Quantitative Information Fusion for Hydrological Sciences

224 Pages·2008·7.562 MB·English

by  Linda See (auth.), Xing Cai, T. -C. Jim Yeh (eds.)

XingCaiandT.-C.JimYeh(Eds.)
QuantitativeInformationFusionforHydrologicalSciences
StudiesinComputationalIntelligence,Volume79
Editor-in-chief
Prof.JanuszKacprzyk
SystemsResearchInstitute
PolishAcademyofSciences
ul.Newelska6
01-447Warsaw
Poland
E-mail:[email protected]
Furthervolumesofthisseries Vol.68.CiprianoGalindo,Juan-Antonio
canbefoundonourhomepage: Ferna´ndez-MadrigalandJavierGonzalez
springer.com AMulti-HierarchicalSymbolicModel
oftheEnvironmentforImprovingMobileRobot
Operation,2007
Vol.58.Jeng-ShyangPan,Hsiang-ChehHuang,Lakhmi
ISBN978-3-540-72688-3
C.JainandWai-ChiFang(Eds.)
IntelligentMultimediaDataHiding,2007 Vol.69.FalkoDresslerandIacopoCarreras(Eds.)
ISBN978-3-540-71168-1 AdvancesinBiologicallyInspiredInformationSystems:
Models,Methods,andTools,2007
Vol.59.AndrzejP.WierzbickiandYoshiteru
ISBN978-3-540-72692-0
Nakamori(Eds.)
CreativeEnvironments,2007 Vol.70.JavaanSinghChahl,LakhmiC.Jain,Akiko
ISBN978-3-540-71466-8 MizutaniandMikaSato-Ilic(Eds.)
InnovationsinIntelligentMachines-1,2007
Vol.60.VladimirG.IvancevicandTijanaT.Ivacevic ISBN978-3-540-72695-1
ComputationalMind:AComplexDynamics
Perspective,2007 Vol.71.NorioBaba,LakhmiC.JainandHisashiHanda
ISBN978-3-540-71465-1 (Eds.)
AdvancedIntelligentParadigmsinComputer
Vol.61.JacquesTeller,JohnR.LeeandCatherine Games,2007
Roussey(Eds.) ISBN978-3-540-72704-0
OntologiesforUrbanDevelopment,2007
ISBN978-3-540-71975-5 Vol.72.RaymondS.T.LeeandVincenzoLoia(Eds.)
ComputationIntelligenceforAgent-basedSystems,2007
Vol.62.LakhmiC.Jain,RaymondA.Tedman ISBN978-3-540-73175-7
andDebraK.Tedman(Eds.)
EvolutionofTeachingandLearningParadigms Vol.73.PetraPerner(Ed.)
inIntelligentEnvironment,2007 Case-BasedReasoningonImagesandSignals,2008
ISBN978-3-540-71973-1 ISBN978-3-540-73178-8
Vol.63.WlodzislawDuchandJacekMan´dziuk(Eds.) Vol.74.RobertSchaefer
ChallengesforComputationalIntelligence,2007 FoundationofGlobalGeneticOptimization,2007
ISBN978-3-540-71983-0 ISBN978-3-540-73191-7
Vol.75.CrinaGrosan,AjithAbrahamandHisao
Vol.64.LorenzoMagnaniandPingLi(Eds.)
Ishibuchi(Eds.)
Model-BasedReasoninginScience,Technology,and
HybridEvolutionaryAlgorithms,2007
Medicine,2007
ISBN978-3-540-73296-9
ISBN978-3-540-71985-4
Vol.76.SubhasChandraMukhopadhyayandGourab
Vol.65.S.Vaidya,L.C.JainandH.Yoshida(Eds.)
SenGupta(Eds.)
AdvancedComputationalIntelligenceParadigmsin
AutonomousRobotsandAgents,2007
Healthcare-2,2007
ISBN978-3-540-73423-9
ISBN978-3-540-72374-5
Vol.77.BarbaraHammerandPascalHitzler(Eds.)
Vol.66.LakhmiC.Jain,VasilePaladeandDipti PerspectivesofNeural-SymbolicIntegration,2007
Srinivasan(Eds.) ISBN978-3-540-73953-1
AdvancesinEvolutionaryComputingforSystem
Design,2007 Vol.78.CostinBadica(Ed.)
