Child pages
  • Tolkoff Final project

Laura Tolkoff

Introduction to GIS

Barbara Parmenter

4 May 2010


Healthy Development in Somerville and Cambridge



The emerging field of healthy city planning symbolizes a growing paradigm shift regarding health and the built environment. It recognizes that the United States is in the midst of a health crisis, one that started with environmental health in the 1960s and 1970s and continues today with the obesity epidemic, lifestyle options, and poor access to healthcare. Healthy city planning recognizes the ways in which the built environment and planning influence health outcomes. For instance, sprawling cities increase dependence on the vehicle, which decreases active transport, increases air pollution and respiratory ill health, and decreases the opportunities for the very young and the elderly to enjoy city amenities. Decreases in affordable and public housing units due to chronic underfunding and gentrification results in displacement, overcrowding in market-rate housing, and strain to pay for other necessities like heating and transportation to jobs, economic opportunities and social networks—all of which result in high levels of stress and increased chances of mental health problems. The World Health Organization’s Social Determinants of Health and the healthy city planning movement acknowledge that interventions, such as healthcare, are unable to meet the growing health disparities and systemic health problems in the city. Instead, it calls for and cooperation amongst municipal agencies, particularly for integration of public health and urban planning, to create more sustainable and healthy cities.


Healthy City Planning and the Healthy Development Measurement Tool

The healthy city planning framework that most strongly influences this spatial analysis of Cambridge and Somerville originates from the San Francisco Bay Area and the Eastern Neighborhoods Community Health Impact Assessment. The Eastern Neighborhoods Community Health Impact Assessment (ENCHIA) resulted from a community-wide response about the redevelopment of San Francisco’s mission district and the resulting gentrification and displacement that neighborhoods feared.  In 2002, the Mission Anti-displacement Coalition’s (MAC) began to develop a “People’s Plan” for the redevelopment of San Francisco’s Mission District that would be an alternative to the Department of City Planning’s plan  for the Mission and the Eastern neighborhoods of San Francisco (Corburn, 163). “The People’s Plan was a land use, zoning, and community development plan drafted by the MAC and endorsed by thousands of local residents…and proposed, among other things, [the promotion of] more affordable housing, [the preservation of] industrial sector jobs, and [to] stop the demolition of existing buildings” (163).              

Upon completion, the MAC asked the San Francisco Department of Public Health to review the health implications of the People’s Plan through a health impact assessment, a review of the likely health impacts of proposed developments and zoning changes. The planning department sought only environmental and social impact assessments for its new plan. Therefore, the MAC and the DPH worked to conduct a health impact assessment (HIA) for both the People’s Plan and the Planning Department’s new plan, although the Planning Department challenged the legitimacy of the HIA made no promises to incorporate the findings of the assessment into their plan (166). Nevertheless, the Health Department sought to make the HIA an inclusive process, emphasizing the World Health Organization’s Gothenburg Statement, “which emphasized democratic participation, equity, and transparency in the analytic process.”

In the end, the ENCHIA is important because it is a different type of “public knowledge-generating process,” embodying the type of scientific experimentation and interdisciplinary nature of healthy city planning. Additionally, the ENCHIA was also important in that it “kept racism part of the discussion” about health and urban form, though some thought the issue “would polarize the group and contribute to participants leaving the process,” while others thought that too often the issue of race was discarded “for the sake of ‘getting things done’” (176). The group agreed that addressing racism was necessary, but that the HIA use the discussion to “focus on how racism might be manifested in land use issues, such as transit access, affordable housing, environmental quality, and economic opportunities” (176). This shows the ways in which these processes can engage with critical discussions of race and equity in ways that are often missing from current planning debates, subsumed under terms like “concentrated poverty” and “socio-economic position.” In the end, the group agreed upon 27 health objectives for San Francisco, which would eventually inform 27 policy briefs to the Planning Department and the Healthy Development Measurement Tool (HDMT) (Corburn, 188).

The Healthy Development Measurement Tool, which I find especially interesting for its unique applicability for vulnerability analyses in Geographic Information Systems, was organized in 2006 after the ENCHIA process and released by the Public Health Department in 2007.  There are three components to the HDMT, the first of which is the Community Health Indicator System, a set of 100 indicators meant to evaluate current conditions in a city and monitor them over time. The second component is a “checklist” to “help stakeholders evaluate specific attributes of development plans and projects” that can act as a guideline for planning processes and development, though it should be adapted to the unique conditions and goals of specific neighborhoods ( ).  The third part is a Menu of Policy and Design strategies, a “list of actions that can be taken by project sponsors/policy makers to achieve health objectives and development goals” ( www.hd m ).

The HDMT used Geographic Information Systems to provide a spatial analysis of health disparities and indicators at the neighborhood level. The “fundamental value behind the HDMT is that all communities should have equal access to health resources”, and a spatial representation of health indicators allows us to see health disparities and better inform decisions about development and city planning. For the scope of this project, I limit my analysis to a few key indicators that were chosen based on interest and on the availability of data. For a full list of indicators, please see .

