Walk Access to BART and Residential Density
12039602240280Walk Access to BART and Residential DensitySherman Lewis and Zoe Roller May 8, 20152787 Hillcrest Ave.Hayward CA 94542510-538-3692sherman@csuhayward.us00Walk Access to BART and Residential DensitySherman Lewis and Zoe Roller May 8, 20152787 Hillcrest Ave.Hayward CA 94542510-538-3692sherman@csuhayward.usSherman Lewis and Zoe Roller Walk Access to BART and Residential DensityMay 8, 20152787 Hillcrest AvenueHayward CA 94542510-538-3692sherman@csuhayward.usAbstractThis paper used new approaches and new Bay Area Rapid Transit District (BART) data to analyze density and walk distance to public transit. We looked at walk access to BART and at density around BART stations with high levels of walk access. BART’s large survey of riders found an average walk from home to station of a little over half a mile. We produced maps showing densities of block groups around high-walk stations. We found no correlation between population densities of block groups close to stations for the 16 highest walk-access stations. When three Central Business District (CBD) stations were removed, however, the correlation improved. When block group census data on means of journey to work by transit, bicycle, and walk was used, the correlation with residential density was very strong. We looked at the potential for walk distances longer than the conventional half mile and found that roughly 16 percent of walk access is 0.89 miles or more. Planners should consider longer walk distances, along with other environmental design and economic incentives to shift travel mode. ObjectiveWalking and other non-auto modes are important for health and for reducing dependence on automobiles and lowering carbon emissions. Dense residential neighborhoods support walking and transit over dependence on private vehicles. Many factors influence mode choices, but this paper only looks at a short list: distance walked to transit, residential density around high walk-access stations, non-auto commute modes, and planning guidelines for walk access. This research supports other papers that have found walk distances to transit over half a mile, but not a good correlation with density around stations. Density was, however, highly correlated with non-auto modes in general.The purpose of the paper is to use the data available from the Bay Area Rapid Transit District (BART) as it relates to density and walk distance. It does not attempt to discuss other variables that also play a role in access to transit, such as availability and cost of parking, design quality of walk paths, and total commute travel time. Literature review A small body of literature exists on walk distance to transit; also referred to as catchment areas. Some of this literature includes reports on surveys of rapid transit riders about their walk distance from home to rapid transit (home-access walk trips). Other topics within this field of study include walks to all kinds of transit, walks at the destination end, and walking in general. Walk to transit is also referred to as a stage in a multi-modal trip. Methodological issues include the use of a radius to define a buffer around a station when walking paths are not direct, quality of self-reporting by respondents, and the use of wearable GPS reporting by respondents.Moran’s literature review “Walking the Walk” (2013) looks at a survey of bus riders in Austin, Texas. Moran challenges the rule of thumb that people will not walk more than five minutes to reach public transit. She concludes that walk access is in fact far more varied, and recommends that further research should not rely on the five minute assumption.In a study of a Toronto suburb, ADDIN ZOTERO_ITEM CSL_CITATION {"citationID":"d9Nmrsqk","properties":{"formattedCitation":"(Crowley, Shalaby, and Zarei 2009)","plainCitation":"(Crowley, Shalaby, and Zarei 2009)"},"citationItems":[{"id":463,"uris":[""],"uri":[""],"itemData":{"id":463,"type":"article-journal","title":"Access Walking Distance, Transit Use, and Transit-Oriented Development in North York City Center, Toronto, Canada","container-title":"Transportation Research Record: Journal of the Transportation Research Board","page":"96-105","volume":"2110","source":"MetaPress","abstract":"This study had two main objectives: to examine how variations in walking distance to rapid transit are related to mode choice as well as to auto ownership and use and to investigate whether temporal changes in the built environment associated transit-oriented development in close proximity to rapid transit (subway) service encourage residents to use transit. The city of Toronto, Canada, and one of the fastest-growing suburban centers of Toronto—the North York City Center located near the northern edge of Toronto on the Yonge subway line—were selected as case studies. With the 2001 Transportation Tomorrow Survey (TTS) data, a quantitative analysis was employed; it that focused on homebased trips by using the city's subway network to demonstrate how walk-access distances to rapid transit are related to subway mode share, auto ownership, and auto use. Further analysis was undertaken to examine the temporal changes in land use and related travel behavior over a 15-year period by comparing TTS data for 1986 and 2001. The results of the analyses illustrate quantitatively the strong association between convenient walk access, lifestyle, and transit use, not only during peak hours but also throughout the day. The results show how the promotion of focused development within a convenient walking distance of rapid transit service in a relatively low-density suburb of Toronto has, over 15 years, been accompanied by a substantial shift in residents' travel behavior toward increased transit