 350 400 2001 2002 2003
2004 2005 2006 2007 2008 2009 2010 2011 APA Triunfo do Xingu 2007
Evolution of deforestation in different regions: 26. Using
institutional arrangements and hybrid automata for regional scale
agent-based modeling of land change 26 1.#Occupa)on# 2.#Expansion#
3.#Restric)on# Ca7le#head#/#deforesta)on#(ha)# Evolution of Cattle
head / deforestation in the So Flix do Xingu. Changing of behaviour
Problem definition 27. Using institutional arrangements and hybrid
automata for regional scale agent-based modeling of land change
27Question Introduction How to develop agent-based models that are
expressive to account for the different strategies of land change
agents in frontier areas? 28. Using institutional arrangements and
hybrid automata for regional scale agent-based modeling of land
change 28Hypothesis Introduction Our hypothesis is that the idea of
strategies (supported by hybrid automata) provides a computational
model that is capable of expressing complex collective behavior of
land change agents. 29. Using institutional arrangements and hybrid
automata for regional scale agent-based modeling of land change
29Hypothesis Introduction Our hypothesis is that the idea of
strategies (supported by hybrid automata) provides a computational
model that is capable of expressing complex collective behavior of
land change agents. Hybrid automata represent strategies that are
loosely coupled to agents and may vary during the simulation in
response to local and institutional variabilitys 30. Using
institutional arrangements and hybrid automata for regional scale
agent-based modeling of land change 30Hypothesis Introduction Our
hypothesis is that the idea of strategies (supported by hybrid
automata) provides a computational model that is capable of
expressing complex collective behavior of land change agents.
Hybrid automata represent strategies that are loosely coupled to
agents and may vary during the simulation in response to local and
institutional variabilitys To validate them, we put up a model for
the area in Amazonia with the highest deforestation rate in the
1990s and 2000s. The model captures large-scale land change during
the 2000s and is used to build scenarios until 2020. 31. Using
institutional arrangements and hybrid automata for regional scale
agent-based modeling of land change 31Relevance for Computer
Science Introduction 3. Computational modeling of complex systems:
artificial, natural, socio-cultural, and human-nature interactions
Grand Challenges in Computer Science Research in Brazil - 2006 2016
(SBC) The complexity of this kind of research grows with the
increase in data volume and/or variables to be considered. Another
complicating factor is the frequent need for combining several
knowledge domains. 32. Using institutional arrangements and hybrid
automata for regional scale agent-based modeling of land change
32Contents 1. Introduction 2. Agent-based models at regional scale
3. Implementation 4. Model description 5. Model simulation and
results 6. Conclusion 33. AGENT-BASED MODEL AT REGIONAL SCALE 34.
Using institutional arrangements and hybrid automata for regional
scale agent-based modeling of land change 34Overview Agent-based
model at regional scale Spatial Units Arrangement #1 Arrangement #2
Agents Strategy #1 Strategy #2 Strategy #3 35. Using institutional
arrangements and hybrid automata for regional scale agent-based
modeling of land change 35Spatial Units Agent-based model at
regional scale Regular cell Farm Regular cells describe the
properties of space. Each cell has information on its land cover
and land use and its spatial properties, such as distance to roads.
Farms are irregular space partitions that belong to agents. A farm
is linked to one of more cells. 36. Using institutional
arrangements and hybrid automata for regional scale agent-based
modeling of land change 36Overview Agent-based model at regional
scale Spatial Units Arrangement #1 Arrangement #2 Agents Strategy
#1 Strategy #2 Strategy #3 37. Using institutional arrangements and
hybrid automata for regional scale agent-based modeling of land
change 37Institutional arrangements Agent-based model at regional
scale We refer to institutional arrangements as deals set up
between interest groups, social movements and state agencies to
respond to rules and norms that are relevant to them (DIETZ;
OSTROM, 2003). These pacts define how agents manage natural
resources (SCOTT; MEYER, 1994). The arrangements evolve on the
time, where for each time period different arrangements are valid
38. Using institutional arrangements and hybrid automata for
regional scale agent-based modeling of land change 38Institutional
arrangements The Institutional Arrangements are represented as
tuples of four components: 1. Name, which distinguishes an
arrangement from the other arrangements. 2. Context, inform a set
of context variables to agents. 3. Condition, the target, which
specifies the farmers that are influenced by the arrangement. 4.
Temporal, the time period for which the arrangement applies.
Agent-based model at regional scale 39. Using institutional
arrangements and hybrid automata for regional scale agent-based
modeling of land change 39Overview Agent-based model at regional
scale Spatial Units Arrangement #1 Arrangement #2 Agents Strategy
#1 Strategy #2 Strategy #3 40. Using institutional arrangements and
hybrid automata for regional scale agent-based modeling of land
change 40Agents Agent-based model at regional scale Agent Farm
Regular cell Agents are farmers that carry out land change. Agents
own farms and have attributes such as capital, technology level and
expansion aims. 41. Using institutional arrangements and hybrid
automata for regional scale agent-based modeling of land change
41Overview Agent-based model at regional scale Spatial Units
Arrangement #1 Arrangement #2 Agents Strategy #1 Strategy #2
Strategy #3 42. Using institutional arrangements and hybrid
automata for regional scale agent-based modeling of land change
42Strategies A set of consistent actions and agent decisions. Each
agent chooses one strategy at a time, which it can change later.
