{"id":5931,"date":"2016-12-11T15:59:01","date_gmt":"2016-12-11T20:59:01","guid":{"rendered":"http:\/\/research.prattsils.org\/?p=5931"},"modified":"2016-12-11T15:59:01","modified_gmt":"2016-12-11T20:59:01","slug":"mapping-superfund-sites-new-york-state","status":"publish","type":"post","link":"https:\/\/studentwork.prattsi.org\/infovis\/visualization\/mapping-superfund-sites-new-york-state\/","title":{"rendered":"Mapping Superfund Sites in New York State"},"content":{"rendered":"

For the final project, I researched the presence and frequency of Superfund sites across New York state, creating a<\/p>\n

dashboard and bar charts, along with a map of the counties in the state with Superfund site(s).<\/p>\n

A Superfund site is \u201cany land in the United States that has been contaminated by hazardous waste and identified by the<\/p>\n

Environmental Protection Agency<\/a> (EPA) as a candidate for cleanup because it poses a risk to human health and\/or the<\/p>\n

environment.\u201d Examples of sites include former chemical or industrial facilities,landfills, and waterways, such as the highly-<\/p>\n

contaminated Gowanus Canal, inBrooklyn, New York.<\/p>\n

In working on this project, I wanted to become more proficient withvisualization programs, and to see what sort of story<\/p>\n

I could tell with the data. Ultimately I wound up using Tableau, for reasons discussed later on.<\/p>\n

The subject matter is something with which I had some familiarity, having worked for the Department of Energy, Office of<\/p>\n

Environmental Management (DOE-EM). That program office is responsible for cleaning up Manhattan project and other<\/p>\n

agency-owned nuclear sites across the country, such as the Hanford complex in eastern Washington state. Working off that<\/p>\n

experience, I shifted focus to Superfund sites and began work on the project.<\/p>\n

 <\/p>\n

\"final-version\"<\/a><\/p>\n

Final\u00a0 Version: https:\/\/public.tableau.com\/profile\/tony.volpe#!\/vizhome\/MappingSuperfundSitesinNYStateFinal\/NYSSuperfundSites<\/a><\/p>\n

In the dashboard<\/a> above, the top rectangular section includes a definition of a Superfund site, followed by a map of the<\/p>\n

state, with Department of Environmental Conservation (DEC) regions, and a list of the counties in the corresponding DEC<\/p>\n

Regions. As noted in the upper-right box, only counties with sites are included in the DEC regions list. Working together, the<\/p>\n

items in the upper third provide an introduction and orientation to the material that follows.<\/p>\n

The second section shows a concentration of sites by county, using a choropleth map. Using that type of map provides a<\/p>\n

ready visual aid that requires minimal text. Additionally, the choropleth mirrors (to some degree) the DEC regions map located<\/p>\n

above, potentially reinforcing understanding of the county geography and where sites are located. The map had an unexpected<\/p>\n

impact for one user, who mentioned that it was rather concerning, as her sister lives on Long Island, the region with highest<\/p>\n

number of Superfund sites in the state.<\/p>\n

The third section (Sites by region) uses a bar chart to present a breakdown of sites by DEC region, allowing for quick<\/p>\n

analysis. \u00a0Last, the fourth section (Superfund sites per county) also uses a bar chart, with the bars running horizontally to avoid<\/p>\n

duplicating the same layout as the chart next to it. Even though the bars are using a different orientati0n, the fourth chart<\/p>\n

presents data in a way that is readily comprehensible for someone viewing it.<\/p>\n

As to the creation process, I had started out with a different idea than what I wound up actually building and presenting.<\/p>\n

The original idea was to make a map of Superfund sites across the country, but I decided to narrow the focus to New York State,<\/p>\n

using Carto for the map portion, and creating bar charts (or other visualizations) in Tableau. I started off creating a map of New<\/p>\n

York in Carto, to get an outline of all the counties, and then imported EPA Superfund data into Carto, with decidedly mixed<\/p>\n

results. I had wanted the sites to appear on a map as points, with the site name and city displayed when one hovered over it.<\/p>\n

While some sites appeared in the appropriate counties, others appeared in entirely different states altogether. The<\/p>\n

data from the EPA lacked certain fields that would enable Carto to properly map the sites.<\/p>\n

After working with Carto some more, I modified my idea, and switched to Tableau. While working with Tableau 9.3, I ran<\/p>\n

into some issues with certain cities not appearing on the map (municipalities under 4,000 were not displaying), so I opted to<\/p>\n

revise again, creating a map showing all counties with Superfund sites. The choropleth map was the best option to show the<\/p>\n

concentration of sites across the state.<\/p>\n

I analyzed the EPA data in different ways (by county, by DEC region) in Excel to provide more context and to present the<\/p>\n

material in a way that worked in concert with the map at the upper third of the dashboard. Some users did note that counties<\/p>\n

with no sites were greyed out on the map, but that was not necessarily a bad thing. The counties with sites were, in a way, more<\/p>\n

visible that way, as they had shading to indicate presence of a site. Once I had the various sheets together in Tableau, I created<\/p>\n

an initial version of the dashboard (Version 0<\/a>), and another (Version 1<\/a>), with the elements arranged in a different manner.<\/p>\n

