SIMULATION OF HUMAN ACTIVITIES DYNAMICS (DAHU)
IN COASTAL SEAS

Matthieu Le Tixerant, Mathias Rouan, Françoise Gourmelon, François Cuq

Laboratoire Géomer (LETG UMR 6554 CNRS), Université de Bretagne Occidentale - Institut Universitaire Européen de la Mer, Plouzane (FR)

Introduction
Coastal seas are the seat of multiple human activities, sometimes conflictual, interacting with environment. Understanding the interactions between these activities and environment is a major objective of research for the sustainable development of human societies [CUQ 00]. In this context, it is notably indispensable to focus on the way human use and exploit environment . It answers to a strong need of scientists, who try to evaluate the part of human activity in ecosystems evolution , and of decision-makers who need synthetic information about social and natural systems. Access to an adapted information is effectively an essential requirement to an "Integrated Coastal Zone Management (ICZM) [COM 00].
While development and implementation of integrated coastal management policies are now established and internationally recognised ideal, the tools and methodologies for achieving such goals are still under development [BAR 00]. These tools must compile numerous multiple origin data and communicate a relevant information. It is notably essential to show the progress in space and time of human activities [HOL 94] [WEB 90].

In this framework, the marine environment poses specific problems. On this form of open space, traditionally considered as "a liberty space ", it is especially difficult to surround and to understand how human activities progress. Stationary physical limits, allowing to assign a space to an activity, don't exist. Different types of activities can coexist, on a same place, at the same moment. Considering the temporal dimension is even more important [ALL 00]. Besides, there are few structured information sources describing marine activities and allowing a global vision of their progress. Existing statistics are often too aggregate (insufficient degree of accuracy) and thus unsuited to the continued objective. They are not always connected to zones and periods of practice (or with inadequate scales). Impact assessment on the environment is then difficult to establish. Marine environment is therefore a space that can be qualified as "blurred" : understanding phenomena related to human activities is then even more complex.

Firstly, this contribution presents the methodology developed for the modelling platform DAHU (Human Activities Dynamics) [CUQ 01]. It is a simulation system based on the coupling of quantitative and qualitative models within a Geographical Information System (GIS). It allows a dynamic description of human activities progress and their environment impact . The module described here concerns specifically the marine environment.
Secondly, an application to a fishing activity (the queen scallop shell dredge fishing) of the Iroise sea (France) is presented.

1 - Modelling principles
Before all modelling procedure, it is necessary to define general principles of the adopted method. First of all, the aim is not to formalise a representation of the progress of virtual activities on a theoretical territory but to propose a descriptive tool founded on the analysis of existing (scientific and administrative) data in order to lead to a "plausible" reality model [TIS 03].

The main specificity of this methodology is to consider the spatial dimension as a simulation constraint. Space constitutes the basis from which are associated temporal and statistical data. Each activity has its own practice territory. But the objective is to simulate their simultaneous progress on a stationary zone previously defined. So, within a unique study zone, practices territories vary and superimpose themselves according to the analysed activities. Thanks to this approach, territory acts as a relevant integrator support since common to the different activities [SED 96]. In an integrated management context , it is necessary to model the territory and its activities in their whole rather than to analyse an isolated action impact on environment [PRE 95] or to answer a sectorial activity needs. This approach tries to encourage a systemic procedure privileging a global and interdisciplinary vision of the territory and enhancing interdependence between the different components of the system.

Professional activities of intensive type are selected in priority. They are generally based on standardised working processes able to be describe on an archetypal way. This report validates the hypothesis of a quasi-determinist approach.

The methodology leans on a concrete case : the Iroise Sea. Situated on the western extremity of Finistère (France), the Iroise Sea is a rich marine ecosystem where multiple practices and potentially conflictual activities can interact with the coastal environment. Only an integrated management of the whole zone would allow to insure its sustainable development [LEG 99].

2- Spatio-temporal modelling
Firstly, a typology of human marine activities is realised. They are structured according to their environment type of use (figure 1): exploitation of the living and non living resources, way of circulation and transportation, waste disposal, protection, research, tourism and recreation [CIC 98].

Then, to understand their progress, it is necessary to disintegrate the initial data in space and time. So for each activity, a coherent modelling environment must be constructed. It integrates all the constraints influencing the simulated activities progress. This modelling environment is constitute by a Potential Practice Territory and by a Potential Practice Calendar: spaces and periods of activities progress are identified by the superimposition of different "spatio-temporal thematic filters" [LET 02b].
Indeed, marine human activities are submitted to a multitude of variables :

Figure 1. Typology of human activities in coastal sea


2-1 Space modelling
Spatialisation is achieved thanks to the projection of constraints influencing activities progress within a database managed by a Geographical Information System (GIS). GIS spatial analysis functions such as union and superposition of coverages allow to identify zones where activities are susceptible to take place. For each activity, a Potential Practice Territory is thus produced (figure 2). It is a coded polygon coverage indicating where the practice is impossible, possible and really practised.


2-2 Temporal modelling
Second stage consists in describing the progress of activities in time. Principle of this description is the same as for space approach: on a one year cycle, superposition of different filters (figure 2) allows to result in a Potential Practice Calendar of the activity, identifying periods during which practice is possible or impossible.

2-3 Assessment
This essential first step allows a previous data formatting which provides bases of simulation of simultaneous progress of human activities in marine environment :
- Relevant scales of time and space are defined for each activity.
- Impact assessment is possible by considering statistical data which indicate the "degree of pressure" of human activities on the environment (figure 2).

