Security Analysis in UBC, Vancouver
To establish my research in context, I first wished to institute the campus' amount of present security amenities. Using data from the Department of Geography I first layered a map of the emergency blue phones, the streetlights and the bus stops as presented in figure 2. (Which you can download and zoom into by clicking it.) Through this overall analysis I hope to establish an uneven distribution of those particular aspects. Bearing in mind that surveyed answers provided insights about the felling of insecurity increasing after 1a.m., bus stops are therefore somewhat irrelevant as close to none buses still run after this time. I will therefore mention this in my discussion section, in terms of economical implements from the city of Vancouver rather than the local authorities of campus.
In this logical, the uneven pattern is quite visible as parts of campus are much more clustered than others. For example, we clearly see that the two first year residences shown in Figure 3 are places of much higher dispersion of all three factors of security emitted through this map.
I pursed this context envisioning through the results of surveying based on ranking of buildings/ areas/ roads that infer insecurity versus security. Creating a second map that I entitled “Physiologically unsecure places at UBC,”(Figure 3) by layering different data after having separated specific buildings from the buildings data of the Department of Geography. Surveyed showed that First year residences Totem Park and Vanier Residences, as well as Geography Building, Forestry building and Music Building, all very remotely located the center of campus, are very much prone to induced insecurity, so are construction zones and parking lots. Figure 3 therefore shows the distribution of those three categories (Parking Lots/Construction Zones/psychological fearful buildings).
The first part to conducting the analysis of the security on the campus of UBC, with 2 principal goals in mind: creating safer travel agendas amongst common locations on campus for students and introducing areas that could benefit such developments; is the obtention of data.
Most of the data obtained comes from the Department of Geography at UBC, through collection from years 2009 and 1012.
The data imported includes
- Benches (2009)
- Trees (2009)
- Streelights (2009)
- Bluelights (Emergency Poles) (2009)
- Soft Landscape (Lawn/Planting Bed/Wild) (2009)
- Translink Stops (2012)
- Roads (2012)
- Buildings : Future, in construction, and actual (2012)
- Parking lots (2012)
- Campus Legal Boundaries 2012
- Landuse (Neighboorhood/Education/ Commerce) 2012
- Basemap (World_Street_Map) 2013
All these layers follow the Universal Transverse Mercator NAD 1983 UTM Zone 10N projection due to being the international standard of coordinate system and the most common Projected based coordinate system, while pertaining as little distortions as possible on the area studied.
Furthermore, data was collected from a small survey of UBC students demonstrating psychological resentment of certain areas of campus, assumed under this study as a lack of security implementations while helping to weigh determinants influencing these fears.
Finally, I have obtained information about locations of recent assaults on campus, through local articles to examine reasons underlying and pertaining a faulted uneven development of security facilities. In this regard I wish to ensure that no where in this study do I wish to put the victims at faults, but rather understand the patterns of spatial distribution to attribute a possible amelioration and eventual complete annihilation of assaults. Valuing ethics of such a sensitive subject stayed in priority during this whole assessment of security.
Data
Part I: Evaluating Campus's security facilities.
Part III: Kernel Density Evaluation of UBC's streelights
Methodology: Data & Part I through VI
Part II: Viewshed Analysis of Emergency Poles
I further developed the idea of viewshed analysis while incorporating it to previously mentionned patterns in PART I, of sensible areas, pcychologically unsure, or sensible areas.
In this next output map i simply analyzed roughly where the past 6 sexual assaults around campus have occured while layering distinct buildings areas and roads thought by UBC students to have negative security attributes.
The workflow of this further viewshed analysis is decribed in figure 7. In order to create sensitive "areas" and not simply the building itself, but the close surroundings as well, i decided to implement buffers on all categories in order to have a larger visualizing as well as the true encompassing of a building and its nearby involvement. Larger buffers were given to the 2 first year residences (50 meters) rather than 10 meters for Geography and Forestry Building, while the roads were only given a buffer of 3 meters. Those specified areas, buildings, and roads were identified through the survey as shown on figure 9. Past assaults were recolted through related articles, denoted idividually from the roads layer by their objective ID through selection by attribute.
