498.txt

Towards Geodesign: Repurposing Cartography and GIS?

Michael F. Goodchild, Ph.D.  |  good@geog.ucsb.edu

Center for Spatial Studies  
Department of Geography
University of California, Santa Barabara 
Santa Barbara, CA  93106-4060

R E P L A C E     W O R D C L O U D

A B S T R A C T

One of the original visions for GIS was as a tool for creating designs, but 
GIS has evolved in numerous other directions. Definitions of geodesign 
are reviewed, together with a short history of the concept. A distinction 
is drawn between Design and design, the latter being addressed through 
spatial decision support systems, and the former being seen as a superset of 
the latter. Geodesign also has a strong and well-defined relationship with 
cartography. The vision of landscape architecture propounded by the late 
Ian McHarg also provides a foundation for geodesign. Two existing gaps 
in the geodesigns computational tools are identified: support for sketch 
and implementation of models representing scientific knowledge of how 
the world works. Two important areas of research are identified that would 
address problems that currently impede geodesign.

I N T R O D U C T I O N

Although there are many historic roots of geographic information systems 
(GIS; Foresman 1998), one of the strongest lies in the notion of making 
design decisions by overlaying maps, each map representing one of the 

Cartographic Perspectives, Number 66, Fall 2010

Article Title  Author Name(s) |  55

factors important in the decision. The net effect of each of the factors would 
be represented by the amount of light penetrating the layers at each point, 
allowing the decision maker to make an intuitive judgment as to the best 
solution. This is one of the central ideas of McHargs (1969) Design with 
Nature, and the stack of layers has become an icon of the entire field of GIS, 
appearing on the front covers of many of its textbooks. One of the strongest 
arguments for GIS has been its ability to place such a simple and intuitive 
concept as overlaying transparent maps on a solid, reproducibleone might go 
so far as to say objective and scientificfooting. Abundant examples of this idea 
can be found in the fields textbooks, ranging from site selection for industrial 
plants to routing of power lines or highway corridors.

However, in the four decades that have elapsed since its birth, this notion 
of GIS as improving the process of design has become less central. GIS has 
evolved into a tool for performing spatial analysis in support of scientific 
discovery, a system for managing inventories of spatially distributed assets, a 
platform for automating the cartographic process and displaying information 
in map form, and a medium for communicating what is known about the 
surface and near-surface of the planet (Sui and Goodchild 2001). Yet we 
increasingly are aware of the planets fragility, and of the need to make wise 
decisions about its future that are informed by evidence and by the best 
scientific knowledge. Now, more than ever, we need a technology of design 
that can work in tandem with human decision-making processes, bringing 
what we know about how the planet works to bear on the decisions that have 
to be made about its future. Humans have the power both to destroy the planet 
and to sustain it. We need tools that can predict for us the effects of tinkering 
with the Earth system, thus helping us to be effective stewards of the only 
planet we have.

This concept of science-based design sits at the interface among several 
disciplines. It involves the disciplines that traditionally have concerned 
themselves with design, including planning and landscape architecture. But 
it also involves the disciplines that acquire and accumulate fundamental 
knowledge about how environmental and social systems operate, including 
geography, ecology, hydrology, earth science, sociology, economics, and political 
science. Finally, it includes the new disciplines of information technology, 
especially geographic information science (GIScience; Goodchild 1992). Input 
from all three of these sets is needed if decisions are to be supported by well-
designed and powerful tools that are easy to use, and by the results of good 
science. 

Over the past decade, there have been several discussions of the need to close 
what many have perceived as the growing gap between GIS and design. 
In January 2001, a workshop was held in Santa Barbara, California on 
Landscape Change, organized by a joint committee of landscape architects 
and GIScientists (http://www.ncgia.ucsb.edu/landscape/landscape.htm). A 
second workshop on Spatial Concepts in GIS and Design was held in Santa 
Barbara in late 2008 (http://www.ncgia.ucsb.edu/projects/scdg/). The term 
geodesign was suggested as a useful umbrella term for this examination of the 
common ground between GIS and design, with its implied emphasis on the 

Now more ThAN 
ever we Need 
A TeChNology 
oF desigN ThAT 
CAN work iN 
TANdem wiTh 
humAN deCisioN-
mAkiNg ProCesses, 
briNgiNg whAT 
we kNow AbouT 
how The PlANeT 
works To beAr oN 
The deCisioNs ThAT 
hAve To be mAde 
AbouT iTs FuTure

56  |  Article Title  Author Name(s)

Cartographic Perspectives, Number 66, Fall 2010

geographic domain and geographic scales. Most recently, in January 2010 
the first GeoDesign Summit was convened in Redlands, California (http://
www.geodesignsummit.com), bringing together GIS and design practitioners 
from academia, non-governmental organizations, government agencies, and 
the private sector, with over 150 participants from across a wide range of 
disciplines.