Examples of strategies include land speculation, intensive farming,
and subsistence agriculture. Agent-based model at regional scale
MIGRATION Search for land SUBSISTENCE Deforest 10%/year CATTLE
Extensive cattle raising Deforestation > 50% Land exhaustion
Farmer gets land 43. IMPLEMENTATION 44. Using institutional
arrangements and hybrid automata for regional scale agent-based
modeling of land change 44 Implementation Model source code model
syntax semantic checking model execution TerraMEINTERPRETER LUA
interpreter TerraMEframework TerraLib database TerraView MODEL DATA
TerraME (Carneiro, 2006) 45. Using institutional arrangements and
hybrid automata for regional scale agent-based modeling of land
change 45Spatial Units Implementation TerraMe Cell Farm TerraME
suports regular and irregular cells. We support the Farm data type,
which includes a set of operations for land market submodels, like
farm creation, farm split and farm merge 46. Using institutional
arrangements and hybrid automata for regional scale agent-based
modeling of land change 46Agents Implementation TerraMe Agent
Farmer Terrame support agents and its operations. We support the
Farmer data type as an extension of the TerraME agent; this type
includes common functions performed by and attributes of farmers.
47. Using institutional arrangements and hybrid automata for
regional scale agent-based modeling of land change 47Strategies
Implementation To implement our proposal, we need software that
expresses strategies. In computational terms, strategies make up a
discrete state machine. When an agent changes his strategy, he
moves from one state to another. Inside a state, an agent carries
out continuous actions. We propose the formalism of hybrid automata
to capture the idea of strategies. 48. Using institutional
arrangements and hybrid automata for regional scale agent-based
modeling of land change 48Strategies A hybrid automaton H has three
parts (HENZINGER, 1996): A finite set of variables X = {x1, x2, ...
xn} which is the automaton internal status. A finite directed graph
G = (V, E). The set of vertices V are states, and the set of edges
E are jumps. Each edge jump connects a source state to a target
state, following a condition. If this jump condition is true, the
automaton discrete state will change from the source state to the
target state. A set of flow rules assigned to each state. When a
flow rule is evaluated it changes the automaton internal status,
defined by the variables {x1, x2, ... xn}. Implementation 49. Using
institutional arrangements and hybrid automata for regional scale
agent-based modeling of land change 49Strategies A hybrid automaton
H has three parts (HENZINGER, 1996): A finite set of variables X =
{x1, x2, ... xn} which is the automaton internal status. A finite
directed graph G = (V, E). The set of vertices V are states, and
the set of edges E are jumps. Each edge jump connects a source
state to a target state, following a condition. If this jump
condition is true, the automaton discrete state will change from
the source state to the target state. A set of flow rules assigned
to each state. When a flow rule is evaluated it changes the
automaton internal status, defined by the variables {x1, x2, ...
xn}. Implementation TerraME supports hybrid automaton. 50. Using
institutional arrangements and hybrid automata for regional scale
agent-based modeling of land change 50Strategies State Flow rule
Jump Condition New state MIGRATION Search for land Farmer gets land
SUBSISTENCE SUBSISTENCE Farmer deforests 10% of land/year
Deforestation > 50% EXTENSIVE CATTLE EXTENSIVE CATTLE Extensive
cattle raising Land exhaustion MIGRATION Implementation Example: A
hybrid automaton model for poor migrant farmer 51. Using
institutional arrangements and hybrid automata for regional scale
agent-based modeling of land change 51Institutional arrangements We
support the institutional arrangement as a new Lua data type. This
code shows an example of an institutional arrangement.
Implementation arrang1 = Arrangement (! {market = HIGH, enforcement
= LOW }, ! function (agent) return agent.capital> 1000 end, !
function (event) return event:getTime() > 2008 end! )! 52. Using
institutional arrangements and hybrid automata for regional scale
agent-based modeling of land change 52Institutional arrangements We
support the institutional arrangement as a new Lua data type. This
code shows an example of an institutional arrangement.
Implementation arrang1 = Arrangement (! {market = HIGH, enforcement
= LOW }, ! function (agent) return agent.capital> 1000 end, !
function (event) return event:getTime() > 2008 end! )! Context
Condition Temporal Name 53. MODEL DESCRIPTION 54. Using
institutional arrangements and hybrid automata for regional scale
agent-based modeling of land change 54 Overview, Design concepts,
and Details The ODD (Overview, Design concepts, and Details)
protocol is used to provide a standardised description of our model
(GRIMM et al., 2006, 2010). 1. Purpose 2. Entities, Attributes and
Scales 3. Process overview and scheduling 4. Initialisation 5.
Input data 6. Submodels Model Description 55. Using institutional
arrangements and hybrid automata for regional scale agent-based
modeling of land change 55 Overview, Design concepts, and Details
The ODD (Overview, Design concepts, and Details) protocol is used
to provide a standardised description of our model (GRIMM et al.,
2006, 2010). 1. Purpose 2. Entities, Attributes and Scales 3.
Process overview and scheduling 4. Initialisation 5. Input data 6.
Submodels Model Description 56. Using institutional arrangements
and hybrid automata for regional scale agent-based modeling of land
change 56Purpose The model attempts to describe landscape dynamics
and replicate the deforestation fluctuations in the time period
simulated by considering changes in the institutional context and
their effects on households strategies. Model Description 57. Using
institutional arrangements and hybrid automata for regional scale
agent-based modeling of land change 57 Overview, Design concepts,
and Details The ODD (Overview, Design concepts, and Details)
protocol is used to provide a standardised description of our model
(GRIMM et al., 2006, 2010). 1. Purpose 2. Entities, Attributes and
Scales 3. Process overview and scheduling 4. Initialisation 5.
