Please reload

Recent Posts

I'm busy working on my blog posts. Watch this space!

Please reload

Featured Posts

Contextualizing Los Angeles’ Park and Open Space Facilities: An Exploration of Park Space and Car-related Crimes

December 17, 2016

 

Research and Conceptual Framework

 

In dense cities where land value is high, competition for acquiring the land to develop into direct profit generator is fierce. But there has been long tradition of viewing park as common goods. I have often heard landscape architects and environmental planners using ‘promoting public safety and decreasing crime rates’ as a generic term to argue for having more public park land in an urban environment because green spaces are more likely to be used by pedestrians and cyclists and create greater ‘natural surveillance’. For instance, Nottingham Trent University’s Trees For Cities interviewed city residents and they reported a stronger sense of safety because criminals are likely to be deterred by the presence of other community members. Some have proved that urban green spaces do not lead to increase in crime rates, if not decrease. In the study of Boston’s South-West Corridor (Linear Parks and Urban Neighborhoods: A Study of the Crime Impact of the Boston South-west Corridor. Journal of Urban Design, Vol. 6, No. 3, 245±264, 2001, KATHERINE CREWE), there was no significant increase in crime fifteen years after the completion of the park and counter to residents’ early anxieties.

 

While at the same time, others have been arguing that open space can easily turn into empty dead zones when there is lack of street activity and pedestrian traffic that can provide surveillance, and these parks become hotbeds for spontaneous crime. In a case study, researchers (http://www.fs.fed.us/nrs/pubs/jrnl/2015/nrs_2015_kondo_004.pdf) performed effects of ‘Lots of Green’ program by Youngstown Neighborhood Development Corporation on crime in and around newly treated lots, in comparison to crime in and around randomly selected untreated lots. They have found a significant increase in motor vehicle thefts at treatment lots compared to at control lots. One possible explanation for this is that residents felt more comfortable parking cars near lots after the lots were greened, thus creating a larger supply of targets.

 

These previous research has been heavily qualitative. There has not been a study that uses data to prove whether this correlation exists, or how robust it is if it does exist. So I started with the question: does having more park space within walking distance significantly decrease automobile crime? What are the spatial distribution of park space and automobile crime in Los Angeles? Can we answer the question with statistically analysis and spatial mapping? What are the planning implications in research for the answer to this question?

 

To start, I gathered Los Angeles’ park boundary published by Department of Parks and Recreation and crime data by LAPD, and used Python geopandas-altair library to plot different types of crime in 2015. Car-related crime really stand out: they are the most common types among all crimes. On a national level, theft from parked cars is one of the most common complaints received by police in residential neighborhoods. According to U.S. Department of Justice statistics, these types of crimes make up some 36 percent of all larcenies reported to the police.  Therefore if green space, along with other factors, are proven to be statistically significant in terms of decreasing automobile crimes, then landscape architects, planners and stakeholders can have a more robust argument when deciding to allocate more land to be park use in the planning process.

 

Before delving into analysis, I have realized that other factors including land use types, population density, income and location of businesses affect automobile crimes as well. These data are downloaded from Los Angeles’ open data portal, LA geohub and census TIGER website. I also retrieved pedestrian count data from other researchers, as well as walkability score, land use mix score, residential density, intersection density and retail density organized by census tract. The criteria of these scores developed to rate the level of mixed-use, street intersection density, retail and residential density by census tract will be explained in the section.

 

Data Cleaning and Statistical Analysis

 

The park data on LA’s open data portal contains lat and long of park. The crime dataset contains location, type and time of each crime. Due to the size of the raw crime dataset, I only chose year 2015. Below is the summary statistics of park square footage within walking distance (0.5 mile) and crime data by census tract in Los Angeles.

 

 

The summary statistics display enormous variation in the park land per census tract and crime counts per census tract: the largest park is of at a scale of about ⅙ of Washington D.C. and the smallest park is the about the size of the house, while the crime counts also vary drastically on the census tract level.

Visualizing Statistical Data

Automobile crime is heavily impacted by built environments and street activities. Research has indicated that most car thefts (37 percent) occurs on the street outside the victim’s home. Therefore, my assumption is that single-family residential land use type is more likely to increase chance of auto thefts, because suburban residential areas are relatively quiet and have less pedestrian traffic. (http://www.popcenter.org/problems/residential_car_theft/print/)

Table 1. Risk of Car Theft by Parking Location in England and Wales (1982-1994)

 

 

Another factor is time. The plot below was made with geopandas again show the automobile-related crime often occur in the early evenings. This is possibly due to that it is the time most cars are present in residential areas, as well as the fact that darkness provides cover for the thieves.

 

 

 

To understand the spatial distribution of car crimes, I first tried to generate heat map with the locations of the car crimes. I tried both geopandas and CardoDB. Points and polygons can give us clear spatial delineation of where each incidents are located, and we can see that most car crimes locate in downtown LA. I enjoy the clear gradient of the heat map by  geopandas, but its lack of interactivity seems to be a drawback. The CartoDB heatmap can make up for this drawback. The census tract layer underneath shows the amount of walkable green space per capita – the darker the more walkable green space. 

 

One takeaway from this set of heatmaps is the fairly uneven distribution of walkable green space in Los Angeles. Where the city designates as park or open space depends on many factors, including land availability, ecological condition, topography and socioeconomic concerns among many other factors. Another takeaway just by looking at the heat map distribution, one can see a lot of overlapping between tracts with lower green space per capita (lighter shade) and higher chance of stolen car incidents.