ISBN978-3-540-72376-9 IntelligentandDistributedComputing,2008
ISBN978-3-540-74929-5
Vol.67.VassilisG.KaburlasosandGerhardX.Ritter
(Eds.) Vol.79.XingCaiandT.-C.JimYeh(Eds.)
ComputationalIntelligenceBasedonLattice QuantitativeInformationFusionforHydrological
Theory,2007 Sciences,2008
ISBN978-3-540-72686-9 ISBN978-3-540-75383-4
Xing Cai
T.-C. Jim Yeh
(Eds.)
Quantitative Information Fusion
for Hydrological Sciences
With81Figuresand7Tables
123
XingCai T.-C.JimYeh
SimulaResearchLaboratory DepartmentofHydrology
P.O.Box134 andWaterResources
1325Lysaker TheUniversityofArizona
Norway Tucson,Arizona85721
[email protected] USA
and [email protected]
DepartmentofInformatics
UniversityofOslo
P.O.Box1080Blindern
0316Oslo
Norway
ISBN978-3-540-75383-4 e-ISBN978-3-540-75384-1
StudiesinComputationalIntelligenceISSN1860-949X
LibraryofCongressControlNumber:2007937225
(cid:1)c 2008Springer-VerlagBerlinHeidelberg
Thisworkissubjecttocopyright.Allrightsarereserved,whetherthewholeorpartofthematerial
isconcerned,specificallytherightsoftranslation,reprinting,reuseofillustrations,recitation,broad-
casting,reproductiononmicrofilmorinanyotherway,andstorageindatabanks.Duplicationof
thispublicationorpartsthereofispermittedonlyundertheprovisionsoftheGermanCopyrightLaw
ofSeptember9,1965,initscurrentversion,andpermissionforusemustalwaysbeobtainedfrom
Springer-Verlag.ViolationsareliabletoprosecutionundertheGermanCopyrightLaw.
Theuseofgeneraldescriptivenames,registerednames,trademarks,etc.inthispublicationdoesnot
imply, even in the absence of a specific statement, that such names are exempt from the relevant
protectivelawsandregulationsandthereforefreeforgeneraluse.
Coverdesign:WMXDesignGmbH,Heidelberg
Printedonacid-freepaper
9 8 7 6 5 4 3 2 1
springer.com
Preface
WaterisvitalfortheEarthandallthelifeformsonit,thustheimportanceof
hydrology–the science of water–goes without saying. Water resources involve
interplay between geologic, hydrologic, chemical, atmospheric, and biologi-
cal processes. To study the occurrence, movement, distribution, and quality
of water throughout our globe is clearly a challenging task, which requires
a joining force between the different branches of hydrology, from hydrome-
teorology, surface water hydrology to hydrogeology and hydrochemistry as
well as hydrogeophysics and hydroecology, to just mention a few. Besides a
conceptual understanding, quantitative monitoring, characterization, predic-
tionsandmanagementmustresorttocollaborativemathematicalmodelsand
numerical algorithms, often together with computer simulations.
In recent years, massive amounts of high-quality hydrologic field data are
being collected at various spatial-temporal scales using a variety of new tech-
niques. Availability of these massive amounts of data has begun to call for
a quantitative integration of geologic, hydrologic, chemical, atmospheric, and
biologicalinformationtocharacterizeandpredictnaturalsystemsinhydrolog-
ical sciences. Intelligent computation and information fusion as such become
a key to the future hydrological sciences. We envision this subject to become
a new research field that will dramatically improve the traditional approach
of only qualitatively characterizing natural systems.
Thiseditedvolumecontains eightchapterswritten bysomeoftheleading
researchers in hydrological sciences. The chapters address some of the most
important ingredients for quantitative hydrological information fusion. The
bookaimstoprovidebothestablishedscientistsandgraduatestudentswitha
summaryofrecentdevelopmentsinthisnewresearchdirection,whileshedding
some light into the future.
The eight chapters can be divided into three mutually overlapping parts.
ThefirstpartconsistsofChapters1and2whichmainlyaddressthemethod-
ological issues. In particular, Chapter 1 discusses different data fusion tech-
niques for integrating hydrological models, where the discussion is carried
out from the perspective of hydroinformatics and computational intelligence.
VI Preface
Chapter2depictsanadvancedcomputationalenvironmentthatenablesinter-
active and real-time 3D groundwater modeling. The combined power of par-
allel computing, dynamic visualization, and computational steering enables a
fusion of flow modeling, transport modeling, subscale modeling, uncertainty
modeling, geostatistical simulation, and GIS mapping.