Input data 6. Submodels Model Description 58. Using institutional
arrangements and hybrid automata for regional scale agent-based
modeling of land change 58 Overview, Design concepts, and Details
The ODD (Overview, Design concepts, and Details) protocol is used
to provide a standardised description of our model (GRIMM et al.,
2006, 2010). 1. Purpose 2. Entities, Attributes and Scales 3.
Process overview and scheduling 4. Initialisation 5. Input data 6.
Submodels Model Description 59. Using institutional arrangements
and hybrid automata for regional scale agent-based modeling of land
change 59Scales Entities, attributes and scales Study&Area&
Main&rivers& Deforesta3on&4&2006&
Main&roads& The model simulates a rural region that covers
60,000 Km2 in the south- east of Par, Brazil. Spatial resolution
225x225 ha, 25x25 ha and 1x1 ha. Temporal extension: 1985 to 2020.
Temporal resolution: 1 year So Flix do Xingu Tucum 60. Using
institutional arrangements and hybrid automata for regional scale
agent-based modeling of land change 60 Overview, Design concepts,
and Details The ODD (Overview, Design concepts, and Details)
protocol is used to provide a standardised description of our model
(GRIMM et al., 2006, 2010). 1. Purpose 2. Entities, Attributes and
Scales 3. Process overview and scheduling 4. Initialisation 5.
Input data 6. Submodels Model Description 61. Using institutional
arrangements and hybrid automata for regional scale agent-based
modeling of land change 61Entities, attributes and scales Migrant
Extensive Farming Extensive Expanding Intensive Farming Speculate
Abandon Rural activity Land use and land market decisions SPATIAL
UNITS: Farms AGENTS: Farmers Context variables Rule parameters
SPATIAL UNITS: Regular cells Environmental, economic and
accessibility information and feedbacks INSTITUTIONALARRANGEMENTS:
RULES AND NORMES 62. Using institutional arrangements and hybrid
automata for regional scale agent-based modeling of land change
62Entities, attributes and scales Migrant Extensive Farming
Extensive Expanding Intensive Farming Speculate Abandon Rural
activity Land use and land market decisions SPATIAL UNITS: Farms
AGENTS: Farmers Context variables Rule parameters SPATIAL UNITS:
Regular cells Environmental, economic and accessibility information
and feedbacks INSTITUTIONALARRANGEMENTS: RULES AND NORMES 63. Using
institutional arrangements and hybrid automata for regional scale
agent-based modeling of land change 63Spatial Units: Farm A farm is
a spatial unit entity, and corresponds to the basic decision unit
referring to Land Market Decisions. The farm attributes include
area, price, quantity of animals, pasture area, area of degraded
pasture and area of remaining forest. Entities, attributes and
scales 64. Using institutional arrangements and hybrid automata for
regional scale agent-based modeling of land change 64Entities,
attributes and scales Migrant Extensive Farming Extensive Expanding
Intensive Farming Speculate Abandon Rural activity Land use and
land market decisions SPATIAL UNITS: Farms AGENTS: Farmers Context
variables Rule parameters SPATIAL UNITS: Regular cells
Environmental, economic and accessibility information and feedbacks
INSTITUTIONALARRANGEMENTS: RULES AND NORMES 65. Using institutional
arrangements and hybrid automata for regional scale agent-based
modeling of land change 65Spatial Units: Regular cells A
RegularCell represents the geographic space attributes, including
biophysical (e.g. slope, soil quality) and accessibility features
(e.g. distance to roads and distance to river). Area of each cell.
Land cover (Forest, Pasture, Secondary Forest, River, Other Number
of animals per cell Minimum Euclidean distance (normalized) to
roads, rivers and urban centers. Slope Entities, attributes and
scales 66. Using institutional arrangements and hybrid automata for
regional scale agent-based modeling of land change 66Entities,
attributes and scales Migrant Extensive Farming Extensive Expanding
Intensive Farming Speculate Abandon Rural activity Land use and
land market decisions SPATIAL UNITS: Farms AGENTS: Farmers Context
variables Rule parameters SPATIAL UNITS: Regular cells
Environmental, economic and accessibility information and feedbacks
INSTITUTIONALARRANGEMENTS: RULES AND NORMES 67. Using institutional
arrangements and hybrid automata for regional scale agent-based
modeling of land change 67Agents Based on fieldwork and literature
review, we defined a agent to be a rural entrepreneur with the
following dynamic attributes: Inclination to obey the law. Number
of farms. Average size of the farm he wants to buy. Technological
capability (Low, Medium, High). Investment capital. Entities,
attributes and scales 68. Using institutional arrangements and
hybrid automata for regional scale agent-based modeling of land
change 68Agents and Strategies Entities, attributes and scales
Migrant Extensive Farming Extensive Expanding Intensive Farming
Speculate Abandon Rural activity A farmer performs land use and
land market decisions on these spatial units. These decisions are
guided by different strategies that an agent can take over time. A
farmer chooses between five strategies: 69. Using institutional
arrangements and hybrid automata for regional scale agent-based
modeling of land change 69Agents and Strategies Entities,
attributes and scales Migrant Extensive Farming Extensive Expanding
Intensive Farming Speculate Abandon Rural activity A farmer
performs land use and land market decisions on these spatial units.
