category: literaturenote citekey: buliunggistoolkitexploringgeographies2006 title: A GIS toolkit for exploring geographies of household activity/travel behavior authors: "Buliung, Ronald N.; Kanaroglou, Pavlos S." year: 2006 date: 2006-01-00 January 2006 doi: 10.1016/j.jtrangeo.2004.10.008 publication: Journal of Transport Geography url: "http://www.sciencedirect.com/science/article/pii/S0966692304000730" zotero_key: M6SVJTPB zotero_storage: D4FEC9U2 collections: magistritöö folder: 001_artiklid firstAuthor: "Buliung, Ronald N."

status: converted

Journal of Transport Geography 14 (2006) 35–51

www.elsevier.com/locate/jtrangeo

A GIS toolkit for exploring geographies of household activity/travel behavior

Ronald N. Buliung *, Pavlos S. Kanaroglou

School of Geography and Geology, McMaster University, Hamilton, Ont., Canada L8S 4K1

Abstract

Recent developments in geographic information systems (GIS) and increasing availability of activity/travel micro-data combine to encourage innovation in visualization and exploration of behavioral outcomes of urban processes. Understanding how households and individuals are using cities to conduct their daily activities has important implications for understanding policy response to urban travel reduction strategies. This paper describes a prototype object-oriented, GIS-based system designed to support exploration of household level activity/travel behavior. The capabilities of the system are demonstrated using households from the 1994/ 1995 Portland Household Activity and Travel Behavior Survey. An illustrative case study demonstrates that exploratory systems of this sort can be useful tools for generating timely activity/travel hypotheses for further study. 2004 Elsevier Ltd. All rights reserved.

Keywords: Activity/travel behavior; Households; Geographic information systems; Object-orientation

1. Introduction

The increased availability of detailed geographic data describing the activities and movement of individuals and households within cities alongside emerging geographic information systems (GIS) and spatial analysis technologies provides researchers with an unprecedented opportunity to study associations between activity/travel outcomes and urban form. The acquisition of activity-based survey data has evolved alongside recognition of the importance of non-work activities and related travel (e.g. Boarnet and Sarmiento, 1998; USDOT FHWA, 1995; Gordon et al., 1988; Greenwald and Boarnet, 2001; Handy et al., 2002). The growing contribution of non-work activities, to levels of urban travel, presents unique challenges related to micro-data management, geovisualization, and explanatory and

E-mail address: buliungr@mcmaster.ca (R.N. Buliung).

predictive modeling of individual and household activity/travel behavior.

The convergence of object-oriented technologies, GIS software, and activity/travel survey data can potentially facilitate a reduction in the complexity of activity/travel behavior data, giving rise to useful spatial and spatiotemporal constructs that can be used to enhance our understanding of household and individual level behavior. These constructs can also be used in empirical study of the sensitivity of micro-level behaviors to characteristics of land use and transportation systems. This paper documents the conceptualization and implementation of several prototype spatial and spatiotemporal tools that facilitate study of household level activity/travel behavior.

Several approaches have been adopted that have their origins in spatial statistics, computational geometry, and time-geography. State-of-the-art GIS-based, object-oriented tools, have been used for spatial database design, implementation and software development. The tools have been prototyped using data from the 1994/1995

^* Corresponding author. Tel.: +1 905 525 9140; fax: +1 905 546 0463.

Portland, Metro Household Activity and Travel Behavior survey. The survey data, alongside other contextually relevant geographic data, have been organized and stored in an object-relational geographic database. This database provides an information foundation for the activity analysis tools. The capabilities of the existing set of tools include generalization of household level activity geocodes to several area-based geometries. These include weighted and un-weighted standard distance circles and standard deviational ellipses. To this end, we also introduce and discuss a new spatial unit that we have termed the household activity space. We have also developed the capability to assemble household level space–time trajectories from the geographic and activity-timing properties defined for activity events. We provide sample applications of these tools and a broader exploratory exercise designed to demonstrate the capabilities of the prototype system for hypothesis generation.

The paper has six sections including the introduction. In Section 2 we review conceptual and applied research that attempts to develop the activity space concept to address spatiotemporal behavioral elements of several substantive research areas. The latter part of the section concentrates on recent convergence of GIS, objectoriented technologies, and concepts from research focused on activity/travel behavior, accessibility, and time-geography. In Section 3 of the paper, we discuss the geographical setting of our research and introduce the data that we are using. In Section 4, the development and implementation of our set of activity/travel behavior tools is discussed in detail. The section addresses both technological and methodological details focusing on database development, assembly of the analytic toolkit, and the origin of adopted approaches. In Section 5 of the paper we develop a case study demonstrating the capabilities of the prototype for geographical exploration of household activity participation. The section ends with identification of several activity/travel hypotheses for future research. The sixth and final section of the paper addresses conceptual, methodological, and software development issues generated by this research and documents several future research objectives.

2. Geographies of activity/travel behavior and modeling tools

Over the last four decades, several studies have attempted to conceptualize and study spatial units describing the actual and potential activity participation of persons and households (Table 1). The early literature can be described as theoretically rich, with considerable focus given to conceptualization of the spatiotemporal behavior of individuals, and to a lesser extent, households. While theory was operating at the micro-level, applications were focused primarily on zone-based implementations of activity-based constructs. This changed in the early 1990s with increased availability of micro-level activity/travel survey data coupled with the spatial data management and analytic capabilities of GIS.

In many ways the literature since the 1990s serves as a response to comments by Beckmann et al. (1983b) with respect to shifting the scale of analysis in the presence of increasingly detailed data describing the urban environment, activities and travel, and behavioral units of interest. Much of the work from the 1990s to the present has focused on specification and implementation of timegeographic, geocomputational algorithms for estimating spatial units that describe either the revealed or potential activity participation of individuals over space. GIS capabilities have been used to support the majority of the research published since 1990 that has been cited in Table 1. The more recent literature remains, at a fundamental level, linked to earlier theoretical and methodological examples included in our review.

While several software environments support visualization, exploration, and modeling of geographic processes (Anselin, 2000; Bailey and Gatrell, 1995; Levine, 1996) few exist that support interactive geovisualization and exploration of activity/travel behavior. In an early example, Makin et al. (1997) combined time-geographic concepts with object-orientation and GIS to develop an operational shopping simulation model. Their model simulates individual shopping behavior in the presence of constraints related to facility operating hours, work schedule, personal mobility, and transport supply. Spatial and temporal entities, and constraints are implemented in an object-oriented GIS called SmallWorldTM.

Efforts have recently been extended to development of prototype database systems that support query and retrieval of activity-based micro-data (Claramunt and The´riault, 2001; Frihida et al., 2002; Wang and Chen, 2001). Claramunt and The´riault (2001) document a process of object-oriented analysis and design (OOAD) that concludes with implementation of an activity/travel database in ArcGIS. Their work serves as a useful demonstration of links between object-oriented database and GIS technologies. Frihida et al. (2002) have developed a prototype system for exploring individual level activity-based travel behavior using data from a 1991 origin–destination survey conducted in the Que´bec City region. Like Claramunt and The´riault (2001), the behavioral unit of interest in their research is the individual. Persons are modeled as complex aggregations of activity/travel concepts (e.g. trips, activity locations, plans and schedules). Their system responds to spatiotemporal queries and, like Makin et al. (1997), has been implemented in Smallworld GISTM.

