发布时间 : 星期四 文章城市发展与生态模拟模型研究更新完毕开始阅读
M.Alberti,P.Waddell/Anintegratedurbandevelopmentandecologicalsimulationmodel219
3.1.UrbanSim
TheUrbanSimmodel,describedindetailelsewhere[42–44],integratesandextendselementsoftheconsumersurplusapproachtakenbyMartinez[29],arealestatestockadjustmentmodel[12],andadynamicmobilityandloca-tionchoicemodelingapproach.Thetheoreticalbasisofthemodeldrawsonrandomutilitytheoryandtheurbaneconomicsoflocationbehaviorofbusinessesandhouse-holds.Thisisembeddedwithinalargersimulationmod-elingframeworkthatdealswithlandmarketclearing,landdevelopment,andaggregatemetropolitanchangesinthedistributionofhouseholdsandbusinessesbytype[42].Themodelpredictsthelocationofbusinessesandhouse-holds,developerchoicestodeveloprealestateonvacantlandortoredevelopexistingbuildings,andthepriceoflandandbuildings.Households,businesses,developers,andgovernment,respectively,makedecisionsaboutloca-tion,production,consumption,andinvestments.Theydy-namicallyinteractinthelandandrealestatemarketsandgeneratephysicaldevelopmentandrelocation.Sinceac-cessibilityisakeyin?uenceonlocationchoices,themodelisinterfacedwithtraveldemandmodelstoaccountforthefeedbackrelationshipsbetweenlanduseandtransportation.UrbanSimisbasedonanobject-orientedframeworkthatmodelsthekeymarketbehaviorsofurbanactorswhichprovideatransparenttheoreticalstructure.Thedemand
componentcurrentlyrepresentsspaceusingzonesthatcor-respondtometropolitantravelanalysiszonesdemand,withwhichitisdesignedtointegrate.Thesupplyoflandisrep-resentedatthelandparcellevel.Themodelusesannualtimestepstosimulatethemobilityandlocationchoicesofhouseholds,thedevelopmentandredevelopmentofreales-tate,andthemarketclearingandpriceadjustmentprocesseswithinthemarket.Travelaccessisupdatedonperiodicyears,triggeredbyatravelmodelsimulationruninyearsthathavesigni?canttransportationsystemchanges.3.2.Modelstructure
Figure1representstheurbandevelopmentandecologi-caldynamicsthattheintegratedmodelwilladdress[1,42,45].UrbanSimpredictsthelocationbehaviorsofhouse-holds,businesses,anddevelopers,andconsequentchangesinlandusesandphysicaldevelopment.Theseareamongtheinputsrequiredtopredictthechangesinlandcoverandecologicalimpacts.OurcurrentstrategyistoextendtheobjectpropertiesandmethodsnowimplementedintheUr-banSimmodel.Insteadoflinkingtheurbanandecologicalcomponentssequentially,weproposetointegratethematafunctionallevel.Weproposetoaddtheproductionandcon-sumptionbehaviorsofhouseholdsandbusinesses,andlinkthesethroughagridrepresentationoflandtoinfrastructureandnaturalsystems.Thestructureoftheintegratedmodel
Figure1.UrbanecologicaldynamicsmodeledinUrbanSim.Note:processesinitalicsarenewmodelcomponentsnotpresentlymodeledinUrbanSim.
Source:WaddellandAlberti(1998).
220M.Alberti,P.Waddell/Anintegratedurbandevelopmentandecologicalsimulationmodel
identi?estheprincipalobjectsashouseholds,businesses,buildings,land,infrastructure,naturalresources,andvari-ousbiophysicalcomponents.
Theprincipalurbanactors,representedinthemodelasobjectscorrespondingtobusinesses,households,develop-ers,andgovernments,eachmakechoicesthatalterthespa-tialpatternsofurbandevelopmentandhouseholdandeco-nomicactivity.Households,businesses,buildingsandlandparcelsarelinkedexplicitlytoindividualcellsofappro-priateresolution.Thechoicesthathouseholdsmakeaboutjobs,location,andconsumptionwillbehandledthroughamicrosimulationapproachtoforecastthedemandforspeci-?edbuildingtypesandlocation,environmentalquality,andservices.UrbanSimwillinterfacewithtravel,environmen-tal,andinfrastructuremodels,toaccountforchangeintransportation,infrastructure,andenvironmentalconditionsandre?ecttheminthemodel.Thesefactorsareoper-ationalizedthroughindicesofaccessibility,infrastructurecapacity,andenvironmentalqualityforagivenlocationcomputedattheappropriatespatialandtemporalresolu-tion.
