Information Technology Reference
In-Depth Information
review several of these in a recent chapter [15]. Berger uses them to represent impacts
of technical innovation and policy changes in Chile, coupling economic impacts for
farmers with new water policy setting in Chilean sub basins [16]. Van Oel and his
colleagues use an ABM to represent dependence of land use decisions in arid
northeast Brazil on practical water availability. Their model of water uses choice
impact on a semi-distributed hydrological model with land cell and rivers represented
as sequences of branches [17].
In several cases these models are considered useful for interaction with
stakeholders, including because of their adaptability to explore various scenarios for
example of water users' preferences or of their context of work such as climate [18].
This interactivity is best represented in participatory modeling cases such as the
KatAWARE model designed in a south African basin [19]. These authors design an
Agent based Model of a river basin taking in charge the suitable entities to cope with
the various viewpoints of stakeholders according to their suggestions in the modeling
workshops. Becu and colleagues [20] have proposed a whole method for eliciting
conflicting views on what drives farmers in their practice up to including these
heterogeneous representations within a single agent based model.
5
The GESPER Model
In this section we describe the GESPER model, based on a cellular automaton for the
physical part and an ABM to include social and behavioral dimensions.
5.1
Physical Part of the Model
The physical layer is made of a grid of cells with a connectivity of four. We assume a
cell representing a square of 500x500 m
2
. Each square cell is made of three
compartments: surface, sub-surface and groundwater. These cells are first described
by their altitude, soil characteristics i.e. a soil water capacity, and ground water
capacity. We apply to these elementary cells the algorithm of MERCEDES model
[10] according to figure 1 below. This means adding further parameters to the cell
characteristics to handle interfaces between compartments: infiltration rate, deep
release rate and superficial release rate.
Surface transfer between cell is adapted from MODCOU algorithm [21-22], a
physically distributed hydrological model with variable scale. This model needs to
predefine two parameters of transfer of the cell, that depend on cell's hydrological
status (part of a river or not). These two parameters, aDist and aVol, entail specifying
the quantity of water is leaving a cell during one time step and the cell this “water
pack” reaches. The distance of transfer (distanceTransfer) and volume of transfer
(volTransfer) of this water pack during one time step is computed according to the
equations below, where slope is the mean slope between initial and final cell and
volStock is the current water level on the initial slope.
/