Asthesecondthemeofthebook,Chapters3-6concentrateonsomemath-
ematicalandnumericalmethods.UsingtheKalmanfilterbasedonKarhunen-
Loe`ve decomposition, the authors of Chapter 3 show how to reduce the
uncertainty in characterizing hydraulic medium properties and system re-
sponses.Chapter4presentsefficientdataanalysistoolsusingtrajectory-based
methods, which also offer insight into inverse modeling of flow and transport.
In close relation, Chapter 5 describes streamline methods that are capable of
reconciling 3Dgeological models todynamic reservoir responses.Another nu-
merical technique in inverse modeling is given in Chapter 6, which addresses
a systematic regularized inversion approach to incorporating geophysical in-
formation into the analysis of tomographic pumping tests.
The third part of the present book focuses on real-life applications of
hydrological information fusion. Chapter 7 is about using satellite rainfall
datasets and hydrologic process controls for flood prediction in ungauged
basins, whereas Chapter 8 reports an engineering case of groundwater man-
agement by integrating large-scale zoning of aquifer parameters and a sedi-
mentary structure-based heterogeneous description of the aquifer properties.
The idea of the present book was conceived following a warm suggestion
by Prof. Dr. Janusz Kacprzyk, Series Editor of Studies in Computational In-
telligenceatSpringer.WearethereforegreatlyindebtedtoProf.Kacprzykfor
hisadviceandencouragement.EngineeringEditorDr.ThomasDitzingerand
Heather King at Springer’s Engineering Editorial Department, in particular,
deserve our sincere thanks for their patient guidance and technical support
throughout the editorial process. We are of course tremendously grateful to
all the contributed authors for carefully preparing their chapters. Moreover,
positive response from numerous researchers to our call-for-chapters is ac-
knowledged, although they were not able to contribute in the end due to the
tight time schedule.
Last but not least, we wish to express our heartfelt gratitude to a large
number of anonymous reviewers, who carefully read through the earlier ver-
sionsofthebookchaptersandprovidedvaluablesuggestionsforimprovement.
There is no exaggeration in saying that this book project has been a team
workfromstarttofinish.Wesincerelyhopethatthisbookwillgivethereader
an equal amount of pleasure as it has given us during the editing work.
Oslo & Tucson, July 2007 Xing Cai
T.-C. Jim Yeh
List of Contributors
Geoffrey C. Bohling Deepak Devegowda
Kansas Geological Survey Department of Petroleum
University of Kansas Engineering, Texas A&M University
Lawrence, KS 66047 College Station, TX 77843
USA USA
[email protected] [email protected]
Yan Chen Jannis Epting
Mewbourne School of Petroleum Department of Environmental
and Geological Engineering Sciences, Applied and
University of Oklahoma Environmental Geology
Norman, OK 73019 University of Basel, 4056 Basel
USA Switzerland
[email protected] [email protected]
Hao Cheng Faisal Hossain
Department of Petroleum Department of Civil
Engineering and Environmental Engineering
Texas A&M University Tennessee Technological University
College Station, TX 77843 Cookeville, TN 38505
USA USA
[email protected] [email protected]
Akhil Datta-Gupta Peter Huggenberger
Department of Petroleum Department of Environmental
Engineering, Texas A&M Sciences, Applied
University College Station and Environmental Geology
TX 77843 University of Basel, 4056 Basel
USA Switzerland
[email protected] [email protected]
VIII List of Contributors
Nitin Katiyar Texas A&M University
Department of Civil College Station, TX 77843
and Environmental Engineering USA
Tennessee Technological University [email protected]
Cookeville, TN 38505
USA Christian Regli
[email protected] GEOTEST AG, 7260 Davos Dorf
Switzerland
Ralph Kirchhofer [email protected]
Department of Environmental
Sciences, Applied Linda See
and Environmental Geology School of Geography
University of Basel, 4056 Basel University of Leeds Leeds
Switzerland LS2 9JT
[email protected] UK
[email protected]
Shu-Guang Li
Department of Civil Natalie Spoljaric
and Environmental Engineering Department of Environmental
Michigan State University Sciences,AppliedandEnvironmental
East Lansing, MI 48824 Geology University of Basel
USA 4056 Basel
[email protected] Switzerland
natalie.spoljaric@stud.
