MBTA Commuter Rail Station Site
Suitability, Everett, Massachusetts
John Taylor | UEP232 Introduction to GIS Final Project | May 7, 2013
Introduction
Everett, Massachusetts is a small inner-core city north of Boston. It has a high-density
residential area north of Route 16 - which bifurcates the city - and a largely industrial and
commercial area south of Route 16. Despite its dense residential makeup and being an
industrial employment hub, public transit is lacking. The city is served by numerous bus routes,
but none of these connect directly with downtown Boston; one must connect via Orange Line
stations in Somerville and Malden. The MBTA Newburyport/Rockport commuter rail line does
pass through the city, however, and Everett planners and policymakers have long discussed
adding a stop in Everett, between North Station and Chelsea. Such a stop also has potential to
encourage new transit-oriented development.
The commuter rail tracks run through two distinct Everett neighborhoods: Lower Broadway and
the Commercial Triangle. A potential station would have to be situated in one of the two areas.
The City is currently engaged in neighborhood master planning processes for both areas,
making it an ideal time to analyze potential commuter rail sites in the two neighborhoods. Lower
Broadway is an appealing site if a proposed casino deal comes through, and is more centrally
located between the current stations at North Station and Chelsea. The Commercial Triangle is
appealing because it is more easily accessible from all points in Everett.
This analysis will seek to identify the most suitable locations for a commuter rail station in
Everett, based on criteria from the literature, parcel cost and ownership data, and a desire from
Everett policymakers for walkable transit access.
Literature Review
Horner, M. W., & Grubesic, T. H. (2001). A GIS-based planning approach to locating urban rail
terminals. Transportation, 28(1), 55-55. Retrieved from
http://search.proquest.com.ezproxy.library.tufts.edu/docview/212427486?accountid=14434.
This article develops criteria for siting park-and-ride rail stations, specifically in Columbus, Ohio.
The model uses Census block level data for the study area to tabulate general population
density. It also looked at socioeconomic data associated with the study area population to
determine areas with the highest concentration of young, well-educated, middle-income
individuals, which they posited were more likely to utilize the new rail station. In addition to
1
population, the model assumed that any new station would have to be built along an existing rail
line (rather than an extension), would require good road access, and that locations could not be
too close the CBD. It delineates a 5-minute travel-time catchment area, and acknowledges that
taking obstructions and connectivity of the catchment-area road network is crucial. While parkand-ride facilities are clearly different than walkable urban rail stations, the general types of
criteria the authors used are helpful to my own analysis. The most transferrable criteria are
travel-time from home to station (5 minutes: 3 miles driving in this case, ¼ to ½ mile walking in
my analysis), road network analysis, and awareness of distance between the station and final
destination.
Rosenberg, J. L., & Esnard, A. M. (2008). Applying a hybrid scoring methodology to transit site
selection. Journal of Urban Planning and Development, 134(4), 180-186. Retrieved from
http://ascelibrary.org/doi/full/10.1061/%28ASCE%290733-9488%282008%29134%3A4%28180%29.
This article discusses development of a scoring approach to compare the suitability of six
potential locations for a new train station in Fort Lauderdale, Florida. The model incorporates
multiple criteria sets to select the most appropriate site. The first is proximity to retail,
government buildings, and housing (using a ¼ mile pedestrian catchment area). The second set
is developability of the site, working with an assumption that “the least expensive properties, the
most immediately available properties, the properties that are most easily attainable (e.g.,
government owned), and specific land uses (e.g., vacant parcels) would be the more
developable.” The model then creates a detailed scoring rubric on a 1-5 scale to convey each of
these criteria. Because worked with a more limited set of input criteria, I did not develop a
scoring methodology for my own analysis, but the specified criteria here were helpful in
developing my own guiding principles.
Santa Clara Valley Transportation Authority. (2007). Station Area Service Guidelines. VTA Transit
Sustainability Policy 2007. Retrieved from
http://www.vta.org/projects/tsp/pdf/8_station_area_service_guidelines.pdf.