These decisions are guided by different strategies that an agent
can take over time. A farmer chooses between five strategies: 70.
Using institutional arrangements and hybrid automata for regional
scale agent-based modeling of land change 70Agents and Strategies
Entities, attributes and scales Migrant Extensive Farming Extensive
Expanding Intensive Farming Speculate Abandon Rural activity A
farmer performs land use and land market decisions on these spatial
units. These decisions are guided by different strategies that an
agent can take over time. A farmer chooses between five strategies:
71. Using institutional arrangements and hybrid automata for
regional scale agent-based modeling of land change 71Agents and
Strategies Entities, attributes and scales Migrant Extensive
Farming Extensive Expanding Intensive Farming Speculate Abandon
Rural activity A farmer performs land use and land market decisions
on these spatial units. These decisions are guided by different
strategies that an agent can take over time. A farmer chooses
between five strategies: 72. Using institutional arrangements and
hybrid automata for regional scale agent-based modeling of land
change 72Agents and Strategies Entities, attributes and scales
Migrant Extensive Farming Extensive Expanding Intensive Farming
Speculate Abandon Rural activity A farmer performs land use and
land market decisions on these spatial units. These decisions are
guided by different strategies that an agent can take over time. A
farmer chooses between five strategies: 73. Using institutional
arrangements and hybrid automata for regional scale agent-based
modeling of land change 73Agents and Strategies Entities,
attributes and scales Migrant Extensive Farming Extensive Expanding
Intensive Farming Speculate Abandon Rural activity A farmer
performs land use and land market decisions on these spatial units.
These decisions are guided by different strategies that an agent
can take over time. A farmer chooses between five strategies: 74.
Using institutional arrangements and hybrid automata for regional
scale agent-based modeling of land change 74Agents and Strategies
Entities, attributes and scales Migrant Extensive Farming Extensive
Expanding Intensive Farming Speculate Abandon Rural activity A
farmer performs land use and land market decisions on these spatial
units. These decisions are guided by different strategies that an
agent can take over time. A farmer chooses between five strategies:
75. Using institutional arrangements and hybrid automata for
regional scale agent-based modeling of land change 75Agents and
Strategies Entities, attributes and scales Migrant Extensive
Farming Extensive Expanding Intensive Farming Speculate Abandon
Rural activity A farmer performs land use and land market decisions
on these spatial units. These decisions are guided by different
strategies that an agent can take over time. A farmer chooses
between five strategies: 76. Using institutional arrangements and
hybrid automata for regional scale agent-based modeling of land
change 76Entities, attributes and scales Migrant Extensive Farming
Extensive Expanding Intensive Farming Speculate Abandon Rural
activity Land use and land market decisions SPATIAL UNITS: Farms
AGENTS: Farmers Context variables Rule parameters SPATIAL UNITS:
Regular cells Environmental, economic and accessibility information
and feedbacks INSTITUTIONALARRANGEMENTS: RULES AND NORMES 77. Using
institutional arrangements and hybrid automata for regional scale
agent-based modeling of land change 77Context variables Context
variable Description
ForestcodeenforcementhowistheForestCodebeingfollowed?
Lawenforcement istherecontroloverpoachingofpubliclands?
Marketforca=le howstrongisthebeefmarketchain? Creditforsmallfarmers
howeasyandcheapisthecreditforsmallfarmers? Creditforlargefarmers
howeasyandcheapisthecreditforlargefarmers? Creditforreforesta?on
howeasyandcheapisthecreditforreforesta?on? Entities, attributes and
scales The agent uses six guidelines to decide on his strategy,
which are: 78. Using institutional arrangements and hybrid automata
for regional scale agent-based modeling of land change 78Entities,
attributes and scales Migrant Extensive Farming Extensive Expanding
Intensive Farming Speculate Abandon Rural activity Land use and
land market decisions SPATIAL UNITS: Farms AGENTS: Farmers Context
variables Rule parameters SPATIAL UNITS: Regular cells
Environmental, economic and accessibility information and feedbacks
INSTITUTIONALARRANGEMENTS: RULES AND NORMES 79. Using institutional
arrangements and hybrid automata for regional scale agent-based
modeling of land change 79Institutional Arrangements We defined six
institutional arrangements in our model: 1. Government-induced
occupation: prevalent from 1970s to mid 1990s, with the government
encouraging people to occupy Amazonia. 2. Beef market chain
organization: From the mid 1990s until today, following initial
occupation with easy access to land, the beef market chain grew. 3.
Deforestation control: From 2005 onwards, law enforcement increased
and the government created new protected areas Entities, attributes
and scales 80. Using institutional arrangements and hybrid automata
for regional scale agent-based modeling of land change
80Institutional Arrangements We defined six institutional
arrangements in our model: 4. Green market: In the late 2000s,
given pressure from consumers, NGOs and public attorneys, part of
the private sector changed. Some farmers and part of the industry
agreed to comply with sustainable practices, in exchange for market
and credit access. 5. Sustainable Development: a possible future
arrangement to bring about equilibrium between social,
environmental and economic goals. This choice combines strong law
enforcement with green market practices. 6. Economic development: a
possible future arrangement based on a return to 1970s model, where
economic growth prevails over environmental or social concerns.