use.","DOI":"10.3141/2110-12","author":[{"family":"Crowley","given":"David"},{"family":"Shalaby","given":"Amer"},{"family":"Zarei","given":"Hossein"}],"issued":{"date-parts":[["2009",12,1]]},"accessed":{"date-parts":[["2014",12,22]]}}}],"schema":""} Crowley, Shalaby, and Zarei (2009) look at the relationships among the built environment, mode choices, and distance to public transit. They found that transit ridership could flourish in low-density areas, even though most riders who walked went only 0.2 miles or less to stations. They also found that car ownership increases as residents live further from a station.Bergman, Gliebe, and Strathman (2011) look at the WES commuter rail in Portland, Oregon, and analyze inter-modal transit choices by distance, trip time, and rider income. They found that median walk distance was 0.54 miles, and over ten minutes. This study mentions that walk access is positively correlated to high densities, and suggests that urban design factors could increase walkability.Research is usually framed in context of helping planners provide transit service, with survey results as a basis for advice. Usually, many variables in addition to walk distance are considered in order to understand a simple act with complex causes. These variables include individual characteristics (income, education, gender, disability, attitudes, vehicle availability), walk route factors (weather, social quality of route, mixed land uses, barriers along route, walk distance, population density), transit features (station spacing, frequency, speed, transfers, wait time, distance to Central Business District (CBD), parking at station), and destination features (employment density, walking at destination).Besides helping planners improve transit and walkability, research is often motivated by a desire to reduce auto dependency for environmental reasons, to promote a more sociable urban alternative to suburbia, to improve pedestrian safety, and to increase walking for health. Since the 1970s, planning standards or guidelines for walking distances, catchment area, or service area to rapid transit stations commonly used a distance of 400 meters (quarter of a mile) or 800 meters (half a mile), which became a conventional wisdom with little empirical basis. More recent discussion are more nuanced, considering not just some average walk distance, but a larger proportion, 75 to 90 percent, of walkers, and measuring the fall off or rate of decay as the distance gets longer.Some older surveys reporting short walk distances refer to access to buses, which have a much shorter walk distance than rapid transit. Since the 1990s, surveys of people walking to rapid transit stations have generally revealed longer walk distances, usually over half a mile, especially as a larger share of walkers is taken in. This summary is largely taken from more detailed discussions by El-Geneidy et al. (2014), Agrawal, Schlossberg, and Irvin (2008) and other articles in the references. In addition, the literature has usually not included how walk access to rapid transit relates to personal travel time budgets, locational decisions, pricing incentives and subsidies in the larger economy, and national household and time use databases.Survey results from the literature are included below along with those from this report. No studies of home origin walk to rapid transit have ever been replicated and all use different variables and methodologies. There is no database where results are compiled into a consistent frame of reference. Some common approaches and systematic compilation into a database would be helpful. The literature on individual neighborhoods around transit stops, with land use maps and details, is almost non-existent. Neighborhoods are usually aggregated and discussed generically and statistically. A recent example is report, The WalkUP Wake-Up Call: Boston, has a real estate development emphasis. One of the sponsors, LOCUS / Smart Growth America, promotes investment in mixed use redevelopment. The study of Boston named 71 “WalkUPs” or “walkable urbanism” of various kinds and defined their boundaries and areas, which included many land uses besides neighborhoods, but not the number of residents. The report makes references to density in general, but no densities of specific neighborhoods. It recommends a gross density for residential units of over 8 per acre with a quarter mile. Since the WalkUPs have many non-residential land uses, there is no way to tell what the density is. The report says there are about 88,000 people per established WalkUP. When I asked Smart Growth America about what this meant, they informed me that the figure was the metro population divided by the number of WalkUPs, and the average WalkUP had under 10,000 people. The WalkUPs average about 28 people per acre, including non-neighborhood uses. The report uses Walk Score, intersection density, and other variables, but does not cover how they relate to walk access to transit.This paper contributes new information about walk distances in more detail than other reports, including a list of stations, the number of walk access respondents, and the mean, median, and standard deviation walk time from home access, showing the large variation in amount of walk access and walk distance that is often hidden in aggregate data. The study compiles results from previous studies and combines them with new analysis. The paper has maps showing block group densities in the half mile around high-walk-access stations, and compares them to walk access and to non-auto mode share of the journey to work, also not found in the literature. The paper has a discussion of how walk access to transit relates to planning. MethodologyBART is an urban transit system serving the San Francisco