3.3.Modelcomponents
ThearchitectureoftheUrbanSimmodeliscurrentlybe-ingredesignedtosupportmicrosimulationofthebehaviorofhouseholds,businesses,anddevelopersandthespatiallyexplicitinteractions.Existingcomponentsarebeingrevisedandnewcomponentsarebeingadded.Figure1depictsthemodelcomponentsintheproposedmodelsystem,andhigh-lightsthenewcomponentsrepresentinglandcover,waterconsumptionandnutrientemissions.
3.3.1.Locationmodels
Urbansimpredictstheprobabilitythatahouseholdandabusinessthatiseitherneworhasdecidedtomovewithintheregion,willchooseaparticularcombinationofloca-tionandbuildingtype.Amultinomiallogitspeci?cationpredictsthejointprobabilityofbuildingtypeandlocation.Thesecomponentsrepresentthedemandforresidentialandcommercialrealestate,thesupplyofwhichispredictedinthelanddevelopmentcomponent.Currentlythelocationmodelcomponentstreatrealestateinamoderatelyaggre-gateform,aggregatinglandparcelsandtheirhousingandcommercialsquarefootageintozonesbytypeofspace.Weproposetomodifythelocationcomponentsusingmi-crosimulation,anddisaggregatingthelocationchoicetothelevelofindividualhousingunitsornonresidentialbuildings,whichareinturnlocatedwithinspeci?cparcelsandonspe-ci?cgridcells.Thisextensionwillallowtheadditionoflocalizedcontextandenvironmentalconsiderationsintothedemandforbuildings.Issuessuchasopenspace,pedes-trianaccessibility,andotherlocalconsiderationscouldbeincorporatedinthisway.
3.3.2.Productionandconsumption
Productionandconsumptionactivitiesofhouseholdsandbusinessesdrivetheinteractionbetweenhumandecisionandenvironmentalprocesses.Productionandconsumptionbybusinesseswillbemodeledusingacombinationoftheaggregateeconomicinput-outputmethodologyandami-crosimulationatthelevelofindividualbusinessestablish-mentsgeocodedtoaparticularlocation.Theinput-outputmodelre?ectsthestructureofconsumptionandproductionwithintheeconomy,aggregatedintosectors.Usingmi-crosimulation,theseaggregate?owswillbeallocatedtoindividualbusinessestablishmentsgeocodedtoacell,ac-cordingtotheindustryandsizeoftheindividualbusiness.Consumptionbyhouseholdswillbehandledthroughami-crosimulationapproachtoforecastthedemandofspeci?edproductsandservices.Thistechniqueis?exibleenoughtorepresentsensitivitytoavarietyoftechnologicalandpol-icyfactorsthataffectconsumerbehavior.Byincorporatingforexamplethelifecycleofproductsandservicesitwouldenableustoaccountfortechnologicalsubstitution.3.3.3.Landuse
ChangesinlanduseinUrbanSimaremodeledthroughalanddevelopmentmodel.Thedemandfornewrealestatedevelopment,orredevelopmentofexistingrealestate,istriggeredbymonitoringthevacancyratewithinrealestatesubmarkets.Whenvacancyratesfallbelowastructuralthreshold,thenpricesbegintorise,andnewdevelopmentisstimulated.TheactualsimulationoflanddevelopmentintheUrbanSimmodelcurrentlyusesindividuallandparcelsastheunitofdevelopment.Themodelpredictsthefea-sibilityofdevelopmentundertheconstraintsofthelanduseplanandotherenvironmentalorregulatoryrestrictions,andthenestimatesthepro?tabilityofdevelopmentintoanyoftheallowedurbanuses.Developmentprojectsthataremostpro?tablearesimulated,addingtotheexistingstockofbuildingsintherealestatemarket,andlinkingthemtospeci?clandparcels.Revisionstotheexistinglanddevel-opmentcomponentwillbebasedonanextendedspatialdatabasethatcross-referencesparcelsandbuildingstoalo-cationgrid,allowingtheuseofspatialmetricstofurtherinformtheexpectedpro?tabilityofdevelopment,basedonthecharacteristicsanddevelopmenttrendsintheproximatearea.Inaddition,themodelwillbefullyconvertedtoalogitformulation,consistentwiththedemandcomponentsinthecurrentmodel.Weplantoaddnewcomponentsformodelingcompetingnon-urbanlandusessuchastim-berandagricultureproductionandrelatedlandmanage-mentpracticeswhichmayaffectthelandmarketattheurbanfringe.Thedevelopercomponentofthemodelistheonemostdirectlyin?uencedbylocalpoliciessuchasthecomprehensiveplan,densityconstraints,theUrbanGrowthBoundary,environmentalconstraints,anddevelopmentim-pactfeesorotherdevelopmentcostssetbylocalgovern-ments.