Qun Liu unibas.ch
Department of Civil
and Environmental Engineering Donald W. Vasco
Michigan State University Lawrence Berkeley National
East Lansing, MI 48824 Laboratory
USA University of California
[email protected] Berkeley, CA 94720
USA
Zhiming Lu [email protected]
Earth and Environmental
Sciences Division Dongxiao Zhang
Los Alamos National Laboratory Department of Civil and Environ-
Los Alamos, NM 87645 mentalEngineeringandMorkFamily
USA DepartmentofChemicalEngineering
[email protected] and Material Sciences
University of Southern California
Dayo Oyerinde Los Angeles, CA 90089
Department of Petroleum USA
Engineering [email protected]
Contents
Data Fusion Methods for Integrating Data-driven
Hydrological Models
Linda See ....................................................... 1
A New Paradigm for Groundwater Modeling
Shu-Guang Li and Qun Liu ....................................... 19
Information Fusion using the Kalman Filter
based on Karhunen-Loe`ve Decomposition
Zhiming Lu, Dongxiao Zhang, and Yan Chen........................ 43
Trajectory-Based Methods for Modeling
and Characterization
D.W. Vasco ..................................................... 69
TheRoleofStreamlineModelsforDynamicDataAssimilation
in Petroleum Engineering and Hydrogeology
Akhil Datta-Gupta, Deepak Devegowda, Dayo Oyerinde,
and Hao Cheng ..................................................105
Information Fusion in Regularized Inversion of Tomographic
Pumping Tests
Geoffrey C. Bohling ..............................................137
Advancing the Use of Satellite Rainfall Datasets for Flood
Prediction in Ungauged Basins: The Role of Scale, Hydrologic
Process Controls and the Global Precipitation Measurement
Mission
Faisal Hossain and Nitin Katiyar ..................................163
Integrated Methods for Urban Groundwater Management
Considering Subsurface Heterogeneity
Jannis Epting, Peter Huggenberger, Christian Regli, Natalie Spoljaric,
and Ralph Kirchhofer.............................................183
Data Fusion Methods for Integrating
Data-driven Hydrological Models
Linda See
School of Geography, University of Leeds
Summary. Thischapterwilladdresstheuseofdifferentdatafusiontechniquesfor
integrating or combining hydrological models. Different approaches will be demon-
stratedusingflowforecastingmodelsfromtheRiverOusecatchmentintheUKfor
aleadtimeof6hours.Theseapproachesincludesimpleaveraging,neuralnetworks,
fuzzy logic, M5 model trees and instance-based learning. The results show that the
datafusionapproachesproducebetterperformingmodelscomparedtotheindivid-
ualmodelsontheirown.Thepotentialofthisapproachisdemonstratedyetremains
largely unexplored in real-time hydrological forecasting.
1 Introduction
Approachestohydrologicalmodelsarevariedandlieonaspectrumthatchar-
acterisesthedegreetowhichtheyencapsulatephysicalprocesses.Ononeend
ofthescalearefullydistributedphysicalmodelsbasedonthelawsofthecon-
servation of energy and mass (e.g. the SHE model of Abbott et al. (1986)).
Conceptual models fall in the middle of the spectrum as parameterisation in-
creaseswhiletheoppositeendisdominatedbydatadrivenmodels(DDM)or
what Wheater et al. (1993) refer to as metric models. As the name suggests,
DDMisbasedonfindingrelationshipbetweentheinputandoutputvariables
of a system without explicit knowledge of its physical behaviour. Physical
models have their limitations because many of the hydrological processes are
complexanddifficulttorepresent.Understandingofthesystemisalsofarfrom
complete so DDM offers an alternative approach to traditional physically-
based models. DDM has been the subject of much research activity in hydro-
logical modelling over the two last decades and includes a range of different
techniques.Thesemainlyoriginatefromthefieldsofcomputationalandartifi-
cialintelligence(Solomatine,2005),andincludetechniquessuchasneuralnet-
works(NN),fuzzylogic,evolutionarycomputingandmachinelearning.Many
L. See: Data Fusion Methods for Integrating Data-driven Hydrological Models, Studies in
ComputationalIntelligence(SCI)79,1–18(2008)
www.springerlink.com (cid:1)c Springer-VerlagBerlinHeidelberg2008