Geoscience Reference
In-Depth Information
farmers whose production is insufficient for surviving the year (Grinblat et al., 2014). The detailed
output includes a field layer, where every field is characterised, for every year of the investigated
period, by the land productivity, crop and crop yield. These spatial data could serve to estimate
wealthy areas and areas of starvation within the province.
To investigate and compare AB or any other models, we have to define them formally. Let us thus
follow Bertalanffy's definition explicitly with respect to AB modelling.
9.1.3 g eograPhic S ySteMS a re c oMPoSed of S Patial e ntitieS : f eatureS
To present the components of a geographic system, including humans, we use the terminology
derived from Geographic Information Science (GISci). GISci considers geographic systems as con-
sisting of discrete spatially located entities, that is, different kinds of 'things that can be uniquely
identified' (Howe, 2001, p. 92). GISci implements this view through a GIS, which is a spatially
enabled extension of a relational database. The theory of relational databases deals with entities and
relationships between entities (Howe, 2001). In terms of database theory, GIS deals with spatial
entities or features .
Land parcels, street segments, buildings, cars, householders, pedestrians, farmers and land
developers are all entities. Entities, regardless of whether they are spatial or not, are characterised
by an identifier and an arbitrary number of attributes . Spatial entities are additionally characterised
by their location.
All real-world entities are located in space, but location information on some of them may be
irrelevant for studying geographic systems. In our example of the savannah horticulture, the loca-
tion of the land-use regulator is irrelevant for understanding the horticultural dynamics of this
system. Important is the location of the farmers and farmers' fields; the further away a farmer is
from his or her field(s), the longer it will take for the farmer to get there, which lowers the chance
that fertiliser will be applied.
It is important to note that a system view of geographic reality includes continuous spatial phe-
nomena, such as the height of the land surface or the concentration of nitrogen in the soil. GISci
and GIS employ grids of artificial cells to represent these phenomena. When defined, however,
the cells of these grids can be considered as standard GIS features (polygons), characterised by
location and non-spatial properties (e.g., the altitude, land use and land cover). Any other poly-
gon coverage can be employed for representing continuous phenomenon, for example, Delaunay
triangles or Voronoi polygons, and GISci provides a series of methods to analyse continuous data
with respect to the chosen partition of space into elementary units (Banerjee et al., 2003; Fischer
and Getis, 2010).
9.1.4 f eatureS and t heir a ttriButeS c hange oVer t iMe
Parameters of geographic features, including the locations of these features, may change over time.
Existing features can disappear and new features can emerge. The incentive for change can come
from the feature itself, as in the case of a moving car, or from the other features, as in the case of
an agricultural parcel sold for the purpose of constructing dwellings. In geographic systems, the
interactions between features strongly depend upon the distance between them. Waldo Tobler's
First Law of Geography : 'Everything is related to everything else, but near things are more related
than distant things' (Tobler, 1970, p. 236) is actually the first law of complex geographic systems
theory. In order to estimate distances and to describe interactions between features as a function of
distance, Tobler's law demands a high-resolution view of systems. Geographic AB modelling builds
on this high-resolution representation. A typical AB geographic model considers spatial and non-
spatial behaviour and decision making of humans or human institutions that influence themselves
and non-animated geographic features; the goal of developing the model is to study the dynamics of
the system, which is driven by the behaviour of the agents.
Search WWH ::




Custom Search