This paper presents one persons view of the nature of geodesign, of its 
objectives, of how the field might be conceptually framed, of its relationships 
to existing fields (especially cartography and GIS), and of research issues that 
need to be addressed if current impediments to effective geodesign are to 
be removed. The remainder of the paper is organized as follows. The second 
section reviews alternative definitions of geodesign, its domain of application, 
and its cognate disciplines. The third section discusses the McHarg vision and 
its limitations, updates it to the present day, and presents a brief critique. The 
fourth section describes the tools and computing environment that would be 
needed to support a fully-fledged practice of geodesign. The final section ends 
with some suggestions for future developments. The discussion is inevitably 
personal, with no implication that all possible topics and arguments have 
been covered. Nevertheless, the paper may provide a useful increment in our 
understanding of the nature of geodesign and of what needs to be done to 
move its agenda forward.

W H AT   I S   G E O D E S I G N ?

D e f i n i t i o n s

The GeoDesign Summit website quotes Carl Steinitz: Geodesign is 
geography by design, a compellingly simple definition. If geography is the 
set of processes that operate on or near the Earths surface, together with 
the forms that result from such processes, then geodesign is concerned with 
manipulating those forms and intervening in these processes to achieve specific 
objectives. Thus, it is normative in the sense that decisions are made about 
aspects of the geographic domain in order to achieve specified objectives, or 
norms. Normative efforts stand in contrast to the traditional aim of science, to 
discover general truths about the world; geodesign is interventionist in contrast 
to the more detached and dispassionate nature of pure science. Geodesign 
seeks to improve the world, whereas traditional science seeks only to provide 
the basis of knowledge on which the world might eventually be improved. 
Pure science is often carefully partitioned from application, and often sees 
its responsibilities as discharged when results have appeared in the pages of 
refereed journals. In that sense, geodesign lies within the domain of applied 
science and engineering, seeking ways of addressing practical problems using 
the scientific method.

Wikipedia defines geodesign as a set of techniques and enabling 
technologies for planning built and natural environments in an integrated 
process, including project conceptualization, analysis, design specification, 

Cartographic Perspectives, Number 66, Fall 2010

Article Title  Author Name(s) |  57

geodesigN 
is PlANNiNg 
iNFormed 
by sCieNTiFiC 
kNowledge oF 
how The world 
works, exPressed 
iN gis-bAsed 
simulATioNs

stakeholder participation and collaboration, design creation, simulation, and 
evaluation (among other stages). The emphasis here is on built and natural 
environments, or Steinitzs geography, and also on the integration of the 
entire design process, presumably through technology. 

Both of these definitions imply a very broad and traditional interpretation of 
the planning process. Others, however, have focused more on how planning 
can take advantage of the capabilities of GIS. Wikipedia also quotes 
Flaxmans address at the GeoDesign Summit: Geodesign is a design and 
planning method which tightly couples the creation of design proposals 
with impact simulations informed by geographic contexts. In other words, 
the ability of modern GIS to create highly compute-intensive simulations 
of the effects of design scenarios provides an additional dimension to 
the traditional planning process, with its emphasis on visual display and 
intuition: geodesign is planning informed by scientific knowledge of how the 
world works, expressed in GIS-based simulations. In a similar, though less 
compute-intensive vein, and quoting Jack Dangermond from the GeoDesign 
Summit website, Imagine if your initial design concept, scribbled on the 
back of a cocktail napkin, has the full power of GIS behind it. The sketch 
goes into the database, becoming a layer that can be compared to all the 
other layers in the database. Clearly, comparison of layers is only one of the 
multitude of functions that are easily invoked with todays GIS. Nevertheless, 
sketch and simulation provide two distinct notions of how the computational 
environment of a GIS might support geodesign.

One might also compare geodesign with other more widely recognized and 
traditional terms, such as computer-aided design (CAD). GIS has often been 
distinguished from CAD (Cowen 1988) by its emphasis on a geographic 
reference system, the richness of the attributes associated with features, its 
ability to deal with continuous fields (Couclelis 1992) in addition to discrete 
objects, and its rich set of analytic and modeling functions. In essence, the 
emphasis in CAD is on designing a structure through digital representation; 
in GIS it is on analyzing and modeling the structures present in the social 
and environmental worlds; and in geodesign it is on user-driven intervention 
in those worlds.

spat i a l   o p t i m i z at i o n

There is a long tradition of finding optimal solutions to design problems in 
the research domain known as spatial optimization. Much of this literature 
concerns finding optimal locations for point-like facilities, such as schools, 
fire stations, retail stores, or restaurants (Ghosh and Rushton 1987). 
Numerous problems have been formulated, depending on the exact nature of 
the application, the objectives and constraints that apply, and the nature of 
the space within which optima are sought. For example, the field of location-
allocation concerns the search for one or more locations for point-like 
facilities to serve a dispersed demand, and solutions involve both the optimal 
locations of the facilities and the service areas that each will cover. 