Entities, attributes and scales 81. Using institutional
arrangements and hybrid automata for regional scale agent-based
modeling of land change 81Arrangements and Variables INSTITUTIONAL
ARRANGEMENTS FOREST CODE ENFORCE LAND POACHING ENFORCE CATTLE
MARKET CREDIT LARGE FARMS CREDIT LARGE FARMS CREDIT REFOREST
Government- induced occupation 2 Beef market chain organization 2 2
1 Deforestation control 1 2 Green market 3 3 3 3 2 Sustainable
development 3 3 1 1 3 3 Economic development 2 2 1 Relevance: 3 =
strong, 2 = medium, 1 = low, no value = no relevance Entities,
attributes and scales 82. Using institutional arrangements and
hybrid automata for regional scale agent-based modeling of land
change 82 Overview, Design concepts, and Details The ODD (Overview,
Design concepts, and Details) protocol is used to provide a
standardised description of our model (GRIMM et al., 2006, 2010).
1. Purpose 2. Entities, Attributes and Scales 3. Process overview
and scheduling 4. Initialisation 5. Input data 6. Submodels Model
Description 83. Using institutional arrangements and hybrid
automata for regional scale agent-based modeling of land change 83
Process Oververview and Schedulling Initialization For each year T
Allocate agents in the landscape Compute attributes at regular cell
level Update agent quantities at time step t (immigration) Update
institutional arrangements acting at time step t Each farmer
execute the land use and land market decisions Compute and
aggregate attributes at farm level 84. Using institutional
arrangements and hybrid automata for regional scale agent-based
modeling of land change 84 Overview, Design concepts, and Details
The ODD (Overview, Design concepts, and Details) protocol is used
to provide a standardised description of our model (GRIMM et al.,
2006, 2010). 1. Purpose 2. Entities, Attributes and Scales 3.
Process overview and scheduling 4. Initialisation 5. Input data 6.
Submodels Model Description 85. Using institutional arrangements
and hybrid automata for regional scale agent-based modeling of land
change 85Initialization In theoretical models, the initialization
may be simple random placement (EPSTEIN; AXTELL, 1996). Empirical
models require much data, such as census and cadastral data. Model
description 86. Using institutional arrangements and hybrid
automata for regional scale agent-based modeling of land change
86Initialization The initialization should answer the following
questions at least: How many farmers are in the study area? What is
the size of each farm? Where are these farms located? Model
description 87. Using institutional arrangements and hybrid
automata for regional scale agent-based modeling of land change
87Initialization The initialization should answer the following
questions at least: How many farmers are in the study area? What is
the size of each farm? Where are these farms located? Model
description 88. Using institutional arrangements and hybrid
automata for regional scale agent-based modeling of land change
88Initialization Area range (HA)QuantitySum of area by range <
5068527625 50-10057444980 100-200628877 200-500309618 500-100096467
> 100014141863 Total1374239430 Model description Number of farms
and total area aggregated for various area ranges (IBGE, 1985). 89.
Using institutional arrangements and hybrid automata for regional
scale agent-based modeling of land change 89Initialization The
initialization should answer the following questions at least: How
many farmers are in the study area? What is the size of each farm?
Where are these farms located? Model description 90. Using
institutional arrangements and hybrid automata for regional scale
agent-based modeling of land change 90Initialization Model
description The deforestation map is derived from classification of
LANDSAT TM images, which use the following classes: forest,
deforestation, not-forest, cloud and river. In 1985, the
deforestation class accounted for an area of 426 km2. Deforested
Area/ Area of farm 1985: 0.17 2006: 0.93 91. Using institutional
arrangements and hybrid automata for regional scale agent-based
modeling of land change 91Initialization Model description 120,000
km2 So Flix do Xing - 1985 92. Using institutional arrangements and
hybrid automata for regional scale agent-based modeling of land
change 92Initialization Model description
SoFlixdoXing-Contexto-1985 Indigenous land 93. Using institutional
arrangements and hybrid automata for regional scale agent-based
modeling of land change 93Initialization Model description
SoFlixdoXing-Contexto-1985 Indigenous land Settlement (1981) 94.
Using institutional arrangements and hybrid automata for regional
scale agent-based modeling of land change 94Initialization Model
description SoFlixdoXing-Contexto-1985 Surge um novo Brasil no Sul
do Par ... Localizado perto de polos importantes como Carajas,
Tucurui, servido por vias de transporte e privilegiado pelas
condies naturais. Panfleto de divulgao do Projeto Tucum, 1981.