Therefore, a schematic description of the qualitative and quantitative relations between society, environment and climate can be considered. It implies to be able to treat, to put in relation and to distribute data in suitable form.

Figure 2 : summary diagram of spatio-temporal modelling principles

3 - Setting in relation of the data: DAHU simulator
Simulator follows a classic working diagram: input data preparation, simulation and results exploitation [CUQ 01] (Figure 3). The system is interfaced with a GIS on which depend both the formatting of initialisation data (pre-processor) and the exploitation of results in exit (post-processor).
The simulation phase is achieved thanks to a non-commercial software developed by the laboratory [ROU 01] [TIS 01]. Specified in United Model Language (UML), it has been implemented in oriented object language (C++). This software has a convivial interface that allows a fast and easy use.


3-1 Preparation of the input data (pre-processor)
Data, pre-treated during the spatio-temporal modelling phase, are then coded with the help of a pre-processor which harmonizes these raw data, so that they are directly exploitable by the simulator (figure 3). It is mainly about relating spatial, temporal and statistical data within a common Relational Database. This Relational Database contains the totality of information, as attribute tables, allowing the activities description.
These tables are structured compare to the attribute polygon codes of the "potential practice territory" coverage. Tables containing temporal and statistical variables are structured in order to be directly associated. Spatial referential acts therefore as the main model calibration data.


3-2 Simulation (processor)
This simulator has an architecture being directly inspired by a Multi-Agents System. Actually, to every activity corresponds an autonomous agent pursuing an objective of optimal practice (figure 3). This agent reacts (presence / absence) according to the practice conditions (environment, weather report, regulation, socio-economy) during the period of selected simulation. Constraints of simulation are calculated by the pre-processor and are introduced in the model via the Relational Database. According to the chosen scenario, agents adapt their behaviour according to a stimuli / answer principle [WEI 99] [BOU 99]. In other words, each agent is forced by the spatial, temporal and statistical variables formatting within the pre-processor.
During a simulation, the system is optimised to select only data necessary to the user's request . A daily iteration calculates the different simulation parameters. Thus, for one selected period, the simulator produces attribute tables describing activities susceptible to be present. Associated variables are the quantitative and qualitative results of simulation for the selected activities. In order to get a synthetic vision of results, the tables are treated to be directly exploitable by a GIS.


3-3 Results exploitation (post-processor)
Results exploitation via a GIS allows cartographic restitution of relevant spatio-temporal representations and constitute a synthetic information useful for integrated management (figure 3). The decision makings often need cartographic documents elaborated from a multitude of spatial data. For decision-makers, maps are an efficient support to inventories and a way of information and communication [GOU 01]. In the DAHU simulator case, synthesis maps are produced and provide elements necessary to a better understanding of the environment practices.


Figure 3 : The DAHU simulator platform

4 - Application to the Iroise sea
According to the methodology, a typology of human activities in the Iroise Sea is first realised. Professional fishing is one of the zone main human activity. The description of this activity and its different subactivities has been studied through a collaboration with the Halieutical Laboratory Resources of Ifremer in Brest.
The objective is to succeed in a description of activities progress according to constraints (environment, weather report, regulation, socio-economy ) governing the progress of practices and from works of investigation achieved by Ifremer [BON 00] [BON 02]. The queen scallop shell dredge fishing is one of the main activity on the zone. Dredge are engines dragged by ships that scrape the seafloor in order to extract shells of it.

The main conditions of practice to be considered are the following :

The spatio-temporal modelling phase (figure 4) :
The Potential Practice Territory integrate rocky seafloor [LEB 99], forbidden fishing zones and zones of " real " practice for the simulated period. The Potential Practice Calendar integrate the forbidden fishing period of practice and weather conditions.

Figure 4. Queen scallop Shell dredge fishing. Schematic representation of the spatio-temporal modelling


The simulation phase (figure 5) :
For a given simulation scenario, the selected data are gathered into the Relational Database and stimulate the agent "Queen scallop shell dredge". The agent's answer are output data exploited by the GIS. The cartographic results depend on the simulation period selected and the chosen simulation conditions. Figure 5 shows four scenarios applied to the case of queen scallop dredge fishing in Iroise sea. It shows that, according to the regulation and to the meteorological conditions, spatial distribution and intensity of this practice can vary considerably.

Figure 5. Some simulation results of the modelling of queen scallop dredge fishing in Iroise sea.

Conclusion / perspectives
A modelling method applied to human activities progress in coastal sea is achieved. Currently, the simulation system allows a daily description of activities progress during the year. While integrating the different practice conditions (environment, meteorology, regulation and socio-economic conditions), it is possible :

This approach can be realised at different scales (spatial and time).

One of the main interest of the method is the possibility to simulate several activities in the same time, on a same space. Besides, the possibility to combine simulation results allows to show spatio-temporal interactions between different activities and to focus on conflictual activities and environmental risks.

For integrated coastal management, this method provides information usable for :

This method could also contribute to the conception of a long term general model describing practices and anthropic pressure on the inshore environment. In the project of an observatory of the inshore domain, it would allow to estimate the relative impact of human activities and climate on the environment evolution [CUQ 02]. In a more general manner, it would bring answers to the present needs of prospective management while estimating the consequences of a decision on the coastal system environment.

References