58  |  Article Title  Author Name(s)

Cartographic Perspectives, Number 66, Fall 2010

More generally, spatial optimization problems can be characterized by 
the type of information represented by the solution. Some problems seek 
optimal locations for points, some for lines (e.g., transmission corridors), 
and some for areas (e.g., optimal allocation of land for specified uses). Some 
problems seek optimal allocations of one set of features to another, as in the 
case of optimal allocation of service areas for schools, or optimal patterns of 
transportation from origins to destinations. In general, a spatial optimization 
problem might find solutions in the form of any augmentation of a GIS 
database. From an object-oriented perspective (e.g., Zeiler 1999), this might 
mean the creation of a new feature class; the addition or deletion of features 
from an existing class; the addition or modification of one or more attributes 
of a feature class; the creation of a new association class representing patterns 
of interaction between existing origins and destinations; the creation of new 
routes that are themselves aggregations of an existing edge feature class, etc. 
In this way, the problems formulated as spatial optimizations can be related 
directly to the elements of a modern GIS database design. More broadly, we 
can see geodesign as transforming an existing database D into a new one D 
through some combination of edits.

Spatial optimization provides a useful framework for geodesign, although it 
is often far too simplistic, as the next section explains. Spatial optimization 
requires an objective function that reflects the goals of the design, expressed 
in numeric form as a function of the solution variables that are available to 
be manipulated by the designer. It requires a solution space that is defined 
by the solution variables and limited by the constraints. The final design will 
occupy one point in the solution space. In geodesign, the solution variables 
all can be found in the database, as attributes or geometries of features, or as 
attributes of association classes. The user is able to interact with the solution 
variables in various ways, such as by using sketch tools to define or edit their 
geometries, or using the keyboard to define or edit their attributes.

  
B i g - D   a nD  s m a l l -D De s i g n

Spatial optimization often is seen as a task to be performed by a machine 
with no human interventionas a fully automated edit of a database. Once 
the objectives and constraints are formulated, and the data are assembled, 
the machine is allowed to take over, producing a solution that by definition 
represents the best possible decision. It is often argued that such formal 
procedures provide a vast improvement over the messy, intuitive process of 
more traditional decision making. Disagreements between stakeholders over 
the objectives and constraints, or over the weights to be applied to different 
factors, can be handled through a variety of equally rigorous and mechanical 
multi-criteria problem formulations (Thill 1999). However, courts have 
sometimes held that a solution can be unacceptable against certain criteria, 
such as racial bias, even though the objective function and constraints 
included no such bias. Moreover, it may simply be naive to believe that 
human rationality, in the form of rigorously formulated optimization 
problems, can ever replace the messy nature of politics. Instead, spatial 
optimization is better seen as a collaboration between human and machine, 

The user is Able 
To iNTerACT wiTh 
The soluTioN 
vAriAbles iN 
vArious wAys, 
suCh As by usiNg 
skeTCh Tools To 
deFiNe or ediT 
Their geomeTries, 
or usiNg The 
keyboArd To 
deFiNe or ediT 
Their ATTribuTes

Cartographic Perspectives, Number 66, Fall 2010

Article Title  Author Name(s) |  59

in which the machines role is simply to perform the calculations and 
iterations that humans find tediouswith the human still firmly in control.
This argument provides a useful basis for distinguishing between two visions 
of the design process. Small-d design takes a simplified viewdesign 
consists of the formulation of an optimization problem with objectives and 
constraints, the collection of data, the execution of a search for the optimum 
solution, and its implementation. In this somewhat naive and simplistic 
view, implementation is seen as inevitable, because all participants agreed 
on the objectives and must therefore accept the result. Small-d design 
most commonly is associated with the disciplines of operations research, 
engineering, and management science.

Big-D Design sees the process as complicated by disagreements among 
stakeholders, difficulties in deciding what is optimal, feedback loops 
that modify objectives, constraints, and data as the process proceeds, and 
uncertainties about implementation. Figure 1, taken from the work of 
Steinitz (1990; Steinitz et al. 2003), structures Design as a sequence of six 

Figure 1. A six-stage framework for the design process, with models at each stage (Courtesy of Carl steinitz).

