Wildlife Strikes


An interactive visualization mapping incidents of wildlife strikes on aircrafts from 1999 to 2012, based on data from the Federal Aviation Administration (FAA). The system allows users to explore the relationships of animal and aircraft collisions over time as well as seasonal changes in animal strikes.

This project was a collaboration with Paul Ellebrecht, Nikki Roda, Jeremy Salfen and Phillip Tularak.

UMSI Exposition'13 Honorable Mention

Interactive Visualization (View)

Graduate Coursework (SI649)

Fall 2012

Examining the Data

Given the complexity of the original dataset, carrying dozens of different fields, we had to narrow the focus of the visualization to build a system that would help users navigate relationships between the primary actors; aircrafts, animals, and time.

Although the dataset starts in 1990, we chose to use only data from 1999 to the present, due to a significant spike in the number of incidents starting in the late 1990s. Our rationale for eliminating earlier data is based on information that we received from the FAA project manager regarding changes, including the increased involvement of biologists, during that time. The increase may have had a significant impact on the quality of the data as well as the quantity.

The histogram of total strikes by year revealed a dramatic increase around 1999.

Constructing Use Cases

As a result of our preliminary analysis of the data, and potential audiences, we settled on two primary use cases around which to focus our design.

For environmentalists, this could be a way to visualize the impact of wildlife strikes on animal populations. The visualization would allow the focus on a single geographic point, stepping through time to examine the numbers and types of wildlife strikes that have occurred there. Once a user finds an animal species of interest, they can then see where else this animal has been involved in strikes.

For the airline industry, there could be a number of uses related to understanding how best to mitigate animal strikes. This visualization could be used by airlines that want to get a better understanding of where their planes are likely to be damaged by wildlife. This could also be used by insurance companies that want to see if they should be charging more for airlines that do significant business in areas where there are large numbers of wildlife strikes.

A considerable amount of brainstorming went to examining how best to represent the data to the user. Focusing first on the most salient retinal variables, and mapping them to the data in question, provided a framework in determining available options.

The main time interface for the visualization. The control panel provides several controls for moving through the time span. Users can "play" the visualization to have it automatically walk through month-by-month, or they can use the scented slider widget to select a specific month or range directly.

Design Rationale

We drew design inspiration from time-focused visualization techniques, considering differences between linear and cyclic time representations and between visualizing time points and time intervals. We explored these various dimensions and decided our system would work best as a linear representation of strike data in time intervals of calendar months. We also considered previous work done on systems that allow users to pivot on various dimensions in the data. In particular, we drew on the work of Marian Dörk et al., in their implementation of PivotPaths.

By allowing users to select an animal of interest, we limit the amount of information on the screen and the amount of information that the user must process. The use of filtering and zooming allows the user to select a given period of time and animate that period or just look at strikes in that period. This also limits the amount of data that users are required to process visually and allows them to focus on an area of interest. By clicking on an airport, the user can see an overlay of detailed information about incidents in that area, but otherwise this information remains hidden. The interactivity thus reduces the cognitive load of exploring this potentially overwhelming dataset.

A contextual popup window, providing detailed information for a specific airport

The visualization's interface for selecting animals is organized into families. Families without strikes in the current date range are grayed out. In addition to being able to select and map individual animals to incidents, a small "eye" pulls in a photo of the animal from Flickr.