For this project, I used a dataset on photos taken by photographer Charles W. Cushman during the year 1966. This dataset was published by the Indiana University Archives, which houses the Charles W. Cushman Collection. For this project, I sought to understand how a sense of placed inspired Cushman’s photographs. How do these photographs look different depending on the environment where they were shot? To answer these questions, I decided to visualize the photographs on a map.
I first used OpenRefine to clean the data and get rid of columns that had no data, i.e. were completely blank. I also deleted columns that were redundant or provided no new information, such as the “Country” column since all of the photos were taken within the United States. By getting rid of these extraneous columns, I was able to focus on the metadata fields that contained important information and showed variation between the images.
To create the presentation, I imported the CSV file of the dataset into ArcGIS Online. Upon mapping the data points, it became clear to me that some of the photographs, particularly those in national parks, had very broad location data and contained only the county and the state, or even just the state in some cases. This led to very imprecise mapping and data clustering, regardless of whether those photographs were actually taken in the same location. I went back into OpenRefine and geocoded the more rural points based on identifiers in the slide description, and added street addresses when possible to city locations. This created a much more accurate data visualization.
I then assigned the data points icons based on the genre of the photograph, as seen in the legend. I decided to use icons, instead of colors, to distinguish between genres because it made the map more visually self-explanatory; I wanted a viewer to be able to look at the map and understand what each point represented without having to constantly reference the legend. I tried to use contrasting colors for the icons when possible so they wouldn’t appear too similar to one another and would be distinct.
I configured the pop-ups to show just the image, its description, and its location, ignoring the other metadata fields. I made this decision because I wanted the visual image to be the focal point of the data visualization and show the viewer how these images changed based on where they were taken. I chose to present the map in ArcGIS Story Maps so I could guide the viewer through the data and point out the key takeaways.
Finally, I decided to add some additional layers to the map that showed clustering. These “Clustered” layers display where particular genres of photographs were most often taken, as the size of the icon corresponds to the frequency of that genre in that location. As expected, the majority of Cushman’s landscape photographs occur in more ‘natural’ places with lots of mountains, such as Utah and Colorado, while his architecture photos and cityscapes occur in more urban locations, like San Francisco. In this manner, Cushman’s environment changed how he photographed his surroundings and influenced his photography.