M.Alberti,P.Waddell/Anintegratedurbandevelopmentandecologicalsimulationmodel221
3.3.4.Landcover
Thelandcovercomponentthatwediscussmoreexten-sivelyinthenextsectioniscentraltointegratingsocio-economicandecologicalprocesses.Landcoverchangeisaffectedbyproductionandconsumptionpatternsofhouse-holdsandbusinesses,theirlocationpreferences,andlanddevelopmentandredevelopment.Changeinlandcoverinturnaffectsthesepatterns,preferences,andultimatelyde-velopment.Wewillmodellandcoverchangeasin?uencedbybotheconomicandbiophysicalprocesses.WeproposetolinkUrbanSimtoaprocess-basedlandscapeandapatch-dynamicapproach.Landconversionismodeledbasedonthechangesinhousingandcommercialbuildingspredictedinthelanddevelopmentcomponent,householdandbusi-nesscharacteristicsoccupyinganybuildingsonaspeci?clandparcel,andotherlandscapecharacteristicsofthepar-cel.Wewillallocatespeci?cbuildingsandassociatedin-frastructuretoindividualcellsofhighresolutiontopredictchangeinlandcoverpatchstructureandfunction.Thepredictedlandusewillconstitutetheinputofaprocess-basedlandscapemodel.Theadvantageofthisapproachistheexplicitrepresentationofthelandscapeprocessesthataffectecologicalconditionswhichwillsuccessivelyfeedbackintothelandcoverandlandusemodel.Asetofspa-tialmetricsofurbandevelopmentandecologicalpatternsofparcelsandneighborhoodswillbeusedtobetterrepresentthelandconversiondynamics.
3.3.5.Wateruse
TheresourcedemandmodelsinUrbanSimwillincludevariousmoduleseachpredictingtheuseofwateranden-ergyonthebasisofconsumption,infrastructurecapacity,andef?cienciesoftechnologies.Ourcurrentfocusisondevelopingawater-usecomponent.Thewaterresourcecomponentwillberepresentedbyawaterdemandmodelthatwillbelinkedtothegridstructureonthebasisofwa-terconsumptionpatternsofhouseholdsandbusinessesandwatersupplycapacity.Acomponentwateruseforecastingmethodwillbeimplementedtoestimatefuturehouseholdwateruseasafunctionofhouseholdcharacteristics(size,income,etc.),parcelcharacteristics(buildingtypology,lotsize,density,etc.),climaticconditions(monthlyprecipi-tationandtemperature),andmarginalpriceofwater.Inadditionnon-pricedemandsidemanagement(DSM)poli-ciesareexpectedtoin?uencewaterdemand.Thepredictedwaterdemandislinkedtotheinfrastructureandnaturalsystemsthroughaspatialgridandcanbeusedtoevalu-atealternativegrowthscenariosandwatersupplysystemcapacity.