Part II: Methodology and Spatial Analysis


The vulnerability analysis presented here is based on the community health indicators in the Healthy Development Measurement Tool. Though the HDMT analyzes determinants at the neighborhood level, the analysis for Somerville and Cambridge is aggregated at the block group level. And while the HDMT encompasses over 100 community health indicators, I looked at six different indicators measured in one or more ways.

The following table summarizes the indicators, or social determinants of health, and the ways in which they are measured.

Table 1:  Indicators of Health and Their Measurements




Access to Public Transit Networks

Proximity to MBTA nodes

Access to Active Transit Networks

Proximity to designated bike lanes

Access to Unhealthy Businesses

Proximity to liquor stores; Proximity to convenience stores; Proximity to grocery stores

Access to Healthcare and Intervention Services

Proximity to community health centers

Socio-economic Information

Percentage of renter occupied housing; Median household income


Each block group is given a score for how it performs on each indicator. The scores are based on the scale of one through four, where four is the highest score or “healthy” and represented by dark green.  A score of three is “relatively healthy” and is represented by sea foam green. Block groups that score a two are “unhealthy” and are coded as the color orange. A score of one is “very unhealthy” and is represented by berry red. Since there are eight measurements of these indicators, block groups can earn a total of 32 points.


Data Sources and Data Quality Assessment

Table 2: Data Sources

Layer Name/Source

Coordinate System

Census 2000 Block Groups

NAD 1983 State Plane


Mass. Community Health Centers

NAD 1983 State Plane


Mass. Infrastructure MBTA Nodes

NAD 1983 State Plane


Mass. EOT Roads

NAD 1983 State Plane


Reference USA Convenience Stores

Code 541103


Reference USA Liquor Stores

Code 445310


Reference USA Grocery Stores

Code 541105


Cambridge City Boundary

NAD 1983 State Plane


Somerville City Boundary

NAD 1983 State Plane


Cambridge Bike Lanes (existing only)



Somerville Bike Lanes (existing only)



Census 2000 Median Household Income Data



Census 2000 Housing Data



Census 2000 Tiger Roads




I was unable to obtain data for bike lane information that was readily usable for GIS. Therefore, Barbara, Abi and I digitized existing bike lanes from the two PDF maps for Cambridge and Somerville’s bicycle networks, which were easily accessible online. However, this introduces some inaccuracy into the analysis. For instance, the bike lanes did not exactly match the Census2000 Tiger Roads layer, nor the orthophotography. Moreover, when I tried to calculate the number of miles of bike lanes per block group, the lanes were not assigned correctly to block groups. The lanes should have perfectly overlaid the major roads and been attributed to the block groups adjacent to the lane/road on both sides. However, inaccuracies in the reference layer (roads) and the digitizing process would have made this calculation inaccurate, misleading, and somewhat inconclusive in assessing health. Instead, I used a proximity analysis to determine the distance to bike lanes.

To simplify my analysis, I created a new layer file from the Block Groups Layer, Infrastructure Layers, MBTA nodes, and Community Health Centers so that the data set would be limited to Cambridge and Somerville, and therefore more manageable.



Each measure involved a proximity analysis. In sum, the steps included:

  1. Enable the Surface Analysis Extensions
  2. Set options: Cell size 10
  3. Select Surface Analysis Distance Straight Line
  4. Zonal Statistics: Mean Distance to ­ _________________ ­ ­ (convenience stores, liquor stores, grocery stores, bike lanes, MBTA nodes)
  5. Add new Field (titled something like ProxMBTA or ProxLQR)
  6. Field Calculator: set new field=zonal statistics Mean
  7. Add new field for indicator score (titled something like ScrMBTA or ScrLQR)
  8. Select by attribute. Since most of these were proximity analyses, I used  the following attributes for scoring:

<400 meters

>=400 meters and <800 meters

>=800 meters and <1200 meters

>=1200 meters and <1600 meters

and in some cases, such as the MBTA measure, X>1200 since there were several values over 1600 meters away from an MBTA node.

  1. Field Calculator to score (eg, attribute<400=4)


Figure 1: Attribute Table Demonstrating the Zonal Statistics and Scoring Results


I repeated that process for all of the proximity/distance indicators.

I also performed some density analyses in order to determine the density of certain factors, such as liquor stores and convenience stores, per block group. However, these were not reported in my poster or scored for the final cumulative map.  The steps for a density map is as follows:

  1. Enable the Surface Analysis Extensions
  2. Set options: Cell size 10
  3. Select Surface Analysis Density


In order to make the PDF maps of bike lanes for Cambridge and Somerville usable, Abi, Barbara and I georeferenced or digitized the two maps. The process is summarized below:

  1. We picked and plotted five points on the EOT roads layer as reference points.
  2. We overlaid the Somerville bicycle lane map over the Cambridge and Somerville block groups and EOT roads layer.
  3. We enabled the georeferencing tool and found the intersections (points) on the scanned maps and connected them to the points we plotted individually. This re-oriented or shifted the map in line with the reference layer. We repeated this for all five points.
  4. I used the Data Editor tool to trace the bike lanes onto the map we are creating.
  5. We created a shape file and added a new field to the attribute table called “Length”
  6. I used Field Calculator to calculate the length of the traced/digitized bike lanes.