Images for each are below.<\/p>\n

Figure 1 \u2013 Version \u20180\u2019 of the dashboard<\/p>\n

\"version-0\"<\/a><\/p>\n

Figure 2 \u2013 Version \u20181\u2019 of the dashboard<\/p>\n

 <\/p>\n

To conduct user testing, I emailed five people, with four responding and providing feedback. The user population, while<\/p>\n

small, consisted of a former manager of mine (who also worked for DOE-EM), a classmate (filling the novice user role), afriend<\/p>\n

with a keen interest in environmental causes and data, and a relative (also another novice user). Requests for feedback went out<\/p>\n

on Friday, the 2nd<\/sup>, with a request to have feedback in by the following Monday so changes could be incorporated into a new<\/p>\n

version. I attached a questionnaire, focusing on three areas: subject matter, aesthetics\/layout, and comprehension, and asked<\/p>\n

that participants provide feedback on those areas. The subject matter asked users to explain the subject of the dashboard in<\/p>\n

their own words (i.e., what is this about, and can you explain to someone if needed?), aesthetics\/layout focused on determining<\/p>\n

which version was more effective and why, and comprehension focused on users\u2019 ability to answer six simple questions, based<\/p>\n

on data presented in the dashboard. Questions and feedback are in the appendix of this report.<\/p>\n

The design of the dashboard evolved, but retained common elements among the three versions (two \u2018test\u2019 and the final),<\/p>\n

including the use of choropleth map, stepped color, and bar charts. Initially, the map had a green color scheme to<\/p>\n

show concentration of sites by county. The stepped color scale ultimately used 5 gradations, after trying out scales with more<\/p>\n

steps. Additional steps may have been effective if the scale were larger, but since it was smaller (0 to 18), a simpler gradation<\/p>\n

scheme worked better for the map. The choropleth map color was changed to match that of the bar charts, to provide a uniform<\/p>\n

look.<\/p>\n

Ultimately, the final version retained more of the initial design (Version 0) in terms of placement of information on the<\/p>\n

dashboard, with some additions to clarify what a Superfund site is, as well as some additional information to clarify what<\/p>\n

exactly is being shown. Previous versions did not indicate that all sites were shown, regardless of status. The final version does<\/p>\n

make a note of that, in the upper third of the dashboard. Looking back, that note would have been beneficial to include, though<\/p>\n

it would not have impacted the layout or comprehension of the subject matter.<\/p>\n

User testing (results were generally positive, with recommendations for improvement. More details can be found<\/p>\n

below.<\/p>\n

In Area I, Subject Matter, all four users were able to state in their own words the purpose of the dashboard, as well as<\/p>\n

explain what was being presented. Asked which version was more effective, one user expressed a clear preference for Version 0 <\/a><\/p>\n

over Version 1<\/a>, saying of the latter:<\/p>\n

\u201cWhile material was clearly labeled, the format and layout created confusion, and the side sweep of the regional interactive map took too much of the page, drowned out the other info in relation to its corresponding info, and also I kept inadvertently dragging the mouse over it, with nauseating zoom effects.\u201d<\/em><\/p>\n

Two users did not specifically choose a version in the subject matter area, but said that Version 0 presented the material in a<\/p>\n

clear and easy to understand manner. The remaining user did not directly choose one over the other, but instead offered<\/p>\n

comments on potential improvements for Version 1.<\/p>\n

Area II focused on Aesthetics\/Layout, with questions focusing on which version (if either) was preferred, and why, as<\/p>\n

well as suggestions for improving the preferred version. Findings align with results described above, with Version 0 emerging<\/p>\n

as the stronger candidate (with revisions) for use in the final product. Recommendations for improvement of Version 0 focused<\/p>\n

mainly around placement of text boxes, and other enhancements to make the visualization clearer. Of version I, one user<\/p>\n

indicated the stepped color scheme was less effective because of the lack of differentiation on the map (users had to select a<\/p>\n

county to see the color and corresponding number of sites, unlike Version 0). \u00a0A note about counties without sites was not<\/p>\n

initially included, but after user testing, I added it to make it clear why certain areas were not showing up on the choropleth<\/p>\n

map.<\/p>\n

The last area (Area III) looked at comprehension and application of material presented on the dashboard. All users<\/p>\n

were able to correctly answer six questions (such as \u201cWhich county has the most Superfund sites?\u201d). The questions were simple<\/p>\n

to keep in tune with the relative simplicity of the dashboard. One user noted a concern about not being able to identify counties<\/p>\n

that were greyed out on the map, but that did not impact overall comprehension of the material.<\/p>\n

Visualization-related findings revealed some unexpected points, including:<\/p>\n