 

 

 

Then I tried to map just using point data with geopandas and CartoDB again to show car crime locations (points) and green space (polygons). The darker magenta indicates higher car crime rate by census tract. It seems like car crimes are likely to happen a small distance from the green spaces. Plus, the huge variation in the size of the park can be misleading because one tends to only see the very large parks on a city level.

 

To test statistical correlation between open space and car crimes that cannot be represented by spatial mapping, I then moved to regression analysis. The first plot is a histogram showing distribution of numbers of car crimes. Most census tracts have between 0 to 500 car related crimes in 2015; few have more than 1500.

 

 

 

More car crimes are about 2000 ft from a nearest park (0.37 mile) rather than within immediate distance, which confirms the observation in the previous spatial mapping. This to some extent explains how car crimes tend to happen not directly next to a park but more tend to happen a distance from it - this disputes the argument that parks are often inductive to car crimes as car crimes according to the histogram below seem to happen a few blocks from the nearest park.

 

The OLS model and plot below indicate that there does not seem to be a strong correlation between green space and car crime rate. Although it does not necessarily confirm many people’s fear of open spaces in urban areas, it still questions many landscape architects’ argument that green space can necessarily improve the public safety. In fact, many parks are truly less used nor are they well lit in the evening hours, resulting in that many of these parks to become more dangerous.

 

 

 

 

 

 

 

 

 

(x: square footage of green space in the 0.5 mile radius ; y: number of car-related crime)

 

 

I speculate that there are many factors that influence the car crime rate around a park. The level of mixed-use land use, for example, is one. I used the Land Use Mix Score published in 2012 as part of the Vision Zero Open Data Plan to test the statistical significance on OLS model. Each census tract is rated by the degree to which a diversity of land use types were present in a block group. The mix measures five land use types: residential, retail, entertainment, office and institutions. Values were normalized between 0 and 1, with 0 being single use and 1 indicating a completely even distribution of floor area across the 5 uses. (http://sallis.ucsd.edu/Documents/Pubs_documents/NQLS_Frank%20et%20al%20published%20walkability%20paper.pdf)

 

 

 

 

 

 

Intersection Density is also statistically significant to impact correlation between green space and auto crimes. Areas with higher density of intersections have more connected street networks, and the ratio between the number of true intersections (3 or more legs) to the land area in areas. A higher density of intersections corresponds with a more direct path between destinations. Auto crimes are less likely to happen in places where streets are well-connected that cul-de-secs because dense street intersections are likely to trigger street activities.

 

 

Surprisingly, residential density, retail density and walkability scores do not affect car crimes as much, at least statistically. 

 

Focused Case Study: 77th Street

 

I tried to map car crime locations at 77th Street neighborhood. The map below is created with leaflet. Orange dots indicate cases still under investigation and green dots indicate cases resolved, and the popup windows show the locations and time of the crime.

 

 

 

 

 

The maps below are created with CartoDB and customized with SQL. Again, one pattern that is more obvious than a city scale map of the location of green space and car crime is that the locations of the car crime are usually not directly near the park but a few blocks away. In the map below, the darker blue shows higher land use diversity score. Surprisingly, there is not a clear distinction of the spatial distribution of car crimes between the lighter shade (predominantly single residential use) and the darker shade (more diversity in land use) for 77th Street neighborhood, which is different from that on a city scale.

 

 

 

 

 

 

77th Car Crime, land use and businesses:

 

After checking the business establishment information (in red dots) and overlaying the land use information, I have realized that many businesses are usually childcare, auto shop and neighborhood scale services and restaurants and they are not intensive enough to attract large pedestrian traffic.

 

 

 

 

 

77th Street Car Crimes, Parks and Traffic Volume

 

 

 

 

77th Street Car Crimes, Population Density and Green Space: it does seem that car crimes tend to happen in census tracts of lower population density

 

 

 

 

 

 

Main takeaways:

  1. Focus on improving safety and surveillance at night through design, programming and management to make sure activities happen into the night through programming, instead of simply proposing more park space that is poorly managed and maintained. In fact, large and poorly-lit park at night could be hugely problematic. Design solutions include increase lighting to avoid dark blind spots at night and avoid large untrimmed vegetated areas that provide natural ‘cover’ for crime.

  2. Increase the diversity in land use. Simply adding density to a single type of land use does not solve the problem. Higher-density in residential uses or retail still constrain the types of users, visitors and inhabitants to these areas, while a higher level of diversity can activate the area during the daylight with office and institutional uses and increase uses in the nightlight time with entertainment. Unlike large regional parks that can be locked down by sunset, open spaces in the urban areas often do not have gates and remain accessible to the public at night. Increase land use diversity around urban parks as much as possible, because car crimes happen on the streets at the edge of the park and urban fabrics.

  3. Make street grids instead of cul-de-secs, and locate parks within the grids.

     

     

     

     

     

     

     

     

     

 

 

Please reload

Follow Us

I'm busy working on my blog posts. Watch this space!

Please reload

Search By Tags
Please reload

Archive
  • Facebook Basic Square
  • Twitter Basic Square
  • Google+ Basic Square

Designed by Chelsea Zhou. 2020 All rights reserved.