Geographic Information System for Identification of High Accident Locations

ABSTRACT





Many highway agencies have been using GIS for analyzing accident 



data.  Identification of problem locations is one of the most important 



aspect of accident studies.  The purpose of the paper is to show how to 



use various modules of ArcInfo to perform operations such as fixed 



segment analysis, floating segment (sliding) analysis and spot analysis 



for identification of high accident locations.  A prototype GIS system 



was developed for the purpose.  Using the system, the user can merge 



accident and roadway data, match the accident data and locations, 



analyze the data using fixed segment, sliding and spot analysis, 



calculate frequency and rate of accidents, select a variable for 



stratification to calculate mean and standard deviation of accident 



rates and frequencies and sort the sections based on selected criteria.  



The GIS will also call external C programs to calculate accident 



characteristics and expert system to identify possible causes. 














INTRODUCTION





Accident data, collected for many years, serves as the ground-base for 



programs designed to reduce the number of traffic accidents.  In 1979, 



the Federal Highway Administration established a Highway Safety 



Improvement Program to aid the states in developing safety programs 



(Zageer, 1981). Early advances in computer technology and 



communications led to simple and rather inexpensive data acquisition 



techniques.  At the same time, several techniques for analyzing high 



accident locations were established (Zageer, 1982).  As a result of the 



availability of data, high-speed computers and software, an extensive 



amount of research was undertaken on statistical analysis of accident 



data.





Highway agencies employ computers to analyze accident data and to 



identify high accident locations.  Intermodal Surface Transportation 



Efficiency Act (ISTEA) requires states to develop, implement and 



maintain safety management systems.  Recently, many highway 



agencies are using GIS for identification of high accident locations.  A 



site that experiences abnormal frequencies, rates, or severities of 



accidents is considered as a high accident location.  Accident and 



roadway files are usually used to identify high accident locations and 



to summarize accident characteristics.  Geographic Information 



Systems has been identified as a potential tool to use for improving the 



accident analysis process.  The trend in analyzing accident data is to: 



(1) identify high accident locations , (2) identify accident 



characteristics, (3) determine the causes (4) identify the 



countermeasures and (5) evaluate the countermeasures.  This process 



is extremely tedious and requires extensive analysis of data.  In this 



study, we developed a prototype Geographic Information System for 



identification of high accident locations using various modules of 



ArcInfo.







STUDY OBJECTIVES





The goal of the study is to develop a GIS system that process and 



analyzes data efficiently in identifying high accident locations.  The 



specific objectives of the study is to develop GIS procedures for the 



following activities:





1. to identify and rank high accident locations using fixed segment, 



    floating segment and spot analysis.



2. to summarize accident characteristics at these high accident 



    locations



3. to identify the causes of accidents by integrating GIS with expert 



    systems







ACCIDENT DATA 





The accident and roadway data used to develop the system were 



obtained from Wake County, North Carolina.  A brief description of 



the "coverages", the "INFO" files and 'route systems' used in this study 



are provided below.





1. Coverages





NHSRD (Line Coverage) - This coverage included only primary 



highways within the county.  This coverage was used to match the 



accident files with the primary highways.  For this coverage, route 



systems were developed.



RD24 (Line Coverage) - This coverage included county boundaries, 



primary and secondary highways and the other roads in the county.  



This coverage was used only for viewing and not for the accident 



matching process.



Brdg (Point Coverage) - This coverage included bridge location and 



attributes.





2. INFO Files





Accident - This file included approximately 20,000 accident records 



from Wake County, North Carolina.  The route number (Route8) in 



this file was used to match the accidents with NHSRD coverage.



Roadway - This file included roadway information such as functional 



classification, terrain, average annual daily traffic (AADT), sight 



distance, shoulder width and others.  The route number, milepost-from 



and milepost-to in this file were used to match with the NHSRD 



coverage.



Pavement - This file included the pavement information of only a few 



primary highways.  The route number (Route8)  in this file was used to 



match the pavement data with the existing NHSRD coverage.





