One of my primary interests as a geographic information systems (GIS) educator is exploring new ways of teaching GIS, and spatial analysis in particular. Lessons about network analysis and habitat modeling are some of the most difficult to explain because of their technical nature – not to mention the complicated process of arranging and changing data to accomplish various analyses.
Therefore, I recently combined my interest in GIS with a growing love of all things zombie. This resulted in a talk at Monmouth University’s GIS Day 2011 titled “Popular Culture and GIS: Using Geospatial Technologies to Model and Prepare for the Zombie Apocalypse.” I also use this talk in my First Year Seminar “Zombies!” and for various classroom presentations. The central goal of this post is the use of a pop culture phenomenon to explore various spatial analysis techniques. In other words, to make learning fun!
A range of groups are utilizing zombies, and more specifically a potential zombie apocalypse, as a teaching tool. This includes the CDC’s use of zombies to teach about emergency preparedness as well as the Missouri Department of Conservation’s use of zombies to talk about invasive species.
So, how can modelling a zombie apocalypse help teach about spatial analysis? In order to answer this question, I will explore a case study for Monmouth County, NJ. The following parameters will form the basis of this exploration:
1) What are the basic characteristics of a zombie horde?
2) Which locations are the most susceptible to zombie outbreaks/contamination?
3) Which form of spatial analysis is most appropriate for modeling a zombie apocalypse?
4) What might the post-apocalypse spatial patterning of zombies look like?
Determining the Basics of a Zombie Apocalypse
There are two central characteristics about zombies which must be addressed prior to undertaking a spatial analysis of their movement. In many ways, these questions are similar to questions asked of a wide range of geospatial phenomena. These characteristics include determining possible outbreak locations and the average speed of a zombie horde. There is ample evidence to consult as we address these fundamental questions.
For instance, in 2007 the City of Austin, TX prepared a risk assessment map of the zombie apocalypse (see above). This pioneering group of spatial analysts calculated the average spread of a zombie horde at approximately 1.5 miles per hour. Of course, the classic debate of zombie speed comes to mind.
I don’t think this debate matters very much in regards to the spread of a zombie horde. A 1.5mph speed is a good place to start given that even fast zombies shuffle from place to place when live humans are not present. Also, as we will see, even at a 1.5mph zombies spread through an area very rapidly.
The second characteristic to address are potential zombie outbreak locations. Zombies have traditionally risen from their graves due to a host of causes (e.g., radiation, vodou). Today, zombies tend to be the result of some kind of pandemic. I have decided to use this second scenario given its increased popularity in recent years.
The above images show the locations of medical locations in Monmouth County, NJ. The left-hand image shows hospitals, the right-hand image shows all emergency medical facilities. I believe it is reasonable to assume that all health facilities would quickly be mobilized in the early stages of a zombie outbreak.
Therefore, we have the basic characteristics of a zombie apocalypse. A zombie horde spreads across a landscape at 1.5 miles per hour. These hordes would most likely radiate outwards from hospitals and emergency medical facilities of all kinds.
Selecting an Appropriate Spatial Modeling Technique
The basic parameters outlined above can be analyzed with various types of geospatial analysis. The trick is to select the appropriate technique and this is where pop culture meets GIS education. The following reviews common geospatial analysis techniques and their appropriateness for analyzing various aspects of a zombie apocalypse.
Buffer Analysis of a Zombie Outbreak
This form of analysis creates a series of polygons around input features at specified distances. Buffer analysis places radiating circles corresponding to expanding distances/times around the hospitals and emergency medical facilities. The innermost circle corresponds to one hour (1.5 miles or 7920 feet) of movement by the zombie horde. The middle circle corresponds to two hours (3 miles or 15840 feet) of movement, and the outermost circle corresponds to three hours (4.5 miles or 23760 feet) of movement.
