Spatial system dynamics

[MOUNTS Tony]

Ilya,

In my research on growth management models, I ran across a paper in the 93
SD conference proceedings titled: Regional Land Use and Infrastructure
Dynamics by Norman Marshall and Stephen Lawe. Their model looked at the
inter-relations between land use and transportation systems. They built a
model of a region in the Maine/New Hampshire boarder region based on GIS
land use data and a transportation model that included trip generation,
distribution over the region, and mode split, among others. The paper
references continuing work in the Tampa region and the possibility of other
such projects.

The listed address is: Resource Systems Group, P.O. Box 1499, Norwich VT
05055

Someone on this list may know whether the authors are still involved in this
work.

Im interested in the development of urban models. Id appreciate learning
more as your project progresses.

Tony Mounts
City of Eugene
Tony.Mounts@ci.eugene.or.us
----------
|
|

[Nathan Forrester <74654.201@Comp]

A few years ago I built a gepgraphically disaggregated model of urban change for
a consulting client in Paris. The model breaks down land area into an arbitraily
large number of fixed parcels on a grid. Each parcel maintains its own
accounting of Residents, Housing, Jobs, Businesses, Occupancy, Density, Job
Availability, Housing Availability, etc. Each parcel interacts with all others
to determine the movement of people and businesses into, out of, and within the
urban area. The trick is in the equations of interaction.

The model was designed to address the feasibility of Prince Charles idea of a
"multicellular" city where people live and work in the same neighborhoods of
limited size (+/- 100,000 inhabitants). The model also explained the effect of
limiting the density of a city core (ala Paris) on surrounding patterns of
decay, congestion, and unemployment. It had no GIS links.

If you are interested in more information on the model, contact me.

Nathan

Nathan Forrester, President
Forrester Consulting
100 Galleria Parkway, Suite 400
Atlanta, GA 30339

Phone: (770) 956-4030
Fax: (770) 956-4031
e-mail: 74654.201@compuserve.com
WWW: http://www.mindspring.com/~nathanf/

["Kent D. Palmer, Ph.D."]

Hi--

Could someone explain how a spatial SD model works. How would one create
a model that is both spatial and temporal? Has any one done that? Is it
possible to have multiple concurent SD models that address different
dimensions?

I find this whole line of investigation fascinating. I never heard of it
before.

Kent

----------------------------------
Kent D. Palmer, Ph.D.
palmer@netcom.com

[Ilya Zaslavsky]

Dear colleagues,

I am new to this group, sorry of this problem has been discussed before.

I am trying to build a simulation model to be used in zoning and land
use management. The model is supposed to trace consequences of land
zoning decisions, and of other events affecting spatial organization of
cities/towns. Zoning and land use information is typically managed with
geographic information systems (GIS). So, I am trying to connect the
system dynamics model with a GIS.

Now, problems:
1. GIS typically dont handle temporal data, at most it is a series of
snapshots (maps) for several dates. Do you know of any attempts to
incorporate continuous time into a GIS, or to incorporate maps
and spatial relations into a system dynamics software? Of course, I can
create a model for a land parcel, duplicate it for other land parcels
(using array and indexing capabilities of some simulation software), and
describe relations between centroids of these parcels. But parcels get
created, subdivided, or eliminated during zoning. Besides, boundary
relations, topological relations, diffusion are beyond this approach.

2. I understand that the main tenet of system dynamics is that a
systems structure determines behavior. What about spatial structure,
and relations resulting from relative location of objects in space?
Such relations would affect some parts of these objects (boundaries, or
along transportation routes) more than others. Can such relations be
represented within system dynamics?

Has anyone looked into this "spatial" system dynamics and faced similar
problems? I will appreciate very much any references or thoughts on
these issues.

Ilya Zaslavsky
Western Michigan University
zaslavsky@wmich.edu

[saeed@ait.ac.th (Professor Khali]

There seem to be two fortunate principles in SD modeling which make it
quite amenable to representing spatial changes:

1. Spatial changes occur over time

2. decision rules depend only on current information and not on time.

Forresters Urban Dynamics model has an implicit spatial dimension since it
tracks various types of neighborhoods and industry. Its application to
specific cases as attempted by Lou Alfeld and others also represent spatial
dynamics more explicitly in terms of changes in land use in the center city
and its suburbs.

