大数据与城市规划 (37).pdf
Evaluating the effectiveness of urban growth boundaries using humanmobility and activity recordsYing Longa,Haoying Hanb,Yichun Tuc,Xianfan ShudaBeijing Institute of City Planning,ChinabDepartment of Urban and Regional Planning,Zhejiang University,ChinacDepartment of City and Regional Planning,University of North Carolina at Chapel Hill,United StatesdDepartment of Land Management,Zhejiang University,Chinaa r t i c l ei n f oArticle history:Received 18 December 2014Received in revised form 3 May 2015Accepted 3 May 2015Keywords:Plan implementation evaluationBig dataSocial networkTransit smartcard dataBeijinga b s t r a c tWe proposed a methodology to evaluate the effectiveness of Beijings Urban Growth Boundaries(UGBs)using human mobility and activity records(big data).The research applied data from location check-in,transit smart card,taxi trajectory,and residential travel survey.We developed four types of measures toevaluate the effectives of UGBs in confining human activities and travel flows,to examine the conformityof urban activities with the planned population,and to measure the activity connections between UGBs.With the large proportions of intra-and inter-boundary travel flows and an overwhelming majority ofcheck-ins inside the UGBs,the research concluded that Beijings UGBs were effective in containing humanmobility and activity.However,the connections between UGBs,indicated by the spatial differentiation ofthe travel flows,were not consistent with the plans intention and strategy.It indicated the potentialunderdevelopment of the public transit serving several new cities.?2015 Elsevier Ltd.All rights reserved.1.IntroductionEvaluation of plan implementation is important because itreflects the extent to which a plan succeeds in predicting,guiding,and controlling future urban development.One common way todetermine what a plan has accomplished is to measure the confor-mance degree between the actual outcomes or impacts and theproposed plans.By doing so,planners can acquire insights onhow the planning decision-making process operates and validatewhether planning efforts do contribute to goal achievement(Alexandar&Faludi,1989;Alexander,2009;Laurian et al.,2004;Talen,1996b).This evaluation helps establish a responsive andaccountableplan-makingand-implementationprocess,thusimproving the overall quality of planning.Since the early 1970s,numerous studies have contributed to the theoretical and method-ological understandings in the field of planning evaluation.A fewstudies have illustrated the evaluation approaches with one partic-ular aspect of planning,including land development(Alterman&Hill,1978;Berke et al.,2006;Chapin,Deyle,&Baker,2008),environmental planning(Brody&Highfield,2005),public facilitiesand infrastructure(Laurian et al.,2004;Talen,1996a),and urbansprawl control(Altes,2006;Brody,Carrasco,&Highfield,2006;Nelson&Moore,1993).In this study we focused on assessing plan implementation interms of the effectiveness of urban growth boundaries.As one ofthe most widely adopted urban containment policy tools,urbangrowth boundaries(UGBs)have been used to control the expan-sion of urban areas,increase urban land use density,and protectopen spaces(Pendall,Martin,&Fulton,2002).The basic conceptof implementing a UGB is to set a physical boundary separatingurban and rural areas.Usually,urban developments are notallowed outside the predefined boundary.Broadly speaking,theimplementation of UGBs also encompasses various regulatorytechniquessuchaszoningandlanddevelopmentpermits.Proponents argue that urban growth boundaries may have at leastthe following six merits(Staley,Edgens,&Mildner,1999):(1)preserve open space and farmland;(2)minimize the use of landgenerally by reducing lot sizes and increasing residential densities;(3)reduce infrastructure costs by encouraging urban revitalization,infill,and compact development;(4)clearly separate urban andrural uses;(5)ensure the orderly transition of land from rural tourban uses;and(6)create a sense of community.An increasingnumber of cities in the U.S.and Europe have regarded UGBs as akey tool in controlling urban sprawl.However,the empiricalhttp:/dx.doi.org/10.1016/j.cities.2015.05.0010264-2751/?2015 Elsevier Ltd.All rights reserved.Corresponding author at:Department of Urban and Regional Planning,AnzhongBuilding,Zhejiang University(Zi-Jin-Gang Campus),Yu-Hang-Tang Road 866,Hangzhou 310058,China.E-mail addresses:(Y.Long),(H.Han),ytulive.unc.edu(Y.Tu),(X.Shu).Cities 46(2015)7684Contents lists available at ScienceDirectCitiesjournal homepage: measuring the effectiveness of UGBs are not common.Thisis partly because that plan implementation evaluation has rarelyattracted adequate attention in the planning profession.It has beenan afterthought to the planning decision-making or implementa-tion framing(Berke et al.,2006;Talen,1996a).The lack of data,robust evaluation theories and methodologies,as well as of thelinkages between theory and practice are among some of the