This document contains rail station design guidelines for California rail systems that are
transferrable to my own analysis. Although this document discusses ridership estimates, which
my analysis will not, it does provide recommended densities within 1/3 mile of a rail station,
recommending at least 10, 20 or 45 dwelling units per acre as the least, acceptable, and ideal
measures, respectively. It also recommends distance between stations of at least one mile.
Finally it recommends station platforms of at least 700 feet in length. To ensure that this
recommendation made sense in the context of the Boston area as compared to the Bay Area, I
looked at recent urban commuter rail station designs for the Fairmount Indigo line in Dorchester,
and the recently opened Talbot Avenue station length is 800 feet
(http://www.mbta.com/about_the_mbta/t_projects/default.asp?id=14582).
2
Weinstein 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. Retrieved from
http://www.tandfonline.com/doi/full/10.1080/13574800701804074.
This article is not a GIS-based analysis, but provides critical information for my own analysis in
determining how far individuals will walk to a transit station. While the general rule of thumb has
tended to be .25 miles, and is still used in many cases, the authors conclude that the average
pedestrian is willing to walk .58 miles to a transit station. The median distance was .47 miles,
indicating that many people actually walked considerably more than .5 miles to transit. The
researchers based their conclusions on a 328-person study, which they specifically designed to
get around the absence/limitations of walking data from the Census, National Household Travel
Survey, and other regional surveys. They found that most of these surveys, if they have walking
distance data at all, tend to round to the nearest mile. Of the five stations in California and
Oregon analyzed, three were situated along major arterials that pedestrians had to cross,
similar to any station in Everett. Survey respondents actually traced their route on a map, which
researchers then used to calculate exact distances, eliminating some of the self-reporting errors
often associated with walking distance.
Methodology
Based on the literature review and best practices, a series of guiding principles were used to
determine suitable sites, primarily using parcel data from the City of Everett, commuter rail data
and 2010 U.S. Census data.
1. The site should be at least one mile away (track distance) from the nearest station in
either direction: Chelsea and North Station
2. The site should be within 100 feet of commuter rail tracks
3. Residential parcels should be excluded
4. Suitable parcels should be at least 0.5 acres to allow for basic station infrastructure
After these parcels were identified, each site was analyzed to show the following:
5. Population within walking distance (0.5 miles)
6. Vacancy and underutilization, using a rough estimate of (assessed building value/total
parcel value)
7. Per acre assessed value, to compare different size parcels by price
Using various GIS tools, including selection by attribute and location, clip, buffer, erase, and
network analyst, all Everett parcels were systematically narrowed down to illustrate only the
parcels meeting the aforementioned criteria. This analysis does not rank the resulting sites, but
presents comparable data points for each so that policymakers can make an informed decision.
3
Data Layers
Layer
Buildings
Parcels
Data Set Name
bldgs_poly.shp
EverettTaxPar.shp
Description
Year
All buildings in
the City of
Everett
2012
All tax parcels in
the City of
Everett
2012
Data
Source
Agency
City of
Everett
City of
Everett
Roads
EOTROADS_ARC.shp
All public, and
most private,
roadways in
Massachusetts
Commuter
Rail
TRAINS_RTE_TRAINS
.shp
All commuter rail
lines in
Massachusetts
2008
MassGIS
Commuter
Rail
Stations
TRAINS_NODE.shp
MBTA commuter
rail stations
2008
MassGIS
Census
Blocks
CENSUS2010BLOCKS
_POLY.shp
Census blocklevel boundaries
from 2010 U.S.