60  |  Article Title  Author Name(s)

Cartographic Perspectives, Number 66, Fall 2010

stages with iterative feedbacksand similar schemata can be found in other 
sources. Each of these stages might be formalized as a model and supported 
by computational tools. Big-D Design most commonly is associated with 
the disciplines of landscape architecture and planning. As the dominant 
paradigm of geodesign, big-D Designrather than small-d designwill be 
implied whenever the term design is used in the remainder of this paper.

The problem of conservation planning provides a useful illustration of the 
difference between design and Design. Conservation planning seeks to 
acquire a set of conservation areas in order to best preserve one or more 
biological species. A set of models is constructed from empirical data 
to predict the ability of a given parcel of land to support a given species. 
A spatial optimization problem is then formulated, seeking the best 
combination of land acquisitions to provide a sustainable population of the 
target species (Hof and Bevers 2002). Issues such as the connectedness of 
parcels are important to allow for interactions between breeding populations. 
The results can be expressed in the form of a new attribute of the land-
parcel feature class, denoting whether or not each land parcel is targeted for 
acquisition. Thorne, Cameron, and Quinn (2006) provide an example plan 
for land acquisitions in Southern California to preserve the mountain lion.

However, it is naive to believe that the publication of such a plan will have 
no impact on the market for land, or on the attitudes of landowners. Instead, 
it commonly creates a strong and sustained reaction among the potentially 
affected landowners. Moreover, parcels inevitably will be acquired over an 
extended period of time, and it is likely that some of the optimal set will 
prove impossible to acquire, and will be replaced by alternative near-optimal 
parcels. In principle, each replacement affects the optimality of the entire 
solution, so the problem needs to be re-solved after every acquisition. In 
reality, then, the simple spatial optimization problem (design) is embedded 
in a much more complex process (Design) that is characterized by large 
amounts of uncertainty. Gallo (2007) shows how important it can be to 
avoid publishing a single, deterministic optimum solution, and instead 
suggests focusing on relative priorities expressed in probabilistic terms.

ge oDe s i g n   a nD  c a r t o g r a p h y

The display of geographic information in map form often is seen as an 
indispensable part of any geodesign process. Geodesign is by definition 
about geographic space, and Dangermonds definition quoted earlier points 
directly to cartography and the role of the computer as transforming 
an informal sketch into an element of a formal database. One of the 
achievements of GIS over the past 45 years has been the development of 
an integrated theory of geographic data representation, in other words, of a 
formal model of phenomena distributed over the surface and near-surface 
of the Earth. It includes discrete objects and continuous fields; points, lines, 
areas, and volumes; approaches to the representation of time; and solutions 
to the problem of representing flows and interactions (Goodchild, Yuan, 
and Cova 2007). GIS has progressed substantially beyond the earlier map 

The disPlAy oF 
geogrAPhiC 
iNFormATioN iN 
mAP Form is oFTeN 
seeN As AN
iNdisPeNsAble PArT 
oF ANy geodesigN 
ProCess

Cartographic Perspectives, Number 66, Fall 2010

Article Title  Author Name(s) |  61

CArTogrAPhy is 
A useFul PArT oF 
geodesigN, buT 
NoT All AsPeCTs 
oF geodesigN 
Are iNhereNTly 
CArTogrAPhiC

metaphor, when a GIS was viewed informally as a computer containing 
maps (Goodchild 1988). Animation, for example, is now almost routine in 
computerized displays, but was impossible as long as maps were confined to 
analog form on paper.

A two-dimensional display of geographic data, with the defining dimensions 
of the display representing (mapping in its mathematical sense) the spatial 
dimensions of the data, is a powerful way of showing the user what is present 
at every location within the extent of the display. Every point on a paper map 
can be printed with any color, and similarly every pixel on a computer display 
can be programmed to display any color. The visual impression of linear 
features is created by using a similar color along a linear sequence of points 
or pixels, and similarly an area is visualized by displaying its boundary, or by 
filling it with a uniform color or pattern. Annotation is communicated by 
linking points or pixels into the form of characters.

Nevertheless, it has always been difficult to use this approach to display 
geographic information that concerns points taken two at a timein other 
words, relationships or interactions between mapped features (Takeyama 
and Couclelis 1997). It is difficult, for example, to label a feature on a map 
according to the several names given to it by the inhabitants of neighboring 
areas, or to show flows of migrants or telephone calls. Such flows occur 
between pairs of features (origins and destinations), are not independent 
properties of either, and therefore cannot be displayed by symbolizing either 
feature alone, or even in combination. Lines might be drawn to connect 
origins and destinations, and appropriately symbolized, but in many cases 
the actual path of flow is not known (Glennon 2010), and a large number 
of such lines can render the map unreadable. In the previous section, this 
type of information was characterized as an association class in the object-
oriented paradigm. In short, while the results of any spatial optimization 
can be regarded as a modification of a database, not all such modifications 
are equally easily visualized cartographically. Sketch is an important way of 
communicating simple geographic features between user and GIS, and other 
important kinds of input can also be captured through interaction between 
the user and a map display. More broadly, cartography is a useful part of 
geodesign, but not all aspects of geodesign are inherently cartographic. 

