Modelling Environmental Dynamics: Advances in Geomatic Solutions

405 Pages·2008·23.463 MB·English

by  Martin Paegelow, María Teresa Camacho Olmedo

Environmental Science and Engineering
Subseries: Environmental Science
Series Editors: R. Allan • U. Förstner • W. Salomons
·
Martin Paegelow María Teresa Camacho Olmedo (Eds.)
Modelling Environmental
Dynamics
Advances in Geomatic Solutions
With 122 Figures, 17 in color
Prof.MartinPaegelow Prof.MaríaTeresaCamachoOlmedo
UniversitédeToulouseII Universidadde Granada
GEODEUMR5602CNRS Dpto.de AnálisisGeográfico
MaisondelaRecherche Regional y Geografía Física
5alléesAntonioMachado CampusdeCartuja,s/n
31058Toulouse 18071Granada
France Spain
[email protected] [email protected]
ISBN:978-3-540-68489-3 e-ISBN:978-3-540-68498-5
EnvironmentalScienceandEngineeringISSN:1863-5520
LibraryofCongressControlNumber:2008931312
(cid:2)c 2008Springer-VerlagBerlinHeidelberg
Thisworkissubjecttocopyright.Allrightsarereserved,whetherthewholeorpartofthematerialis
concerned,specificallytherightsoftranslation,reprinting,reuseofillustrations,recitation,broadcasting,
reproductiononmicrofilmorinanyotherway,andstorageindatabanks.Duplicationofthispublication
orpartsthereofispermittedonlyundertheprovisionsoftheGermanCopyrightLawofSeptember9,
1965,initscurrentversion,andpermissionforusemustalwaysbeobtainedfromSpringer.Violationsare
liabletoprosecutionundertheGermanCopyrightLaw.
Theuseofgeneraldescriptivenames,registerednames,trademarks,etc.inthispublicationdoesnotimply,
evenintheabsenceofaspecificstatement,thatsuchnamesareexemptfromtherelevantprotectivelaws
andregulationsandthereforefreeforgeneraluse.
CoverDesign:Integra Software Services Pvt. Ltd.
Printedonacid-freepaper
9 8 7 6 5 4 3 2 1
springer.com
Foreword