Census
2010
MassGIS
P1 Total Population
Total population,
by Census block,
of Middlesex
County,
2010
2010 U.S
Census,
American
Fact
Census
Block
Population
2012
MassGIS
URL
n/a
n/a
http://www.mass
.gov/anf/researc
h-and-tech/itserv-andsupport/applicati
on-serv/officeof-geographicinformationmassgis/datalay
ers/eotroads.ht
ml
http://www.mass
.gov/anf/researc
h-and-tech/itserv-andsupport/applicati
on-serv/officeof-geographicinformationmassgis/datalay
ers/trains.html
http://www.mass
.gov/anf/researc
h-and-tech/itserv-andsupport/applicati
on-serv/officeof-geographicinformationmassgis/datalay
ers/trains.html
http://www.mass
.gov/anf/researc
h-and-tech/itserv-andsupport/applicati
on-serv/officeof-geographicinformationmassgis/datalay
ers/census2010.
html
http://factfinder2
.census.gov/fac
es/nav/jsf/pages
/index.xhtml
Key
Attribute
Variables
none
BLDG_VAL
TOTAL_VAL
LOT_SIZE
USE_CODE
SITE_ADDR
STYLE
none
COMM_LINE
STATION
none
BlockPop
4
Massachusetts
from 2010 U.S.
Census
Finder
Open
Space
OPENSPACE_POLY.s
hp
All open space in
Massachusetts
2012
MassGIS
http://www.mass
.gov/anf/researc
h-and-tech/itserv-andsupport/applicati
on-serv/officeof-geographicinformationmassgis/datalay
ers/osp.html
Everett
Border
city_boundary.shp
City border of
Everett
2012
City of
Everett
n/a
All
Borders
Water
Ortho
TOWNS_POLY.shp
All
Massachusetts
city and town
borders
201
MassGIS
HYDRO25K_POLY.shp
Includes lakes,
ponds,
reservoirs,
wetlands, bogs,
flats, rivers and
streams in
Massachusetts
2010
MassGIS
coq08_30cm_7.sid
30cm
orthographic
imagery for
Boston’s north
shore
2008
MassGIS
http://www.mass
.gov/anf/researc
h-and-tech/itserv-andsupport/applicati
on-serv/officeof-geographicinformationmassgis/datalay
ers/hd.html
http://www.mass
.gov/anf/researc
h-and-tech/itserv-andsupport/applicati
on-serv/officeof-geographicinformationmassgis/datalay
ers/hd.html
http://www.mass
.gov/anf/researc
h-and-tech/itserv-andsupport/applicati
on-serv/officeof-geographicinformationmassgis/datalay
ers/colororthos2
008.html
none
none
none
none
none
* All layers contain vector data, except the ortho photo.
Base Data Layer Processing
Buildings
1. Downloaded bldgs_poly.shp from City of Everett GIS file
2. In ArcGIS, renamed layer “Buildings”
5
Parcels
1. Downloaded EverettTaxPar.shp from City of Everett GIS file
2. Selected by attribute to exclude road and rail right of way polygons, along with water
polygons: “POLY_TYPE” = ‘FEE’
3. In ArcGIS, renamed layer “Parcels”
Roads
1. Downloaded EOTROADS_ARC.shp from MassGIS on Tufts M: drive
2. Clipped data with FinalProject_Boundary.shp layer to create layer of roads in Metro
Boston
3. Projected to NAD 1983 StatePlane Massachusetts Mainland FIPS 2001 (Feet)
4. In ArcGIS, renamed layer “Roads”
Commuter Rail
1. Downloaded TRAINS_RTE_TRAINS.shp from MassGIS on Tufts M: drive
2. Clipped data with FinalProject_Boundary.shp layer
3. Projected to NAD 1983 StatePlane Massachusetts Mainland FIPS 2001 (Feet)
4. Selected by attribute to select only Newburyport/Rockport MBTA Line, which runs
through Everett: "COMM_LINE" = 'NEWBURYPORT/ROCKPORT'
5. In ArcGIS, renamed layer “Commuter Rail”
Commuter Rail Stations
1. Downloaded TRAINS_NODE.shp from MassGIS on Tufts M: drive
2. Clipped data with FinalProject_Boundary.shp layer to create layer of stations in Metro
Boston
3. Projected to NAD 1983 StatePlane Massachusetts Mainland FIPS 2001 (Feet)
4. Selected by attribute to select only first stations on either side of Everett: "STATION" =
'CHELSEA' OR "STATION" = 'NORTH STATION'
5. In ArcGIS, renamed layer “Commuter Rail Stations”
Census Blocks
1. Downloaded CENSUS2010BLOCKS_POLY.shp from MassGIS on Tufts M: drive
2. Clipped data with Everett Border layer to create layer of census blocks located in Everett
3. Projected to NAD 1983 StatePlane Massachusetts Mainland FIPS 2001 (Feet)
4. Created new layer named “Census Blocks”
Census Block Population