The coloration of each map of individual measure can be confusing. For instance, the areas that are closest to indicators of ill health—namely, convenience stores, liquor stores, and fast food restaurants—are marked in berry red. The areas that are further away from these establishments are dark green. In contrast, other establishments or indicators of positive health—such as MBTA nodes, bike lanes and community health centers—are coded as dark green.

Figure 3. Proximity is Coded as Berry Red Because Convenience Stores are elements of Ill Health



Some of the maps of individual indicators demonstrate that a few social determinants of health are well regulated. For instance, the density map of liquor stores shows that they are not concentrated in any one block group or area, as shown below.

Figure 4.  Map of Density of Liquor Stores per Block Group


Figure 5. Density Map of Convenience Stores per Block group





The final, cumulative map shows that the highest score in a block group is twenty-six points out of thirty-two points possible. The areas in red are “very unhealthy” and should be prioritized for development that promotes health. However, since the block groups that are dark green and score as “healthy” could still use improvement and healthy development. Nonetheless, the spatial analysis presented in this project is limited to only five indicators and eight measurements of the indicators. A more comprehensive approach that can analyze the full 100 community health indicators will give planners, developers, and public health promoters a more concrete guide as to which block groups should be targeted for healthy development and in which ways.


Figure 6. Cumulative Map: Healthy Development Score









Buzzelli, Michael and Gerry Veenstra.  2007. New approaches to researching environmental

      justice: combining critical theory, population health and geographical information

     science (GIS).  Introduction, Health and Place 13:1. March 2007, pp. 1-2.


Jerrett, Michael, Gale, S., C. Kontgis, and MS Student. 2009. What GIS tells us about

      environmental and public health. University of California, Berkeley .

This article reviews multiple approaches for quantitative and spatial assessments of environmental health risks and exposures. It starts with a brief review of land use decisions and urban planning issues, such as urban sprawl and auto-dependence, and their relationships to environmental health sciences. The authors make clear that they utilize environmental justice and “geography of risk” frameworks for their spatial conceptions of environmental health (4). The authors look closely at other environmental health articles that use GIS for their spatial analyses of environmental risks, specifically in vulnerability analyses for pollution and climate change. They identify specific methodologies such as cluster analysis to show disease patterns and exposure patterns as well as to identify environmental justice communities (11). They also show the benefits of GIS statistical queries for environmental health data. The authors further point to a conurbation scale risk assessment to compare various scenarios of risk and adaptation (18). The article warns, however, that mapping in environmental health is susceptible to logical fallacies; it is difficult to “control simultaneously for all known risk factors…and analysts may have to temper conclusions,” showing association and not causation (22).


Maantay, Juliana. 2007. Asthma and air pollution in the Bronx: Methodological and data

     considerations in using GIS for environmental justice and health research.  Health and

     Place 13:1. March 2007, pp. 3-56.


Maantay’s analysis of asthma and air pollution in the Bronx illustrates some of the problems and complications of mapping health outcomes and health data in GIS. Some of these problems include the appropriate scale of analysis and spatial resolution, data accuracy including attribute and positional distortions, the reliability of self-reported health data, confidentiality, among many others. 


MacKendrick, N., J. Parkins, M. P. B. Initiative, and M. P. B. I. PO. 2005. Social dimensions of

     community vulnerability to mountain pine beetle.”

This article builds on risk assessment and vulnerability analysis for community vulnerability analysis, adaptation, and resiliency (8). This article has a strong literature review regarding public perceptions of risk and vulnerability analyses frameworks, including physical, political and economic dimensions (19). Though this article does not utilize geographic information systems, it is a good example of social science methodology for developing indicators and other study elements. It is also extremely informative about vulnerability analysis in environmental health.


Malhotra, Shriya. 2009. Conflating boundaries to Envision Public Health. Parsons Journal for

     Information Mapping 1:3.

This article presents the general needs for spatial analysis and GIS in Public Health, citing that visualization of patterns allows for greater communicability and education (2).


San Francisco Department of Public Health. The Healthy Development Measurement Tool. (accessed 12 April 2010).


Sengupta, S., et al. 1996. Assessment of population exposure and risk zones due to air

      pollution using the geographical information system. Computers, Environment, and

     Urban Systems 20: 3, pp. 191-199.


Tim, U. Sunday.1995. Review: the application of GIS in Environmental Health Sciences:

     Opportunities and Limitations. Environmental Research 71, pp. 75-88.


This article “examines major issues related to the application of GIS in environmental health sciences,” pointing to its benefits and to its limitations. Among the standard complexities of data quality such as currency, completeness, and “distortions in position”, there are other problems regarding data security and confidentiality. The author notes that environmental health spatial analysis “presents a conflicting dichotomy between the need to allow the free access or exchange of information and the desire to maintain confidentiality,” particularly when aggregating information from multiple data sets (83).