majorreasons for its limited applications in planning practices(Brody,Highfield,&Thornton,2006;Laurian et al.,2004;Oliveira&Pinho,2010;Talen,1996a,1996b).In addition to these general issues,the development of UGBsimplementation evaluation has also been constrained by the over-simplified evaluation dimension.To date,most relevant studiesfocused on assessing the physical outcomes,that is,the degree towhich the actual urban extent and development layout conformto the proposed UGBs.For instances,several studies utilizedremote sensing images and geographic information system to trackland use/cover changes(e.g.Hasse,2007;Hepinstall-Cymerman,Coe,&Hutyra,2013).Among them,Han,Lai,Dang,Tan,and Wu(2009)examined the effectiveness of the UGBs in Beijing overtwo planning implementation periods,19831993 and 19932005,and concluded that the UGBs failed to contain urban growth.Some studies focused on analyzing the driving forces of the urbanexpansion(Boarnet,McLaughlin,&Carruthers,2011;Brueckner&Fansler,1983;Burchfield,Overman,Puga,&Turner,2006;Long,Gu,&Han,2012).Using quantitative techniques such as regressionmodels,these studies helped identify the effects of particular vari-ables(e.g.planning and political elements like UGBs,built environ-ments,and socioeconomic attributes)on urban expansion or landdevelopment.Ideally,one could look into the land use data toexamine the land use changes.However,in China,an accuratelyand timely monitoring of land use changes is never an easy task.A comprehensive land use survey of a Chinese city may take aslong as 10 years,and even longer in some large cities.Even afterplanners acquire the results of the most recent land use survey,they may find that the data are either inadequate or inaccurate.Polygons in land use maps are usually very big,omitting much use-ful information.Also,some areas that have been lately developedas urban uses or urban infrastructures may still be marked as agri-cultural use(Long&Liu,2013).Due to the burdensome task to pro-vide real-time changes of land uses,a relatively easier way toacquire a city-scale change of human activities would be a helpfulsupplement to the traditional land use examinations with poorreliability.Moreover,one of the major problems associated withthese studies is that they simply equal urban expansion to thechanges in land cover or use.What has been ignored is the assess-ment of how human activities actually react to the UGBs whenpeople utilize urban spaces and development where UGBs intendto regulate.What are the relations between urban activities andUGBs?Do the UGBs really work on shaping and controlling humanmobility and activities?Unfortunately,previous studies have pro-vided few clues or solutions to these questions.In this study,we evaluated the effectiveness of UGBs from theperspective of human mobility and activities using locationcheck-ins from social network,taxi trajectories from GPS devicesequipped by a large number of taxis,and smartcard records frompublic transit system.The increasing availability of these urbanbig data has provided unprecedented opportunities for urbanresearchers and planners to better understand and manage urbansystems.These data have enabled us to describe and analyzereal-time human behaviors and movements in a more precise,reli-able,and economic way.We also see the potential of applyingthese data in planning evaluation,particularly in developing coun-tries where official statistics are less sufficient or reliable.Based onthe analysis of the massive data on human mobility and activities,the study aims to(1)evaluate the effectiveness of UGBs inconfining human mobility and activities,(2)examine whetherthe intensity of urban activities correlate to that of planned popu-lation across UGBs,and(3)measure the interconnections betweenUGBs and examine whether they conform with plan intentions.This study selected Beijing as a case to illustrate how the eval-uation is developed.Beijing has undergone rapid urban develop-ment in the past two decades and can be regarded as arepresentative among rapid-developing cities.Considering the nat-ure of the methodology adopted in this study,it can also be appliedto developed cities.In Section 2,we introduced the study area anddata sources.In Section 3,we elaborated the methodology and pre-sumptions,as well as the evaluation results.In Section 4,we dis-cussed the findings in details.In Section 5,we concluded bysummarizing our findings,suggesting the strength and weaknessof our study and giving recommendations for potential subsequentstudies in future.2.Study area and data2.1.Beijings recent urban planningAs the capital of China,Beijing is one of the most populous citiesin the world.The population at the end of 2013 was 21.15 million.The Beijing Metropolitan Area(BMA)is 16,410 square kilometers.AccordingtolandusedatasetofBeijingInstituteofCityPlanning,the total urban area as of 2012 was 1675 square kilome-ters.The BMA currently comprises 16 administrative subdivisions(districts),as illustrated