There are a number of drawbacks to using buffer analysis. Planning movement and actions based on buffer analysis is hindered because exact measurements within the various circles are not available. If a location falls within the outermost circle, the user has to visually gauge if the horde will arrive between two hours and two hours and fifty-nine minutes. Another form of analysis is required to create specific measurements.
Euclidean Distance Analysis of a Zombie Outbreak
Euclidean analysis calculates, for each point in space, the actual distance from a source point. This form of analysis provides more accurate data, and users can see the exact euclidean distance between various locations and any given outbreak point.
Euclidean analysis has its drawbacks as well. While returning a more accurate measurement, euclidean distance analysis still treats the landscape as a cartesian plan. This is a measure of distance “as the crow flies.”
Network (Service Area) Analysis of a Zombie Outbreak
A network analysis creates an area (polygon) which encompasses all accessible streets within a given distance or time. Sometimes referred to as a service area analysis, this form of spatial analysis takes into account the placement of streets. Network analysis is often used for logistics planning. This type of analysis is particularly useful for determining which warehouses are best positioned to re-supply specific stores.
This form of analysis is appropriate for analyzing the spread of a zombie horde. Zombies are more likely to wander along streets then cross open fields, bodies of water, or other areas. Network analysis not only takes features of the landscape into account, it returns exact measurements like euclidean distance analysis above.
This type of analysis provides a preferable method for predicting the spread of a zombie horde. This is particularly true if the above conditions of a 1.5mph rate of movement radiating out from medical facilities holds true.
While predicting the movement of zombies at the beginning of the zombie apocalypse is certainly useful (e.g., planning your way home), survivors will be faced with additional obstacles in the days and months following the initial outbreak and subsequent apocalypse. This is where site suitability analysis comes in handy.
A Preliminary Zombie Habitat Model: Site Suitability Analysis
Suitability modeling “typically answers the question, Where is the best location? — whether it involves finding the best location of a new road or pipeline, a new housing development, or a retail store.” (ESRI). Or where are zombies most likely to congregate following an outbreak.
Suitability analysis takes many forms, and is a higher class of analysis than those discussed above. Several variables are taken into account and modeled in relation to one another. I chose to use fuzzy logic to model the post-apocalypse distribution of zombies. Fuzzy logic is useful when the definition of phenomena and things affecting them are imprecise. As such, it is increasingly used to model suitable habitats for plants and animals, which do not follow mapped features the way living humans do.
I chose three variables (or datasets) for this suitability analysis.
Landuse – classifies every spot on the landscape. This type of data tells you what kind of use – if any at all – exists at a given location. I reclassified this data by giving higher weights to areas of human disturbance. In other words, places like shopping centers and industrial areas have a higher likelihoods of zombies being there. Alternatively, open fields and swampy areas have less weight because it is less likely that zombies would congregate in those areas.
Bodies of Water – I classified these areas as deterrents to zombies, causing them to slowly shuffle away given the lack of human population. While I agree with Max Brooks’ view that some zombies would eventually wander into bodies of water, I chose to give these landscape features a low likelihood of zombies being present.
Major Highways – I classified these areas as highly likely to have zombies since every film, book, and comic always shows these areas as the location of secondary outbreaks.
The following graphic shows the results of the analysis using an intuitive color scheme. Red represents areas with the highest probability of zombies following the initial outbreak.
I have found the above example of using spatial analysis to model the zombie apocalypse an entertaining and useful introduction to spatial analysis for my students at Monmouth University. This example even allows us to arrive at traditional conclusions. A network (service area) analysis is the most appropriate for modeling the initial spread of zombies – if they do indeed move primarily along roadways at an average speed of 1.5mph and spread out from health care facilities. Fuzzy logic analysis is the most suitable for modeling the eventual spread of zombies in that they behave more like animals and do not follow man-made features in the same ways living humans would.
Thanks for reading,
Educators, interested in giving a lecture on zombies and spatial analysis? Download the following template and insert similar analysis images in it for use in your class.