Bay Area. The 2008 BART Station Profile Study (BART, 2008) and BART’s research department provided rider survey data on median, mean, and standard deviations of walk distance from home origins to BART stations (details in Table 3 in the Data Appendix). BART conducted interviews at many different hours of the day and in a number of languages. The BART Station Profile survey included 5,974 riders who walked from home. We believe their samples are representative of the areas around the stations. The paper uses unweighted data from BART and does not go into methodological questions about it. The BART data is probably robust due where it has 100 or more respondents per station, a very large sample for individual stations from any rapid transit system. There are also limits on the validity of accuracy block groups with small populations. The purpose of the paper is not statistical accuracy but to approximate findings for policy purposes. To look at decay rates or fall off of walk access with distance, we used station data for median times and standard deviations and plotted second order polynomial fit lines, as shown in Figure 1. Standard deviations were not used as statistics, but as useful number for a policy estimate. Strict accuracy about density is not useful because design and pricing factors are also important. To look at density around stations, we selected the sixteen stations with more than 100 respondents who walked from home to the station. We also used census data on block group population and Google satellite images of land use around BART stations to determine density. Census tracts were not used because they are often too large to use for walking distances. Cervero et al (1995, p. 41) says “In lower-density areas...census tracts generally increase in size...In some suburban and exurban parts of the Bay Area, for instance, large amounts of open space are within several census tracts, producing large territorial units. Even if one access trip origin is in one of these zones, using our criteria, the zone will be added to the catchment, thus skewing the estimate of land coverage. ...Virtually all census tracts that met the catchment criteria had far more land that was developed than undeveloped. Still, ideally, smaller geographic units, like block groups or even blocks, would be used in defining catchments. ... SEQ CHAPTER \h \r 1The use of census tracts for defining catchment areas posed problems for studying walk-on trips in suburban areas where census tracts can be large, often with dimensions well beyond the one-quarter- to one-half-mile distance normally considered to be the maximum distance Americans will walk. In more urbanized areas, especially downtown San Francisco, census tracts (sometimes as small as four or five city blocks) are more suitable for studying the catchments for pedestrian access trips.”34671000Pleasant Hill Station Area Catchment, Cervero et al. (1995).Pleasant Hill Station Area Catchment, Cervero et al. (1995).The map on Pleasant Hill from Cervero et al. using tracts illustrates the problem. The shaded area is well over half a mile and even goes beyond the next station. Another issue is whether to define the area by distance or by some percent of all the walk access trips using the station. The larger the percent, the larger the area around the station and the more likely the density will be lower. Cervero et al. used both census tracts and 90 percent of the walk access, resulting in larger areas with lower densities: “The catchment areas for walk-on trips to BART stations were defined as the census tracts encompassing the origins of 90 percent of all access trips made by foot.”Block groups also have problems. They are lumpy; they are irregularly shaped and meander inside and outside the half-mile circle. We included data from block groups which are more than 50 percent within the circle, so the density estimate is approximate. Note that the half mile is not about walking distance. A buffer analysis of walking distance using the road network is more accurate, although the gain in accuracy may not be meaningful for policy purposes. The half-mile circle actually approximates a bit over half a mile walking because the route follows streets, not a straight line. The circle is used identify block groups, which also is an approximation given the data available (better than tracts, not as good as building data, which is not easily available.) So the question is if imperfect data can still yield useful results in an efficient use of research time. Walk distance and density are affected by residential densities and by non-residential uses that may intervene between the neighborhood and the station. Non-neighborhood uses, like parking lots, offices, and institutions, have the effect of diluting the apparent density of the neighborhood as it relates to access to transit. We looked at Google maps to determine land uses within the block groups. If most pedestrians had to walk through a mostly non-residential block group to reach a stations, it affected the functional density of the station area and we kept them in the density estimate. By contrast, a few mostly non-residential block groups did not affect common walking routes to the station, so we did not include them in the density estimate. We excluded, for example, the University of California campus near the Downtown Berkeley Station, Glen Canyon Park near Glen Park Station, the Westfield Shopping Center near Powell Station, and the City College of San Francisco campus near Balboa Park Station. The station area maps below show block group densities in color in the half-mile circle around the sixteen selected BART stations. Table 3 has station data and Table 4 has the block group data for the maps. We then look for