A few years ago, one of my former students addressed the problem of
unbalanced regional growth in Thailand using a SD model. Another linked
land use in Pattaya to a map of spatial changes using StellaStack which
produced a delightful spatial display that changed in real time when
simulated.

A GIS data base could possibly help to delineate a spatial reference mode
for a modeling exercise. The model, based primarily on behavioral
considerations could then be used to study future spatial scenarios and
search for behavioral policies that might achieve desired scenarios. I must
think harder to conceive other types of linkages between SD and GIS, but I
found this this be a stimulating query.

Khalid

Khalid Saeed
Professor and Program Coordinator
Infrastructure Planning & Management Program
School of Civil Engineering
ASIAN INSTITUTE OF TECHNOLOGY
G.P.O. Box 2754, Bangkok, THAILAND
phones: (66-2)524-5681, (66-2)524-5785; fax: (66-2)524-5776
email saeed@ait.ac.th

Visit our program website at: http://www.ipm.ait.ac.th/

[jimhines@interserv.com]

On Sat, 28 Sep 1996, saeed@ait.ac.th (Professor Khalid Saeed) wrote:
>There seem to be two fortunate principles in SD modeling which make it
>quite amenable to representing spatial changes:
>
>1. Spatial changes occur over time

.... [major snip] ...
......
> I must
>think harder to conceive other types of linkages between SD and GIS, but I
>found this this be a stimulating query.


On a less practical plane, there is another way to think about the spatial
connection:

We usually think about the indendent variable in our models as representing
time. But, there is no reason that it couldnt represent a spatial
dimension, instead. If for example, you were interested in how the
characteristics of a river change as a result of being connected spatially,
(rather than temporally), you could have the "time unit" of the model be,
say, miles. I heard once that Barry Richmond had constructed such a model.

This would be something like having a continous-space finite element model
operating in one dimension and only "down stream" (i.e. there can be no
upstream influence).

Jim Hines
LeapTec and MIT
JimHines@Interserv.Com

[davet@NREL.ColoState.EDU (David]

Ilya and others:

I have seen a few articles exploring spatial system dynamics:

Costanza, R., Sklar, F.H., and White, M.L. 1990. Modeling Coastal Landscape
Dynamics. BioScience 40:91-107.
This approach uses a Stella model which describes the interactions of one
cell of a spatial grid, and outputs the differential equations into a GIS
to run it spatially. A neat approach, but I think the iterative and
interactive power of system dynamics (like Stella II) is lost here.

Ruth, M. and Pieper, F. 1994. Modeling Spatial Dynamics of Sea-level Rise
in a Coastal Area. System Dynamics Review 10:375-389.
Shows how Stella can be used in a spatial sense, but this approach is
limitied to *small* spatial systems.

What I think is needed is a fuller fusion of system dynamics and GIS
concepts, not simply an integration of the two. I outlined some of my
thoughts in the following paper:

Theobald, D.M. and Gross, M.D. 1994. EML: A Modeling Environment for
Exploring Landscape Dynamics. Computers, Environment and Urban Systems
18:193-204.


The great opportunity of spatial system dynamics efforts is to provide a
means to *easily* and *rapidly* define a system and see the consequences of
different assumptions (the kind that Stella II makes easy). For the last
year or so, I have been programming a land use change model in Arc/Info AML
(the predominant GIS system). Many times I have felt I have lost the forest
for the trees, taking over a week sometimes to program the changes in a
model, only to forget why I was creating the model in the first place.
Frustrating! The emphasis in current GIS software and research is on data
- data accuracy and manipulation of very large spatial data sets, worthy
topics, but I think users conceptual understanding of system structure is
lost.


Cheers,

Dave Theobald
_____________________________

David M. Theobald
Natural Resource Ecology Laboratory, NESB A210
Colorado State University
Fort Collins, CO 80523-1499
(970) 491-5122 FAX (970) 491-1965
davet@NREL.colostate.edu

[GregNorris@aol.com]

Dear all,

The folks at Resource Systems Group are still active in the area of
spatial SD models of urban systems, with a strong transportation modeling
emphasis.
One could contact Norman Marshall: nmarshall@rsginc.com.
Regards,
Greg Norris
504 Nelson Drive
Vienna, VA 22180
703-319-3944
703-319-3943 (fax)
GregNorris@aol.com