Using remote sensing data to derive built-form indexes to analyze the geography of residential burglary and street thefts

Figures & data

Figure 1. The study area and the six planning areas.

Figure 2. a) Cadastral polygons and buffer zones of residential building; b) fishnet point and buffer zone of OSA fishnet.

Figure 3. a) Crime rate of residential burglaries; b) Crime rate of street theft.

Figure 4. Scatterplots of crime rates a) residential burglaries and b) street theft against their respective GHSL scores by basområden.

Table 1. Testing for spatial autocorrelation in the residuals of a normal linear regression model.

	Model 1 Moran’s I	Model 2 Moran’s I	Model 3 Moran’s I
Residential burglary	0.237**	0.217**	0.183**
Street theft	0.386**	0.355**	0.338**

**Statistically significant at the 1% level.

Table 2. Model results for residential burglaries.

	Model 1		Model 2		Model 3
	Coefficient.	Significance.	Coefficient	Significance	Coefficient	Significance
Intercept	0.12	0.58	1.18	0.00	0.54	0.21
GHSL_B	0.05	0.00**	–	–	–	–
GHSL_B Planning A	–	–	0.04	0.00**	0.03	0.01**
GHSL_B Planning B	–	–	0.02	0.70	0.01	0.65
GHSL_B Planning C	–	–	0.06	0.36	0.61	0.28
GHSL_B Planning D	–	–	0.05	0.77	0.04	0.69
GHSL_B Planning E	–	–	0.02	0.72	0.03	0.97
GHSL_B Planning F	–	–	0.01	0.18	0.03	0.14
Transport Hubs	–	–	–	–	−0.00	0.70
Young male population	–	–	–	–	0.02	0.26
Unemployment rate	–	–	–	–	0.06	0.15
Popualtion turnover	–	–	–	–	−0.00	0.45
Average Income	–	–	–	–	0.01	0.02**
Lambda	0.44	0.00**	0.12	0.41	0.22	0.10*
Diagnostics
R^2	25.5%		30.9%		33.0%
AIC	1627.52		1612.13		1609.81
Breusch-Pagan test	210.944	0.000**	277.153	0.000**	7277.848	0.000**
Likelihood Ratio Test	9.301	0.002**	0.53	0.466	1.829	0.176

**Significant at the 5% level, *Significant at 10% level.

Table 3. Model results for street theft.

	Model 1		Model 2		Model 3
	Coefficient	Significance	Coefficient	Significance	Coefficient	Significance
Intercept	−24.10	0.42	−77.85	0.10*	−16.44	0.78
GHSL_O	6.69	0.00**	–	–	–	–
GHSL_O Planning A	–	–	9.94	0.00**	9.66	0.00**
GHSL_O Planning B	–	–	−0.72	0.01**	−0.12	0.01**
GHSL_O Planning C	–	–	0.64	0.01**	0.19	0.00**
GHSL_O Planning D	–	–	1.36	0.01**	−0.58	0.00**
GHSL_O Planning E	–	–	0.10	0.01**	−0.01	0.01**
GHSL_O Planning F	–	–	0.57	0.00**	−0.82	0.00**
Transport Hubs	–	–	–	–	17.58	0.00**
Young male population	–	–	–	–	0.27	0.93
Unemployment rate	–	–	–	–	−0.11	0.06*
Popualtion turnover	–	–	–	–	−0.10	0.43
Average Income	^−	–	–	–	−0.93	0.21
Lambda	0.47	0.00**	0.28	0.03**	0.25	0.07*
Diagnostics
R^2	25.0%		29.6%		34.7%
AIC	5876.99		5867.68		5846.22
Breusch-Pagan test	828.038	0.000**	1009.349	0.000**	967.143	0.000**
Likelihood Ratio Test	12.212	0.000**	3.047	0.080*	2.268	0.132

**Significant at the 5% level, *Significant at 10% level.

Table

Name	Description
A. Dense Cityscape (Tätstad)	Mainly closed neighborhoods with a strict division between public and private. Despite the closed neighborhoods, the fine-meshed street network offers many choices for those who move, a variation between public and private land.
B. Large Scale	Traffic-separated environments in large-scale buildings or terraced and semi-detached house areas. Pedestrian and vehicle traffic systems completely separate. The walkways go through the buildings or vegetation. At meetings between pedestrian and vehicle traffic, pedestrians are often led in tunnels.
C. Funkis (Functionalism)	“House-in-park” (buildings freely located in a nature-like environment). Varying location in relation to the street, often far into the plot with entrances facing inwards. The boundary between streets about neighborhood land/plots is often unclear. Walkways are usually located on sidewalks along the streets.
D. Subway City	Further development of “house-in-park” with some impact from older environments. Buildings grouped around court yards or parks. Varying approach to the street, in many cases the buildings and entrances turn inwards or away from the street, in other cases the buildings turn more clearly toward the street. Walkways both located on sidewalks along the streets and on separate lanes through nature-like parkland.
E. Single Family Houses	Mainly one- and two-dwelling houses in garden environments. In garden cities and small cottage areas, the street is defined by buildings bordering the street, hedges and avenues. In residential environments, there is a wide garden/courtyard zone facing the street. The location and design of the buildings and greenery vary.
F. Other/Mix	Random areas in the city, industrials cites, harbor areas, airports and funparks among other things.

Table

Type of data		Description	Year	Source	Variables in the model
Crime statistics	Residential burglaries	Crime codes: 825, 857, 874, 9801, 9802	2019–2020	BRÅ, Police Authority	Relative risk rate of residential burglaries per building per basområden
	Street theft	Crime codes: 801, 802, 803, 804, 805, 806, 808, 809, 892, 9803, 9806, 9808, 9810, 9812, 9817, 9818, 9819, 9820, 9828, 9829, 9830			Rate of street theft crimes per basområden open space area (sqkm)
Land use data	Buildings	Polygons of residential buildings		Open geodata	Buildings with at least one residential apartment/household
	Transport Hubs	Location (x,y) of trains, subway, and bus stations			Proportion of major Transport nodes per basområden population
	Apartments	Location (x,y) of buildings and count of apartments			Number of apartments per building
	GHSL_B	Pixel value of GHSL index per apartment building		Sentinel 2A/Own calculation	Average GHSL pixel value of apartment buildings per basområden for a 50 m radius
	GHSL_O	Pixel value of GHSL index per neighborhood unit Open space area	2020		Average GHSL pixel value per basområden Open Space Area for a 50 m radius
	Water bodies	Areas with water on the surface (lakes, rivers, sea)
	City center	City center (point) of City center (point)	2020	Own calculation	Distance in Km of each basområden to the city center
	Planning areas	Polygons of 6 categories of urban planning development	2021	Own calculation	Dummy variables of urban planning areas present in each basområden
Demographic, social & economic data	YoungMale	The count of male pop. (ages 15–24) per neighborhood unit	2020	SCB	Proportion of male pop. (age 15–24) per basområden population
	Unemployed	Count of unemployed population per neighborhood unit			Proportion of unemployed population by basområden workforce
	Pop_turnover	Population changes between 2018, 2019 and 2020			Difference of population moving in and moving out per basområden
	Income	The average income per neighborhood unit			The average income per basområden for population older than 20 years old.
	Non-Swedish born	Count of population born abroad per neighborhood unit			Proportion of population born abroad per basområden population.
	Single family houses	Count of population per single family houses per neighborhood unit			Proportion of single-family houses population per neighborhood unit population.

Data availability statement

The crime data utilized in this study cannot be made publicly available given the current General Data Protection Regulation (GDPR).