correlations among residential density, number of walk accesses, and walking distance by station. FindingsWalk distanceIn 2008, 31 percent of BART riders walked to stations, with a median distance of 0.540 miles. The longest average walk was 1.53 miles to reach the Dublin/Pleasanton Station, but with a small sample of 19 respondents. (The Dublin/Pleasanton median was 1.36 miles and the standard deviation was 0.58 miles.) The shortest average walk was 0.456 miles to reach 16th St. Mission Station, with a large sample size of 405 respondents. (The median was 0.395 miles and the standard deviation was 0.58 miles.) The BART data can be compared to other findings from other paring this BART data to other survey results shows that BART’s findings are inconsistent with some, usually older, research showing walks of a quarter to half a mile, and consistent with several recent papers showing over half a mile. Table 1 shows all the studies we could find and get access to on the web that reported survey result on walks from home to rapid transit. Table 1: Survey results, home to rapid transit walk distancesSourcePlacesample sizetype of railmean distancemedian distancelongest walks, % of populationDistanceAgrawal, Schlossberg, and Irvin, 20083 stations in Portland, 2 in SF Bay Area328LRT0.520.4775th percentile0.68Alshalalfah, B. and Shalaby, A., 2007Toronto?subway?0.2220 percentover .31North Toronto?subway?0.28??Etobicoke?subway?0.25??Bergman, Gliebe, and Strathman, 2011Portland?WES commuter rail?0.54 incl. wait85th percentile3.00Burke and Brown, 2007Brisbane10,931train0.650.5585th percentile0.98Crowley, Shalaby, and Zarei, 2009North York City Center5,090subway?over .2512% of sampleover 1.00Cervero et al., 1995BART????90th percentile1.51Daniels and Mulley, 2011Sydney667 tripstrain0.50.472.4 percentover 1.20 Ditto?????75th percentile0.63El-Geneidy et al., 2014Montreal ?commuter train0.510.4985th percentile0.78 Dittoditto?metro subway0.350.3385th percentile0.54Ker and Ginn, 2003Perth?suburban rail??55th percentile0.62O’Sullivan and Morrall, 2007Calgary1,800LRT0.4?75th percentile0.52Stringham, 1982Toronto and Edmonton?suburban rapid transit??"well over" 90th percentile0.28TCRP, 1996Chicago?Metra0.75???Lewis and Roller [unpublished]SF Bay Area?5,974BART0.5980.54est. of 84th percentile0.89C:\Users\Sherman\Dropbox\Mobility Analysis\BART walk article\Supporting files\Lit matrix.xlsxDrop off with distanceBART provided standard deviations for the distribution of walk access to each station. To estimate the drop off with distance walked, or distance decay, we plotted the station medians minus half a standard deviation, the medians, and the medians plus one standard deviation, from shortest to longest walk, with 100 percent of access for the shortest distance dwindling down to a very small percent for the longest walks. This use of standard deviation is not meant to be statistically valid; it is only for policy estimates of shorter and longer walking distances.center3849370Figure 1: Distance decay and walk distance from home to BART by StationSource: BART walk density data.xlsx | decay functionFigure 1 shows the amount of variation by station, different levels of walk access by distance, and several criteria for planning. Density and walk accessAfter walk distances, we looked at land use around high-walk-access stations. The maps used Census 2010 accessed through Social Explorer and satellite image from Google maps. The maps show a half-mile radius around each station with block groups shaded for density in persons per acre. -74295387921500Maps 1 and 2: Powell St. and Civic Center Stations, San FranciscoMaps 3 and 4: 16TH Mission and 24TH Mission Stations, San Francisco-12446012730950Maps 5 and 6: Glen Park and Balboa Park Stations, San Francisco0165735Maps 7 and 8: 12TH Street and 19TH Street Stations, Oakland0914400Maps 9 and 10: Lake Merritt and Macarthur Stations, Oakland0174991Maps 11 and 12: Rockridge Station, Oakland, and Ashby Station, Berkeley0914400Maps 13 and 14: Downtown Berkeley and North Berkeley Stations09144000312158Maps 15 and 16: El Cerrito Plaza, El Cerrito, and Pleasant Hill StationsTable 2 and Figure 2 show the data for the maps above. Density and high walk access were not correlated; the correlation was minus .0885. While stations with many walkers – 24th Street and Mission and 16th Street and Mission – are situated in a dense neighborhood, the results from other stations do not have consistent correlation between density and walk access, or between the walk distance and the numbers of walkers. For example, Glen Park and North Berkeley stations show that people are willing to walk in lower density neighborhoods. At those stations, the lack of parking may be a major explanation for high walk access.Table 2: Station data, 16 highest walk access stationsStationPedes-trians surveyedPeople per acreMedian distance walkedMean distance walkedStandard deviationMedian distance + SDPopu-lationBlock group acres24th St and Mission595590.450.510.3210.7720,937355Glen Park446270.430.460.2750.717,942292Downtown Berkeley439350.580.620.3520.939,07625616th St and Mission405520.400.460.2910.6919,726378North Berkeley333210.580.630.3530.927,527363Ashby330240.490.520.270.7611,38247519th St – Oakland308250.760.730.3161.077,818314El Cerrito Plaza299150.560.570.3440.895,881398Rockridge290180.490.520.3250.819,344433MacArthur220180.480.550.3260.819,585544Lake Merritt211280.500.590.3670.8710,159361Civic Center1641070.530.590.3640.8926,648249Balboa Park161310.540.630.3530.899,03929112th St – Oakland149300.380.480.4110.7912,148402Pleasant Hill104140.400.490.3380.738,293600Powell101850.560.640.3970.9520,887246Total4,55533.210.5050.554195,5745,889BART walk density data.xlsx | BART walk top 16 Information on density around