Arquivo pessoal de Rosngela Sampaio 95. Using institutional
arrangements and hybrid automata for regional scale agent-based
modeling of land change 95Initialization Model description
SoFlixdoXing-Contexto-1985 96. Using institutional arrangements and
hybrid automata for regional scale agent-based modeling of land
change 96Initialization Model description 97. Using institutional
arrangements and hybrid automata for regional scale agent-based
modeling of land change 97Initialization Model description 0 2000
4000 6000 8000 10000 1985 1988 1991 1994 1997 2000 2003 2006 2009
2012 (a) Deforested areas in 1985 (b) Number of farmers from 1985
to 2010 (b) Distribution of farms Deforestation - 1985 98. Using
institutional arrangements and hybrid automata for regional scale
agent-based modeling of land change 98Initialization Model
description 0 2000 4000 6000 8000 10000 1985 1988 1991 1994 1997
2000 2003 2006 2009 2012 (a) Deforested areas in 1985 (c) Number of
farmers from 1985 to 2010 (b) Distribution of farms Deforestation -
1985 99. Using institutional arrangements and hybrid automata for
regional scale agent-based modeling of land change 99Input data
1985 < 50 50-500 500-2000 > 2000 total qt 685 666 12 11 1374
area 27625 63475 10844 137486 239430 1995 < 50 50-500 500-2000
> 2000 total qt 871 1811 93 41 2816 area 42650 274944 103678
320823 742095 2005 < 50 50-500 500-2000 > 2000 total qt 3895
1756 353 105 6109 area 85006 270872 363661 738062 1457601 Model
description 100. Using institutional arrangements and hybrid
automata for regional scale agent-based modeling of land change
100Input data 1985 < 50 50-500 500-2000 > 2000 total qt 685
666 12 11 1374 area 27625 63475 10844 137486 239430 1995 < 50
50-500 500-2000 > 2000 total qt 871 1811 93 41 2816 area 42650
274944 103678 320823 742095 2005 < 50 50-500 500-2000 > 2000
total qt 3895 1756 353 105 6109 area 85006 270872 363661 738062
1457601 Model description 101. Using institutional arrangements and
hybrid automata for regional scale agent-based modeling of land
change 101 Overview, Design concepts, and Details The ODD
(Overview, Design concepts, and Details) protocol is used to
provide a standardised description of our model (GRIMM et al.,
2006, 2010). 1. Purpose 2. Entities, Attributes and Scales 3.
Process overview and scheduling 4. Design ideas 5. Initialisation
6. Input data 7. Submodels Model Description 102. Using
institutional arrangements and hybrid automata for regional scale
agent-based modeling of land change 102Submodels: Pasture
degradation Model description For cattle famers, the productivity
of a farm is characterised by the number of animals per ha. This
variable is complex and depends on many different factors. We
simplified this variable by making it a function of only technology
level and pasture age: Animal unit = degradation (technology level,
pasture age) Where: technology level = {low, medium, high} 103.
Using institutional arrangements and hybrid automata for regional
scale agent-based modeling of land change 103Submodels: Pasture
degradation Model description The degradation is a discrete
function that has a single value for each technology level and
pasture age. This function may be calibrated for based on the
region. In our experiments, we use the discrete function
represented in Figure 59. 0 0,2 0,4 0,6 0,8 1 1,2 1,4 1,6 1 2 3 4 5
6 7 8 9 10 11 12 high medium low 104. Using institutional
arrangements and hybrid automata for regional scale agent-based
modeling of land change 104Submodels: Farm locator/creator Model
description This submodel creates farms, both at initialisation and
during the simulation. For each new farm in the model, it 1. first
produces a sample of 500 cells. 2. it calculates the potential for
each cell (x, y) using the multi-criteria evaluation (MCE). 3.
chooses the cell with the greatest potential. 4. The chosen cell is
marked as the location of the agents house. 5. The Farm Creator
operator takes the chosen cell and a specified farm area as
parameters. 6. Finally, the submodel calculates and updates the
spatial relations amongst farms using the Neighbourhood operator.
MCE 105. Using institutional arrangements and hybrid automata for
regional scale agent-based modeling of land change 105Submodels:
Farm locator/creator Model description This submodel creates farms,
both at initialisation and during the simulation. For each new farm
in the model, it 1. first produces a sample of 500 cells. 2. it
calculates the potential for each cell (x, y) using the
multi-criteria evaluation (MCE). 3. chooses the cell with the
greatest potential. 4. The chosen cell is marked as the location of
the agents house. 5. The Farm Creator operator takes the chosen
cell and a specified farm area as parameters. 6. Finally, the
submodel calculates and updates the spatial relations amongst farms
using the Neighbourhood operator. MCE 106. Using institutional
arrangements and hybrid automata for regional scale agent-based
modeling of land change 106Submodels: Farm locator/creator Model
description 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0
0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 1 1 0 0 0 0 0 1 1 0 0 0 0 1 1 0 0 0
0 0 0 0 0 0 0 0 0 0 Create (1) Create operator, which instantiates
a new farm given an initial settlement cell and a farm area. 0 0 2
2 0 0 0 0 2 2 0 0 3 3 1 1 0 0 3 0 1 0 0 0 4 0 0 0 5 5 4 4 0 0 5 5 0
0 2 2 0 0 0 0 2 2 0 0 3 3 1 1 0 0 3 0 1 0 0 0 4 0 0 0 5 5 4 4 0 0 5
5 Neigh (1) Neighbourhood operator, which identifies the spatial
adjacency relations amongst the farms. Farm operators used for farm
locator/creator submodel: 107. Using institutional arrangements and
hybrid automata for regional scale agent-based modeling of land
change 107Submodels: Farm locator/creator Model description 1.
Receives an initial cell and farm area. The choice of initial cell
may depend on factors such as the proximity of rivers or roads. 2.