Adopting an extended entity relationship (EER) approach, Wang and Chen (2001) have developed a

Table 1 Spatial and spatiotemporal models of individual and household activities

Citation	Representation	Implementation	Categorya
Lewin (1951)	Lifespace	Field theory	6
Wolpert (1965)	Action space	Theoretical framework influenced by Lewin (1951)	1
Wolpert (1967)	Metropolitan migration fields	Projected directional sectors coupled with exponential distance zones	1
Hurst (1969)	Movement space: core, median, extensive	Conceptual framework	5
Moore (1970)	Urban contact fields	Probability density functions	6
Ha¨gerstrand (1970)	Constrained spatiotemporal trajectories	Theoretical framework	6, 3
Brown and Moore (1970)	Aspiration region, awareness space, indirect contact space	Theoretical framework	1
Horton and Reynolds (1971)	Activity space action space	Zone-based familiarity index	5
Johnston (1972)	Sectoral activity space uniform action space for homogenous groups of individuals	Residential desirability surfaces (trend surfaces)	1
Lenntorp (1976)	Space–time prism, potential path space (PPS), potential path area (PPA)	Conceptual and operational definition	2, 3
Burns (1979)	Space–time prism, PPA	Conceptual space–time prism (conic) with elliptical PPA	2, 3
Zahavi (1979)	Travel probability fields	Standard deviational ellipse	4
Beckmann et al. (1983a,b)	Travel probability fields	Theoretical framework and empiricism (ellipse, density functions)	4
Miller (1991)	PPA	GIS: network algorithm	2, 3
Golledge et al. (1994)	Feasible activity locations	GIS: shortest path, buffers	5
Newsome et al. (1998)	Observed activity space	GIS: elliptical model of home-work trip chains	5
Dijst (1999)	Actual action space	GIS: circular, elliptical, linear	5
Kwan (1999a,b)	PPA, Daily PPA (DPPA)	GIS: DPPA	2, 3, 5
The´riault et al. (1999)	Sets of geographical entities	GIS: convex hull, dispersion ellipse	5
Miller and Wu (2000)	Space–time accessibility measures (STAMS)	GIS: econometrics, space–time utility functions	2, 3
OSullivan et al. (2000)	Space–time isochrones, space– time locations	GIS: isochrones	3
Weber and Kwan (2002)	PPA, DPPA	GIS: DPPA	2, 3
Kim and Kwan (2003)	Space–time prism, PPA	GIS: PPA, constrained service areas	2, 3
Scott (2003)	Space–time prism, PPA	GIS: PPA	2, 3, 5
Scho¨nfelder and Axhausen (2003)	Observed activity space	GIS: kernel, ellipse, minimum spanning trees	5
Miller (in press)	Space–time path, prism, lifeline, stations, bundling, intersections	Theoretical and analytical development of concepts	2, 3, 4

^a Substantive focus: 1. urban migration; 2. time-geography; 3. accessibility; 4. trip-based travel demand; 5. activity/travel behavior; 6. other.

spatiotemporal data model for exploring the daily activity/travel behavior of individuals in space and time, alongside several characteristics of the activity/travel setting. The database is implemented in a GIS environment with additional development of a user-interface that facilitates time, activity, person, and location-based query of database content. Prototype implementation includes a case study focused on the activity/travel behavior of individuals drawn from six families in Hong Kong. The system is comprehensive, supporting exploration of both the activity/travel behavior outcomes of household members and the spatial and temporal characteristics of urban opportunities. For example, individuals can be described in terms of their sociodemographic characteristics, activity programs (defined as a list of pre-planned activities), and activity participation over specified time intervals. Background activity/travel environment queries can be used to identify locations that support specific types of activities (e.g. shopping) coupled with timing information (e.g. hours of operation).

Schwarze and Scho¨nfelder (2001) have developed the VISAR (Visualisierung von Aktionsra¨umen) extension for ArcView. In a recent study, Scho¨nfelder and Axhau sen (2003) use VISAR to characterize individual activity spaces with data from the Mobidrive travel diary survey conducted in the German cities of Halle/Saale and Karlsruhe in the autumn of 1999. They emphasize the potential of the ''activity space'' concept as a tool for examining the process of social exclusion. Adopting a more general substantive focus, Shaw and Xin (2003) have developed a prototype GIS-based system for exploring spatial and temporal interactions between transportation and land use systems. System capabilities are demonstrated using data from Dade County, Florida. Analysts can use the application to study relative changes in transportation and land use characteristics at different points in time.

Liu and Zhu (2004) have developed an extension to ArcView GIS called the ''Accessibility Analyst''. The extension is a unique and comprehensive collection of tools for measuring accessibility and travel impedance, coupled with advanced spatial interpolation and geographic visualization capabilities. The tools are applied in a case study examining accessibility impacts of new mass rapid transit (MRT) lines on access to the working population living in public housing estates in Singapore. While the system is quite comprehensive, the authors note that its use is currently limited to large scale, zone-based transportation planning. They also indicate that the extension cannot presently accommodate micro-level, activity-based spatiotemporal behavior. The most recent studies that we have reviewed in this section are similar with respect to substantive focus, but also share common ground in terms of developmental approaches. Similarities include, in some manner, adoption of object-oriented design and programming approaches, and incorporation of the spatial data management, visualization, and analysis capabilities of GIS.

3. Geographic setting and data sources

Transportation and land use data have been drawn from two sources in Oregon State. These include the 1994/1995 Household Activity and Travel Behavior Survey and geographic data layers from the Regional Land Information System (RLIS). These data are managed by a regional government organization known as Metro. The jurisdiction of Metro includes the urbanized portions of three counties in Oregon State (Washington, Clackamas, and Multnomah) and a population of approximately 1.3 million. Metro is the acting Metropolitan Planning Organization (MPO) for the region and is the only elected regional government organization of its kind in the United States. With a population of 529 121 (2000 US Census) Portland is the largest city within Metros jurisdiction.

Population and employment projections for Metro suggest that over the next 20 years increasing demands will be placed on existing transportation and land use systems (Metro, 2000). Metro has adopted an integrated transportation and land use planning strategy in an attempt to accommodate urban growth while meeting fiscal, environmental, and social objectives. To support investigation of transport and land use policy alternatives, Metro has engaged in collection and study of detailed micro-data capturing the mobility and activities of households within the region. The 1994/95 Household Activity and Travel Behavior survey emerges as the most comprehensive survey conducted in the region during the last decade.

The survey includes detailed description of the activity/travel behavior of 4451 households (63% of 7090 recruited households). At the individual level, 9471 respondents reported a total of 122 348 activities and 67 891 trips (Cambridge Systematics, 1996). The survey was completed during the spring and autumn of 1994 and the winter of 1995. The survey data have been used extensively in research ranging from identifying determinants of various activity/travel behavior outcomes (Golob and McNally, 1997; Golob, 2000; Greenwald and Boarnet, 2001; Kim and Kwan, 2003; Reiff and Kim, 2003; Sun et al., 1998; Weber and Kwan, 2002) to development of operational activity-based model systems (Mark Bradley Research and Bowman, 1998; Wen and Koppelman, 1999).

The survey contains characteristics of households, individuals, modes of travel, and the activity/travel behavior of household members. Households recorded activities and travel for two consecutive days. Activity characteristics include type (e.g. meals, work, shopping etc.), location, timing, mode, and vehicle availability. Recorded activities also include those that did not directly require travel. The end result has been expansion of detail with respect to activities that occur at the home location. Respondent activity locations were geocoded using intersection and address-based methods, with the exact approach depending on detail and accuracy of respondent reported location information. Geocodes are reported to be accurate to approximately 200ft of actual activity locations (Cambridge Systematics, 1996). From these geocodes, researchers have been able to construct and study spatial patterns of household and individual activity participation (e.g. Buliung, 2001; Kwan, 2000).

We have also retrieved geographic data from the Metro Data Resource Centers (DRC), Regional Land Information System (RLIS). The RLIS system was developed during the 1980s to support planning initiatives and regional land use and transportation modeling. It is an integrated, detailed, regional geographic database that includes data from 24 cities and 3 counties. Regional traffic analysis zones (TAZ), planning districts that can be used to subdivide the region into suburban and urban zones, and political boundary layers have been extracted from RLIS. Both survey data and a subset of the RLIS data can be acquired from Metros web space.

