Cellular automata and agent base models for urban studies: from pixels to cells to hexa-dpi's - Urban Remote Sensing: Monitoring, Synthesis and Modeling in the Urban Environment - page 308

22.1 Introduction

surveying and the geographic wideness it could cover (Fig. 22.1).
Each image was treated as an array of values corresponding to
the spectral channels, and the pixel was the basic/minimum unit
of the image, each pixel's location was identified by row-column
coordinates (Fig. 22.2).
If image acquisition in remote sensing allowed to define
new data structures (raster/matrix/pixel), image classification
of remotely sensed data through the use of image classifiers
(computer programs that would implement processes to classify
images - convert numeric images to classified images) started to
develop a new field that would allow to make sense of the vast
amounts of data arriving in a matrix of numbers.
The classified images would then feed other computer applica-
tions that would describe attributes, perform statistical analysis,
spatial descriptors, search spatial relationships, and perform spa-
tial analysis and simulation. These new opportunities tended to
link cartographic production of base maps resulting from these
remotely sensed data to the production of thematic maps for
multiple uses. A name that started to be associated with the pro-
cess of producing/analyzing digital maps was commonly known
as geographic information systems (GIS). A toolbox:

This chapter focus is on the development of digital data struc-
tures and main models during the last decades; in particular
it will emphasize the importance of the development of raster
data structures though remote sensing that were pivotal in the
development of key artificial intelligence models, such as cel-
lular automaton (CA), agent base models (ABM) and genetic
algorithms (GA).
It will emphasize the interaction between remote sens-
ing and CA-ABM; in particular it will focus on the trilogy
pixels-cells-hexa-dpi, it will explainwhy this is important for CA
and ABM and how it can be enhanced by remote sensing. It will
present today's pixel-cell interaction and potentialities and it will
also point out future directions by exploring the opportunity of
integrating remote sensing, CA, ABM, and other AI approaches.
In order to fully understand where we are today in terms
of the multiple attempts to integrate different approaches, a
set of phased steps will tell the ''history'' from the pixel to
the cell, from the cell and agent's individual development to
its recent integrated approaches; this will be followed by a
discussion of one of the main challenges at the present moment
(integration of spatial and aspatial data structures and modeling
environments), the chapter finishes with a discussion towards a
proposal of a hexa-dpi data structure that will close the circle of
dpi-pixel-cell-hexa-dpi and will allow to integrate spatial and
aspatial dimensions in time and space. A hexa-dpi structure is a
hexagonal spatial matrix that will work as the ''virtual'' magnetic
field where spatial cell-based-dots can interact in different
time/scale dimensions. Allowing spatial and aspatial attributes
to converge and synchronize in dynamic spatial representations.

a GIS can be seen as a set of tools for analyzing spatial data.
These are, of course, computer tools, and a GIS can then
be thought as a software package containing the elements
necessary for working with spatial data

(Clarke, 1999).

From this initial definition GIS, migrated towards more inte-
grated approaches, linking basic statistic analysis with more
sophisticated data mining approaches, and advanced model-
ing approaches (Maguire, Batty and Goodchild, 2005; Haining,
2003), contributing towards what is nowadays considering a
geographic information science (Goodchild, 2009).
Therefore, new technologies, new methodologies and a new
language started to emerge from the previous developments. For
the purpose of this chapter, the development of the pixel-matrix
data structure and the development of computer algorithms that
would explore pixels in a matrix in order to assign meanings to
numbers and classify patterns allowed, in a subsequent phase,
the implementation of new emerging theoretical developments
and the creation of the ''cell.''

22.2 Computation: the
raster - pixel aproach

Aerial photos, ''dots in a paper,'' had a profound impact on
cartography and its affiliated subjects. Gray tones or false colors
that represented different land attributes and the development of
stereoscopic pairs that would allow three-dimensional visualiza-
tion transformed the science of cartography, geography, urban,
and environmental planning.
Remote sensing as:

22.3 Cells: migrating from
basic pixels

the practice of deriving information about earth's land and
water surfaces using images acquired from an overhead
perspective, using electromagnetic radiation in one or more
regions of the electromagnetic spectrum, reflected of emitted
from the earth's surface

If the pixel was the basic unit of a matrix and this allowed to
scan/convert and produce digitalmaps,witheachpixelhavinga
specific numeric attribute (i.e. resulting froma spectral signature,
or from a gradation of tones in a color/grayscale table) and if
computation was already allowing moving windows to operate
simple/complex operations in a matrix or between matrixes, why
not allow a pixel interaction at a deeper level? And why would
this be needed?
Starting to answer this question: why would this (interaction
between pixels) be needed? There are at least three obvious
reasons for this: (1) computer supported technology was able
to do more than simple scans of the Earth and re-produce
base-cartography; (2) the development of analytical thematic
cartography required more that a simple overlap of maps; (3)

(Campbell, 1996, p. 5)
was at the basis of one of the most important revolutions
in data collection and analysis - plains and satellites sensed the
earth for multiple purposes. Less known among the research
community is the importance of remote sensing to the develop-
ment of matrix-base data structures and associated algorithms
(Fig. 22.1).
Gathering information at a distance using satellites with
spectrometers that surveyed the earth by means of reflected or
emitted electromagnetic energy and that would send information
to earth codified as raster images, allowed increasing the pace of