1. Downloaded Table P1 Total Population from American Fact Finder, with the following
filters:
Dataset
2010 SF1 100% Data, Population Total
6
Geography
All blocks within Middlesex County, Massachusetts
2. Converted .csv to .xlsx file, renamed file, and active sheet,
CensusBlockPopulationMiddlesex
3. Changed column headings as follows: GEO.display-label = CensBlock; D001 =
BlockPop
4. Changed GEO.id2 cell format to “number”
5. Joined table to “Census Blocks” shapefile in ArcGIS
Open Space
1. Downloaded OPENSPACE_POLY.shp from MassGIS
2. Clipped data with Everett Border layer
3. In ArcGIS, renamed layer “Open Space”
Everett Border
1. Downloaded city_boundary.shp from City of Everett GIS file
2. In ArcGIS, renamed layer “Everett Border”
All Borders
1. Downloaded TOWNS_POLY.shp from MassGIS on Tufts M: drive
2. In ArcGIS, renamed layer “All Borders”
Water
1. Downloaded HYDRO25K_POLY.shp from MassGIS on Tufts M: drive
2. Clipped data with FinalProject_Boundary.shp layer to create layer of roads in Metro
Boston
3. Projected to NAD 1983 StatePlane Massachusetts Mainland FIPS 2001 (Feet)
4. In ArcGIS, renamed layer “Water”
Orthography
1. Downloaded coq08_30cm_7.sid (2008-2009 30cm mosaic for north of Boston) from
MassGIS on Tufts M: drive
2. In ArcGIS, renamed layer “EverettOrtho”
Data Analysis
Commuter Rail
1. Use network analyst tool, setting the Newburyport/Rockport commuter rail line as the
network
2. Within network analyst, set one mile distance from facilities (Chelsea station and North
station) to select portions of track extending 1 mile in each direction from each station
3. Save as new layer
7
4. Use erase tool to erase those portions of commuter rail line located within 1 mile of
either station, leaving just the stretches of track greater than 1 mile from the nearest
station
5. Save as new layer of commuter rail line
6. Clip new layer to Everett boundary, creating new layer “Everett Commuter Rail Tracks
Feasible for Station”
Parcels
1. Select by location all parcels within 100 feet of new layer “Everett Commuter Rail Tracks
Feasible for Station”, located in Lower Broadway or Commercial Triangle (excludes
Gateway Shopping Center)
2. Select by attribute all nonresidential parcels, including industrial, commercial, cityowned, and MBTA-owned
a. All use codes greater than or equal to 300 in parcel dataset are non-residential,
including industrial, commercial, and government-owned
b. Select: GIS_Assess.USE_CODE >= ‘300’
3. Select by attribute all parcels at least 0.5 acres
4. Save as new layer “Nonresidential Parcels Within 100 ft of Tracks, at least half acre”
Utilization
1. Add new field to “Nonresidential Parcels Within 100 ft of Tracks, at least half acre”
calculating rough estimation of land utilization, dividing assessed building value by total
land value (BLDG_VAL/TOTAL_VAL)
2. Value of 0 represents completely vacant land; values close to 1 indicate that building
value is largely responsible for most of total land value
Total Value per Acre
1. Add new field to “Nonresidential Parcels Within 100 ft of Tracks, at least half acre”
calculating total value per acre (TOTAL_VAL/AREA_ACRES)
2. Provides better potential cost comparison over different sized parcels
Potential Stations
1. Create new shapefile, placing a point along commuter rail line for each suitable parcel
2. Conduct spatial join of new potential station shapefile with “Nonresidential Parcels Within
100 ft of Tracks, at least half acre,” giving each point the attributes of the polygon closest
to it
3. Save new layer “Potential Stations Parcels_within_100_ft_nonres_halfacre”
Population within 0.5-mile radius of potential stations
1. Use buffer tool for “Potential Stations Parcels_within_100_ft_nonres_halfacre,” setting a
distance of 0.5 miles
2. Clip new layer to Everett boundary, as Everett residents will likely be the primary users