T H E   MC H A R G   V I S I O N

Reference has already been made to the early days of GIS and the 
importance of design based on multiple layers of input. McHargs vision for 
his school of landscape architecture at the University of Pennsylvania used 
the stack of layers as a metaphor for the organization of the school (McHarg 
1996). Each layer corresponded to one discipline whose subject matter was 
important to landscape design, including ecology, hydrology, and geology. 
Overlaying the layers symbolized the simultaneous attention that needed to 
be paid to each of these as a plan was developed. Each layer would be shaded 
according to the weight to be assigned to the corresponding disciplines 

62  |  Article Title  Author Name(s)

Cartographic Perspectives, Number 66, Fall 2010

issues, and the relevant factors present at each point. 

This is only one aspect of McHargs contribution, of course, but it is 
emphasized here because of the way it links design and GIS, and thus 
relates to the topic of this paper. Suppose, for example, that the impact of 
a proposed pipeline at location x is determined to be ze(x) per unit area, 
when measured from the perspective of ecology. Similarly, the impact of the 
pipeline at x from the perspective of hydrology might be zh(x) per unit area, 
and the economic cost of acquiring the necessary land might be zc(x). The 
three measures are incommensurate, of course, so weights must be assigned 
to reduce them to a common metric, and to rate their relative importance. 
Define these weights as we, wh, and wc respectively. Then the problem can 
be formulated as finding a route such that the total weighted cost along the 
route is minimized. If S denotes the solution set, that is, the set of points 
along the route, then the task is to minimize:

 

Z

=



x
S


[
zw
e

e

( )
x

+

zw
h

h

( )
x

+

zw
c

c

]
( )
x

In the analog method described in Design with Nature (McHarg 1969), both 
w and z must be captured by the darkness of the corresponding layer at point 
x, and the optical process of overlaying layers replaces the summation in the 
equation by a multiplication. In reality, of course, the kind of rigor exhibited 
in the equation was never intended to be imposed in the analog method, but 
the more formal GIS overlay process forces the user to address all of these 
issues explicitly. For example, the weights w might be assigned using Saatys 
Analytical Hierarchy Process (AHP), or a variety of other methods that are 
described in the standard texts on multicriteria decision analysis (Malczewski 
and Rinner 2010; Thill 1999). Similarly, it is common to address the apples-
and-oranges issue of non-commensurate variables by normalizing each to 
the range 0 to 1. But since the observed range depends on the exact extent of 
the study area, there are obvious logical flaws in this practice.

Despite the informality of the analog model, McHarg clearly intended his 
design process to be informed by science, and achieved this by constructing 
each layer according to the knowledge base of the corresponding discipline. 
Moreover, the school included representatives of each of those disciplines on 
its staff, forcing intensive engagement and interaction. This is a very different 
approach from that commonly followed in most universities, where the 
science disciplines are separated from the design disciplines, often across the 
boundaries between colleges or faculties. One of the underlying themes of 
geodesign is its potential to reduce that separation.

McHargs vision is now more than four decades old, so it makes sense to ask 
whether and how it should be updated to the present. Enormous advances 
have been made in the disciplines that study social and environmental 
processes on the Earths surface, and an argument can clearly be made for 
including all of them to the extent that they are relevant to a specific design 
question. But in addition, we know far more now than we did then about the 

Cartographic Perspectives, Number 66, Fall 2010

Article Title  Author Name(s) |  63

process of decision making, and particularly about the role of uncertainty. 
Thus, it would seem important to include decision scientists and statisticians 
in the mix, especially spatial statisticians. We also would need to include 
the computer scientists and information scientists who address issues of 
representation and develop the algorithms needed to implement scientific 
knowledgeespecially geographic information scientistsand the experts 
in remote sensing and sensor networks who address issues of spatial data 
acquisition. We also would need to include the cartographers and specialists 
in spatial cognition who address human factors in the interactions between 
designers and tools, and the social psychologists who study processes of 
group interaction.