Figure 2: Representing a taste of what the analysis is about to demonstrate, the uneveness of distributions of secure areas, leading to unwished similar patterns of consequences
Figure 3: Data comes from survey of UBC students as they interpret patterns of uneasiness created throughout campus's activities.
Figure 4: Results of surveying: Emergency Poles Use
Figure 6: Workflow of Viewsheld Analysis Final Output
Figure 5: Results of surveying : Emergency Polls availability
The second part of this overall analysis of uneven patterns of security’s amenities involves a deeper look at the emergency blue poles. After having surveyed students, I realized that the majority (STAT) did acknowledge the insufficient aspect of blue poles, although (STAT) of the interviewee had never even used them in the past, as observed from figures 4 and 5. These results led me to reassess the value of the implemented emergency blue phones through a Viewshed Analysis.
As described by ArcGIS help, "viewshed identifies the cells in an input raster that can be seen from one or more observation points or lines." In this case we are inputting cells from the Campus Legal Boundary that can being seen from the Roads, as the polyline.
PROCESS: The workflow of the analysis is presented in figure 6. In order to interpret the visibility of blue phones I took into account trees and buildings (present ones, in construction, as well as future ones – these were taken into account for 2 reasons, the data of 2012 might not be accurate today and even if it were, this can create great knowledge for future references) taking into account their heights as barriers to visibility. Giving an average height for all buildings on campus of 30 meters is a big assumption, as some ranges much higher and others much lower. For the purpose of this study I took an average number that would level off the outliers otherwise inevitable on both end of the height spectrum. The same process was done for trees with an averaged value of 4meters. Both new added fields were done in the attribute tables of both layers, before merging the two together. Campus Legal Boundary layer was given a height of 0 meters for the purpose of having a similar field to be linked to the newly created Buildings and trees merged layer. The latter was then erased from the former in order to merge it back together this time with the height value field added in the whole of the Campus Legal Boundary layer. Before performing the viewshed analysis the layer has to be transformed into a raster model, as the criteria is required for successful analysis
.
Figure 7: Output of Viewshed Analysis of Emergency Poles layered with Sensitive Areas and Past Assaults
In this third part, I will put forward the assumptions given by the survey and myself, that darker spaces provides not only more vulnerability to students both in terms of safety and fear.
Kernel Density Evaluation takes into accounts distances parameters as it evaluates the density of features in a neighborhood around those features. In this case as it calculates the magnitude per unit area of streetlights across the campus it displays a smooth tapered surface to each of those points.
PROCESS: Streetlights are being evaluated with search radius of 10 meters, as it is assumed that streetlights can only brighten outwards to a certain extent.
Due to its attributes in the allocation of density in certain places I decided to, following the same processes as with the viewshed analysis, add a layering of past assaults and sensitive area on top of the Kernel Density Evaluation output.
Figure 8: Output from surveying. (The comment section of this section lead to major comments about the darkness in general anywhere affecting their "unsafe feeling", which leads into Part III)
Part IV: Weighed Factors of UBC's Security for Least-Cost Paths creations
Figure 9: Workflows of the creation of a map of weighed factors of security
The process by which we create a least cost path revolves around the idea to create the most suitable route to go from source to destination features. In this case, studying the most common itinerary students go through on a daily basis I revolved the sources and destinations interactions between sources of endangered security as well as “safe zones,” that were discussed in the results of the survey. Those involve the library of Irving; safe spot quoted my many due to its closeness to the center of the campus, and mainly the SUB. The two first year residences of Totem Park and Place Vanier, which as will be discussed in results, are main places of security imbreachment. The two educational buildings of Geography and Forestry, which also were repeatedly stated by many as having, lower safety after a certain time (Usually associated with light deficiency). And finally I added a sixth location, the bus loop, as it is the mean of transportation to and away from the campus, and relating to 4 others.