stations from other studies is limited. Cervero et al. (1995) grouped several stations together based on cluster analysis, so there is, for example, no information on Pleasant Hill itself. They also used census tracts and 90 percent of walk access, leading to finding low density. Pleasant Hill was in their “Suburban Center” group, which had a density of 6 persons per acre. Our research using block groups for Pleasant Hill found a density of 14 per acre. This disparity may also be due to outdated information in Cervero et al.The lack of correlation of density to walk access was also caused by three large CBD stations. The two BART stations with the highest densities – Powell and Civic Center – had very few pedestrians going to BART and a longer than average walk distance. Large, dense populations near transit do not guarantee that many people will walk to transit. Yet another example is 19th Street Oakland and 12th Street Oakland. At 19th Street Oakland the median walking distance is a long 0.76 of a mile. On the other hand, at 12th Street Oakland, just a few blocks away and with a similar density, survey respondents walked a very short median distance, 0.38 of a mile. The 19th Street Oakland Station had the 7th highest walk access while 12th Street ranked much lower, 13th highest and about half as many in number. Figure 2: Walk Access by Residential Density Much of the lack of correlation is due to these three outliers, Civic Center, Powell, and 12th Street Oakland, all CBDs. The low correlation is probably due to the fact that many people are already close where they need to go and thus do not need transit. These three stations serve the largest employment centers in the Bay Area, so many nearby residents are likely to walk to work and don’t need BART. Figure 3 removes these outliers. Without Civic Center, Powell, and 12th Street Oakland, there is a correlation of 0.709 between density and walk access. Another clue is time of entry, with high entries in the morning indicating people coming from home, and high entries in the afternoon indicating people leaving work to go home. All three downtown stations had about 70 percent of entries being for the home-bound trip.Figure 3: Walk Access by Residential Density Revised Density and Non-auto Journeys to WorkIf walk to transit is only one kind of trip related to density, the relationship of density to all non-auto modes for all trip purposes might be stronger. Such data is not available, but the census does have block group data on mode of journey to work. American Fact Finder gave access to this data better than Social Explorer. Non-auto modes of transit, bicycle, and walk were combined and calculated as a percent of non-auto modes plus car modes. The percent of non-auto modes was plotted against density to see what the correlation might be. Figure 4 shows the result. Figure 4: Density and Non-Auto Journey to WorkThe correlation is 0.847 for the 16 stations, an improvement over using walk access to BART minus three CBD stations and, in fact, a high correlation. It is interesting to get a good correlation with density data alone despite lumpy geography and a margin of error of about 33% in block group populations. More refined statistical analysis with more variables would produce a stronger correlation. Planning GuidelinesPlanning guidelines consider empirical data about walk distances and policy ideas about how to improve transit ridership. Empirical data showing short walks favor policy in response, such as closer transit stops to reduce walk distances or increased density within one-fourth mile to increase riders. Empirical data showing longer walks favor policy affecting a larger land area for densification and improving the walk experience.The issue of catchment areas for transit overlaps with the general issue of how far people are willing to walk in general. In this context, it is important to distinguish the trip to work from other kinds of trips. Commuters are willing to spend more time to get to work than for any other common travel purpose, and walk time is part of that time. Walking to rapid transit supports more work trips than other kinds of walking, such as for routine shopping, services, socializing and recreation such as characterized by the Walk Score. In delineating neighborhoods, Leinberger and Lynch (2015) said “a single walkable place tends not to exceed 600 acres, based upon experience and the limitations people are willing to walk, generally agreed to be between 1500 and 3000 feet.” Those distances are .28 and .57 miles respectively, and may be accurate for general walking while missing the longer walks to transit which are perhaps about 40% of walk accesses. There is no agreement about how to define the walk access catchment area. Including all riders produces an impractically large area with few riders coming from the outer distances. Some outer limit between 75 and 95 percent of riders would be helpful, without implying it as a strict planning guideline. BART, for example, had a few people walking over two miles to reach a station, outliers are not useful for understanding walk access for planning purposes. Cervero et al (1995) say “The use of the 90 percent rank to demarcate catchment areas was chosen to represent the distances at which the vast majority of access trips are drawn. … Beyond 90 percent, most access trips fall toward the extreme tail of the distance distribution.”Another issue is what proportion of people to plan for. Crowley, Shalaby, and Zarei (x) ask, “How far will most transit users and potential transit users walk to access transit from (to) their homes and to (from) their workplaces, schools and other non-residential locations?” “There has been a general recognition that for a transit service to