C O M P U TAT I O N A L   S U P P O R T   F O R   G E O D E S I G N

In the complex process represented by Figure 1, geodesign can be partitioned 
into a series of stages, each underlain by a model and each supported by 
computational tools. This is very different from the conceptualization of 
small-d design, in which the entire process occurs in a single stage, and in 
which a large proportion of control is surrendered to the computational 
system and its task of finding the best solution. The field of spatial decision 
support systems (SDSS) has long addressed the kinds of computational tools 
needed to support design decisions, and has accumulated a substantial 
literature (Leung 1997; Sugumaran and Degroote 2010). Li and her 
collaborators have recently constructed a very substantial collection of Web 
resources (http://www.institute.redlands.edu/sds), including an ontology 
of SDSS, in an effort to address the varying use of terms and to clarify the 
fields relationship with other cognate fields.

What then is the relationship between SDSS and geodesign? SDSS has its 
roots in the early 1990s (Densham and Goodchild 1990), and in a desire 
to apply GIS tools to a host of problems of spatial optimization. SDSS has 
always had a strong science base, so one might see geodesign as an effort to 
expand SDSS to include some of the design problems that have traditionally 
made less use of scientific knowledge to simulate the effects and impacts 
of plans. In other words, traditional SDSS may best be seen as a subset of 
geodesign, if any distinction is needed.

Reference was made earlier to the notion that there are two areas where 
geodesign tools need further development:

sk e t c h   t o o l s

The first key area of support for geodesign is sketch, or the ability of the user 
to create informal renderings of points, lines, and areas in geographic space, 
and to have the computational system capture, formalize, and store these. 
ESRIs ArcSketch already offers some of these capabilities, and Googles 

64  |  Article Title  Author Name(s)

Cartographic Perspectives, Number 66, Fall 2010

iN CoNveNTioNAl 
ediTiNg, The user 
seeks To APProximATe 
A TRUTH, As exPressed 
PerhAPs by A sourCe 
doCumeNT, whereAs 
No suCh TruTh exisTs 
iN The CAse oF skeTCh, 
whiCh is iNhereNTly 
vAgue

SketchUp extends them to the third spatial dimension. 

In essence, sketch tools would allow the user to edit a GIS database by 
inserting new point, line, and area features. These might be added to existing 
feature classes or might be captured as entirely new classes. For example, a 
user seeking the best locations for a number of new retail stores to add to 
an existing chain might sketch potential locations. The system then might 
evaluate these locations based on a predictive model of store sales, or use 
them as the starting points for an optimal search procedure. From this 
perspective, sketch tools are augmentations of existing GIS database editing 
functions. But the emphasis is rather different; in conventional editing, the 
user seeks to approximate a truth, as expressed perhaps by a source document, 
whereas no such truth exists in the case of sketch, which is inherently vague.

si m u l at i o n   t o o l s

The second area is simulation, or the examination of design scenarios 
by simulating their impacts based on sound scientific knowledge. For 
example, the impacts of a proposed new highway might be examined by 
simulating its effects on the pattern of traffic in the surrounding area; on 
the downstream effects on local hydrology; and on noise and atmospheric 
pollution. Calibrated models exist for each of these sets of impacts as a result 
of basic scientific research. Moreover, in many areas large numbers of models 
exist, based on different sets of assumptions, requiring different inputs, 
and yielding different answers. One of the most valuable outputs of a GIS 
simulation may lie in the uncertainties associated with predictions, based on 
uncertainties within each model and on variation across models.

Many successful efforts have been made to integrate models of social 
and environmental processes into GIS, and the results are described in a 
substantial literature (Goodchild, Parks, and Steyaert 1993; Skidmore 2002). 
Many models are difficult to integrate with GIS and with other models 
because of lack of standards governing data formats, and there is also a 
need for greater standardization in the languages in which model software 
is written. Both of these factors impede the goals of geodesign, because 
they make it difficult to implement many models as simple functions of a 
geodesign software environment.

More specifically, research is needed to address two issues of major 
importance:

1. Models need to be encapsulated easily within GIS, so that they can be 
executed and the results analyzed within the workflow of a geodesign 
process. This implies that data inputs and outputs need to follow GIS data 
format standards so they can be integrated readily with GIS databases, 
and that model parameters be exposed to the user through a GIS 
interface.

2. Models need to be written in a common language, so that their 

component parts can be reassembled and reused readily. In practice, 

Cartographic Perspectives, Number 66, Fall 2010

Article Title  Author Name(s) |  65

models are commonly written in a range of computing environments, 
from source languages such as C++ to scripting languages such as Python. 
Efforts to develop a common, uniform language for GIS have made only 
limited progress in the past (e.g., the Map Algebra of Tomlin 1990), 
though van Deursens scripting language for PCRaster (van Deursen 
1995) offers a comprehensive solution at least for simulations over a 
cellular landscape. A comprehensive solution to this problem would be a 
major contribution to the goals of geodesign.