4. System characteristics and implementation

Associations emerge from the activity/travel literature between activity patterns, urban spatial structure, and the characteristics of individuals and households. With respect to urban spatial structure, Zahavi (1979) comments that the changing state of an urban area can impact patterns of travel and other properties that are now generally considered part of the activity paradigm (e.g. mode choice, distances traveled, activity type distributions etc.). One way to describe household activities using a geographic framework is to identify spatial locations, visited by household members, over a particular time period. Taken together, these separate locations describe spatial patterns of household activity participation. Coupled with activity timing information (e.g. start, end, duration), these patterns provide an opportunity to study the spatiotemporal behavior of households. With this in mind we have implemented three area-based approaches and a fourth, time-geographic approach, for exploring the collective behavior of household members.

A set of exploratory analysis tools has been developed, in a GIS-based environment, which access and process activity/travel data stored in an underlying object-relational geographic database. The activity/travel outcomes are processed using methods adopted from spatial statistics, computational geometry, and timegeography. Two and three-dimensional geometries are produced that facilitate state-of-the-art exploration of household level spatiotemporal behavior. When discussing household activity patterns in the remainder of this section we are referring to the set of spatial locations visited by household members on survey days and corresponding timing characteristics.

4.1. Data integration and modeling

The intrinsic complexity and potential for continued use of the Portland survey, and RLIS data in ongoing research lead us to adopt a formal object-oriented analysis and design (OOAD) approach to database development. Recent advances in GIS and object-oriented technologies were additional factors contributing to this decision. Our application of OOAD principles contributes to recent literature, documenting links between geographical research focused on human activities and object-orientation (Bedard, 1999a; Claramunt and The´ riault, 2001; Frihida et al., 2002; Makin et al., 1997). Out of this process emerged a database design model. This model documents the content and structure of an activity/travel behavior spatial database that has been implemented as an ArcGIS Geodatabase (Fig. 1).

The data model was developed with the aid of the Unified Modeling Language (UML) notation and the Microsoft Visio visual modeling tool (Booch, 1994; Rumbaugh et al., 1999; Taylor, 1998). The design model has been specified to logically separate geographic (Feature Classes) and non-geographic (Tabular Classes) content. Boxes in the design model are classes that, when implemented, become tables in an objectrelational database. Line segments connecting classes identify relationships that can be used to navigate through database content. Attributes have been defined on all classes but have been suppressed for some classes in the diagram to facilitate display of the entire data model. In addition, the operations box on most classes has been suppressed, as we have not yet identified requirements for extending classes with custom behaviors beyond those inherited from ArcObjects.

In earlier work we have described, in detail, the process involved in designing a similar database (Buliung and Kanaroglou, 2004). The schema reported in this paper is a revised form of an earlier database. The revision process was stimulated by the presence of additional data from Metro, and the evolution of our application development objectives. A geographic data layer of regional planning districts (PlanningDistrict) has been included alongside several, household level, urban diversity and design variables. These measures have been programmed as properties of the PlaceofResidence class (Fig. 1).

Urban diversity measures include variables describing the distribution of households and employment opportunities around geocoded residential locations. For example, at a radial distance of one mile around each residential geocode, the HHHM property records the number of households, while RETEMPMI records the number of retail employment opportunities. These variables fall under the general class of cumulative opportunities measures in the accessibility literature (Handy and Niemieier, 1997; Handy and Clifton, 2000). Urban design or connectivity measures focused on local assessment of network geometry have also been provided and include the number of intersections (INTHM, INTMI) and cul-de-sacs (CULHM, CULMI) at radial distances of half and one-mile respectively around residential sites. The planning district data facilitate allocation of household residential locations to urban or suburban jurisdictions, while the urban diversity and design measures provide an opportunity for future exploration of associations between activity/travel behavior, transport supply, and land use patterns (e.g. Boarnet and Greenwald, 2000; Friedman et al., 1994; Handy, 1993; Hess and Ong, 2002; McNally and Ryan, 1993).

The approach to database development that we have adopted holds both conceptual and software development advantages. Jackson (1994) has argued that, conceptually, object-orientation provides a set of constructs that support intuitive abstraction of regional science concepts and processes. Recent research lends support to Jacksons position, demonstrating the utility of object-orientation as a more instinctive approach to abstraction of activity/travel concepts for behavioral research (Claramunt and The´riault, 2001; Frihida et al., 2002; Makin et al., 1997). The data model that we have developed contains several activity/travel, and urban

Fig. 1. Activity/travel database schema.

geographical concepts, specified in a logically consistent manner. Translating concepts such as traffic analysis zones, households, and persons, into classes of the same name, containing several descriptive attributes, is a relatively basic task that results in a powerful information foundation for ongoing application development and empirical research.

The adoption of an object-oriented approach is also advantageous from a software development perspective. First, the process of database revision is enhanced by the presence of the original data model, the visual modeling environment, and the re-use and extension properties that characterize object-oriented approaches to system design. Second, we have programmed detailed class and relationship descriptions directly into the design model. These descriptions include spatial characteristics of the data, information sources and contacts, and detailed field descriptions. The design model serves as both a structural diagram of database content, and as a rich metadata repository. Third, using relationships programmed between classes during design, we are able to operationalize object collaboration in support of query, retrieval, and assembly of individual and household level activity patterns. In this sense the approach to data management that we have adopted supports rapid development of exploratory tools. Lastly, as Bedard (1999b) has suggested, the use of a standardized notation for database design, coupled with detailed class and relationship descriptions, potentially reduces dependence of future researchers on the experience and knowledge of the original database programmer.

4.2. Activity analysis toolkit

To date, research characterizing the geography of activity/travel behavior has primarily focused on individual household members. Interest in household level modeling is supported by research demonstrating relationships between household characteristics and personal activity/travel behavior (e.g. Bhat, 1996a,b; Bhat and Koppelman, 2000; Hanson and Hanson, 1981; Lu and Pas, 1999; Sun et al., 1998; Yee and Niemeier, 2000), and increasing recognition that complex household decisions processes influence the success and

Fig. 2. Activity analysis tools.

evaluation of travel demand management (TDM) strategies (e.g. Bhat, 1996a,b; Gliebe and Koppelman, 2002; Golob, 2000; Golob and McNally, 1997; Lu and Pas, 1999; Scott and Kanaroglou, 2002). While we typically do not observe household decision making, we can explore and measure the outcome of this process at the individual and household level.

We have developed a set of analytic tools for spatial and spatiotemporal exploration of household level activity/travel behavior (Fig. 2). The tools are unique in the expression of activity/travel behavior at the household level. Area-based measures provide an opportunity to explore coverage, dispersion, and orientation of household activity patterns. Implementation of timegeographic concepts facilitates visualization of cases where household members engage in activities co-located in space and time. We have not referred to co-located activities as joint activities because, while multiple householders might be at the same location in space and time, they could be engaged in separate activities. The household space–time trajectories that we have developed also support visualization of the geometrical complexity of household level activity patterns.

The time-geographic component of our toolkit represents an incremental step toward exploration and measurement of activity/travel interactions between household members. Beyond substantive considerations, households are used as a data management device for organizing individual patterns in a logically consistent manner. From a behavioral perspective, this facilitates effective visualization and exploration of activity-based interactions between household members while describing the spatial scope and orientation of household level activity patterns.

The tools that we have developed serve as an analytic interface to the underlying activity-based, objectrelational spatial database, discussed in the previous section. Activity patterns can be explored using a variety of approaches originating in spatial statistics, computational geometry, and time geography. Development of these tools involved programming of ESRI ArcObjects using Visual Basic for Applications (VBA). ArcObjects is an extensive set of Component Object Model (COM) compliant objects that facilitate customization of ArcGIS to meet the demands of specific applications (Zeiler, 2001). ArcObjects can also be extended using any COM-compliant programming language (Zeiler, 2001). The end result is a powerful GIS-based development environment that encourages interoperability and re-use of software components (Zeiler, 2001).