be accessible, the planning area for a TOD should extend to between a quarter-mile (400 m) and a half-mile (800 m) from a transit station, roughly the distance associated with a leisurely 5- to 15- minute walk.” It makes sense to plan for most users, but does it make sense to plan for more than most? Planning for over half leaves out almost half of potential riders, but planning for all potential riders would not be cost effective. The literature has articles supporting longer and shorter guidelines. Daniels and Mulley (x p. 5) say, “A consistent finding of walking distance research including Agrawal et al. (2008) in California and Oregon, Alshalalfah and Shalaby (2007) in Toronto Canada, and Ker and Ginn (2003) to access transit in Perth, Australia is that people walk considerably further to access public transport than commonly assumed “rules of thumb”. This finding has implications for both transport and land use planning, including transit oriented developments (Canepa 2007). People also walk further than assumed for purposes other than access to public transport (Iacono et al. 2008, Larsen et al. 2010).”Some of the differences over a guideline distance relate to the question being asked. “What is most desirable?” yields a quarter of a mile. How far are people willing to walk so we can plan for more people in a larger area? Yields half a mile or more. Distance decay is a measure of attractiveness of a short walk diminishing to willingness over distance. We wanted some idea of a maximum walk distance for a larger percent of riders, well over 50 percent but less than 100 percent. The application of guidelines should be influenced by cost-effectiveness, which varies among places. A guideline greater than half of potential users should not be rigid, but should encourage planners to look at areas farther than a quarter mile or half mile from a station.The BART research found a median walking distance of over a half mile. In a normal distribution the range from zero to the median plus one standard deviation includes 84 percent of the population, so planning for that number would reach more people than planning for half, assuming cost-effectiveness. The BART walk access distribution is not a normal curve, but rather skewed to the right. We can still, however, make a rough estimate that planning for about 84 percent of walk access would allow planning out to 0.89 miles from the station, a longer distance than is commonly accepted.Crowley, Shalaby, and Zarei (x) also say, “Maximizing subway ridership requires that development be more concentrated within a convenient walking distance of transit (within 400m preferably). The strong subway use was observed not only for the AM peak period but throughout the day as well.” While “requires” is too strong a term, the fact that more ridership is generated within a quarter mile is a valid point. Planning guidelines can indicate where to look for opportunities, with decisions guided by cost-effectiveness. Close-in densification seems sure to work, while farther out would need more design support to have attractive walking and discourage autos. Conclusions and further researchBART data show that the average rider walks more or less half a mile to a station in all sorts of neighborhoods, ranging from CBDs to dense neighborhoods like the Mission District to sprawling suburban areas like Pleasant Hill. The data also show wide range of walk distances. This differs from the five-minute rule described by Moran (2013), confirms Bergman, Gliebe, and Strathman (2011), and challenges pessimistic assumptions about willingness to walk to get to a transit station. High density around a station supports, but does not correlate with, increased walk access to transit and short walk distances. We found too many anomalous stations – with low density and high walk access, or vice versa – to support a positive relationship. Residential density needs to be combined with other factors to explain the walk to transit.Further research which would leave walk-to-BART to one side and look at density in relation to all non-auto modes combined, including other transit modes, walk, and bike, and how a rapid transit station could increase the length and amount of walking. More research could look at all the BART stations. Factors besides density influence mode choice to reach transit, many of which are included in mode choice computer models: parking availability, parking cost, tolls, car costs, transit costs, total travel time, total travel cost, auto availability, and many more. Over the last few years, BART has implemented parking charges at all stations, which could be studied to see the impact on ridership and on mode of access to stations. For example, very high parking charges at West Oakland BART have not discouraged ridership there, but may not have led to more non-auto access to the station. West Oakland avoids the congestion of driving into San Francisco, the bridge toll, and parking costs, and has very frequent trains, but the station area has low density and unattractive walking.While barriers and lack of safety discourage walking, we don’t know how much a pleasant, safe walking environment might encourage it. Increased safety, pleasant street lighting, pedestrian separation from traffic, and commercial activity with people on the street could make a difference, making distances over half a mile quite feasible. ReferencesBART. (2008). BART Station Profile Study [Online]. [Accessed: 25th February 2015.] Also email from BART planning staff.Agrawal, A., Schlossberg, M., & Irvin, K. (2008). How far, by which route and why? A spatial analysis of pedestrian preference, Journal of Urban Design, 13 (1), 81-98. . [Schlossberg et al. & Weinstein et al. are earlier versions of the same