Behind this need for a common language lies a much more fundamental 
problem, that of defining a standard set of GIS operations. While various 
taxonomies have been published, it is regrettably true that after 45 years of 
GIS development there exists no standard set that is defined on a rigorous 
conceptual basis. Instead, the sets of functions offered by popular GIS 
packages, such as the ArcToolbox, are the result of a haphazard historical 
process of development. There are no universal standards and no rigorous 
concept of granularity, making it difficult to discover functions offered on the 
Web and undermining the entire concept of service-oriented architecture. A 
conceptual framework for the structuring of GIS functionality would be an 
enormously valuable contribution, enabling a new level of interoperability 
across the GIS field.

D I S C U S S I O N   A N D   C O N C L U S I O N

It will be obvious from the preceding sections that geodesign is not new, but 
instead represents a re-examination and perhaps a repurposing of a number 
of established fields. In the case of GIS, this re-examination is prompted by 
a perceived lack of attention to the use of GIS in design, and to its potential 
role in improving the geographic world. In the case of spatial optimization, it 
is prompted by the perception that design problems are more complex than 
simple mathematical formulations, and that the political process of decision 
making is more complex than the execution of a single optimization. In the 
case of landscape architecture, it is prompted by the notion that science can 
play a much stronger role in informing important decisions over the use of 
land, and that GIS is a valuable platform for integrating scientific knowledge 
into the design process.

The design of tools is driven by a constant tension between the specific and 
the general: between the scale economies that result from a one-size-fits-all 
solution, and the speed with which a targeted solution to a specific problem 
can be constructed. In the 1970s, GIS emerged as a generic solution to a set 
of requirements that ranged from cartographic editing to land-use planning 
and the administration of the census. Today, a suite of integrated geodesign 
tools may emerge from the realization that a host of geographic design 
problems share a common structure, and rely on access to a common GIS 
database. Just as with GIS, the attendant economies of scale in software 

TodAy, A suiTe 
oF iNTegrATed 
geodesigN Tools 
mAy emerge From 
The reAlizATioN 
ThAT A hosT oF 
geogrAPhiC 
desigN Problems 
shAre A CommoN 
sTruCTure, ANd 
rely oN ACCess 
To A CommoN gis 
dATAbAse

66  |  Article Title  Author Name(s)

Cartographic Perspectives, Number 66, Fall 2010

Now more ThAN 
ever, we seNse The 
Need For eFFeCTive 
Tools ThAT CAN 
helP us To eNsure A 
desirAble FuTure For 
The PlANeT, ANd gis 
CleArly CoNTAiNs 
The FouNdATioN For 
suCh Tools

production, training, and documentation would be enormous.

For that to happen, however, several issues have to be resolved. Two have 
already been mentioned: the lack of interoperability between existing model 
codes, and the lack of a language within which an integrated vision of 
reusable codes could be implemented. In addition, however, it is important to 
address the question of how a suitable computing environment, once defined, 
might be widely adopted. Several successful models can be found in the 
history of GIS:

1. The commercial software route. Commercial software developers have the 

development staffs and the necessary mechanisms for promotion, training, 
and support to turn a design into a widely adopted reality. An advisory 
group of geodesigners might define the framework, and ensure that it was 
successfully implemented in tools.

2. The open-source route. GRASS was an early and highly successful effort 

to develop a comprehensive GIS for environmental modeling, based 
on open-source code and a network of researchers who added routines 
within a loosely defined set of standards. The role of the US Army Corps 
of Engineers in providing the initial foundation was critical, and suggests 
that a suitable strategy would be to obtain a major grant from a funding 
agency to construct the framework and to build the initial community of 
contributors and users.

3. The research center route. GeoDa (http://geodacenter.asu.edu) is another 
example of a highly successful package of tools, in this case addressing the 
needs of social scientists for easy-to-use software for spatial analysis. It 
was developed under a major center grant from the US National Science 
Foundation, which funded not only the code but also tutorials and 
workshops that publicized its applications.

Design was clearly an early objective of GIS, but as argued earlier, it tended 
to lose its centrality as GIS evolved to serve more lucrative and immediate 
markets. Now more than ever, we sense the need for effective tools that can 
help us to ensure a desirable future for the planet, and GIS clearly contains 
the foundation for such tools. The concept of geodesign presents a simple 
banner for a renewed effort to emphasize the value of cartography and GIS 
as tools for improving and sustaining the surface and near-surface of the 
Earth.

ACkN O W L E D G E M E N T

This paper has benefited from numerous discussions over the past few 
months, most notably with Bill Miller (ESRI), Naicong Li (Redlands 
Institute), and Carl Steinitz (Harvard University). The opinions and 
conclusions expressed are entirely those of the author, however.
 