Embedding the prototype system within the ArcGIS environment provides access to tools that are native to the existing GIS in addition to our custom analytic capabilities. By extending the core capabilities of Arc-GIS our application is an example of the encompassing framework to coupling spatial analysis capabilities and GIS (Anselin and Getis, 1992; Anselin, 2000). Historically, customizations of this sort have relied on proprietary GIS-based scripting or macro languages (Anselin, 2000). This has presented interoperability challenges due to functional and semantic variation across languages (e.g. MapBasic, AML, Avenue etc.). Convergence on VBA and COM makes the approach that we have followed more accessible to analysts, programmers, and across GIS systems.

Two different approaches can be used to begin investigation of household activity patterns. The most general of these involves interactive spatial selection of place-of-residence objects from a place-of-residence geographic data layer using the Household Activity Pattern tool (identified on the toolbar in Fig. 2 using a point feature icon). This event initiates a search through the database to retrieve related SpaceTimePath objects from the underlying database (Fig. 1). The location coordinates for selected activity objects are converted to point features and stored in household activity layers. Separate layers are automatically constructed for each survey day along with a third layer containing the joint distribution of activities. These separate layers represent point patterns of household activities open to further exploration using the other activity analysis tools.

The second method involves selection of specific types of households using the Household Selection tool (Fig. 3). Households can be chosen from a three-dimensional household structure typology based on the presence of adults (20 6 age 6 59), children (age 6 12), and adults with or without paid employment (full or part-time). At this stage in development, the typology is meant as an illustrative device that can be adjusted to accommodate a broader range of household types. In addition, the selected set can be reduced to include households that are living in planning districts defined by Metro as predominantly urban or suburban. House-

hold activity patterns can then be constructed by applying the Household Activity Pattern tool to selected households. This capability facilitates exploration of the geographic characteristics of activity participation for similar types of households and exploration of the spatial properties of household activity patterns in urban and suburban subregions of the study area. The conceptual and mathematical foundations of the adopted pattern exploration approaches are discussed in greater detail in the next section.

4.3. Spatial statistical approaches to activity pattern description

The spatial organization of functions within cities, tastes and preferences of individuals and households, and the presence of constraints (e.g. Ha¨gerstrand, 1970), may combine to influence the selection and performance of activities at locations that are nonuniformly distributed in space. Spatial statistics provides several approaches that can support exploration of patterns of point events materializing from individual and household decision processes. Our application can generate the mean center, standard deviational distance (standard distance), and standard deviational ellipse for exogenous daily household activity geocodes (Fig. 4). These measures can be weighted to control for properties such as activity duration (Fig. 2). The computational formulae for these measures are well known (Bachi, 1963; Ebdon, 1988; Earickson and Har lin, 1994; Levine, 2002; Yuill, 1971) and not reported here. To be clear, we have adopted the formulae specified by Levine (2002) to produce unbiased estimates of standard distance and the standard deviational ellipse. Weights are applied to the standard distance estimation as in Earickson and Harlin (1994) and to the standard deviational ellipse estimation as in Yuill (1971).

Standard distance is the standard deviation of the distance of points, in a point pattern, from the mean center of the set (Bachi, 1963). A large value for the standard distance implies a relatively disperse pattern of point events. In our application, we estimate the standard deviation of the distance of household activity location geocodes to the mean center of the household activity pattern. We then represent this standard distance spatially by generating circles with radii set to the standard distance estimate (Fig. 4). Using this approach provides a useful, simple geometry, for comparing the dispersion of activity events for different types of households.

While standard distance provides a reasonable single measure of dispersion, its estimation tends to exaggerate the effect of spatial outliers. This is problematic when studying household spatial behavior due to the effect of distant activity locations on interpretation of activity Fig. 3. Household selection tool. pattern dispersion. The standard distance might suggest

Fig. 4. Area-based models of household activity patterns.

that a household has a dispersed pattern of visited activity sites when in fact the observed effect could be attributed to the presence of infrequently occurring, distant activities. Additionally, and perhaps of greater consequence, is that standard distance cannot be used to investigate pattern orientation or shape under conditions where a spatial process gives rise to a directionally biased event distribution (Ebdon, 1988; Levine, 2002; Yuill, 1971). Intuitively, we might expect individual and household activity patterns to be directionally skewed given the spatial characteristics of land uses and transportation systems within cities. Responding to this, we have implemented the standard deviational ellipse as an additional approach to investigating household activity patterns (Fig. 4).

The standard deviational ellipse (SDE) is a useful construct for exploring spatial patterns of household and individual activities (Fig. 4). While working from a trip-based perspective, Zahavi (1979) suggested that one advantage of this type of approach is the powerful visualization capability offered by a rather simple geometry, used to characterize the spatial distribution of many trips. We go further suggesting that the ellipse, and more generally, area-based measures of activity patterns, can serve as useful mechanisms for describing the orientation and spatial scope of sets of dissimilar, yet potentially interdependent, spatially linked activities. In our approach, ellipses are focused on the mean center of household activity patterns. Elliptical properties are estimated using the location data associated with the daily activities of individuals within households. In this sense, the resulting ellipse characterizes the spatial behavior of households (Fig. 4). Our tool reports elliptical properties that can be used to describe the statistical and geometric characteristics of household activity patterns. These properties include: mean center, area, perimeter, length of axes, eccentricity, ratio of major to minor axis, number of activities, and angle of orientation (Fig. 2).

The use of the SDE for describing spatial patterns of events is not a recent development. In fact, initial specifications emerged during the 1920s (Furfey, 1927; Lefe ver, 1926). Yuill (1971) develops examples related to spatial commerce and agriculture and appears to be responsible for introducing the SDE concept to geographical research. More recently, ellipses estimated using a variety of approaches, have been used to characterize patterns of human activities and travel (Table 1). The behavioral, methodological, and implementation foci of our approach sets it apart from these applications. Most recent examples have used ellipses to examine activity spaces of individuals (Dijst, 1999; Newsome et al., 1998; Scho¨nfelder and Axhausen, 2003) or to typify movement patterns for sets of individuals (The´riault et al., 1999; Zahavi, 1979). These are useful exercises complemented by our exploration of activity/travel behavior at the household level using geocoded microdata describing the activity locations visited by household members.

With respect to implementation, our approach enables interactive GIS-based visualization and exploration of household activity patterns in a computationally efficient manner (Fig. 4). Household ellipses can be simultaneously drawn on screen and stored in geographic data files for future reference. Elliptical properties are also reported and stored as attributes of features stored in the geographic data files. Due in part to the absence of mature computing and spatial technologies, Zahavi (1979) for example, would not have been able to achieve this at the time of his innovative work. The exploratory capabilities of the pattern dispersion measures discussed in this section are demonstrated further in the case study discussed in Section 5 of the paper.

4.4. Household activity space

So far, we have discussed applications of descriptive spatial statistical procedures for studying spatial distributions of point events. There exists no de facto standard for depicting household activity patterns using area-based geometries. From this perspective, it is useful to explore a range of possible measures. Dijst (1999) for example, constructs circular, elliptical, and linear action spaces to explore the activity/travel behavior of individuals in two-earner families. Interestingly, his study implicitly demonstrates that behavioral insights drawn from activity/travel measures are not necessarily consistently observed across different measures.

Considering the spatial behavior of individuals, Hor ton and Reynolds (1971) developed the action space and activity space concepts. The former describing a choice set of urban alternatives for which individuals possess sufficient knowledge to assign preference, and the latter comprising a subset of activity locations visited by an individual during participation in daily activities. Informed by the action space/activity space literature (Table 1) we are introducing the household activity space, as an area-based measure for studying daily household activity patterns. We define the household activity space as a convex hull containing activity locations, visited by the members of a household, during the course of a single day.