3. Route System





HWY (route system for NHSRD coverage) - This route system 



included all primary highways in Wake County.  The route number 



(RTJOIN) was used to match with accident and roadway files. 







ACCIDENT REFERENCING/VIEWING SYSTEM 





AML programs were developed to view the accident, bridge and road 



coverages. The developed system would allow the user to:



- display the selected coverages.



- zoomin, zoomout, pan on the screen.



- add or move accident point using ARCEDIT.



- display the selected data in a coverage, for example: the accidents on 



   route I-64 alone can be displayed.



- select the linesymbol and markersymbol for the display of line and 



   point coverages.



- query the selected coverages.







ACCIDENT MATCHING PROCESS





Dynamic segmentation module was used to match the accident file 



with NHSRD or RD24 for primary highways or roads, respectively.  



The NHSRD coverage which included only primary highways was 



used to match with the accident data.  It would have been more 



meaningful to use RD24 coverage, to maximize the matching process.  



Unfortunately, for the RD24 coverage, route systems had not been 



developed at the time of this study.  In order to use the accident data 



for further analyses, the matched accidents were converted into point 



coverage using EVENTPOINT command in dynamic segmentation.







METHODOLOGY





In this study, the accident data was analyzed by segment (fixed or 



floating) or spot analysis.  





Fixed Segment Analysis: A segment of fixed length is used to query 



accident file for locations exceeding certain number of accidents. For 



example using 0.5 mile segment, a search can be made for each route 



from mile post 0 to 0.5, 0.5-1.0, 1.0-1.5 etc. to identify of high 



accident segments. (Zageer, 1982)







Floating segment or Sliding analysis: This is a special case of segment 



analysis in which segment length floats or moves sequentially by 



milepost. Such floating segments usually advance in 0.1 mile 



increments through the accident file.  For example a search can be 



made for each route 0-0.5, 0.1-0.6, 0.2-0.7 etc. (Zageer, 1982)





Spot Analysis: A spot is short segment of a highway (generally 0.3 



miles or less) used to identify high accident spots such as bridges, RR 



crossings, intersections etc. (Zageer, 1982)





Accident characteristics summary: The next step after identification of 



high accident locations is to summarize accident characteristics at 



these sites by severity, collision type, road condition, light condition, 



road character and alcohol related etc.  This will help to determine 



abnormal accident patterns and lead to the selection of appropriate 



safety improvements.  Due to limitations of INFO module, C programs 



were developed to summarize accident characteristics and to rank and 



sort the sites.





Accident Causes: Once accident characteristics are identified, the next 



step is to determine possible causes.  Expert systems are the most 



suitable for determining the cause of accidents.  In this study, the 



expert system is developed only to demonstrate the prototype and does 



not contain heuristic knowledge.  The expert system shell CLIPS (C 



Language Integrated Production System) developed by NASA was 



used for this study.





Statistical Methods: One of the methods used to identify problem 



locations is to compare accident rates of individual locations with the 



accident rates of all similar sites in the region over the same period of 



time.  In this study, the user can group the sites of similar 



characteristics by selecting a variable.  Mean  and standard deviation 



of the accident rate and frequency were calculated for each group 



using STATISTICS command.  A site is identified as hazardous, if its 



accident rate exceeds mean accident rate plus a multiple of standard 



deviation of similar sites.







FIXED SEGMENT ANALYSIS





The first step in fixed segment analysis is to create an INFO file which 



contains segments of fixed lengths as specified by the user for each 



route.  This  event data  is used along with roadway characteristics 



event data to perform line on line overlay analysis.  The results of the 



overlay contains roadway characteristics for each segment and is 



stored in a event table.  Next, overlay this event table with accident 



location table to find the condition of roadway at each of the accident 



sites.  The resulting overlay event table is used to find number of 



accidents on each segment using FREQUENCY command.  This 



process of fixed segment analysis is shown in Figure 1.