paper.][Schlossberg, M., Agrawal, A., Irvin, K., & Bekkouche, V. (2007). How far, by which route, and why? A spatial analysis of pedestrian preference. MTI Report, 06-06. San José, CA: Mineta Transportation Institute & College of Business, San José State University.][Weinstein, A. SEQ CHAPTER \h \r 1; Bekkouche, V., Irvin, K., & Schlossberg, M. (2006). How Far, by Which Route, and Why? A Spatial Analysis of Pedestrian Preference. TRB 2007 Annual Meeting CD-ROM. . ]Alshalalfah, B., & Shalaby, A. (2007). Case Study: Relationship of Walk Access Distance to Transit with Service, Travel and Personal Characteristics. Journal of Urban Planning and Development, pp. 114-118. , ?, John Gliebe, J., & Strathman, J. (2011). Modeling Access Mode Choice for Inter-Suburban Commuter Rail. Journal of Public Transportation, 14 (4): 23-42, 2011. . Burke, M. & Brown, A. (2007). Distances people walk for transport. Road and Transport Research, 16 (3), 16-29.Crowley, D., Shalaby, A., & Zarei, H. (2009). Access Walking Distance, Transit Use, and Transit-Oriented Development in North York City Center, Toronto, Canada. Transportation Research Record: Journal of the Transportation Research Board, pp. 96-105.Cervero, R., Round, A., Goldman, T., & Kang-Li Wu, K. (1995). Rail Access Modes and Catchment Areas for the BART System Institute of Urban and Regional Development University of California Berkeley, Working Paper UCTC, No. 307, Berkeley: University of California. Cervero, R. (1993). Ridership Impacts of Transit Focused Development in California. Working Paper UCTC, No 176, Berkeley: University of California.Daniels, R., & Corinne, M. (2011). “Explaining walking distance to public transport: the dominance of public transport supply,” World Symposium on Transport and Land Use Research, Whistler Canada.El-Geneidy, A., Grimsrud, M., Wasfi, R., Tétreault, P., & Surprenant-Legault, J. (2014). New evidence on walking distances to transit stops: Identifying redundancies and gaps using variable service areas. Transportation, 41(1), 193-210. Ker, I., & S. Ginn. (2003). Myths and Realities in Walkable Catchments: The Case of Walking and Transit. Road and Transport Research. Australian Road Research Board ARRB Group Limited. Leinberger, Christopher & Patrick Lynch. (2015). The WalkUP Wake-Up Call: Boston. George Washington University School of Business. , Maarit Marita. (2013). Walking the Walk?: An Assessment of the 5-Minute Rule in Transit Planning. ’Sullivan, S. & J. Morrall. (1996). Walking Distances To and From Light-Rail Transit Stations. Transportation Research Record, No. 1538, pp. 19-26. . Stringham, M. (1982). “Travel Behavior Associated With Land Uses Adjacent To Rapid Transit Stations.” ITE Journal 52(4) pp. 18–22.TCRP. (1996). TCRP Report 16: Transit and Urban Form, Volume 1. Part I: Transit, Urban Form, and the Built Environment: A Summary of Knowledge. Transportation Research Board, National Research Council, Washington, D.C. SEQ CHAPTER \h \r 1Two reports prepared for this project but not published are Mode of Access and Catchment Areas for Rail Transit, and Influence of Land Use Mix and Neighborhood Design on Transit Demand.Untermann, Richard. (1984). Accommodating the Pedestrian: Adapting Towns and Neighborhoods for Walking and Biking. Van Nostrand Reinhold. Data AppendixSupporting files are in a Dropbox folder at (Sign in is not really required.)Table 3: BART 2008 data on walk access from home originsUnweighted sample size correlates to level of walk accessBART 2008 Station Profile Study????StationUnweighted sample sizeMiles from home to BARTStandard DeviationMedianMean24th Street Mission595.446.507.321Glen Park446.433.464.275Downtown Berkeley439.583.627.35216th Street Mission405.395.456.291North Berkeley333.579.639.353Ashby330.486.517.27019th Street/Oakland308.755.738.316El Cerrito Plaza299.559.576.344Rockridge290.485.520.325MacArthur220.483.546.326Lake Merritt211.500.591.367Civic Center164.530.595.364Balboa Park161.538.627.35312 Street/Oakland149.381.480.411Pleasant Hill104.395.486.338Powell101.556.646.397Daly City96.493.629.380Fremont92.765.802.405Hayward87.634.716.479El Cerrito del Norte85.581.647.409San Leandro83.602.652.385Montgomery79.562.682.463Richmond79.699.721.430Fruitvale76.735.826.503West Oakland75.445.488.303Union City73.735.768.310Walnut Creek54.641.733.430Lafayette54.481.581.427Concord54.697.737.377Castro Valley51.724.831.402San Bruno50.744.868.409Colma48.625.648.313Bay Fair46.612.755.379SSF42.524.749.525South Hayward40.668.654.515Embarcadero36.478.647.518Millbrae27.755.882.411Coliseum26.671.964.699Pittsburg/ Bay Point231.0021.077.511Dublin/ Pleasanton191.3641.537.581North Concord/ Martinez18.876.806.305Orinda6.610.747.467Summary59740.540.598 .349Mean plus std deviation???.889Each station is weighted by its sample size for an accurate system-wide figure.Table 4: Census 2010 residential density within 0.5 miles of BART stationsStationsBlock groups within 0.5 milePopulationBlock group acresPeople per acre People per square mile12th St - OaklandBG 1, CT 40301,56936.243.327,701 BG 1, CT 40312,23885.726.116,712?BG 2, CT 40281,91776.125.216,127 BG 1, CT 40281,42822.463.940,870?BG 4, CT 40349949.4106.067,862?BG 1, CT 40291,43496.214.99,539 BG 2, CT 40341,34924.555.035,214?BG 2, CT 40301,21951.123.915,281Station summary:?12,148401.630.3 ??? ?16th St and MissionBG 2, CT 2011,33023.157.736,923?BG 3, CT 2011,48217.186.855,531?BG 4, CT 2012,06521.297.462,358?BG 1, CT 2081,40016.783.753,540?BG 2, CT 2081,93721.490.658,010?BG 1, CT 2072,29534.466.742,715?BG 3, CT 2022,52234.872.546,407?BG 2, CT 2022,57729.288.156,401?BG 1, CT 2021,17019.659.738,236?BG 1, CT 2011,29548.826.516,978?BG 2, CT 1771,653111.714.89,469Station summary:?19,726378.052.2 ?StationsBlock groups within 0.5 milePopulationBlock group