Adds the available neighbouring cells until a given area is
attained. 108. Using institutional arrangements and hybrid automata
for regional scale agent-based modeling of land change
108Submodels: Land-market Model description Puts on sale 109. Using
institutional arrangements and hybrid automata for regional scale
agent-based modeling of land change 109Submodels: Land-market Model
description Puts on sale Farms on sale given a set of restriction
(price and area)? 110. Using institutional arrangements and hybrid
automata for regional scale agent-based modeling of land change
110Submodels: Land-market Model description Puts on sale Farms on
sale given a set of restriction (price and area)? Rank-order-select
Farm buy 111. Using institutional arrangements and hybrid automata
for regional scale agent-based modeling of land change
111Submodels: Land-market Model description 0 0 0 0 0 0 0 2 1 1 0 0
0 2 2 1 1 0 0 2 2 1 1 0 0 0 0 0 0 0 0 0 0 0 0 0 Merge (1,2) 0 0 0 0
0 0 0 1 1 1 0 0 0 1 1 1 1 0 0 1 1 1 1 0 0 0 0 0 0 0 0 0 0 0 0 0
Merge operator, which joins two farms and updates the farm
boundaries. Farm operators used for land-market submodel . 0 0 0 0
0 0 0 2 3 3 0 0 0 2 2 3 3 0 0 2 2 3 3 0 0 0 0 0 0 0 0 0 0 0 0 0
Split (1) 0 0 0 0 0 0 0 1 1 1 0 0 0 1 1 1 1 0 0 1 1 1 1 0 0 0 0 0 0
0 0 0 0 0 0 0 Split operator, which divides a given region into
several farms of specified areas. 112. Using institutional
arrangements and hybrid automata for regional scale agent-based
modeling of land change 112Submodels: Land use Model description In
our experiments, land-use decisions refer to deforestation and
pasture creation, which depend on the technology level. Land-use
decisions also include pasture reformation and reforestation. These
decisions may yield to different trajectories (A, B and C), These
trajectories reflect the different strategies that farmers can
employ, each strategy constrains how much, where and when the
agents deforest, create pasture and manage the land. 113. Using
institutional arrangements and hybrid automata for regional scale
agent-based modeling of land change 113Submodels: Land use Model
description We propose using a land-allocation submodel (similar to
pattern models) to define what cells are changed. In this model,
each agent performs a land allocation, using a method such as rank-
order-change or a clues-like algorithm, under the constraint that a
farm can have at most 60,000 cells. In our simulation, we use a
simple allocation strategy in which we rank the cells using the MCE
to calculate the potential of each cell. The factors considered
include the slope and distance to the house. Then, we order the
cells and finally, change the land use. 114. MODEL SIMULATIONS AND
RESULTS 115. Using institutional arrangements and hybrid automata
for regional scale agent-based modeling of land change
115Simulations Simulation Period Institutional Arrangements
Temporal range of arrangement 0 (Calibration) 1985-1997
Government-induced occupation 1970-1996 1 (Validation) 1985-2010
Government-induced occupation 1970-1996 2 (Validation) 1985-2010
Government-induced occupation Beef market chain organization
1970-1996 1997-2010 3 (Validation) 1985-2010 Government-induced
occupation Beef market chain organization Deforestation control
1970-1996 1997-2010 2005-2010 Model simulation and results 116.
Using institutional arrangements and hybrid automata for regional
scale agent-based modeling of land change 116Simulations Simulation
Period Institutional Arrangements Temporal range of arrangement 0
(Calibration) 1985-1997 Government-induced occupation 1970-1996 1
(Validation) 1985-2010 Government-induced occupation 1970-1996 2
(Validation) 1985-2010 Government-induced occupation Beef market
chain organization 1970-1996 1997-2010 3 (Validation) 1985-2010
Government-induced occupation Beef market chain organization
Deforestation control 1970-1996 1997-2010 2005-2010 Model
simulation and results 117. Using institutional arrangements and
hybrid automata for regional scale agent-based modeling of land
change 117Results Model simulation and results 0 100000 200000
300000 400000 500000 600000 700000 1985 1986 1987 1988 1989 1990
1991 1992 1993 1994 1995 1996 1997 Observed Simulated Deforestation
rate 0 10000 20000 30000 40000 50000 60000 70000 80000 1986 1987
1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 Observed
Simulated Total deforested area 118. Using institutional
arrangements and hybrid automata for regional scale agent-based
modeling of land change 118Simulations Simulation Period
Institutional Arrangements Temporal range of arrangement 0
(Calibration) 1985-1997 Government-induced occupation 1970-1996 1
(Validation) 1985-2010 Government-induced occupation 1970-1996 2
(Validation) 1985-2010 Government-induced occupation Beef market
chain organization 1970-1996 1997-2010 3 (Validation) 1985-2010
Government-induced occupation Beef market chain organization
Deforestation control 1970-1996 1997-2010 2005-2010 Model
simulation and results 119. Using institutional arrangements and
hybrid automata for regional scale agent-based modeling of land
change 119Results S1, S2 and S3 Model simulation and results 0
20000 40000 60000 80000 100000 120000 140000 160000 180000 200000
1986 1988 1990 1992 1994 1996 1998 2000 2002 2004 2006 2008
Observed S1 S2 S3 0 500000 1000000 1500000 2000000 2500000 1985
1987 1989 1991 1993 1995 1997 1999 2001 2003 2005 2007 2009
Observed S1 S2 S3 Deforestation rateTotal deforested area 120.
