Introduction: Defining Project Goals

The national parks stand as crucial habitats for a wide variety of animal species. This project’s goal is to visualize the relationship one of these species – birds, have with the 63 national parks that the country has set aside to provide the public with access to natural resources. The visualization was first attempted as a map visualization created through Tableau that visualized the relationship between the number of bird species at each of the parks and manifested in a poster that colored the parks a darker shade of red if more diversity was present.

When revisiting this project, I wanted to look at the project more holistically, adding additional variables that help to further describe the relationship and are important to visitors of the parks and bird watchers alike. These additional variables are the different biomes that exist in the parks (shrubs, grasslands, forrest, or tundra), the majority bird type (migratory, resident, vagrant, or breeding), and the average threat level these bird species face at each of the parks (taken as an average level of threat for all bird species that are found in each of the parks). When dealing with these four primary variables and two additional variables (physical space and park names), I decided to make two different visualizations that would serve two different audiences and use cases. The first, another poster visualization that is made to be distributed to those who have a general familiarity with birding and a relatively strong interest in visiting the national parks. The second, an augmented reality (AR) visualization intended to prototype the type of visualization a natural science museum or park museum could feature to allow further exploration of the visualization. Besides audience and use case, the major difference between the two visualizations is the relationship between space as the AR visualization allows interaction with the 3D plane and therefore can introduce height as a further way to visualize the variables. 

When thinking of how to visualize four unique variable types, I turned to the inspiration presented by a visualization of individuals who make money by collecting cans (often called “canners”) around NYC. This visualization invites the user to spend more time understanding the different variables and communicates the data by creating unique icons for each canner with demographic variables making up different parts of the icon template.

Taking Flight: Creating the Visualization(s)

Using this inspiration, I thought about how I could use this same visualization method and apply it to the subject matter at hand. One thing that was really important to me was staying true to the visual design style already established by the national park service and turned to Google to find symbols the parks already use to communicate information to guests. The symbol that stuck with me was the trail head – as it already has the association of helping orient hikers. Once this symbol was chosen, I started thinking about how the data could be translated using this visual key. The data itself, acquired from the open-source platform Kaggle.com, was vast, with a wide range of information of each bird species, the park they were associated with, the number of sightings they had at these parks, and the threat they faced. I used Illustrator to start designing the trailhead that I would use to translate information to the viewer. 

Since these trailheads would serve as my visual key, I wanted them to be simple enough to read clearly while still delivering all the information needed to understand the relationship between the variables at play. I started by using the visual cues users already know intrinsically. What stood out immediately was using stoplight colors to represent threat level – with red representing a high threat and green representing a low threat. Other variables were more complicated and therefore needed to be considered with more nuance. Looking at the data, I knew that I would have to translate the different bird types (migratory, breeding, vagrant, and resident) through iconography as a general audience wouldn’t know how to associate a color with any one of these types. For this I used Illustrator where I developed an icon for each type of bird. Migratory, breeding, and resident made the most immediate sense as I decided to use a bird flying to represent migratory birds, a bird’s nest to represent breeding birds, and a birdhouse to represent resident birds. In order to visualize vagrant birds, I needed to ensure that I fully understood the definition of what made a bird vagrant before thinking of which icon would represent it. Once I knew that vagrant birds are those that show up outside their normal range, it made sense to create an icon that represents the concept of being lost (a signpost pointing in opposite directions). 

Visualizing geographic biomes was another aspect of the visualization that required a bit more reflection. Again, there was a variable that immediately made sense to me – coloring the tundra biome type as white since I felt there was a clear association between snow and ice and the color white. The other biome types (shrub, grassland, and forest) on the other hand were more difficult as all could justifiably be represented using the color green. With this challenge, I decided to start by differentiating them the best I could using the colors yellow, green, and brown respectively and made a note to validate these choices with users later. Once these trailheads were created, I kept the overall format of a map visualization and placed them on the map corresponding to the location of the park each trailhead represented. 

With these decisions made, the only other variables I had to fit on the badge were the park name and the number of different bird species that called each park home. For both of these variables, I decided to simply use text to translate these concepts. However, in thinking more about bird species I thought about how 3D space could help communicate how the number of bird species differ from park to park and turned to creating an AR visualization that kept the same concept of badges to translate the other variables to the user. In this visualization, created through Adobe Aero, I placed the trailhead on a post and made the post’s height directly correlated to the number of bird species. All posts were then placed on a map of the United States corresponding to the geographic location of the park they represented. 

As this visualization was rather complex, I next worked on a key to help users decode the information each trailhead communicated. In creating this key I again turned to the inspiration presented by the Canning NYC visualization. In addition to the key, I also included a glossary to help further explain some of the variables I thought users might need further explanation on.

Testing & Iterating

Once these three pieces were complete (the poster visualization, the AR visualization, and the key that described the trailheads used in both), I needed to think through how I would validate these visualizations. When deciding who I wanted to test with, I thought about the types of people who would find these visualizations useful. In the end, I ended up testing with four individuals. This group of four participants was split by their familiarity with interpreting data visualizations – with two unfamiliar and two extremely familiar.