acresPeople per acre People per square mile24th St and MissionBG 1, CT 2091,78425.470.244,910?BG 4, CT 2091,05025.940.625,993?BG 3, CT 20959512.946.229,599?BG 4, CT 2531,56936.443.127,599?BG 4, CT 21085316.950.432,269?BG 3, CT 2101,34925.952.033,286?BG 2, CT 21098317.157.336,685?BG 1, CT 2101,01117.258.837,628?BG 2, CT 2071,99934.458.137,176?BG 3, CT 2081,27418.469.144,240?BG 4, CT 2081,96628.668.643,933?BG 1, CT 228.031,86434.054.935,117?BG 1, CT 229.011,77225.469.744,616?BG 2, CT 229.011,69422.575.248,128?BG 3, CT 229.011,17413.686.155,109Station summary:?20,937354.859.0 ??? ?AshbyBG 2, CT 4239.011,10663.417.411,158?BG 3, CT 400568722.730.319,411?BG 1 CT 4239.0191434.226.717,113?BG 3, CT 42351,06445.323.515,024?BG 2, CT 423585649.017.511,190?BG 1, CT 42351,19862.719.112,229?BG 3, CT 42341,63161.326.617,036?BG 4, CT 42341,29549.526.216,759?BG 1, CT 4240.0188434.825.416,265?BG 2, CT 4240.0168017.938.024,296?BG 3, CT 4240.011,06734.031.420,079Station summary:?11,382474.724.0 ??? ?Glen ParkBG 2, CT 2171,11143.325.616,405?BG 4, CT 21897334.728.017,929?BG 1, CT 3111,00738.426.216,786?BG 3, CT 2171,62379.920.313,007?BG 3, CT 21868920.433.821,602?BG 1, CT 2551,37739.135.222,551?BG 2, CT 2551,16236.132.220,610Station summary:?7,942291.927.2 ??????StationsBlock groups within 0.5 milePopulationBlock group acresPeople per acre People per square mileMacarthurBG 4, CT 401071034.920.313,007?BG 3, CT 40101,26058.421.613,818?BG 6, CT 401095550.418.912,123?BG 5, CT 40101,03154.818.812,046?BG 1, CT 40141,09558.218.812,041?BG 1, CT 40131,041110.39.46,038?BG 3, CT 40111,04751.820.212,948?BG 2, CT 40111,32145.229.218,697?BG 4, CT 40111,12580.414.08,955Station summary:?9,585544.417.6???? ?Pleasant HillBG 1, CT 3382.031,982205.49.66,175?BG 2, CT 3382.031,822106.617.110,934?BG 2, CT 3400.012,309189.312.27,807?BG 3, CT 3240.011,94156.634.321,965?BG 4, CT 3240.0123942.25.73,628Station summary:?8,293600.113.8???? ?RockridgeBG 1, CT 40031,07865.716.410,508?BG 2, CT 400295674.012.98,272?BG 4, CT 40031,49873.720.313,005?BG 3, CT 40031,09172.215.19,673?BG 2, CT 40041,17261.019.212,300?BG 3, CT 40041,11048.822.714,546?BG 1, CT 40041,42164.122.214,195?BG 1, CT 40021,01873.113.98,914Station summary:?9,344532.517.5???? ?El Cerrito PlazaBG 2, CT 3891944114.68.25,273?BG 1, CT 38911,05852.320.212,954?BG 2, CT 38801,33392.314.49,240?BG 2, CT 390183277.210.86,896?BG 2, CT 389297036.726.416,894?BG 1, CT 389274424.430.519,522Station summary:?5,881397.514.8???? ?StationsBlock groups within 0.5 milePopulationBlock group acresPeople per acre People per square mileNorth BerkeleyBG 1, CT 42221,10076.414.49,212?BG 3, CT 42221,05451.020.713,225?BG 2, CT 422299039.325.216,110?BG 3, CT 42231,10139.228.117,995?BG 2, CT 42231,40359.823.515,016?BG 1, CT 422388339.422.414,329?BG 2, CT 421999657.617.311,063Station summary:?7,527362.820.7???? ?Downtown BerkeleyBG 2, CT 42292,34795.224.615,770?BG 1, CT 42291,98939.250.732,433?BG 2, CT 422489821.142.627,285?BG 2, CT 42301,52059.725.416,284?BG 3, CT 42282,32240.457.536,774Station summary:?9,076255.735.5???? ?19th St - OaklandBG 2, CT 40281,91776.125.216,127?BG 1, CT 40281,42822.463.940,870?BG 1, CT 40291,43496.214.99,539?BG 1, CT 4035.012,00066.330.219,312?BG 3, CT 40131,03953.019.612,538Station summary:?7,818314.024.9???? ?Lake MerrittBG 1, CT 40332,420184.813.18,381?BG 3, CT 40348849.395.561,092?BG 4, CT 40349949.4106.067,862?BG 2, CT 40341,43926.255.035,214?BG 2, CT 40301,21951.123.915,281?BG 1, CT 40301,56936.243.327,701?BG 2, CT 40331,63444.037.123,746Station summary:?10,159360.928.1???? ?PowellBG 2, CT 11797671.713.68,711?BG 2, CT 1211,10815.272.746,533?BG 2, CT 123.021,31015.385.954,957?BG 1, CT 123.011,50118.780.251,332?BG 2, CT 123.011,2133.8318.7203,937?BG 1, CT 125.021,9607.6258.8165,651?BG 2, CT 125.021,8617.6245.9157,345?BG 2, CT 125.011,54714.1109.870,280?BG 1, CT 125.013,78819.4195.0124,796?BG 2, CT 176.012,80127.0103.666,321?BG 5, CT 176.011,36519.271.145,511?BG 1, CT 178.011,45726.555.035,204Station summary:?20,887246.184.9???? ?StationsBlock groups within 0.5 milePopulationBlock group acresPeople per acre People per square mileCivic CenterBG 2, CT 123.011,2133.8318.7203,937?BG 1, CT 125.021,9607.6258.8165,651?BG 2, CT 125.021,8617.6245.9157,345?BG 1, CT 125.013,78819.4195.0124,796?BG 5, CT 176.011,36519.271.145,511?BG 2, CT 176.012,80127.0103.666,321?BG 2, CT 122.011,8687.5247.6158,480?BG 2, CT 124.013,13011.4275.5176,316?BG 1, CT 124.011,94511.4171.1109,532?BG 1, CT 124.021,06024.443.427,751?BG 3, CT 124.021,93315.5124.979,934?BG 2, CT 124.0298148.820.112,867?BG 3. CT 176.012,74345.959.838,283Station summary:?26,648249.4106.8???? ?Balboa ParkBG 2, CT 312.021,74345.338.524,633?BG 3, CT 2611,20730.239.925,554?BG 4, CT 2611,31566.519.812,648?BG 5, CT 2551,49434.243.727,988?BG 1, CT 312.012,46246.852.733,697Station summary:?8,221223.036.9???????Source: Social Explorer; Census 2010; American Fact Finder???Table 5: Monthly Ridership Report Nov. 2014 weekdaysMaps credit: Cheyenne Concepcion, May 2015, using Illustrator with data from US Census Data Center TIGER files and American Community Survey Social Explorer, exported to jpgs. ................
................
In order to avoid copyright disputes, this page is only a partial summary.
To fulfill the demand for quickly locating and searching documents.
It is intelligent file search solution for home and business.
Related searches
- access to education for children
- access to education global
- lack of access to education
- access to internet google
- access to education in america
- free access to unclaimed money
- xfinity e mail access to my account
- access stored usernames and passwords
- china access to medical care
- how to find volume from density equation
- free access to scholarly journals
- what to wear for bone density test