We have adapted the analytical expression for a convex hull to reflect the household activity space concept. For a household, h, consider the set of daily activity locations, S = {l1h,l2h,...,lnh}, where lih for i = 1,...,n represents the geocodes for the locations of the n activities of household h. If we drop the index h from all expressions for brevity, then the daily household activity space A, is the set defined as

$$A \equiv \left{ \sum{i=1}^{n} \lambda{i} l{i} : \lambda{i} \geqslant 0 \text{ for all } i, \text{ and } \sum{i=1}^{n} \lambda{i} = 1 \right}$$

The expression implies that all of the weights, ki, take on values between 0 and 1, inclusive. A specific combination of the ⁿ weight values provides a point ^Pⁿ ⁱ¼¹kili in Euclidean space. The collection of all such points for all possible combinations of the weights provides the convex hull. In practice, convex hulls are typically constructed using computational geometry algorithms. While several algorithms have been developed for this purpose, we have solved the problem by applying an ArcObjects hull method to a geometry object containing household activity location objects (Fig. 4). A Quickhull algorithm (Barber et al., 1997) has been implemented in ArcObjects as the computational method for generating convex hull polygons.

Taking a household activity pattern for a particular survey day as input, the Household Activity Space tool (Fig. 2) creates geographic datasets containing geometries and properties that can be explored further using the geovisualization and mapping capabilities of Arc-GIS (Fig. 4). Properties include: area, perimeter, number of household activities, and a diagnostic code used to identify errors that can occur during hull creation. While the household activity space is a useful construct for visualizing the spatial scope of daily patterns of household activities, its geometric properties are affected by the location of spatial outliers. That is, distant activities in a pattern are more likely to influence hull construction than geographically clustered activities (Fig. 4). This becomes less of an issue for households with few distant activities but could lead to inappropriate conclusions with respect to household activity patterns in the presence of distant, but less frequently occurring activities. These types of effects could potentially be controlled through longer periods of observation. Unfortunately, while the Portland data are extensive, household activity/travel behavior has only been recorded for two consecutive days.

With this in mind, the household activity space concept has been translated to a basic geometry that can support visualization and exploration of the spatial characteristics of household activity participation. While the household activity space has been developed to measure daily activities it can be conceptually and practically refined to include activities recorded over a longer time period.We have also developed batchprocessing capabilities for this, and the spatial statistical measures, to support future research that will examine the sensitivity of household activity patterns to urban form and transport supply characteristics. Theoretically, empirical study of the area-based measures will inform the current debate focused on the effectiveness of land use policy as a tool for reducing urban travel. Our a priori expectation is that households living in areas characterized by dense, mixed-use development will have, on average, smaller activity spaces than other households. Having detailed the area-based approaches, we conclude this section with a discussion of our implementation of time-geographic concepts.

4.5. Household trajectories in space–time

Depicting daily movements or trajectories of individuals, in an urban setting, using three-dimensional paths in space and time emerged as part of the timegeographic approach during the 1970s (e.g. Burns, 1979; Ha¨gerstrand, 1970; Lenntorp, 1976; Lenntorp, 1978; Pred, 1977). Researchers have implemented and studied time-geographic concepts in efforts designed to measure potential opportunities available to individuals in space–time given external and personal constraints (e.g. Kim and Kwan, 2003; Kwan, 1999a,b; Miller, 1991). In a recent study, Dijst (1999) develops and studies the actual action space concept for individuals from twoearner families in two communities in the Netherlands. The actual action space is used to characterize activity locations visited by individuals that have been recorded in an activity/travel survey.

Actual or revealed activity/travel behavior of individuals has also recently been investigated through GISbased implementations of space–time path concepts (Kwan, 1999a,b, 2000). While Kwan demonstrates the utility of individual space–time paths as an exploratory mechanism, the potential of time-geographic concepts for interactive geovisualization of individual behavior has not been investigated. Additionally, the potential for visualizing and exploring interactions within households has not explicitly been addressed in recent research. Household level implementation promotes discovery of interesting interactions between individuals that can potentially be attributed to household roles and responsibilities. These types of geovisualization exercises possess considerable potential for communicating, in an intuitive manner, what at times may appear to be somewhat abstract discussions of gender relations, household roles, and urban travel.

While space–time paths can be studied at different spatiotemporal scales, we are interested in exploring the daily trajectories of households and individuals. At this scale of analysis, geocoded activity sites visited by household members define daily locations in planar space while the timing (start and end times) information for these activities define locations in time. Classical time geography introduces the concept of the bundle to describe the grouping of several space–time paths (Ha¨gerstrand, 1970; Miller, in press). We have implemented the bundle concept, providing an innovative approach for visualizing household space–time behavior as an assemblage of the space–time paths of individual household members. Using a selected household activity layer as input, our Household Trajectories tool (Fig. 2) automatically constructs household space–time path data structures for all survey days. The approach facilitates exploration of joint activity/travel behavior in a computationally efficient manner. Our implementation also eliminates interaction with complex three-dimensional activity/travel data stored in the underlying database.

Resulting path structures can be studied in planar or three-dimensional space (Fig. 5). The example that we provide demonstrates that the spatiotemporal household path is more clearly understood when rendered in three-dimensions. A single weekday of activities, for two households, is shown in Fig. 5. Household (A) has a residential location in the Portland CBD, while household (B) is located in a suburban planning district. The sample households clearly have very different trajectories. The central household has a

Fig. 5. Household activity trajectories in space–time.

less dispersed pattern that is focused around the home location while the suburban household has a male-commuter traveling toward a central workplace destination. While male and female trajectories are similar for the centrally located household this is not the case for the suburban household. Closer inspection of the activity information indicates that the female householder is conducting more household maintenance activities, closer to home, than the male householder. It is also of interest to note that the central householders are selfemployed, whereas the suburban case consists of a male with full-time, paid employment, outside the home and a female with self-employment. The illustrative cases provide a useful spatiotemporal characterization of literature dealing with gender, household roles, and employment related travel (e.g. Blumen and Kellerman, 1990; Johnston-Anumonwo, 1992; Madden, 1981) but can also serve to advance the debate by focusing on a wider range of activities beyond journey to work (e.g. Law, 1999).

5. Exploratory case: households, life-cycle stage, and residential location

In this section of the paper we report on a demonstrative exploratory investigation of differences in dispersion and orientation of activity patterns for households with and without young children, living in urban or suburban planning districts. Household composition is based on the presence of two adults (20 6 age 6 59) with paid employment (full or parttime), and the presence or absence of young children (age 6 12). The exercise focuses on description of household activity patterns using the un-weighted, standard deviational ellipse. We vary the residential location to illustrate the capabilities of GIS-driven analytic tools for investigating and generating hypotheses concerning associations between urban form and activity/travel behavior outcomes. The exercise culminates in several timely activity/travel hypotheses developed for future investigation (Fig. 6).

Fig. 6. Household activities, life-cycle stage, and residential location.

The decision to control for family status and residential location has been informed by the literature. Empirical evidence suggests that, among other determinants, individual and household level activity/travel outcomes can be partially explained by household roles, gender, presence of children, timing and duration of activities, and the accessibility of residential locations (Bhat, 1996a,b; Boarnet and Greenwald, 2000; Golob and McNally, 1997; Golob, 2000; Greenwald and Boarnet, 2001; Hanson and Hanson, 1981). Further, recent studies attempting to establish statistical associations between aspects of neo-traditional design and travel behavior suggest that activity patterns of urban residents are likely to contrast those of similar households located in suburban areas (Boarnet and Greenwald, 2000; Greenwald and Boarnet, 2001; Crane, 2000). The identification of urban and suburban planning districts has been informed by discussions with Metro. Higher levels of residential and employment density and mixed-use development generally characterize the urban planning districts. This is not to say that the suburban planning districts do not contain pockets of mixed-use, dense development as well.

