Scribd Logo
Загрузить

Добро пожаловать в Scribd!

  • Case of The Great Lakes Region

    Загружено:

    operas69
    5 просмотров35 страниц
    ARTICULO CIENTIFICO

    Оригинальное название

    Case of the Great Lakes Region

    Авторское право

    © © All Rights Reserved

    Доступные форматы

    PDF или читайте онлайн в Scribd

    Этот документ был вам полезен?

    Это неприемлемый материал?

    Пожаловаться на этот документ
    ARTICULO CIENTIFICO

    Авторское право:

    © All Rights Reserved
    Скачайте в формате PDF или читайте онлайн в Scribd
    Отметить как неприемлемый контент
    5 просмотров35 страниц

    Case of The Great Lakes Region

    Загружено:

    operas69
    ARTICULO CIENTIFICO

    Авторское право:

    © All Rights Reserved
    Скачайте в формате PDF или читайте онлайн в Scribd
    Отметить как неприемлемый контент
    из 35
    18112016 Exploring the Ink between urban form and work related transportation using combined seteltimage and census information: Case ofthe Great lakes re, Dowsloss POF | Exar Adve sch Arte culreis essing Intemational Journal of Applied Earth Observation and Geoinformation bs Volume 47, May 2016, Pages 139-150 Exploring the link between urban form and work related transportation using combined satellite image and census information: Case of the Great lakes region Ying Zhang +, Bert Guindon, Krista Sun Show more 4 40.1018),o920°5.12.008, Highlights + Acomprehensive bi-national urban database, the Great Lakes Urban Survey (GLUS), comprising all cites with populations over 200,000 in the Great lakes region was generated. + Asuite of analysis tools are proposed to utilize information sets such as GLUS to investigate the Ink between urban form and work-related travel + Anew indicator, the Employment Deficit Measure, is proposed to quantity the balance between employment and worker availabilty at diferent ransithorizons. corresponding to modes of transportation + Asspatial interaction models developed with accurate prediction of ravel distance ‘aggregated atthe census tract level, which is used to explore the reduction in travel distance associated with citferent urban forms. Abstract, ‘Aspects of urban transportation have signficantimplications for resource consumption and environmental quality. The level of ravel activity the viabilly of various modes of transportation and hence the level of ransportalion-related emissions are influenced by the structure of eos, Le. thelr urban forms. Whi itis widely recognized that satelite romote sensing can provide spatial information on urban land cover and land use, is effective use for understanding impacts of urban form on issues such as transportation requires that this information be integrated with relevant demographic information. A comprehensive bisnational urban database, the Great Lakes Urban Survey (GLUS),, comprising ll cies with populations in excess of 200,000 has been created from Landsat imagery and national census and transportation survey information from Canada and the United States. A suite of analysis tools are proposed to uilize Information sets such as GLUS to investigate the link between urban form and work related travel. Anew indicator, the Employment Deficit Measure (EDM) is proposed to {uaniiy the balance between employment and worker availabilty a different transit horizons and hence to assess the viabilty of altemate modes of transportation. Itis argued that the high degree of residential and commercialindustvial land uses greatly Impact travel to work mode options as well as commute distance. Aspatlalinteraction ‘models developed and found to accurately predict travel distance aggrogated at the census tractlevel. We argue tha this model could also be used to explore the relative levels of travel activity associated with different urban forms. haipuhwunseioncodirectcom remot dgh ar. mis encolariclop/S03002424 15300884 p=y 1198 181812016 Exploring the Ink between urban form and work related transport using combined stele image and census information: Case ofthe Great lakes re, Keywords Urban land use; Urban form; Urban transportation; Indicator quantification; Travel modeling 1. Background Urbanization isa global trend. According to United Nation's World Urbanization Prospects publications (United Nations, 2002 and United Nations, 2007), the percentage ofthe word's population ving in urban areas increased from 28% in 1960 to 50% in 2008 and can be expected to reach 70% by 2050. Urbanization is generally associated with increasing affluence which is reflectedin high automobile ownership and use. Therefore, Urbanization can be expected to resultin increased energy demand (Transport Canada, 2008, United Nations, 2007, Wang et al, 2012 and Han etal, 2014). Inthe case ofthe United States and