Before testing, a series of questions were developed along with a series of simple tasks I wanted to make sure the participants could complete. Overall, since testing was done after a visualization was created, the main goal of testing was to make sure that the choices I had made would resonate and to tweak how I was delivering information to ensure it was as clear as possible. Some variability also had to be accounted for as these user tests were done virtually and therefore the visualizations were being viewed by the participants in a format different than what was intended. For the poster visualization this meant hovering around a PDF shown by using the screen share function on the video conferencing platform Zoom and for the AR visualization this involved showing them a short video of how the AR visualization looked, but didn’t allow any interaction. The following is the general structure I used in both interviews:

Introduction: 

1. General explanation of visualization content and goal 
2. Explanation of how the visualization was intended to be viewed 

Observation:

1. Observation of how the participants understood the key 
2. Observation of how the participants understood the variables 
3. Observation of how the participants understood the glossary

Pre-Test Questions: 

1. Are there colors that you feel better represent the different biome types?
2. Are there icons that you feel better represent the different bird types? 
3. Is there any additional information you were hoping to find in the glossary? 

Testing: 

1. Can you explain to me what this badge tells you about [sample park]? 
   • Success: The participant is able to correctly identify all visual cues used in the trailhead.
2. Can you tell me what general trend you can identify about the national parks of Alaska?
   • Success: The participant is able to identify that all Alaskan parks have a majority of breeding birds and that these birds have a low threat.
3. Can you tell me the difference between the birds in the national parks in Alaska and Colorado
   • Success: The participant is able to understand that Colorado has more diversity of bird type than Alaska, but the birds face a much higher threat.

Post-Test Questions: 

1. Which of the four variables (threat level, bird type, number of bird species, or biome type) is most important to you? 
2. How could you see yourself using these visualizations? 
3. What would you change in order to make this visualization more useful to you?

Overall, these tests proved that the visualization choices I had made were largely successful, with users able to grasp the general concepts presented in each of the visualization types. All four of the users were also happy with the overall design of the visualizations and once they saw the poster could easily interpret the AR visualization as the design was kept extremely similar with only the way the number of bird species was visualized changing. From these tests, I was able to observe the following and address some of the concerns brought up. 

Finding 1: The Map Legend Should be More Specific

Besides the typo of not writing “# of bird species”, users wanted more explanation of what threat level and bird type meant and what data was being translated. All users pointed out that they wanted it to be clearly stated that “Threat Level” meant “Average Threat Level” and that “Bird Type” meant “Majority Bird Type.” In addition, one of the users wanted the arrows to be pointing more clearly as they were confused between the difference of the arrows pointing to threat level and bird type. Additionally, another user wanted more information about threat level and wished ranges were used for “High” and “Low” threats. 

How it was addressed: The key was changed in order to address the requests of the participants. Additionally, ranges were used in order to more clearly communicate to viewers what the different average threat levels meant in the data. 

Finding 2: Colors Should be More Distinct 

Two of the users found difficulty in understanding the difference between the colors used for biome type and threat level and wished that there was more distinction between the colors used to visualize these variables. Additionally, there was some disagreement amongst participants of which colors should be used to represent the different biomes. Overall, there was an equal split of those that green should represent grassland vs. shrub, but there was unanimous agreement that the color used for the forest biome should be more brown than red. 

How it was addressed: The color of the threat level was brightened in order to help it stand out against the different colors used to represent the biomes. In addition, the colors used to represent the forest biome was made brown as per participant suggestion. In order to change the colors used to represent the other biomes I would want to do further testing to understand which colors a larger audience feels represents the shrub and grassland biomes. 

Finding 3: The AR Visualization Should Communicate as much as Possible 

One surprise from the user testing was how much participants responded to the novelty of the AR visualization. All participants were able to understand the way they would themselves interact with the visualization and could now better identify which parks had more diversity of bird species than others. When looking at the AR visualization three of the participants brainstormed the additional information they wanted it to communicate – such as coloring the post based on the biome type of the park or splitting the bar chart up by the percent of each bird type that called the park home. 

How it was addressed: In the end, time prevented me from making any significant changes to the AR visualization. As a future step, I would want to test the AR visualization with individuals in-person so I could test how they interacted with it before making any of the changes suggested by this round of user testing. 

Apart from these findings, participants were satisfied with the icons used to communicate the different bird types and were happy with the overall style of the map. The visualization developed from these findings can be seen below:

Before finishing the project, it was presented in front of a group of my peers in order to understand what additional changes could be made to improve the visualization. The following insights were identified:

• The trouble identifying the difference between the biome colors and threat colors was reiterated. In order to address this, it was proposed that colors other than stoplight colors shouldn’t be used to identify the different biomes.
• Along this line, it was also suggested that the same brightness scale was used for all biome types, as the poster currently puts priority on the shrub biome as its brightness stands in the foreground.
• One suggestion posed was making the weights of the different states greater if the state has more parks – in order to deemphasize the states that have no national parks within their borders.
• Commenting on the AR visualization, it was asked if any consideration would be given to gamifying the visualization. This follows with other suggestions of interaction as different users want to pull more information from the data (primarily all the different bird species in each park).

With these suggestions in mind, the following small poster samples represent attempts to solve the suggestions peers and participants suggested.