Using GIS to Identify Areas for DUI Enforcements and Analyzing Impacts Michael McGahee Washington College Maryland Department of Transportation Maryland State Police

Introduction Washington College is supporting the Maryland State Police’s DUI team with geographic data and analysis to reduce impaired driving in the Baltimore and D.C. metropolitan areas. We are using multiple data sources to develop a model to make risk projections for linear roads.

Introduction The goal is to help identify road segments that should be targeted for saturation patrols in order to reduce DUI incidents.

Data Our predictive linear risk terrain model utilizes three main data sets. We are grouping the data in three to five year increments to provide a representative time frame for our model. Citations Crashes Liquor Licenses Our source for DUI citation data is the Maryland Judiciary and District Courts Web Portal Electronic Traffic Information Exchange (E-TIX). We have gathered statewide E-TIX data and selected out those citations with offense codes related to impaired driving and DUIs. All crash data comes from the State Highway Administration of Maryland.

Data Washington College has developed a database of liquor licenses throughout the state of Maryland that can be used in conjunction with our other data sources to provide further analysis of impaired driving patterns. We have obtained data regarding active liquor licenses and violations from county liquor boards and license commissioners throughout Maryland. Our data continues to be updated as we receive new information from each of our liquor contacts. With this database, we are able to target drinking establishments with liquor license violations, identify the areas served by these establishments, and locate potential crash hotspots.

Methodology Using network analyst and other GIS tools, we are able to display the data in terms of linear road segments. Our model considers each of the data inputs and returns high risk road segments for DUI enforcement. Note: in order to protect sensitive information, the location data shown has been randomly created for this example.

Methodology There are three components to our model, one for each of the input data sets: DUI Citation Route Density Crash Counts by One Mile Road Segments Liquor License Service Areas

DUI Citation Route Density A network route data layer was created that connects the DUI citation location to the violator’s home address based on the fields collected in the data. Citation Location Home Addresses This was accomplished by running a python script with the Network Analyst Extension tool in ArcMap. After pathways were mapped between the citation location and violator’s home address, a linear density analysis using spatial joins of the route layer was completed.

DUI Citation Route Density After running the E-TIX routing python script, routes have been created connecting ticket locations with the corresponding home addresses. Note: in order to protect sensitive information, the location data shown has been randomly created for this example.

Crash Counts by One Mile Road Segments A state road file was first broken into one mile sections using a custom python script. Then, alcohol-related crashes were joined to these road segments to create a crash count for each segment. Alcohol Crash Counts One Mile Road Segments Alcohol-related crashes are defined as incidents where a driver was determined to be alcohol-impaired through the driver condition, blood alcohol content, substance use detected and contributing factor fields on the Maryland crash report.

Crash Counts by One Mile Road Segments The routes were then used to clip the road network and were spatially joined to create road segments with a count of overlapping routes. Note: in order to protect sensitive information, the location data shown has been randomly created for this example.

Liquor License Service Areas We then assigned a priority ranking to our liquor licenses and utilized network analyst to create 15-minute service areas for the highest priority bars and restaurants. Bars listed by Priority Service Area Travel Route We identified bars and restaurants that close before 10pm as low priority, those that close between 10 pm and midnight as medium priority, and those that are open past midnight as high priority, based on closing times for Friday and Saturday nights. The priority thresholds were established based off of impaired crash data time of incident which showed peak crashes occurring Friday and Saturday nights from midnight to roughly 4am. Service area = network based radius of roads with possible DUI traffic; road segments lying within multiple service areas were assigned an index value

Liquor License Service Areas The TGI Friday’s in Annapolis shown with 5 minute, 10 minute, and 15 minute service areas. Note: in order to protect sensitive information, the location data shown has been randomly created for this example.

Results By combining each of the input data layers, we are able to produce a Predictive Linear Risk Terrain Model for DUI Enforcement. This shows the number of DUI related E-TIX routes that overlap on each road segment; a hotspot display for five, ten, and 15 minute service areas around a bar; and crash counts by one mile road segment data. We use graduated color symbology to categorize road segments and service areas by their relative risk potential for DUIs.

Results Using this analysis, Washington College can inform the Maryland Department of Transportation and the Maryland State Police where high risk road segments for DUI traffic exist, and help target these areas for DUI enforcement saturation patrols.

Products Our model allows us to generate a series of routine DUI analysis products: Quarterly ETIX Analysis Post-Patrol Analysis Holiday Analysis Special Requests We can adjust the time frame and geographic extent of our input data to suit the needs of our customers Requests can incorporate any or all of our primary data sets

Our Customers Maryland Highway Safety Office (MHSO) Maryland State Police (MSP) Local police departments and sheriff’s offices Local liquor licensing agencies The MSP DUI detachment team’s media campaign advertised these images on websites, billboards, gas station pumps, and bus stop shelters.

Acknowledgements Washington College GIS Lab Maryland Department of Transportation Maryland State Police Maryland Judiciary and District Courts Maryland State Highway Administration

Washington College GIS Lab Contact Washington College GIS Lab http://www.washcoll.edu/centers/ces/gis Michael McGahee GIS Analyst II Washington College mmcgahee2@washcoll.edu 443-282-0019