Canada, transportation accounted for 27.2% and 29.0% respectively ofoveral energy consumption in 2001 (TEDB, 2007 and Cuddihy ota, 2005), ‘While the extent of urban areas and theirintensiy are fundamentalland information items, an understanding of urban impacts on energy use and the environmentrequires a ‘more comprehensive and integrated information base. A relevant characterization of Urbanization is urban form’ We define urban form as the combined spatial stributions ofland uses and demographic-related attributes (e.g, overall population, dwelling counts, employment, etc.). thas long been recognized thatthe structure of cites, Le., thelr urban ‘form’, greatly influence inva-urban human actives and consequently levels of energy consumption (e.g, Kenworthy, 2003), In the ease of transportation, the level of lravelactivity (Le, distances traveled, accessibility and feasibilty of public transi, etc) are influenced in partby urban attributes such as land use mix, population density, etc (Crane, 1999, Badoe and Miller, 2000 and Ratner and Goetz, 2013), Effective use of Urban form databases requires a suite of analysis tools to quantify relative transportation energy efficiencies of ities through a combination of sustainabily indicators and travel modeling Romote sensing has the potential to provide urban land surface Information. There are two key challenges in effective use of satellite imagery for these urban applications. First, hile there is a need to map temporal changes in overall urban extent 2.9.,.Kosin otal 2000), energy applications require more detailed intraurban land use information. For example, to address transportation energy use associated with work-related travel, “residential and commerciallindustra land use categories provide spatially explicit locations of tip orgins and destinations respectively. Satelite image classification suffers from spectral confusion between bull areas and other non-vegetated surfaces such as fallow agricultural land. This problem s especialy acute for large area mapping intiatves, for example the generation ofthe National Land Cover Datasets (NLCD) (Vogelmann et al,.2001 and Homer eal, 2007) as opposed to studies that restiict ‘mapping to only areas around cites (e.9., Lo and Chol, 2004, Lu and Weng, 2008 and Yuan sta, 2008). In the former case, since urban land only contributes a small fraction of land caver, commission efrors can be severe. To overcame this problem, hybrid processing involving satelite image classification, census and nightime light data have been developed (e.g., Vogelmann etal., 1998). The second challenge involves integration ofland surface information wth ‘demographics (e.g., population counts, employment statistics, etc.) Asuitable level of census detall must be selected thatis commensurate with the spatial mapping resolution of satelite data. Also, since census reporting areas are based on population count criteria, they are nol of constant areal extent nor are they ‘pure'in terme of land use. For ur synoptic information creation and analysis goals, we have selected census Information at the census tract level. Census tracts (CTs) tonto exnibitincreasing areal ‘ze with distance from city cores, reflecting radia decline in population density with distance from the city center. In addition, while core CTs encompass only urban land Uses, finge CTs can include significant contrbutions from non-urban land uses. The objectives ofthis paper are (a) to descrive the design and creation ofan urban haipuhwwnseioncodirect.com remo dgh ar. mis encolarictop/S03002424 15300884 p=y 208 181812016 Exploring the Ink between urban form and work related transport using combined stele image and census information: Case ofthe Great lakes re, haipuhwunseioncodirectcom remot dgh ar. mis encolariclop/S03002424 15300884 p=y information base thats tallored for analysis ofthe ink between the urban form and lransportation, (bt identify issues related tothe creation of equvalentinformation for Urban areas of ifferent countries thus supporting international comparisons and (c)to demonstrate the application ofthese information bases through the derivation and anlaysis of a relevant sustainabilly indicator and through travel acivty modeling. In Section 2 we describe the creation of abi-national database of urban form, the Great Lakes Urban Survey (GLUS). The following two sectionsllustrate the utlization of GLUS to study the urban form—transportation link. In Section 3 we propose anovel sustainability indicator that targets the feasillty of alternate modes of transportation Section 4 presents a modeling scenario o ilusirate how EO-based information can be exploited to quantily the impacts of urban form attributes on travel distance. Finally, Section 5 provides the conclusions resulted from