Using institutional arrangements and hybrid automata for regional
scale agent-based modeling of land change 120Results S1, S2 and S3
Model simulation and results 0 20000 40000 60000 80000 100000
120000 140000 160000 180000 200000 1986 1988 1990 1992 1994 1996
1998 2000 2002 2004 2006 2008 Observed S1 S2 S3 0 500000 1000000
1500000 2000000 2500000 1985 1987 1989 1991 1993 1995 1997 1999
2001 2003 2005 2007 2009 Observed S1 S2 S3 Deforestation rateTotal
deforested area 121. Using institutional arrangements and hybrid
automata for regional scale agent-based modeling of land change
121Results S1, S2 and S3 Model simulation and results 1200000
1400000 1600000 1800000 2000000 2200000 2400000 2005 2006 2007 2008
2009 2010 Observed S1 S2 S3 122. Using institutional arrangements
and hybrid automata for regional scale agent-based modeling of land
change 122Results S3 Model simulation and results 123. Using
institutional arrangements and hybrid automata for regional scale
agent-based modeling of land change 123Results S3 Model simulation
and results 124. Using institutional arrangements and hybrid
automata for regional scale agent-based modeling of land change
124Scenarios Simulation Period Alternative Institutional
Arrangements Temporal range of arrangement 4 (Scenario A) 1985-2020
Government-induced occupation Beef market chain organization
Deforestation control Green market 1970-1996 1997-2008 2005-2020
2009-2020 5 (Scenario B) 1985-2020 Government-induced occupation
Beef market chain organization Deforestation control Sustainable
development Green market 1970-1996 1997-2008 2005-2012 2013-2020
2009-2020 6 (Scenario C) 1985-2020 Government-induced occupation
Beef market chain organization Deforestation control Economic
development Green market 1970-1996 1997-2008 2005-2012 2013-2020
2009-2020 Model simulation and results 125. Using institutional
arrangements and hybrid automata for regional scale agent-based
modeling of land change 125Scenarios A, B and C Model simulation
and results Pasture area (ha) Forest area inside farms (%) 0 0,1
0,2 0,3 0,4 0,5 2010 2011 2012 2013 2014 2015 2016 2017 2018 2019
2020 A B C 1480700 1580700 1680700 1780700 1880700 1980700 2080700
2180700 2280700 2380700 2009 2010 2011 2012 2013 2014 2015 2016
2017 2018 2019 2020 A B C 126. Using institutional arrangements and
hybrid automata for regional scale agent-based modeling of land
change 126Scenarios A, B and C Model simulation and results 0 1000
2000 3000 4000 5000 6000 A B C 0 50 100 150 200 250 300 350 400
201020122014201620182020 A B C 0 200 400 600 800 1000 A B C 0 500
1000 1500 2000 2500 3000 A B C (a) Extensive Capitalizing (c)
Intensive (d) Abandoning Rural Areas (b) Extensive Expansive 127.
CONCLUSION 128. Using institutional arrangements and hybrid
automata for regional scale agent-based modeling of land change 128
Using empirical ABM for socio-ecological system is a great
challenge: Agent-based modelling meets an intuitive desire to
explicitly represent human decision making. (...) The question is
whether the benefits of that approach to spatial modelling exceed
the considerable costs of the added dimensions of complexity
introduced into the modelling effort. Helen Couclelis, 2002. We
believe that the ideas of institutional arrangements and hybrid
automata can help reduce the complexity of agent-based modelling of
land change. Conclusion 129. Using institutional arrangements and
hybrid automata for regional scale agent-based modeling of land
change 129Conclusion The thesis shows how to express the ideas of
strategy and institutional arrangements in computer simulations.
Strategies are loosely coupled to agents and may vary during the
simulation in response to changes in local context and
institutional arrangements. The hybrid automata formalism
(available in TerraME) turned out to be a good way to express
decision-making when agents change their strategies over time.
Conclusion 130. Using institutional arrangements and hybrid
automata for regional scale agent-based modeling of land change
130Conclusion This thesis presents a method for building
agent-based models of land change. Most agent-based models in the
literature deal with individual decisions in small areas, based on
field surveys. Our approach extends agent-based models to larger
areas, based on collective behaviour. We assume that farmers have a
limited set of strategies to manage land. Over time, a farmer may
change his strategy depending on external conditions. Conclusion
131. Using institutional arrangements and hybrid automata for
regional scale agent-based modeling of land change 131Future work
This model was very important to validate our concepts. however,
this work is an is a first step of a long journey where we may
suggest the following future work Calibration of pasture
degradation submodel. Land use submodel should consider other land
use types, allowing build different scenarious. Include actors,
like miners and loggers. Road submodel, creation of roads should be
modeled as endogenous process in future version. Connection to
market. Normative agents Conclusion 132. Using institutional
arrangements and hybrid automata for regional scale agent-based
modeling of land change 132 Obrigado 133. Using institutional
arrangements and hybrid automata for regional scale agent-based
modeling of land change 133 StrategiesandJump Conditions State Jump
Conditions Migrate 1. When agent gets a farm and has little
capital, jump to Extensive Farming. 2. When agent gets a farm and
has much capital, jump to Extensive Expanding. 3. If agent is risk
prone, jump to Speculate. Extensive Farming 1. If agent gets
capital, jump to Extensive Expanding. 2. If law enforcement is
high, pasture is degraded and there is credit for small farmers,
jump to Intensive Farming. 3. If law enforcement is high, pasture
is degraded and there is no credit for small farmers, jump to
Abandon Rural Activity. Extensive Expanding 1. If law enforcement
is high, pasture is degraded and there is credit for large farmers,
jump to Intensive Farming. 2. If law enforcement is high, pasture
is degraded and there is no credit for large farmers, jump to
Abandon Rural Activity. Speculate If law enforcement is high, jump
to Abandon Rural Activity. Intensive Farming If there is no credit
for large farmers and the beef market chain is weak for more than 5
years, jump to Abandon Rural Activity. Abandoning Rural Activity
(final state, no jump conditions)