Zahavi (1979), in his early experimentation with travel probability fields, observed that (1) the orientation of zone-based trip destinations tended toward major regional urban centers; (2) the length of the major axis of a travel field is proportional to the distance from origin zone centroids to major regional centers; (3) the area of auto-based travel fields is larger than for transit modes suggesting greater dispersion of auto-based trip destinations; and (4) the orientation of the field is affected by proximity to transport supply. Like Zahavi (1979), there is some preliminary evidence that the orientation of elliptical activity spaces, for households of both types, tends toward a major regional center, the Portland central business district (Fig. 6). This finding potentially reflects the diversity and frequency of certain types of employment and recreational opportunities that are available as one travels toward central Portland. What is less clear is the extent to which elliptical orientation is affected by proximity to transport supply. Zah avi (1979) estimated ellipses for aggregate household destinations modeled as traffic zone centroids and displayed these ellipses against a backdrop of major transportation routes. In the presence of micro-data, we have shifted the scale of analysis to the household level and have represented activity sites using geocoded activity locations. The relationship between elliptical orientation and transport supply, at least where major routes are concerned, is not as clear in our exploratory exercise.

An effect that appears to be more dominant for households without children is that urban cases have smaller activity ellipses than their suburban counterparts suggesting greater dispersion of activities for suburban households. The extent to which this effect can be associated with residential self-selection or is determined by urban form remains an area for future investigation. Contrasting urban households with and without children we begin to observe larger activity spaces for those households with young children. This could reflect increased levels of non-work activities generated by the presence of children (e.g. education, recreation, daycare).

The ellipse with the longest major axis belongs to a centrally located household with young children, and one family member commuting to a suburban workplace. This type of household behavior perhaps reflects a desire to trade residential context with a longer journey to work. The absence of employment duration information does not allow us to examine the possibility that the job location changed in the presence of a fixed residence. The workplace destination is also located close to a major interstate. From this, we might want to investigate whether interaction between transport supply and workplace location influences the residential location choice process of households with outbound commuters. This case raises interesting questions with respect to levels of outbound commuting, from central locations, which are potentially attributable to residential location choice. The contribution of households with this type of activity pattern to aggregate VMT could potentially match similar, suburban households. In other words, dense, mixed-use residential locations cannot automatically be associated with net reductions in household VMT.

In the remainder of this section we introduce a list of eight activity/travel behavior hypotheses that will serve as principle questions for future research. These questions have emerged through exploration of the household activity patterns depicted in Fig. 6. We are not suggesting that all of the questions are new, but that their composition has been aided by the exploratory exercise.

1. Patterns of urban development, levels of employment and residential density, and proximity to transport supply affect the dispersion, orientation, and spatial scope of household activity patterns.
2. Urban households, with children, have activity patterns characterized by higher levels of spatial complexity than their suburban counterparts. This complexity is reflected in higher activity frequency and the spatial networks used to link activity destinations.
3. Suburban households have more dispersed activity patterns than urban households due in part to geographical distributions of potential activity sites.
1. Household activity patterns tend to be oriented toward regional centers.
1. Higher levels of household auto-ownership are associated with spatially disperse household activity patterns.
1. Transit dependent households have activity patterns that are more focused around the home location, oriented along major transport routes, and toward regional centers.
1. The uneven distribution of household responsibilities contributes to residentially focused activity patterns, higher levels of joint activities with children, and more out-of-home maintenance activities for adult female householders.
1. Residential location preferences for dense, mixed-use neighborhoods, gives rise to outbound commuting from central neighborhoods to suburban workplace locations.

This exploratory exercise has been conducted with recognition that we have estimated ellipses for a small set of households only. With this in mind, the effects that we have discussed are essentially working hypotheses designed for further investigation in the presence of a larger number of observations. It is our intention to engage in a broader multivariate study of the sensitivity of household activity patterns to several urban form, transport supply, and household socio-demographic characteristics.

6. Discussion and concluding remarks

We have developed an extensible platform that facilitates exploration of household activity/travel behavior using multi-day survey data stored in an underlying object-relational geographic database. The capabilities of our prototype system for hypothesis generation have been demonstrated in an exploration of household level activity/travel behavior using data from the 1994/1995 Portland, Metro Household Activity and Travel Behavior survey. The range of current capabilities includes estimation of un-weighted and weighted standard distance circles, standard deviational ellipses, implementation of household activity spaces, and the assembly of three-dimensional household trajectories as collections of individual space–time paths.

Overall, we have found that the GIS-driven development environment provides an opportunity to effectively reduce the complexity of household level activity/travel micro-data to geometries that facilitate construction and exploration of timely activity/travel hypotheses. We plan continued refinement and extension of this analytic platform with several directions for future research emerging from the experience reported in this paper. We conclude with a discussion of what we view as being desirable conceptual and methodological refinements, and further software development initiatives.

The current approach attempts to characterize the observed behavior of households as essentially the outcome of the aggregate behavior of individual household members. Descriptive geometries are assembled that portray actual or observed activity patterns and spaces for survey households. To broaden the application of our analytic environment we will begin investigating integration of models that endogenize the potential activity space (PPA, DPPA) given observed activity data, exogenous distributions of land use, and transport supply characteristics. This objective will also involve explicit integration of the mode choice dimension of household activity/travel behavior. While mode choice exploration is not explicitly available through the exploratory tools it has been included in household activity/travel measurement routines not yet coupled with the existing software.

Motivated by renewed interest in time geography and the emergence of location aware technologies (LAT) and location-based services (LBS), Miller (in press) has recently advanced a measurement theory for time geography. He has developed rigorous analytical definitions for several core time-geographic concepts including, space– time paths, prisms, stations, bundles, and intersections (Table 1). His work holds several potential implications for our research. For example, we anticipate that extension of the household trajectory component of our system, to include spatiotemporal intersection of householder space–time paths, will advance measurement and exploration of joint activity/travel behavior within households. We expect that the measurement theory proposed by Miller (in press), combined with recent thinking concerning identification of joint activities from activity/travel survey data (e.g. Gliebe and Kopp elman, 2002; Scott and Kanaroglou, 2002), will inform extension and refinement of our toolkits current exploratory capabilities.

Spatial outliers can impact estimation of area-based activity pattern geometries in undesirable ways. For example, the major axis of the standard deviational ellipse has a tendency to extend into regions without household activities when estimated in the presence of spatial outliers. One potential solution could be to force elliptical foci to known locations. For example, New some et al. (1998) estimate journey to work ellipses, for individuals using work and home zone centroids as elliptical foci. While this solution works quite nicely at the individual level, we are faced with the challenge of building a representative, non-home focal point, to complement the home location when estimating ellipses for household activity patterns. It is our intention to test and implement alternative elliptical structures within the current analytic environment.

We have presently imposed a somewhat fixed, but informative typology of households within the prototype. Assembling this household typology was informed by the literature and is meant to highlight the activity/ travel behavior of potentially active households. A future objective is to transition this fixed typology to what can best be described as a household assembler that enables retrieval of households from the underlying database through user-defined compositional criteria implemented through a graphical user interface. Related to households, is the possibility that an effective abstraction of the household activity pattern will not emerge from one or two days of activity/travel data. Scho¨nfelder and Axhausen (2003) have discussed this issue with their approach benefiting from the presence of six weeks of information for survey respondents. Our tools currently facilitate construction of multi-day activity geometries but would have to be adjusted to accept longitudinal survey data.

From a software development perspective we would like to reduce system overhead by exploring implementation outside of the ArcGIS environment. While embedding the tools within ArcGIS has provided application development and analytic advantages, our research objectives do not demand the full range of capabilities offered by this GIS environment. One potential solution could involve utilization of the recently released ArcGIS Engine, a set of embeddable ArcObjects components or, an alternative open and object-driven GIS environment like CommonGIS (e.g. Jankowski et al., 2001).

We would also like to extend the system to support activity/travel data stored in alternative formats (e.g. shapefiles, coverages, other RDBMS), while maintaining geodatabase capabilities. We are currently working toward seamless three-dimensional visualization and exploration capabilities. Presently, household trajectories are assembled as three-dimensional shapefiles and visualized in a separate software extension. Lastly, we are currently modifying the processing capabilities of the system to facilitate assembly of data tables, containing various activity measures, for subsets of selected households. This capability will be used to support empirical study of associations between comprehensive measures of household level activity/travel behavior and patterns of land use, transport supply characteristics, accessibility, mobility, and household sociodemographic characteristics.