this work 2. Creation of the Great Lake Urban Survey The Great Lakes are an important water resource, containing 84% of North America's and 18% of the work! fresh surface water resources, While the land in the northern part ofits watershed remains largely forested, he Souther portion is highly developed both in terms of urbanization and widespread agricultural use. The total urban population inthis rogion is over 3 milion, which includes roughly 10% ofthe US population and 30% of the Canadian population This paper builds upon earlier work involving the creation of a Canadian urban form information base, the Canadian Urban Land Use Survey (CUrLUS) (Zhang etal, 2010), and its use in analysis of he link between Canadian urban form and aspects of works related travel Znang and Guindon, 2006 and Guindon and Zhang, 2007), Based on ‘encouragement from the Air Qualty and Science Advisory Boards of the Intemational Joint Commission, the CUrLUS concopt was oxtended to create abi-national (Canada United States) information base, the Groat Lakes Urban Survey (GLUS) that would provide consistent urban form and travel information on all22 urban centersin the ‘watershed with populations in excess of 200,000 in the year 2000 (Table 1), These urban areas are designated as Census Metropoltan Areas (CMAs) in Canada and Metropolitan Statistical Areas (MSAs) in the United States, Both CMAs and MSAS include CTs of a ‘major city and eurrounding ural land and satelite communities. Fig. 7 llustrates the ‘extent ofthe overallwatershad and the rectangular portion of the watershed (1973 « 577.8 kmin extentin the east and north directions respectively) that has been ‘mapped using Landsatimagery for this study. Fig. 2ilustrates the outine of our information generation and integration strategies forthe creation of GLUS and for subsequent transportation-related analysis. sto major wan areas relent Great Lakes Ubon Suey. cuss Populon CMSA Pputaton (hleage-Cary, LN 9.197.540 Kalamazco-Batle Cook, MI 452,851 Clovelard-Asron, OM 2,988,891 Kicherer Wesco, ON e429 Makes Racine, 1,699,572 Saginaw-Bay City, Ml 409.070 Butale-Niagera Fas, NY SATO11- St, Catharnes-Nagara Fall, ON 377,006 Recher, NY 4,908,201 Windsor ON sorarr (rand Rapis-Muskegor-Holord MI 088.514 Oshaws, ON 296,298 amilon, ON 552401 Southend, IN 295.58 Teledo, OH 618.203 Gren Bay, 226,778 181812016 Exploring the Ink between urban form and work related transportation using combined steltimage and census information: Case ofthe Great lakes re, Fit Map shoting ho parton ofthe Great Lakes watorshed mappos with Landsat Themati Mapper rapes (octanglerbox) andthe locion af major esos (ed ets ncutea no Graal Las Urban Land Use So on ae noon I Soo rare mena ea _ aan Urban Fe icator Map aan co Gershciiy as i ee eg ‘and rots ae shown as gay, ght ray ad write boxes respectively 2.1. Generation of land cover land use information Land cover and land use (LCLU) information was extracted trom summer Landsat ‘Thematic Mapper images (summer acquisition dates inthe interval June 8, 1998-August 22,2001) using a sem automated classification methodology proposed earlier (Suindon ‘stal,, 2004). The approach uliizes a rule-based approach to combine pixel-based and haipuhwunseioncodirectcom remot dgh ar. mis encolariclop/S03002424 15300884 p=y 181812016 Exploring the Ink between urban form and work related transport using combined stele image and census information: Case ofthe Great lakes re, sogmentlovel classifications to (a) distinguish betwoen non-builturban areas (0.9., Parklands) and rural vegetated areas (b) to generalize urban land uses, such as residential suburbs and (c) to identity and compensate for large shadows incty cores. Classification was aided using 1:50,000 and 1:100,000 topographic maps and higher resolution satellite and aerial images as reference material Forty-six Landsat TM scenes, geocoded in the Albers Equal Area projection, were procured, classified and composited (Guindon and Edmonds, 2002) o create a seamless land cover land use product ofthe southern portion ofthe Great Lakes watershed that included the 22 urban centers of interest (Fig.2). The product includes four classes of urban land use, residential, commerciatindustal, "urban opentand’ and ‘ransitonat’. The frst two categories are particularly appropriate for travel modeling as they provide spatially «explicit information on trip origins and destinations for work-related commuting, Earlier accuracy assessment work (Guindom etal, 2004) found the praducer accuracies of these two classes to be 78 and 73% respectively. Openiand encompasses non-buit areas including parks, school yards and vacantland within an urban environment Transitional land is land undergoing urban development atthe time ofthe satellite image acquisition, Fig. 3 shows