Cartographic Perspectives, Number 66, Fall 2010

Article Title  Author Name(s) |  67

RE F E R E N C E S

Couclelis, H. 1992. People manipulate objects (but cultivate fields): beyond 
the raster-vector debate in GIS. In A.U. Frank and I. Campari, eds., Theories 
and Methods of Spatio-Temporal Reasoning in Geographic Space. Lecture Notes 
in Computer Science 639:6577. Berlin: Springer.
Cowen, D.J. 1988. GIS versus CAD versus DBMS: What are the 
differences? Photogrammetric Engineering and Remote Sensing 54:15511554.
Densham, P.J. and M.F. Goodchild. 1990. Research Initiative 6: Spatial 
decision support systems: report for the specialist meeting. Technical Report, 
90-5. Santa Barbara, CA: National Center for Geographic Information and 
Analysis.
Foresman, T.W. 1998. The History of Geographic Information Systems: 
Perspectives from the Pioneers. Upper Saddle River, NJ: Prentice Hall PTR.
Gallo, J. 2007. Engaged conservation planning and uncertainty mapping 
as a means towards effective implementation and monitoring. PhD diss., 
University of California, Santa Barbara.
Ghosh, A. and G. Rushton, eds. 1987. Spatial Analysis and Location-
Allocation Models. New York: Van Nostrand Reinhold.
Glennon, J.A. 2010. Creating and validating object-oriented data models: 
modeling flow within GIS. Transactions in GIS 14(1): 2342.
Goodchild, M.F. 1988. Stepping over the line: technological constraints and 
the new cartography. American Cartographer 15:311319.
Goodchild, M.F. 1992. Geographic information science. International Journal 
of Geographical Information Systems 6(1): 31-45.
Goodchild, M.F., B.O. Parks, and L.T. Steyaert, eds. 1993. Environmental 
Modeling with GIS. New York: Oxford.
Goodchild, M.F., M. Yuan, and T.J. Cova. 2007. Towards a general theory 
of geographic representation in GIS. International Journal of Geographical 
Information Science 21(3): 239260.
Hof, J.G. and M. Bevers. 2002. Spatial Optimization in Ecological 
Applications. New York: Columbia University Press.
Leung, Y., 1998. Intelligent Spatial Decision Support Systems. Berlin: Springer 
Verlag.
Malczewski, J. and C. Rinner. 2010. Multicriteria Decision Analysis in 
Geographic Information Science. Berlin: Springer.
McHarg, I.L. 1969. Design with Nature. Garden City, NY: Natural History 
Press.

68  |  Article Title  Author Name(s)

Cartographic Perspectives, Number 66, Fall 2010

McHarg, I.L. 1996. A Quest for Life: An Autobiography. New York: Wiley.
Skidmore, A. 2002. Environmental Modelling with GIS and Remote Sensing. 
New York: Taylor and Francis.
Steinitz, C. 1990. A framework for theory applicable to the education 
of landscape architects (and other environmental design professionals). 
Landscape Journal 9:136143.
Steinitz, C., H. Arias, S. Bassett, M. Flaxman, T. Goode, T. Maddock, D. 
Mouat, R. Peiser, and A. Shearer. 2003. Alternative Futures for Changing 
Landscapes: The San Pedro River Basin in Arizona and Sonora. Washington, 
DC: Island Press.
Sugumaran, R. and J. Degroote. 2010. Spatial Decision Support Systems. Boca 
Raton, FL: CRC.
Sui, D.Z. and M.F. Goodchild. 2001. Guest Editorial: GIS as media? 
International Journal of Geographical Information Science 15(5): 387389.
Takeyama, M. and H. Couclelis. 1997. Map dynamics: integrating 
cellular automata and GIS through Geo-Algebra. International Journal of 
Geographical Information Science 11:7391.
Thill, J.-C. 1999. Spatial Multicriteria Decision Making and Analysis: A 
Geographic Information Sciences Approach. Brookfield, VT: Ashgate.
Thorne, J.H., D. Cameron, and J.F. Quinn. 2006. A conservation design for 
the central coast of California and the evaluation of mountain lion as an 
umbrella species. Natural Areas Journal 26:137148.
Tomlin, C.D. 1990. Geographic Information Systems and Cartographic 
Modeling. Englewood Cliffs, NJ: Prentice Hall.
van Deursen, W.P.A. 1995. Geographic Information Systems and Dynamic 
Models: Development and Application of a Prototype Spatial Modelling 
Language. PhD Dissertation, Utrecht University.
Zeiler, M. 1999. Modeling Our World: The ESRI Guide to Geodatabase Design. 
Redlands, CA: ESRI Press.

Cartographic Perspectives, Number 66, Fall 2010

Article Title  Author Name(s) |  69