3.3.6.Nutrientloads
TheemissionmodulesinUrbanSimsimulatetheemis-sionofvariouspollutantsintotheatmosphere,water,andsoil.Theseemissionmodeloutputsultimatelywillservetoconstructwholeurbanecosystemmassbalancesofma-terialsandrelativecontributionsfromthevariousmedia.Ourcurrentfocusisonmodelingnutrient(phosphorous
andnitrogen)exportfromurbanlanduses.Wewillbuildaspatially-explicitnutrientexportmodelforthePugetSound.Nutrientloadsaremodeledbasedonrunoffcoef?cientsandexpectedpollutionconcentrationsrelatedtolanduse,den-sity,typeofbuilding,typeofbusiness,andtreatmentplant.Theexportcoef?cientmodelisparticularlysuitableasaba-sisforestimatingnutrientloadingbecauseitusesthearealextentofdifferentlandusetypesandestimatedrunoffco-ef?cients.3.4.Feedback
Theproposedmodelframeworkisdesignedtotakeintoaccounttheinteractionsbetweentheecologicalimpactsandurbanprocesses.Theoutputoftheintegratedurbanmodelwillserveastheinputtoseveralbiophysicalmod-elsaddressinghydrology,hillslopestability,waterquality,atmosphere,andaquaticandterrestrialecosystems.Ecolog-icalchangeswillfeedbackonthechoicesofbothhouse-holdsandbusinesslocations,andavailabilityoflandandresources.Weproposetouseasetofparcel-andcell-basedenvironmentalqualityindices(e.g.,airquality,waterqual-ity,noise,etc.)andpotentialriskorhazardindices(e.g.,?oods,landslides,etc.)thatin?uencelocationchoicesandpro?tabilityofdevelopment.Inadditionasetofspatialmetricswillbeimplementedtoinformlandusedemandanddevelopmentbytakingintoaccountspatialproximitytoamenitiesanddisamenitiesandneighborhoodeffects.4.Landuse–coverchange:astrategyformodelintegrationLanduseandlandcoverdynamicsareatthecoreoftheintegratedurbanecologicalmodelingframework.Theyaredistinctbutcloselylinkedprocesses.Landcoverchangeisdrivenbybothbiophysicalandsocio-economicforcing.Changesoflandcoverdrivenbysocio-economicforcingarecurrentlythemostimportantandmostrapidofallchanges.Biophysicalprocesses,suchasvegetationdynamics,in-volvealterationsincoverduetonaturalchangesinclimateandsoils.Wedistinguishtwotypesofchangesinlandcover:conversionandmodi?cation[37].Landconversionisachangefromonecovertypetoanother.Landmodi?ca-tionisachangeinconditionswithinthesamecovertype.Hereweconsideronlytheeffectsoflandusechangeonlandconversion.Wedevelopastrategyformodelinglanduseandlandcoverdynamicsbybuildingonvariousmod-elingapproachesrootedineconomics,landscapeecology,andcomplexsystemscience.Ourhybridmodelstructurecombines:(a)amicrosimulationofactorchoices(location,housing,travel,productionconsumption,andlanddevelop-ment);(b)aprocess-basedmodelofphysicalandecologicaldynamics(hydrology,nutrientcycling,primaryproduction,andconsumerdynamics);and(c)aspatially-explicit,grid-basedmodelstructurewhichrepresentsthedynamicsinher-entinlanduseandlandcoverchangeanddetailedspatialqueriesandsimulation.
222M.Alberti,P.Waddell/Anintegratedurbandevelopmentandecologicalsimulationmodel
4.1.Socio-economicprocesses
Thedemandforbuiltspaceforvariousactivitiesisgen-eratedthroughmicrosimulationofdemographicandeco-nomicprocesses,andlocationchoicesofhouseholdsandbusinesses.Theseareintegratedwithamarket-clearingcomponent.Landusechangeismodeledthroughaspatiallyexplicitmicrosimulationoflanddevelopmentandredevel-opment.Wedevelopahybridspatially-explicitmicrosimu-lationstructurebycombiningthecurrentUrbanSimbehav-ioralapproach,whichexplicitlyrepresentsthelanddevel-opmentprocess,withaspatiallyexplicitapproachwhichexplicitlyrepresentslocalspatialdynamicsandneighbor-hoodeffectsofbothlanduseandlandcover.
4.1.1.Demographicandeconomicprocesses
Themodelingofeconomicanddemographicprocesseswithinmetropolitaneconomiesaremostcommonlydonethroughmacro-economicmodels,usingeitheranin-put/output,structuralequations,orhybridapproach.De-mographicprocessesareoftenmodeledasafunctionofcohortsurvivalthatinvolvesageingofanage-sexpopula-tionpyramid,withfertilityanddeathprobabilitiesappliedtothepopulationcountsineachage-sexcohort,withage-sexspeci?cnetmigrationratespredictedasafunctionofemploymentopportunities.HybridapproachessuchastheWashingtonSimulationandProjectionmodeldevelopedbyConway[8]provideareasonablyrobustmeansoflinkingthesimulationoftheevolutionoftheeconomicstructureofaregionorstatetobroadernationalandglobaleconomictrendsandin?uences.