Acknowledgments

We would like to thank Metro for the provision of data and supporting documentation and Bill Stein for taking the time, on several occasions, to discuss substantive and technical aspects of our research. The second author gratefully acknowledges the financial assistance of the Social Sciences and Humanities Research Council (SSHRC) Canada Research Chairs (CRC) program.

References

Anselin, L., 2000. Computing environments for spatial data analysis. Journal of Geographical Systems 2, 201–220.
Anselin, L., Getis, A., 1992. Spatial statistical analysis and geographic information systems. Annals of Regional Science 26, 19–33.
Bachi, R., 1963. Standard distance measures and related methods for spatial analysis. Papers of the Regional Science Association 10, 83–132.
Bailey, T.C., Gatrell, A.C., 1995. Interactive Spatial Data Analysis. Addison-Wesley Longman, Essex.
Barber, C., Dobkin, D., Huhdanpaa, H., 1997. The quickhull algorithm for convex hulls. ACM Transactions on Mathematical Software 22, 469–483.
Beckmann, M.J., Golob, T.F., Zahavi, Y., 1983a. Travel probability fields and urban spatial structure: 1. Theory. Environment and Planning A 15, 593–606.
Beckmann, M.J., Golob, T.F., Zahavi, Y., 1983b. Travel probability fields and urban spatial structure: 2. Empirical tests. Environment and Planning A 15, 727–738.
Bedard, Y., 1999a. Visual modelling of spatial databases: towards spatial PVL and UML. Geomatica 53, 169–186.
Bedard, Y., 1999b. Principles of spatial database analysis and design. In: Longley, P.A., Goodchild, M.F., Maguire, D.J., Rhind, D.W. (Eds.), Geographical Information Systems. vol. 1: Principles and Technical Issues. Wiley and Sons, New York, pp. 413– 424.
Bhat, C.R., 1996a. A Hazard-based duration model of shopping activity with nonparametric baseline specification and nonparametric control for unobserved heterogeneity. Transportation Research B 30 (3), 189–207.
Bhat, C.R., 1996b. A generalized multiple durations proportional hazard model with an application to activity behavior during the evening work-to-home commute. Transportation Research B 30 (6), 465–480.
Bhat, C.R., Koppelman, F.S., 2000. Activity-based travel demand analysis: history, results and future directions. Paper presented at 79th Annual meeting of the Transportation Research Board, Washington, DC.
Blumen, O., Kellerman, A., 1990. Gender differences in commuting distance, residence, and employment location: metropolitan Haifa 1972 and 1983. Professional Geographer 42 (10), 54–71.
Boarnet, M.G., Greenwald, M.J., 2000. Land use, urban design, and nonwork travel: reproducing other urban areas empirical test results in Portland, Oregon. Transportation Research Record 1722, 27–37.
Boarnet, M.G., Sarmiento, S., 1998. Can land-use policy really affect travel behavior? A study of the link between non-work travel and land-use characteristics. Urban Studies 35 (7), 1155–1169.
Booch, G., 1994. Object-Oriented Analysis and Design with Applications, second ed. Addison-Wesley, Mass.
Brown, L.A., Moore, E.G., 1970. The intra-urban migration process: a perspective. Geografiska Annaler Series B, Human Geography 52 (1), 1–13.
Buliung, R.N., 2001. Spatiotemporal patterns of employment and nonwork activities in Portland, Oregon. In: Proceedings, 2001 ESRI International User Conference. ESRI, San Diego. Available from: <http://gis.esri.com/library/userconf/proc01/professional/papers/ pap1078/p1078.htm> (Accessed 19 July 2004).
Buliung, R.N., Kanaroglou, P.S., 2004. On design and implementation of an object-relational spatial database for activity/travel behavior research. Journal of Geographical Systems 6, 237– 262.
Burns, L.D., 1979. Transportation, Temporal and Spatial Components of Accessibility. D.C. Heath and Company, Lexington.
Cambridge Systematics, 1996. Data collection in the Portland, Oregon metropolitan area: case study. Report DOT-T-97-09. US Department of Transportation.
Claramunt, C., The´riault, M., 2001. A UML-based modelling approach for analysing travel behavior. Paper presented at Eurosim 2001, Delft, NL.
Crane, R., 2000. The influence of urban form on travel: an interpretive review. Journal of Planning Literature 15 (1), 3–23.
Dijst, M., 1999. Two-earner families and their action spaces: a case study of two Dutch communities. GeoJournal 48, 195–206.
Earickson, R.J., Harlin, J.M., 1994. Geographic Measurement and Quantitative Analysis. Macmillan College, New York.
Ebdon, D., 1988. Statistics in Geography, second ed. Blackwell, Oxford.
Friedman, B., Gordon, S.P., Peers, J.B., 1994. Effect of neotraditional neighborhood design on travel characteristics. Transportation Research Record 1466, 63–70.
Frihida, A., Marceau, D.J., The´riault, M., 2002. Spatio-temporal object-oriented data model for disaggregate travel behavior. Transactions in GIS 6, 277–294.
Furfey, P.H., 1927. A note on Lefevers standard deviational ellipse. The American Journal of Sociology 33 (1), 94–98.
Gliebe, J.P., Koppelman, F.S., 2002. A model of joint activity participation between household members. Transportation 29, 49–72.
Golledge, G.R., Kwan, M.-P., Garling, T., 1994. Computational process model of household travel decisions using a geographical information system. Papers in Regional Science 73 (2), 99–117.
Golob, T.F., 2000. A simultaneous model of household activity participation and trip chain generation. Transportation Research B 34, 355–376.
Golob, T.F., McNally, M.G., 1997. A Model of activity participation and travel interactions between household heads. Transportation Research B 31, 177–194.
Gordon, P., Kumar, A., Richardson, H.W., 1988. Beyond the journey to work. Transportation Research A 22 (6), 419–426.
Greenwald, M.J., Boarnet, M.G., 2001. Built environment as determinant of walking behavior: analyzing nonwork pedestrian travel in Portland, Oregon. Transportation Research Record 1780, 33– 42.
Handy, S., 1993. Regional versus local accessibility: implications for nonwork travel. Transportation Research Record 1400, 58– 66.
Handy, S.L., Clifton, K., 2000. Evaluating neighborhood accessibility: issues and methods using geographic information systems. Research Report SWUTC/00/167202-1 Southwest Region University Transportation Center, Center for Transportation Research, The University of Texas at Austin. Available from: <http:// swutc.tamu.edu/reports.html> (Accessed 19 July 2004).
Handy, S.L., Niemieier, D.A., 1997. Measuring accessibility: an exploration of issues and alternatives. Environment and Planning A 29, 1175–1194.
Handy, S.L., DeGarmo, A., Clifton, K., 2002. Understanding the growth in non-work VMT. Research Report SWUTC/02/167222 Southwest Region University Transportation Center, Center for Transportation Research, The University of Texas at Austin. Available from: http://swutc.tamu.edu/reports.html (Accessed 19 July 2004).
Hanson, S., Hanson, P., 1981. The travel-activity patterns of urban residents: dimensions and relationships to sociodemographic characteristics. Economic Geography 57, 332–347.
Ha¨gerstrand, T., 1970. What about people in regional science? Papers of the Regional Science Association 24, 7–21.
Hess, D.B., Ong, P.M., 2002. Traditional neighborhoods and automobile ownership. Transportation Research Record 1805, 35–44.
Horton, F.E., Reynolds, D.R., 1971. Effects of urban spatial structure on individual behavior. Economic Geography 47 (1), 36–48.
Hurst, M.E., 1969. The structure of movement and household travel behavior. Urban Studies 6, 70–82.
Jackson, R.W., 1994. Object-oriented modeling in regional science: an advocacy view. Papers in Regional Science 73, 347– 367.
Jankowski, P., Andrienko, N., Andrienko, G., 2001. Map-centered exploratory approach to multiple criteria spatial decision making. International Journal of Geographical Information Science 15 (2), 101–127.
