GIS, Crime Analysis, and Risk Terrain Modeling

Penn IUR and MUSA (Master of Urban Spatial Analytics) hosted a recent lunch series on GIS, Crime Analysis, and Risk Terrain Modeling, featuring Dr. Joel Caplan, Associate Professor of Criminal Justice and Deputy Director of the Center on Public Security at Rutgers University.

A consultant for agencies throughout the United States and internationally on matters of public safety and national security, Dr. Caplan presented his work to an audience of Penn graduate students and Philadelphia area practitioners, and discussed the development of Risk Terrain Modeling, which employs GIS and spatial analysis in evaluating illegal behavior, crime patterns, and spatial risks.

Dr. Caplan stressed the behavior settings for crime and the environmental context of low- versus high-risk crime neighborhoods as well as the importance of spatial influence in risk terrain modeling. “A sidewalk or bush might be considered a benign feature,” said Dr. Caplan, but “sidewalks surrounded by bushes might be perceived as potentially risky areas.” He went on to discuss environmental features that create spaces in which criminals feel comfortable and can hide.

However, Caplan explained that vector points and reliance on GIS can be poor indicators of criminal activity and don’t necessarily communicate the naturally occurring features of the real world. Whereas risk analysis that manifests features of landscape can be used to compute probability of criminal behavior occurring – risk terrain modeling creates models of spatially vulnerable areas. 

With knowledge of environmental factors, interventions designed to suppress crime in the short-term can achieve greater success by mitigating spatial factors in high-crime areas, in an effort to make them less attractive to crime in the long-term. For example, “creating a community garden where vacant buildings once were helps police and other stake holders mitigate spatial risk factors,” said Dr. Caplan.

Spatial dynamics of crime aren’t the same in all neighborhoods, and Dr. Caplan explained that traditional approaches to tackling crime have limitations. They often follow “standard” patterns that are believed to exist across multiple environments, when in fact such patterns are rarely consistent in varied environments.

“Only when underlying characteristics of hot-spot areas of crimes [are identified] do we realize that places of incidence are very, very different” he explained. Caplan contrasted several models of hot-spot mapping across the nation. For example, bars are computed into the risk model that police use in Chicago, but not in Kansas City, where crime and bar-laden neighborhoods are not correlated.

Dr. Caplan shared several hot-spot maps with the audience, pointing out that while hot-spot mapping has proven accurate, to an extent, there’s an inherent assumption that crime won’t move from where it’s taken place in the past, even after police intervene. What happens when police are successful? Why does repeat offending happen even when cops have patrolled the same hot spots for decades? These are the questions, said Dr. Caplan, which spatial diagnoses need to answer.

The issue with risk terrain modeling, suggested Dr. Caplan, is that it’s very predictive, making it important to consider the impact of physical environments and how these influence human interactions within spaces. “We need to think differently,” he said, and “stop playing whack-a-mole” with crime. Targeting and ameliorating criminal behavior requires concerted efforts to focus on “mechanisms that enable hot spots to emerge, persist, and ultimately desist over time.”