Humans have always created and used environmental models, whether the
models are expressed as traditional fables that are designed to convey an-
cient wisdom concerning how we should interact with nature or as com-
puterized algorithms that are designed to give advice concerning the how
we should draft environmental legislation. In both cases, the models ex-
press what we think we know about human-environment interactions and
what the implications are for natural resource management. In both cases,
scenarios are commonly used as communication device, since a scenario is
a story about the way the future could unfold, based on the consequences
of human decisions. A major difference between today’s models and pre-
vious models is that digital technology is now available to help to organize
the story teller’s thoughts and to communicate the teller’s messages. To-
day’s digital modelling techniques can have substantial influence on both
the people designing the stories, e.g. the scientists, and the people inter-
preting the stories, e.g. the decision makers and the public.
Some of today’s models, such as those that produce weather forecasts,
are so common that both the scientists and the public seem to interpret
them with ease. For example, people who have access to the internet can
easily see a map of a weather forecast and usually know intuitively how
much trust they can have in the prediction for their particular region. We
has been able to develop this intuition because we can engage in a fre-
quently repeated validation exercise by seeing such forecasts and compar-
ing them to the weather that we personally experience. We have had much
less direct experience with other types of simulation models that produce
maps. For example, some models produce maps that predict the environ-
mental impacts of anticipated global climate change. It is not immediately
clear how humans should interpret maps from models that attempt to pre-
dict phenomena that have never happened before. Even scientist them-
selves are challenged by creating and interpreting models concerning proc-
esses that either have not occurred before or occur gradually over long
time scales, such as climate change and land change. Nevertheless, it is
precisely these types of changes that are important to model, because many
of their consequences are large and practically irreversible.
This book presents some of the most recent work in modelling the inter-
action between humans and the environment, especially where anthropo-
genic land change is a central focus. Land change is important to model due
its particular characteristics. First of all, many of the effects of land change
vi Foreword
are physically irreversible, such as loss of biodiversity, deposition of haz-
ardous wastes, or construction on high quality agricultural soil. For those
land transitions that are theoretically reversible, many are unlikely to be re-
versed for social reasons, since many actors become rapidly invested finan-
cially and legally in an established land use pattern. Furthermore, land use
has substantial implications for how individual lifestyle choices are con-
strained. For example, the landscape in the United States is designed for
travel by car, not by bicycle or by foot. The consequences are that green-
house gas emissions from American cars have become a threat to global
health, and obesity in the United States has become a primary threat to
Americans’ personal health. Many other countries are presently developing
similar landscapes, and are likely to face similar consequences if they
choose to follow the American style of suburbanization and urban sprawl.
How should scientist and the public address such issues? Land change
modelling offers a potentially useful set of tools, while such modelling has
its challenges. Scientists, policy makers, and the public do not have as
much experience in interpreting output from land change models over sev-
eral decades as they do for other types of forecasts. Land transformation
processes are relatively slow compared to weather changes, so it is difficult
for scientists and non-scientist to develop an intuitive feel for how much
trust we should have in such models. Calibration of these models can be
complicated in some places where we would like to simulate a phenomenon
that has never occurred before, such as building entirely new roads through
virgin forests. Validation can be complicated by the fact that the processes
during the calibration time interval may be different than the processes dur-
ing the validation time interval, due to a sudden change in agricultural pol-
icy for example. Moreover, there are a seemingly infinite number of ways
to calibrate a model and to measure its accuracy during a validation step.
Some of the measurements of accuracy that are most intuitive initially, such
as percent of pixels classified correctly, can be extremely misleading. The
method by which the model is used to inform policy may be even more im-
portant than the particular computer algorithm used. It is helpful when
models can offer insight to a wide variety of decision-makers, while we
cannot expect non-experts to grasp immediately the differences among neu-
ral nets, cellular automata, agent based models, fuzzy logic, and logistic re-
gression. However, if the model design allows for it to be used in a partici-
patory fashion, then a larger number of people can be involved in an
interactive process of modelling and decision-making.
Regardless of what you may think of environmental models, they are
here to stay and will become even more important in the coming decades.
Land change models in particular will become ever more influential as our
global society wrestles with policies to reduce greenhouse warming. One
Foreword vii
proposed policy calls to implement carbon offset projects that would offer
financial incentives to reduce deforestation. This plan is being carried out
via a carbon credit trading system, where a credit is awarded for a conser-
vation project that prevents deforestation that would have occurred, had it
not been for the project. How do we compute how much deforestation the
conservation project prevented from occurring? The answer is land change
modelling. In the coming years, billions of dollars will be exchanged an-
nually for greenhouse gas credits that are based on output from such envi-
ronmental models. There is big money on the line based on the output
from environmental models. More importantly, the viability of our most
precious ecosystems is also on the line.
Some critics express concern that such models contain substantial uncer-
tainties. Uncertainties will always exist in any type of model. But one thing
is for certain: if we wait for the uncertainties of these models to be elimi-
nated before we implement policy to preserve valuable ecosystems, then
those ecosystems and their accompanying services will be lost, because the
processes of land disturbance are already well underway in some of our
planet’s most important locations. Therefore, we need methods to design, to
use, and to interpret models in intelligent ways that both appreciate the
level of the model’s uncertainty and acknowledge the urgency of our envi-
ronmental challenges. This book marks a major step forward in advancing
the agenda concerning the implementation and interpretation of environ-
mental models. Whereas our traditional models in the form of fables ex-
pressed general guiding principles for behaviour, we need for our next gen-
eration of models in the form of computer code to give specific guidance
concerning our options for environmental management. Hopefully, humans
will learn how to use these models to integrate and to communicate impor-
tant lessons of ancient and modern wisdom, before it is too late.

Robert Gilmore Pontius Jr.

