The document discusses a simulation model called SIMARIS that couples a geographic information system (GIS) and an agent-based model (ABM) to simulate marine activities. It aims to represent multiple activities simultaneously, evaluate their impacts on marine protected areas, and identify potential conflict zones. The model integrates the GIS GRASS with the ABM platform GAMA using Python. Preliminary results from a test simulation in a 40km2 area of France show maps of boat traffic patterns over time and the consumption and regeneration of fishing resources like king scallops.

1. Simulation of marine activities by coupling
Geographical Information System and Agent Based
Model: improvements and technical achievements
Annalisa Minelli, Cyril Tissot, Mathias Rouan, Matthieu Le Tixerant
12 October 2016, Perugia (IT)

2. Outline:
1. MAS & GIS integration
2. The SIMARIS project: aims and description
3. Methods: GAMA, GRASS and the Python
4. Technical configuration of the integration
5. Some results and conclusions
OGRS 2016, 12 October, Perugia (IT)

3. ● Simulation occurring in georeferenced environment (for MAS simulation)
● Time variable integration (for GIS modeling)
● Representation of “intelligent” spatio-temporal operators (for GIS
modeling)
MAS & GIS integration: Why?
OGRS 2016, 12 October, Perugia (IT)

4. Tight coupling (*)
● Integration of geographical operations in MAS, but..
○ Limited quantity
○ Quite limited performance
● Integration of intelligent agents in GIS, but..
○ Time variable issue
○ Few experiments
○ Lack of a dedicated AB infrastructure
MAS & GIS integration: How?
OGRS 2016, 12 October, Perugia (IT)(*) from Karadimas et al., 2006

5. Loose coupling(*)
● Only data coming from GIS models or ABM simulation are integrated into
the other system
Direct cooperative coupling(*)
● Client-server architecture
MAS & GIS integration: How?
OGRS 2016, 12 October, Perugia (IT)(*) from Karadimas et al., 2006

6. Indirect cooperative coupling(*)
● A third software structure recalls single functionalities of both simulation
and geographical modeling
MAS & GIS integration: How?
OGRS 2016, 12 October, Perugia (IT)(*) from Karadimas et al., 2006

7. Indirect cooperative coupling(*)
● A third software structure recalls single functionalities of both simulation
and geographical modeling
MAS & GIS integration: How?
OGRS 2016, 12 October, Perugia (IT)(*) from Karadimas et al., 2006
SIMARIS:
Simulation du déroulement d'activités marines

8. Simulation of human-environment interaction in the near sea.
It is..
● Geographically based
● Multi-scale and multi-level
● Marine activities
● Highly automated
● Ant colony algorithm implemented
The SIMARIS model
OGRS 2016, 12 October, Perugia (IT)

9. Simulation of human-environment interaction in the near sea.
It is..
● Geographically based
● Multi-scale and multi-level
● Marine activities
● Highly automated
● Ant colony algorithm implemented
The SIMARIS model
OGRS 2016, 12 October, Perugia (IT)
It aims to..
● Represent simultaneously several
activities
● Evaluate impacts on marine
protected areas
● Individuate possible conflict zones
between activities

10. OGRS 2016, 12 October, Perugia (IT)
The SIMARIS model

11. OGRS 2016, 12 October, Perugia (IT)
The SIMARIS model
GAMA platform

12. OGRS 2016, 12 October, Perugia (IT)
The GAMA platform
● GIS Agent-Based Modeling Architecture
● Spatially explicit and Open Source
● Models in .gaml, an high level and intuitive language
● Headless mode supported
● Multi-level and multi-scale
● Integration libraries with R, PostGres, SQLite etc.
● Many data type supported (es. OSM)
● Developed by the IRD/UPMC international research unit UMMISCO

13. OGRS 2016, 12 October, Perugia (IT)
GRASS GIS
● Geographic Resources Analysis Support System
● 34 years of life
● Many development centres all around the world
● Many programming languages
● Over 350 modules for different purposes
● Managing, analysis, visualization, storage and creation of spatial data
● Big data support and topological 2D/3D engine
● Full temporal framework