of any station
3. Conduct spatial join with Census Block Population layer
8
a. Estimate may be slightly higher than actual due to census block boundaries, but
it is a close approximation
4. Save new layer as “Half_Mile_Catchment_Area_with_Census_Data”
Final Mapping
1. Create map showing all 10 potential station parcels
2. Create map showing utilization of each parcel, defined as building value/total land value
3. Create map showing total land value per acre for each parcel
4. Create groupings for each potential site, including the station location, parcel, and 0.5
mile catchment area (with census data); display each of the 10 groupings in its own map
5. Create table of key attributes for each parcel, including assessed building value,
assessed land value, utilization, ownership, acreage, population within walking distance
(poster space permitting)
Limitations
Data quality in this analysis was generally high. The only parcel that raised a potential red flag
was the one closest to the Chelsea city border (Site 10). The parcel had one building, covering
less than 5% of the total land area, but the building value was an unusually large percentage of
the total parcel value. It was unclear from the data if this was an accurate reflection of economic
activity on the site, or if it was an error in the data.
The major limitations of this analysis were related to the Census block data and 0.5-mile
pedestrian catchment area. First, the population counts within each catchment area in the
analysis are likely slightly higher than actual. This is because on the outer edges of the
catchment area, if the 0.5-mile radius touched any part of a Census block, it included the whole
block’s population in the total count, even if only a fraction of the block population is actually in
the 0.5-mile radius. Still, the blocks in the affected areas were relatively small, and as noted
from the Weinstein article, pedestrians may be willing to walk an average of 0.58 miles to a
transit station.
Second, the 0.5-mile circular catchment area likely overestimates population because using the
existing road network is almost always a greater distance than as the crow flies. Using a circular
buffer, though, rather than road network analysis, was a conscious decision. It is likely that new
developments in Lower Broadway and the Commercial Triangle will include improved road
accessibility, so the method used gives a fuller accounting of the potential populations.
Conclusions and Further Research
As shown in the accompanying maps, 10 sites met the criteria for commuter rail station
suitability. Each has a different combination of parcel affordability and population access, but a
few sites stand out for doing well on both fronts. Site 4, located at the nexus of Lower Broadway
and the Commercial Triangle, has low per acre value, low utilization, and has a 0.5-mile
catchment area including over 8,000 people. Site 9, in the middle of the Commercial Triangle, is
9
almost as undervalued and underutilized, with over 7,000 people in the 0.5-mile catchment
area. These two sites stand out, but again, all 10 sites have potential based on the priorities of
residents and officials. Site 1 on the Mystic River has a low residential population nearby
currently, but as the site of a potential casino, it could accommodate a station that would serve
visitors and a new residential population as Lower Broadway develops.
As the planning processes for Lower Broadway and the Commercial Triangle continue, and as
the fate of the casino is decided later in 2013, it would be useful to do more analysis to estimate
new residential populations and visitor demand, to see the impact on site suitability. It would
also be useful to analyze potential improvements in the road network and model how those
improvements increase accessibility to potential stations throughout Everett.
10