Thealternativetoamacroeconomicapproachtomodel-ingtheseeconomicprocessesisthedevelopmentofapurelymicrosimulationapproachtoeconomicanddemographicprocesses.Thereisestablishedworkonthemicrosimula-tionofhouseholdevolutionthataddressesthedemographicprocessesofageing,birth,death,andhouseholdforma-tionanddissolution.Wegener[49]suggeststhatmanyoftheseprocessesaretypicallymodeledastransitionproba-bilitiesusingMonteCarlosimulationratherthanchoicestobemodeledusingabehavioralspeci?cation.Verylittleworkhasbeenpublished,todate,onthemicrosimulationofeconomicprocessesatthelevelofthe?rm.
Oneofthemajorchallengesfacedinthefuturedevel-opmentofafullmicrosimulationmodelofthetypeweproposehereisthereconciliationofmicrosimulationoflanddemandandsupplyprocessesasdescribebelow,withmacroeconomicprocesses.WeproposeatthistimetoadopttheapproachtakenintheWashingtonEconomicSimula-tionmodel,whichcontainsahybridinput/outputandstruc-turalequationsmodelofthemacroeconomyoftheStateofWashington,andtolinkthistothemicrosimulationoftheprocessesdescribedbelow.Thebasisforthischoiceisthatthereisasubstantialdegreeofremainingresearchtobedonebeforeamicrosimulationofthemacro-behaviorofaregionaleconomycanbeeffectivelyimplemented,inawaythatcapturesinteractionsnotonlywithintheregion,
butalsointeractionwitheconomicprocessesinthenationandtheworld.
4.1.2.Householdsandbusinessdemand
ThemodelingofhouseholdandbusinessdemandforrealestateatdifferentlocationswillbemodeledusinganextensionoftheexistingframeworkalreadyimplementedinUrbanSim.Theprimarydifferencewillbetheimple-mentationofafullsampleenumerationofhouseholds,andbuildings,inthedemandcomponentsofthemodel.Withthelinkageofbuildingstoalocationgrid,andaninfrastruc-tureforspatialqueryandanalysisthatisindevelopment,weplantoenrichthedemandcomponentsofthemodeltoincorporatemorespatially-explicitattributesoflocationdemand.
Withreferencetoresidentiallocationdemand,themicro-locationattributesmayincludeattributesofthepedestrianenvironment,suchasthestreetpattern,theavailabilityofopenspace,andthewalkaccesstoshoppingandentertain-mentopportunities.Inaddition,thespatialmetricstobedescribedlatermaybeadaptedtodescribeaspectsofthespatialpatternoflanduse,andenvironmentalcharacteris-tics,thatcouldinformthedemandfunctionsofdifferenttypesofhouseholds.Thislaststepwillopenthepossi-bilitytoestablishfeedbackfromthebiophysicalprocessesandlandcoverchangetodemandforresidentiallocations.WeproposetoimplementamicrosimulationofthedemandcomponentsofthemodelfollowingtheapproachdevelopedbyWegeneretal.[50].
Withreferencetobusinesslocation,theopportunitytodevelopmorespatiallydetailedandexplicitrepresentationofthecontextsurroundingavailablesitesshouldsigni?-cantlyincreasethefeasibilityofenrichingthedemandbe-haviorofthemodel.Sitecharacteristicssuchasfrontageorproximitytotransportationfacilities,thepedestrianen-vironment,andotherenvironmentalqualitycharacteristicswouldbepotentialin?uencesonbusinessdemandforlo-cationsthatcouldbeexplored.
4.1.3.Marketclearingandpriceadjustment
ThemarketclearingandadjustmentofpricesoflandandbuildingsiscurrentlyaddressedinUrbanSimthroughaprocessthatmatchesmovinghouseholdsandbusinessestovacanciesinthehousingornonresidentialbuildingstockbasedontheconsumersurplusofthematch.Consumersurplusmeasuresthedegreetowhichthewillingnesstopayforanalternativeexceedsitsmarketcost[14,29,42].Oncethematchingofactiveconsumersandavailableva-canciesiscompleted,pricesintherealestatesubmarketsareadjustedaccordingtotheirrelationshipbetweenthecur-rentvacancyrateinasubmarketandthestructuralvacancyrate,followinganapproachdescribedbyDiPasqualeandWheaton[12].Whencurrentvacancyratesdipbelowthestructuralvacancyrates,pricesadjustupwards,andcon-versely,exceptionallyhighvacancyratespullpricesdown-ward.