Johnston, R.J., 1972. Activity spaces and residential preferences: some tests of the hypothesis of sectoral mental maps. Economic Geography 48 (2), 199–211.
Johnston-Anumonwo, I., 1992. The influence of household type on gender differences in work trip distance. Professional Geographer 44 (2), 161–169.
Kim, H.-M., Kwan, M.-P., 2003. Space–time accessibility measures: a geocomputational algorithm with a focus on the feasible opportunity set and possible activity duration. Journal of Geographical Systems 5, 71–91.
Kwan, M.-P., 1999a. Gender and individual access to urban opportunities: a study using space–time measures. Professional Geographer 51 (2), 210–227.
Kwan, M.-P., 1999b. Gender, the home-work link and space–time patterns of non-employment activities. Economic Geography 75 (4), 370–394.
Kwan, M.-P., 2000. Interactive geovisualization of activity-travel patterns using three-dimensional geographical information systems: a methodological exploration with a large data set. Transportation Research C 8, 185–203.
Law, R., 1999. Beyond -Women and Transport: towards new geographies of gender and daily mobility. Progress in Human Geography 23 (4), 567–588.
Lefever, D.W., 1926. Measuring geographic concentration by means of the standard deviational elilipse. The American Journal of Sociology 32 (1), 88–94.
Lenntorp, B., 1976. Paths in space–time environments: a time geography study of movement possibilities of individuals. Lund Studies in Geography No. 44. The Royal University of Lund, Sweden.
Lenntorp, B., 1978. A time-geographic simulation model of individual activity programmes. In: Carlstein, T., Parkes, D., Thrift, N. (Eds.), Human Activity and Time Geography. Wiley and Sons, NY, pp. 162–180.
Levine, N., 1996. Spatial statistics and GIS: software tools to quantify spatial patterns. Journal of the American Planning Association 62 (3), 381–391.
Levine, N., 2002. CrimeStat II: A Spatial Statistics Program for the Analysis of Crime Incident Locations (version 2.0). Ned Levine & Associates, TX/National Institute of Justice, Houston, Washington, DC.
Lewin, K., 1951. In: Cartwright, D. (Ed.), Field Theory in Social Science: Selected Theoretical Papers. Harper and Row, NY.
Liu, S., Zhu, X., 2004. Accessibility analyst: an integrated GIS tool for accessibility analysis in urban transportation planning. Environment and Planning B 31, 105–124.
Lu, X., Pas, E.I., 1999. Socio-demographics, activity participation and travel behavior. Transportation Research Part A 33, 1–18.
Madden, J.F., 1981. Why women work closer to home. Urban Studies 18, 181–194.
Makin, J., Healey, R.G., Dowers, S., 1997. Simulation modelling with object-oriented GIS: a prototype application to the time geography of shopping behavior. Geographical Systems 4 (4), 397– 429.
Mark Bradley Research and Consulting, Bowman, J., 1998. A System of Activity-Based Models for Portland, Oregon. US DOT, Washington, DC.
McNally, M.G., Ryan, S., 1993. Comparative assessment of travel characteristics for neotraditional designs. Transportation Research Record 1400, 67–77.
Metro, 2000. 2000 Regional Transportation Plan. Metro, Oregon. Available from: <http://www.metro-region.org/article.cfm?ArticleID= 236> (Accessed 19 July 2004).
Miller, H.J., 1991. Modelling accessibility using space–time prism concepts within geographical information systems. International Journal of Geographical Information Systems 5 (3), 287–301.
Miller, H.J., Wu, Y.-H., 2000. GIS software for measuring space–time accessibility in transportation planning and analysis. GeoInformatica 4 (2), 141–159.
Miller, H.J., in press. A measurement theory for time geography. Geographical Analysis.
Moore, E.G., 1970. Some spatial properties of urban contact fields. Geographical Analysis 2, 376–386.
Newsome, T.H., Walcott, W., Smith, P.D., 1998. Urban activity spaces: illustrations and application of a conceptual model for integrating the time and space dimensions. Transportation 25, 357–377.
OSullivan, D., Morrison, A., Shearer, J., 2000. Using desktop GIS for the investigation of accessibility by public transport: an isochrone approach. International Journal of Geographical Information Science 14 (1), 85–104.
Pred, A., 1977. The choreography of existence: comments on Hagerstrands time-geography and its usefulness. Economic Geography 53 (2), 207–221.
Reiff, B., Kim, K.-H., 2003. Statistical analysis of urban design variables and their use in travel demand models. Report prepared for: Performance Measures Subcommittee of the Oregon Modeling Steering Committee. Oregon Department of Transportation. Available from: <http://www.odot.state.or.us/tddtpau/modeling. html> (Accessed 19 July 2004).
Rumbaugh, J., Jacobson, I., Booch, G., 1999. The Unified Modeling Language Reference Manual. Addison-Wesley, Mass.
Scho¨nfelder, S., Axhausen, K.W., 2003. Activity spaces: measures of social exclusion?. Transport Policy 10, 273–286.
Schwarze, B., Scho¨nfelder, S., 2001. ArcView-Extension VISAR— Visualisierung von Aktionsra¨umen, Version 1.6, Arbeitsbericht Verkehrs- und Raumplanung, 95, Institut fu¨r Verkehrsplanung, Transporttechnik, Strassen- und Eisenbahnbau, ETH, Zu¨rich.
Scott, D.M., 2003. Comparison of two GIS-based algorithms for generating potential path areas. Paper presented at the 50th Annual North American Meeting of the Regional Science Association International.
Scott, D.M., Kanaroglou, P.S., 2002. An activity-episode generation model that captures interactions between household heads: development and empirical analysis. Transportation Research B 36, 875–896.
Shaw, S.-L., Xin, X., 2003. Integrated land use and transportation interaction: a temporal GIS exploratory data analysis approach. Journal of Transport Geography 11, 103–115.
Sun, X., Wilmot, C.G., Kasturi, T., 1998. Household travel, household characteristics, and land use: an empirical study from the 1994 Portland activity-based travel survey. Transportation Research Record 1617, 10–17.
The´riault, M., Claramunt, C., Villeneuve, P.Y., 1999. A spatiotemporal taxonomy for the representation of spatial set behaviors. Lecture Notes in Computer Science 1678, 1–18.
Taylor, D.A., 1998. Object Technology: A Managers Guide, second ed. Addison-Wesley Longman, Mass.
US Department of Transportation, Federal Highway Administration (USDOT FHWA), 1995. Our Nations Travel: 1995 NPTS Early Results Report. Washington, DC.
Wang, D., Chen, T., 2001. A spatio-temporal data model for activitybased transport demand modeling. International Journal of Geographical Information Science 15, 561–585.
Weber, J., Kwan, M.-P., 2002. Bringing time back in: a study on the influence of travel time variations and facility opening hours on individual accessibility. Professional Geographer 54, 226– 240.
Wen, C.-H., Koppelman, F.S., 1999. Integrated model system of stop generation and tour formation for analysis of activity and travel patterns. Transportation Research Record 1676, 136–144.
Wolpert, J., 1965. Behavioral aspects of the decision to migrate. Papers of the Regional Science Association 15, 159–169.
Wolpert, J., 1967. Distance and directional bias in inter-urban migratory systems. Annals of the Association of American Geographers 57 (3), 605–616.
Yee, J.L., Niemeier, D.A., 2000. Analysis of activity duration using the Puget sound transportation panel. Transportation Research Part A 34, 607–624.
Yuill, R.S., 1971. The standard deviational ellipse; an updated tool for spatial description. Geografiska Annaler Series B, Human Geography 53 (1), 28–39.
Zahavi, Y., 1979. The -UMOT Project Report. US Department of Transportation DOT-RSPA-DPB-20-79-3. US Department of Transportation, Washington.
Zeiler, M. (Ed.), 2001. Exploring ArcObjects. ESRI Press, Redlands, CA.

buliunggistoolkitexploringgeographies2006.md 84 KB Permalink Riwayat Mentahan