An Analysis of Campus Safety Infrastructure: Streetlights and Emergency Callboxes

Preface / Description of Data:

The data manipulated in this project was drawn from numerous sources. The majority of the data was borrowed from a campus-wide geodatabase managed by the Infographics Lab – save for a section of point features representing streetlights that have been ground-truthed with a GPS unit.

The newly ground-truthed data was recorded as a series of waypoints and imported into QGIS as point features, then combined with an existing vector layer to form a fully comprehensive vector layer with features that represent streetlights across campus. I avoided conflict when merging datasets by ensuring the two sets referenced the same datum, NAD 83. If this were not the case, the most recently ground-truthed waypoints would not accurately represent the location of the streetlights with respect to the other datasets (although this could still be the case due to human error). In general, referencing the same datum raises the accuracy of the spatial location of our features, and thus grounds our analysis in legitimate, accurate spatial data.

The data borrowed from the campus geodatabase is presented in its default coordinate system – Lambert Conformal Conic: State Plane South. This coordinate system was likely chosen for the minimal distortion it presents in the Southern Willamette Valley. By contrast, the GPX file we import from the GPS unit is in an unprojected Coordinates Referencing Systems WGS 84.

Approach:

The following spatial analysis is primarily inspired by the question of how to make common walking routes around campus safer for the passerby in light of theft, assault and harassment – crimes that have been known to occur on campus and the surrounding area. I focus my analysis on verifying two solutions: (1) to deter criminal activity by increasing nighttime visibility with streetlights, and (2) to increase victim and witness accessibility to direct-line emergency call boxes. Both of these infrastructural units currently exist on campus, and my aim is to uncover specific locations where more units are needed.

To facilitate a clearer understanding of visibility and emergency call box accessibility on campus, two maps were created drawing emphasis to areas that lack these traits. The map below displays key campus features including campus routing lines – common walking routes around campus. The paths are styled to show where a pedestrian would not be able to locate an emergency callbox by sight.

A buffer of 150ft was calculated around the existing emergency callboxes to account for general visibility – both in the daylight and at night. Actual visibility may warrant a larger buffer radius, but was compromised to account for obstacles and other visual impediments. The buffer was then subtracted from the routing lines in a vector overlay operation to provide a cleaner map product, and to highlight pedestrian paths where these callboxes were not visible or accessible.

The map above helps us decipher areas that may be in need of increased light infrastructure. Similar to the analysis shown in the first map, 30-foot buffers have been calculated around the merged light feature class to form map of estimated illumination.

In light of the previous two calculations, the final map product has been created highlighting key areas of concern around campus. This has been completed by combining the two buffer feature classes via union overlay to create a polygon feature class representing both the callbox and street light buffers.

This feature class was then used in another overlay (this time, employing a difference calculation) to obtain sections of path that are not in the appropriate distance of either a light or callbox. The resulting line features were named ‘hazardous paths’ to account for their distance from safety infrastructure. The map above utilizes this new feature class to highlight the primary areas that lack accessible emergency callboxes, and areas where nighttime visibility is low.(Note: the union buffer class is not displayed in this map for sake of simplicity)

The final map displays recent campus crime data from the University of Oregon Cleary Report. The hope of this map is to highlight areas subjected to these incidents.

However, relating this information to the spatial argument proved to be a challenge.

Findings:

From the final map product (the third map), we are able to locate areas around campus furthest from the safety infrastructure. Namely, the map shows that walking routes bordering south campus, as well as the section of campus bounded by 13th, 15th, Kincaid and University as being large key areas where this safety infrastructure is inaccessible to pedestrians. The analysis leads us to believe that increasing the amount of emergency callboxes and streetlights in these areas will help provide a strong sense of security that is uniform across campus.

By far the most glaring issue with this analysis is the limitation of the accuracy of crime locations. The Cleary Reports provides us with useful documentation, but the spatial referencing information given to pair with these crimes is mediocre at best; crimes are spatially referenced by a ‘general location’ field which generally consists of a building name. This puts our argument at a disadvantage, because the most commonly referenced crime in the space is bike theft (which generally occurs outdoors). Thus, our map falsely implies crimes are happening indoors, which provides our analysis with little edge when trying to assess the need for outdoor safety infrastructure.

To clarify our argument, a larger dataset of on-campus crimes is desired along with a more robust method of spatially referencing said events. If this were available, spatial computations (such as a density calculation) could be employed to find areas of high crime. This could be compared with the current infrastructural arrangement and could be used to scrutinize it. Such an analysis is not carried over in this project due to the lack of features, and a lacking spatial referencing system.