in Fig.1.Since the latest adjustment of the Beijing administrative bound-aries in 1958,five urban master plans have been drafted in 1958,1973,1982,1992 and 2004 respectively.Each master plan includesan official land use map.Individual land parcels in the map wereassigned according to a classification of either urban(residential,commercial,industrial,public green land,and mixed-use land)ornon-urban(farmland,forestland,and wetland)uses(Long et al.,2012).The map guided the future urban development,and useswere expected to conform to the plan.The BMA has experienced an unprecedented increase in popula-tion growth and urban development since early 1990s.By the year2003,Beijings population and urban built-up area had already sur-passed the capacity set forth in the 19922010 Master Plan.Tocope with new challenges in the future,the Beijing MunicipalCommission of Urban Planning updated the citys master plan fora 2020 planning horizon.Approved in 2005,the revised plan wassought to outline general principles and create new guidelinesfor Beijings long-term economic,social,and physical development(Ding,Song,&Knaap,2005).In this new plan,the projected population of Beijing was 18 mil-lion in 2020.From a spatial perspective,the plan promotes atwo-axes,two-belts,and multi-sub-centers urban developmentpattern.A total of 1650 square kilometers of planned urbanbuilt-up area would be allocated to the central city and elevennew cities.Urban developments were planned to occur withinthe planned urban construction areas.The boundaries of theseareas can be regarded as the Chinese UGBs which functioned in asimilar way as the UGBs in the U.S.The issuance of land use per-mits outside these boundaries was generally forbidden in orderto curb urban expansion and protect open spaces.Four types ofUGBs are identified,including those in the central city,new cities,towns,and other small isolated areas.2.2.Date sources2.2.1.Location check-in dataCompared to traditional approaches to obtaining information ofurban activities,the use of data acquired from mobile devicesY.Long et al./Cities 46(2015)768477enablesareal-timerepresentationofurbandynamicsandtheirevo-lution over time and space(Ratti,Pulselli,Williams,&Frenchman,2006).In this study,we used location check-in data provided bySina Weibo(having a similar function as Twitter in China)to proxythe actual urban activity.A total of 890 million check-in recordswere collected during the time period.These check-in records fromMay 16 throughout July 28,2013(74 days)were linked to a total of102,826 Point-of-interests(POIs).POIs generally have eight typesbased on land use classifications.They are(1)shopping,(2)enter-tainment,(3)hotel and public,(4)sports,(5)firm,(6)residential,(7)educational institute,and(8)restaurant.The check-in datasetwas transformed into a POI-based attribute table,in which eachPOI record was comprised of a full range of information includingland-use classification,latitude and longitude,number of totalcheck-ins,and some other geographic features.2.2.2.Transit Smart Card DataTransit smart card system of Beijing was introduced in 2003.By2005,over 90%of bus/subway riders in Beijing had used the transitsmart card for payment.The transit smart card system records aset of cardholders information including trip origins and destina-tions,boarding and/or alighting time,card numbers,and card types(e.g.student card or regular card).Unlike a subway system whichrequires a swipe in and out,not every bus trip keeps an alightingrecord.That is because Beijings bus system uses two fare schemes a flat fare scheme and a distance fare scheme.In a flat farescheme,a 0.40 CNY is charged for every single trip and it doesnot require riders to swipe the card on the way out.As a conse-quence,the Smart Card Data(SCD)for the flat fare lines doesntstore any information about trips arrival time or destination stop.A distance fare line requires cardholders to swipe twice both onboarding and alighting a bus,so the SCD contain trips fullinformation.In this study,the SCD of bus and subway system were obtainedfrom Beijing Municipal Administration&Communications Card Co.(BMAC).These data were collected from April 5 to 11 in 2010.Forbus system,though incomplete flat fare trips information mightresult in the failure of identifying travel patterns,it would be usefulwhen a flat fare trip was taken as a transfer between two distancefare trips.Therefore,data from both flat and distance fare schemeswere used for human mobility analysis.A total of 97.9 millionbus/subway trips were generated by 10.9 million cardholders dur-ing the time period.The summary of bus and subway SCD is shownin Table 1.2.2.3.Taxi trajectoriesTaxitrajectorieswerecollectedwithinoneweekfromNovember 7 to 13 in 2011,with a total of 2,254,068 origindestination trips from over 20,000 taxis.The location of each tripsorigin and destination were stored in the system.2.2.4.Resident travel surveyIn addition to the aforementioned three types of urban big data,we also
