Investigating the Influences of Tree Coverage and Road Density on Property Crime

Chengming Ye, Yifei Chen, Jonathan Li

TL;DR

This study investigates how urban tree coverage and road density influence property crime rates in Vancouver, Canada. Using high-resolution LiDAR data and GIS analysis, the researchers found a significant negative correlation between both tree coverage and road density with property crime, suggesting that urban planning interventions like tree planting and complex road networks could contribute to crime reduction.

What Problem Does This Paper Address?

Urban crime is a persistent challenge, and understanding its spatial patterns is crucial for effective prevention strategies. While various factors influencing crime have been studied, the specific roles of urban vegetation (tree coverage) and road network characteristics remain debated. Some research suggests vegetation can provide cover for criminals, while others argue it promotes natural surveillance and a sense of order. Similarly, road networks can facilitate or hinder criminal activity depending on their complexity and accessibility. This paper aims to clarify these relationships in the context of Vancouver, a city known for both its livability and its crime rates, by examining property crime specifically.

Methods

The researchers employed a multi-faceted statistical approach using Geographic Information Systems (GIS). They began with Ordinary Least-Squares (OLS) regression to establish initial correlations. However, recognizing the inherent spatial autocorrelation in crime data (i.e., crime tends to cluster), they moved to more sophisticated models:

  • Spatial Lag Model: This model accounts for the influence of neighboring areas on the crime rate of a specific location, addressing the spatial dependency.
  • Geographically Weighted Regression (GWR): This advanced technique allows for the examination of spatial non-stationarity, meaning the relationships between variables can vary across different geographic locations within the study area. GWR generates local coefficients, revealing how the influence of tree coverage and road density on crime changes from one neighborhood to another.

Datasets

  • Vancouver Property Crime Data: Obtained from the City of Vancouver Open Data catalogue, this dataset included crime type, year, month, neighborhood, and coordinates for property crimes (theft, breaking and entering, and mischief) from 2003 onwards. For this study, 2013 data was used.
  • High-Resolution Tree Crown Data: Extracted from airborne LiDAR data collected in February 2013. This provided detailed information on tree canopy coverage.
  • Road Network Data: The 2013 road network data was obtained from Statistics Canada.
  • Ancillary Data: This included population density, unemployment rate (2006 data due to missing 2011 data), percentage of lone-parent families, percentage of low-income families, number of street lights, and number of graffiti incidents. Census data from 2011 was used for dissemination area boundaries.

Key Insights and Ideas

  • Green Spaces as Crime Deterrents: Contrary to some theories, the study found that increased tree coverage is associated with lower property crime rates. This suggests that the benefits of vegetation, such as promoting outdoor activity and natural surveillance, may outweigh its potential as criminal cover in this urban context.
  • Complex Road Networks Reduce Crime: A higher road density (indicating more road segments and potentially more complex networks) was also linked to a decrease in property crime. This aligns with the idea that complex routes can make it harder for criminals to escape or find targets.
  • Spatial Variation is Key: The GWR analysis revealed that the influence of tree coverage and road density on crime is not uniform across Vancouver. The strongest negative correlations were observed in the downtown core and surrounding neighborhoods, highlighting the importance of localized urban planning strategies.
  • BNE is Different: Breaking and Entering (BNE) showed a less pronounced negative relationship with tree coverage compared to theft and overall property crime. This suggests that BNE might be influenced by a different set of factors not fully captured by tree coverage or road density alone.

Key Results

  • Negative Correlations: Both tree coverage and road density showed statistically significant negative correlations with property crime rates, theft rates, and BNE rates.
  • GWR Performance: GWR models demonstrated improved explanatory power (higher adjusted R²) compared to OLS and Spatial Lag models, confirming the spatial heterogeneity of the relationships.
  • Downtown Focus: The GWR maps indicated that the negative impact of tree coverage on property crime was strongest in the downtown Vancouver area and its surrounding neighborhoods.
  • Accuracy: The extraction of tree covered areas from LiDAR data achieved high accuracy, with an overall accuracy of 98.4%, lending confidence to the subsequent regression analyses.

Limitations and Future Work

  • Cross-Sectional Analysis: The study is cross-sectional, meaning it captures a snapshot in time. This limits the ability to establish definitive causal relationships. Future research should employ temporal trend analysis to explore causality.
  • Crime Aggregation: The study aggregated certain crime types. Future work could benefit from analyzing specific crime types (e.g., commercial vs. residential BNE) separately, as their spatial patterns may differ.
  • Exclusion of Violent Crime: Due to data limitations, violent crime was not analyzed. Given Vancouver's high Crime Severity Index, future research should incorporate violent crime data.
  • Road Network Complexity: The study used road density as a proxy for road network characteristics. Future research should explore other factors like road width, connectivity, and complexity.
  • Data Limitations: The use of 2006 unemployment data and potential changes in 2013 census data are acknowledged limitations.
  • Urban Park vs. Street Trees: The study did not differentiate between trees in parks and street trees, which may have varying impacts on crime.
  • 3D Information: The potential of using the 3D information from LiDAR data for more sophisticated crime prevention applications remains largely unexplored.

Summary updated

Abstract

With the development of Geographic Information Systems (GIS), crime mapping has become an effective approach for investigating the spatial pattern of crime in a defined area. Understanding the relationship between crime and its surrounding environment reveals possible strategies for reducing crime in a neighborhood. The relationship between vegetation density and crime has long been under debate. The convenience of a road network is another important factor that can influence a criminal’s selection of locations. This research is conducted to investigate the correlations between tree coverage and property crime, and road density and property crime in the City of Vancouver. High spatial resolution airborne LiDAR data and road network data collected in 2013 were used to extract tree covered areas for cross-sectional analysis. The independent variables were inserted into Ordinary Least-Squares (OLS) regression, Spatial Lag regression, and Geographically Weighted Regression (GWR) models to examine their relationships to property crime rates. The results of the cross-sectional analysis provide statistical evidence that there are negative correlations between property crime rates and both tree coverage and road density, with the stronger correlations occurring around Downtown Vancouver.
BibTeX
@article{ye2018investigating,
  author = {Chengming Ye and Yifei Chen and Jonathan Li},
  title = {{Investigating the Influences of Tree Coverage and Road Density on Property Crime}},
  journal = {ISPRS Int. J. Geo Inf.},
  year = {2018},
  url = {https://openreview.net/forum?id=xKzy0fZ130T},
}
Figure