14. ..all wrapped in Python

15. 1. Preprocessing of SIMARIS inputs by GRASS (no GUI):
reshaping of input layers in relation to the chosen spatial extension
2. Inputs are integrated to gama-headless within the code of the model:
SIMARIS set the right spatio-temporal resolution, calibrate and runs the
analysis
3. Outputs are produced both during the simulation and at the end of the
processing
..all wrapped in Python
OGRS 2016, 12 October, Perugia (IT)

16. Set up the integration
OGRS 2016, 12 October, Perugia (IT)
SIMARIS takes as input:
● Bathymetry of the zone - required
● Fishing calendars - required
● Departing and unload ports (for each fishing activity) - required
● Fishing zones - optional
● Estimated fishing potential map - optional
● Tide levels - required

17. Set up the integration
OGRS 2016, 12 October, Perugia (IT)
SIMARIS gives in ouput:
● Map of attendance (in terms of boat traffic)
● Map of attractiveness (using the ant colony algorithm)
● Charts of resource consumption
● Fishing balance in time
● Snapshots of maps at different timesteps

18. Set up the integration
OGRS 2016, 12 October, Perugia (IT)
Geographical bbox of the zone

19. Set up the integration
OGRS 2016, 12 October, Perugia (IT)
Inputs of SIMARIS model

20. Set up the integration
OGRS 2016, 12 October, Perugia (IT)
Path to GRASS database

21. Set up the integration
OGRS 2016, 12 October, Perugia (IT)
GAMA headless experiment
configuration files

22. Set up the integration
OGRS 2016, 12 October, Perugia (IT)
Geographical bbox of the zone
Inputs of SIMARIS model
Path to GRASS database
GAMA headless experiment
configuration files
Model
output

23. Simulation description
OGRS 2016, 12 October, Perugia (IT)
Elaboration details:
● ~40 km2
in the Brest bay (Brittany region, France)
Spatial resolution = 20 mt
● 9 boats for 3 types of fish (King Scallop, algae and clams)
● Simulation over four months (fishing season - sept/dec)
Temporal resolution = 30’
● Shapefile of fishing zones given

24. Simulation description
OGRS 2016, 12 October, Perugia (IT)
Studied zone:
~40 km2
● Iroise Sea, Brest Bay (France)
● Marine Protected Area
● Many human activities in the near sea (naval military
base, commerce, stakeholdes, plaisance etc.)

25. Simulation results
OGRS 2016, 12 October, Perugia (IT)
From snapshots:
Animation showing boats
operating in the fishing zones
and tide evolution in time
(timestep = 30’)

26. Simulation results
OGRS 2016, 12 October, Perugia (IT)
From shapefiles:
Animation of most
frequented zones in time
(capture rate = 1 week)

27. Simulation results
OGRS 2016, 12 October, Perugia (IT)
Resource consumption and regeneration in time (king scallop example)

28. Simulation results
OGRS 2016, 12 October, Perugia (IT)
Growth rate
decreases in the
adult phase of
King Scallop life
Resource consumption and regeneration in time (king scallop example)

29. Simulation results
OGRS 2016, 12 October, Perugia (IT)
Fishing balance

30. Conclusions
OGRS 2016, 12 October, Perugia (IT)
Strength points:
● Rapid execution of the analysis since no GUI is loaded
● Better performance in leaving geographical operations executed by GIS and
simulation executed by MAS (time and quality of the output)
● Good automatisation level guaranteed by the Python script
● Easiness in configuration for a future WPS

31. Conclusions
OGRS 2016, 12 October, Perugia (IT)
Points to improve:
● Better connection with a specific library (GAMA to GRASS or viceversa)
● GAMA headless configuration is a little complicate (a lot of different
configuration files)
Ongoing work:
● Generalisation of the model (more maritime activities)
● Strengthening of the multi-level infrastructure

32. Thanks to you all for the attention
OGRS 2016, 12 October, Perugia (IT)
annalisa.minelli@gmail.com
All the software is available at:
https://github.com/annalisapg/maritimeSimulation