Department of International Development, Community and Environment,
Graduate School of Geography, Clark University, Worcester, United States
of America
http://www.clarku.edu/~rpontius
Acknowledgements
The editors are grateful to the Ministerio de Educación y Ciencia (BIA
2003-01499), Plan Nacional de Investigación Científica, Desarrollo e In-
novación Tecnológica and the FEDER, whose funding for supported this
research. The editors are also grateful to ECOS (Evaluation-orientation de
la COpération Scientifique), Programme de coopération et de partenariat
universitaire technique et scientifique France – Amérique latine, action
MH5AH4 France – Mexique.
The editors are very thankful to all of the contributing authors. The
strength of this book lies in the international cooperation, which creates a
rich diversity of modelling approaches and themes.
We are also indebted to Emily Zack for her in-depth English revision
and to Enrique Fernandez Seguí for his patient work formatting the text.
Finally, the editors are obliged to their research centres: GEODE UMR
5602 CNRS, University of Toulouse II, France and Department of Análisis
Geográfico Regional y Geografía Física, University of Granada, Spain.
Preface
This book begins with the observation that modelling environmental dy-
namics becomes more urgent everyday. It becomes all the more pressing
because modelling results can contribute to a better understanding of cur-
rent, complex phenomena and time projections using predictions and sce-
narios which can resolve many of the challenges that occur in daily life:
global climatic changes, biodiversity, deforestation, risk prevention and
land planning at the local level, etc. A great deal of research has been
completed. In spite of this, actual modelling tools remain exploratory
rather than operational and most of them can’t be applied in common pol-
icy instruments intended to avoid or even resolve the above-mentioned
problems. Also today’s accessibility of user-friendly modelling tools also
brings with it some risks; it seems easy to model practically anything. Par-
ticularly newsworthy topics such as global change links to the increase of
temperature, the melting of ice caps or the discussed cessation of the Gulf
Stream have become subject to numerous publications using models and
predicting more or less catastrophic simulations. In spite of numerous, se-
rious research results and crucial planetary stakes we have to remain hum-
ble and critical and ask ourselves about the degree of our understanding of
complex environmental systems, the amount and the quality of the data
used and the validation of model results. These are major topics in this
book, which presents various research results in modelling environmental
dynamics in a transparent way focussing on result validation.
What is this book about?
The main objective of this book is to contribute to advances in modelling en-
vironmental dynamics involving both: the spatial and the temporal dimen-
sion. The goal is to perform simulations either as probabilistic predictions or
scenarios showing ‘what will be if’. The aims of modelling are various and
cover a wide range stretching from better comprehension to decision support.
During the last few years, the modeller’s toolbox has become signifi-
cantly enriched by novel methods such as fuzzy logic, multi-agent systems
or neural networks to resolve geographical problems. This book, starting
with an introductory overview about the challenges and modelling ap-
proaches, provides a sample of actual research results using a variety of
modelling methods and tools applied to an assortment of environmental
dynamic situations. It also shows a wide range of model results and topical
modelling conceptualisation like participatory modelling. All of these
xii Preface
contributions follow the same structure and emphasize mainly the meth-
odological aspects such as model calibration and model validation.
What is this book not about?
Modelling is, from a conceptual point of view, a current and important is-
sue in many research areas. A lot of concepts and methods are emerging.
This book doesn’t offer new conceptual or methodological advances but it
shows validated modelling results based on innovative methods like neural
network, multi-agent system, cellular automaton, fuzzy modelling and
more traditional, mostly stochastic, approaches.
For this reason, this is a first actual set of case studies, and some theo-
retical aspects can’t be discussed in depth. However, in part A of this
book, there is an attempt at creating a synoptic summery with numerous
references to help the user find further reading.
However with the harmonized presentation of the contributions, this
book is neither a manual nor a tutorial.
How to use this book
This book is written for academics, students and professionals belonging to
a wide range of disciplines like geography, geomatics, environmental sci-
ences, land planning and urbanism with at least an initial experience with
spatio-temporal data, GIS and modelling. It also may be a welcome appli-
cation example for specialists the in computer sciences dealing with spa-
tio-temporal data. The gradual concept of the book and the presentation of
performed research results, which are presented using the same structural
set-up in each chapter, may make it useful for more thematic experts too.
Since the audience has various levels of knowledge and experience in
geomatics and modelling and different academic and professional back-
grounds, the book starts with a succinct overview about modelling (what,
with what and for what?). Advanced readers may skip this introductory
part and turn their attention directly to the following case studies. Each of
them provides a large list of references for further reading.
Structure of the book
This book contains two main parts: a brief introduction to modelling and a
set of case studies.
Part A launches basic ideas about modelling environmental dynamics
starting with its challenge. In this chapter the reader will find the scientific
context of this work and its objectives: What? Environmental dynamics.
With what? Geomatics solutions. For what? Outcome, modelling for simu-
lation. This first chapter is completed with a summary of the opportunities
created by this book as well as references to some earlier works.