LCLU output of portions of four example cites in the region. The Cities include Chicago (54 x 54 km), Milwaukee (54 x 54 xm), Toronto (90% 90 km) and Hamilton (35 « 35 km). In allimage cases presented here andin subsequent images the artical image axes correspond to the north direction. (Craw onrons—) >. Information integration The integration of LCLU information with census data is key step in quantiying urban {form but presents a numberof challenges. One involves tne selection of an appropriate reporting evel thats commensurate withthe minimum mapping unit of the satelite imagery, While block level statistics may best match the spatial resolution of Landsat, two factors proclude its use, namely, cost of procurement and availabilty of ancillary information, such as transportation statistics, a this high level of dota Utilization of coarser census levels is complicated by the fact that these urban census reporting areas, are variable in spatial size since they are based on a population criterion. Typically they are smallestin city cores where population density is highest and gradually increase in ize ou from the city center. In addition, reporting areas at urban fringes tend to include both urban and rural land uses, haipuhwunseioncodirectcom remot dgh ar. mis encolariclop/S03002424 15300884 p=y 98 181812016 Exploring the Ink between urban form and work related transport using combined stele image and census information: Case ofthe Great lakes re, We have selected the census tract (CT) as our demographic unit forthe following (@) There is areasonable similarity between Canadian and US definitions of census {ract’. Canadian consus tracts are defined as ‘usually have a population of2500— ‘8000. They are located in large urban areas that must have an urban core population ‘50,000 or more’. The US Bureau of Census defintionis, ‘census tracts constitute reasonably compact, continuous land areas. Population size criteria are: Optimum 4000 Minimum:1600 Maximum:8000" ©) ‘The number of consus tracts comprising the cites in our study candidate lists large ‘enough to provide a reasonably detailed representation of urban form (ranging from 40-50 CTs for cities witn populations of 200,000 up to 2000 forthe Chicago MSA). Finally, the demographic parameters for each census tract used in this study include: (2) Total population (@.) Number of workers resident in he census tract (6) Number ofjobs (@.) Median commute distance (6) Number of workers using various modes of transportation for commuting The value of earth observation (EO) data in contributing to urban form characterization bestunderstood by studying the spatialimitations of demographic reporting areas such 8 CTs, Fig 4ilustrates, for the Rochester MSA, the variations in CT area and fraction of CTarea thatis actually urbanized (.e., classed as ‘residential, commercialindustral or “urban openland!) asa function of distance from the center ofthe city. There are two distinct radial distance regimes. At distances ess than approximately 20km i.e, the city core, CTs tend to be small and primarily urban as they constitute the city of Rochester proper. Beyond 20 km, a mixture of CTs is found, Most are much larger in extent than the ‘ore C7 and the fraction oftheir areas which is urban is small, There are also a small numberof CTs which are comparable in size to core CTs and have high urban content, These correspond to central regions of satellte towns. In conclusion, CT area and location definition are too coarse to properly characterize the actual extent of Urbanization and hence urban form, LCLU derived fram EO data, on the other hand, can provide refined spatial explictness of residential and employment areas within CTs, This refinementhas a number of key advantages. Firs, estimations of attributes such as residentiaemployment densities are greatly improved since the land areas of the land ses of interest are properly accounted for. Second, travel distance estimations are improved. Finally, the impacts on travel atvibule estimation arising from Canadian and US census tract differences can be expected to be minimized a) Cr Area em) é a ‘tance cy cate) Distance eCity Come em) Plots of CT aoa (et) ans CT urban action (ih) the constituent CTs ofthe Rachester New Yo MSA. of tance om te enter ne MEA for haipuhwwnseioncodirectcom remot dgh ar. mis encolariclop/S03002424 15300684 p=y 98 18/3016 Exploring the lnk between urn form and work related ransperation using combined satelite Image ard census information: Case of the Great lakes ro. Fig. 5 prosants examples of integrated urban form—transportation information n the {otm of color-coded maps for the example cities shown in Fig. 3. The information Portrayed, ie, median commute distance (Fig. SA) and percent of workers who drive to ‘work (Fig, 5B) is hased on place of residence statistics hence only pixels classed as residential have been color-coded, Commercialindustrial pixels are shown in grey while allruralpixels are black and water is shown in white for context purposes, These cles ‘oxhibit commute patterns typical of North American urban centers, namely the incroasing travel distance and private vehicle use with increasing distance from the ety center. PEW | a. Finally, we note that pixels within CTs thet have been classified as water have been ‘excluded from further consideration in the computation of indicator values and in travel haipuhwunseioncodirectcom remot dgh ar. mis encolariclop/S03002424 15300884 p=y 198 181812016 Exploring the Ink between urban form and work related transport using combined stele image and census information: Case ofthe Great lakes re, model calculations 3, The Employment Deficit Measure Widespread automobile use and long travel distances are two key contributors tothe high levels of pr capita transportation energy use and as such can be viewed as surrogate measures of energy use. Inthe past, policy makers have relied primarily on scalar indicators, such as overall ty population density to gauge urn transportation energy efficiency. These measures have two principal imitations. Fist, they lack sufficient spatial explcitness to analyze the effects ofintra-urban variation in indicator values, Second, that they lack a direct ink to transportation attributes. In this section we ‘demonstrate how GLUS can be exploited to quantify an indicator that provides a beller link between intra-urban land use mix and the feasibiliy of various modes of transportation Reports on sustainable transportation (¢.g., Center for Sustainable Transportation (CST), 2002) frequently propose land use mix or mixed use as a potentially useful indicator of travel effort, The CST study defines ‘mixed use’ as the degree to which units, of various urban land use classes are spatially interspersed. The implicit assumption behind ths indicators that he level of mix impacts opportunities fr local employment and therefore work-related travel patterns and the feasibilty of various modes of transportation. With the integrated information of GLUS, we are able to devise mix measures that better target work-related transportation. We propose a measure that can be quantified at different scales thereby assessing the feasbilly f alternate modes of transportation. This measure hereafter referred to as the ‘Employment Deficit Measure (EDM), can be expressed as: (MeN) cower) - SM) a EDM(R) estimates the numbers of jobs (N) and workers (N) within a given transit horizon, R, surrounding each residential location (pixel). The value of EDM is then a measure of the shortage (deficit) of jobs within this search radius. if EDM is high, this Implos that thore are many more workers than jobs hence It is Iikoly that travel to work ‘must involve commute distances greater than the transit horizon, Since the travel horizons of interest are comparable to the linear scale of CTs, itis necessary to take into account the intra-CT residential and employment distrbutions. This is accomplished, in the emplayment case, by assigning an employment weight to each commercialfindustrial pixel (ie. @ place of employment) equal o the total number of jobs in the CT of interest by the total number of commerciallindustrial pixels. A similar ‘worker weighting is applied to each rasidontial pixel. Fig. 6llustrates sults forthe three largest urban areas (Chicago, Detroit and Toronto) in terms of EDM, averaged over the whale CMA/MSA, as a function of transit horizon for horizon values ranging from 1.5 o 7.Skm. Two points are noted. Fitst, EDM decreases ‘with increasing transit horizon. Outside the cores of North American cities, residential and employment land uses are highly segregated on scales on the order of kilometers. ‘Asa result, fortransit horizons smaller than this segregation scale, EDM values can be ‘expected tobe high, Second, although there are significant overall density diferences between Toronto and the two US cities, EDM is similar at small tansithorizons. This suggests that these cities have similar low lovels of walking and cycling ‘riendlinoss ‘rom an uroan form perspective. tis only at high transit horizon values that Toronto's, EDMis lower than either Detroitor Chicago. At these scales, private vehicle use is the mode of choice. haipuhwunseioncodirectcom remot dgh ar. mis encolariclop/S03002424 15300884 p=y 95 181812016 Exploring the Ink between urban form and work related transportation using combined steltimage and census information: Case ofthe Great lakes re, haipuhwunseioncodirectcom remot dgh ar. mis encolariclop/S03002424 15300884 p=y Employmere Defet “Tanit Horizon