PROCESSES of LEAST COST PATHS: In order to achieve routes between all those places, a long and complex, yet repetitive procedure entails. I will just summarize one of them as the concept is the same for all locations. I first began by converting the Land Factors vector layer into a raster through a polygon to raster method. Using the cost distance tool, I created a costLU and costBL for each five of the stated locations. These two layers were then used in the tool of cost path which was repeated 5 times for each locations (eg: Totem to Irving, Totem to Vanier, Totem to the bus loop, Totem to Forestry or Totem to geography) The raster path had to be converted to a vector polyline due to the mixed pixels problem, where raster overgeneralized the representation of ground feature thus leading to a non real path that often is not visible on the map, or doesn’t even reach fully the destination. By merging all five paths to a single map for each source, I created what I hope to be the best path to undertake for students considering all factors weighed.
The 4th part of this analyses focuses less on individual actors of security stress but rather of a communal engagement of a wide variety that have shown to induce or deter security mindsets. Weighing factors are stepping-stones to creating least cost paths. In the survey, people were asked to rank factors according to “how safe” it made them feel. Results showed wide agreement on light’s valuable aspect, as well as bars and benches relating to people or noise. Benches are usually a way for, as Jane Jacobs mentions in her great novel The Life and Death of American Cities, permitting an “eye on the street” at all time from the community themselves, resulting in increasing security. On the other end of the spectrum for factors’ security, parking lots, construction zones and trees/bushes ranked last due to their lack of the higher-ranking “secure” factors.
In order to establish a pattern in the different factors and understand how important some factors are over others is crucial to the metabolism of theories of unevenness in the data for security amenities, and most importantly, in order to learn how to adjust and find potential sites that could be ameliorated. A friction cost is thus given to each rank according to how “hard” it becomes to get to highest friction costs without feeling unsafe. Indeed the higher the cost, the “more unsafe one would feel walking by the, say “category.”
The friction costs associated with each of the above listed factors involved in the creation of a least cost path have a range of 109. The reason for this is that a smaller range would result in a more Euclidean path, which is not the topic of this research, as they focus on going straight from point A to B. Furthermore, the range in frictions is almost linear, which was not what was aimed for at first, however, in order to produce an outcome, all the land must be given a cost. Thus, such factors as “other buildings” encompass a wide range of differences that might not be accounted for here, as they were given a quite “average” friction cost.
PROCESSES OF WEIGH FACTORS: The map of weighed factors was created as shown through the workflow in figure 9 by first by getting the necessary data. Some merging (between future buildings and construction as well as open land), selection by attribute (for neighborhoods, bars, libraries) and erasing (for other buildings) processes were created in those instances. While all factors got added a new field for their respective “cost,” only the point features got buffered (in relation their actually impact, eg: streetlights have wider literal ranges of complementary safety outlets than emergency phones do), due to the necessary of them equaling an area rather than a point for the success of their merging. All the factors merging create LandFactors, shown by the output map in figure 10.
Figure 10: The output map of the selected 12 weighed factors of security on campus. Blue and green area show high security while yellow to orange and especially red depicts low security
Part IV: Potential Sites to be Improved
Finally, the last map I tried to create was a proposition for amelioration of security allocation to specific patches of the campus’s ground. I decided to combine back the two main areas studied in earlier parts of this study : Streetlights, through the Kernel Diversity Evaluation and emergency phones, through viewshed analysis. The workflow of figure 12 displays the process.
PROCESS: Because Kernel Diversity Evaluation was a floating pixel and using raster calculator to convert it to an integer did not succeed, I opted for another simpler option. By buffering the streetlights to 10meters, which would be the maximum their range of light would provide and merging those to buildings, which I then erased from the campus legal boundaries I created unlit areas. Layering the result to viewshed analysis I was able to examine land that need special attention.
Figure 12: Workflow for output map of Potential Sites for Amelioration