on) Fio.6 Employment deft a function tral horzon for Chicago, Det ane Tern, Finally, Fig. 7 lustrates the relationships between EDM and population forthe two ‘extreme transithrizons of 1.5 and 7.5 km. Neither case exhibits a significant rend between these two variables. tis of interest to nate that fr the 1.5 km transit horizon case, US and Canadian cites exhibit similar levels of EDM regardless oftheir size. tis only at large horizon values that there is some suggestion that Canadian cities exhibit better land use mix (ie, lower EDM values). The one exception s the Oshawa CMA although tis an anomaly since a large portion ofits workers is employedin the Toronto (CMAhence EDM cannot reach low values, TransitHorizen=15km TransitHorizon=75 km Employment defi measure as ne horizon of 1.5 km Jf} and 7.5 gh on of poulaton for CMAs and MSA in GLUS for cass wth ran 4. Travel madeling using GLUS information GLUS includes @ level ofinformation detail that can support the developmentand assessment of more sophisticated tools improve understanding ofthe urban forr™— travel ink including modeling suilable for forecasting and visualization o effectively convey key results, 4.4. Example travel model Modeling isa more powerlul approach to travel analysis because of is potential for quantitative understanding of urban form factors influencing travel activity and for {forecasting applications. Travel modeling can be approached from micro or macro-based strategies. Inthe micro case city evel travel attributes are arrived at based on aggregation of ingividual tips. Our Urban databases are synoptic and not amenable to modeling althis level of detail. both in terms of spatial detall and supporting household travel surveys). Amacro 181812016 Exploring the Ink between urban form and work related transport using combined stele image and census information: Case ofthe Great lakes re, approach is therefore demonstrated here. We again emphasize that our objective isnot to develop definitive travel models but rather to demonstrate how integrated EO— demographic information bases such as GLUS can be effectively used in a modeling environment. ‘Since our focus is on travel prediction based on geospatial aspects of urban structure or {form we have selected agravty or spatial interaction model formulation for our iMustrative purposes. There are numerous publications describing the theoretical foundation of such models (Fatheringham, 1983, E:lander and Stewart, 990, Fotheringham and Wegener, 2000, Gitlesen and Thorsen, 2000 and Zhang et a., 2010). ‘Ageneral formulation forthe model in the case of transportation can be written as: tT, Way 2 where Ty s the expected travel flow from zone Ito zone, and Vj, W, and Qy are ‘measures of tip origin propulsion, destination atraction and trip deterrence respectively. In our workerelated travel application each CT can be considered a zone withthe land use ‘residential’ and ‘commercialindustral’ pixels acting as tip origins and destinations respectively. In complex modeling, the deterrence term can incorporate many factors other than simple distance between zones, for example, destination accessibiliy (Fotheringham, 1983) and hierarchical ordering or competition among potential destinations (Gitlesen and Thorsen, 2000), Because our goalis to focus on assessing urban form atributes on travel, we have limited deterrence tobe simply dstance-dependent. In addition, the destination attraction is deemed to be proportionalto 9, the number afjobsin the th employment canter. Asa result, our spatial interaction formulation can be writen as: I @ ar The parameter B encapsulates n an abstract manner a complex set of attributes that govern intra-urban mobilly such as efficiency of street networks, flexibly of available transportation modes, congestion, ec, This simple model cannot account for soclo- ‘economic factors such as income that can vary across accity and are known to impact, mobility. Itisimportantto address a number of practical issues related to application of gravity models at the MSA/CMA level. Spatial interaction models implicitly assume that each MSA/CMAis an isolated entity (Le, there is no flow of workers into and out of the MSAICMA) and that the mader' attraction and deterrence forces apply to each worker and ob, These constraints are not completely metin practice, ‘We wilillustrate these points using the Rochester, New York MSA.as an example. Fig. 8A ilustrates the LCLU product covering the whole extent ofthe Rochester MSA, (137 «137 km), This is typical in the sense that MSA/CMA areas greatly exceed the bull- Up area of the core city and thus include extensive rural lands. The smaller communities within a MSA/CMA are diverse, some of which may be bedroom’ communities and others self

    Вам также может понравиться