The Effect of Roadside Elements On Driver Behavior and Run-Off-the-Road Crash Severity

University of Massachusetts Amherst
ScholarWorks@UMass Amherst
Masters Theses 1911 - February 2014 Dissertations and Theses
2013
The Effect of Roadside Elements on Driver

Behavior and Run-Off-the-Road Crash Severity
Cole D. Fitzpatrick
University of Massachusetts Amherst, cole@fitzpat.com
Follow this and additional works at: http://scholarworks.umass.edu/theses

Part of the Other Civil and Environmental Engineering Commons
Fitzpatrick, Cole D., "The Effect of Roadside Elements on Driver Behavior and Run-Off-the-Road Crash Severity" (2013). Masters
Theses 1911 - February 2014. 1037.
http://scholarworks.umass.edu/theses/1037
This thesis is brought to you for free and open access by the Dissertations and Theses at ScholarWorks@UMass Amherst. It has been accepted for
inclusion in Masters Theses 1911 - February 2014 by an authorized administrator of ScholarWorks@UMass Amherst. For more information, please
contact scholarworks@library.umass.edu.
THE EFFECT OF ROADSIDE ELEMENTS ON DRIVER
BEHAVIOR AND RUN-OFF-THE-ROAD CRASH
SEVERITY
A Thesis Presented
by
COLE D. FITZPATRICK
Submitted to the Graduate School of the
University of Massachusetts Amherst in partial fulfillment
of the requirements of the degree of
MASTER OF SCIENCE IN CIVIL ENGINEERING
MAY 2013
Civil Engineering
THE EFFECT OF ROADSIDE ELEMENTS ON DRIVER
BEHAVIOR AND RUN-OFF-THE-ROAD CRASH
SEVERITY
A Thesis Presented
by
COLE D. FITZPATRICK
Approved as to style and content by:
_________________________________
Michael A. Knodler, Chair
_________________________________
John Collura, Member
_________________________________
Matthew R. E. Romoser, Member
__________________________________________
Richard Palmer, Department Head
Civil & Environmental Engineering
ABSTRACT
THE EFFECT OF ROADSIDE ELEMENTS ON DRIVER BEHAVIOR AND RUN-
OFF-THE-ROAD CRASH SEVERITY
MAY 2013
B.S.C.E OREGON STATE UNIVERSITY
M.S.C.E UNIVERSITY OF MASSACHUSETTS - AMHERST
Directed by: Michael A. Knodler Jr. Ph.D.
Roadside vegetation provides numerous environmental and psychological benefits to
drivers. Previous studies have shown that natural landscapes can effectively lower crash
rates and cause less stress and frustration to the driver. However, run-off-the-road crashes
resulting in a collision with a tree are twice as likely to result in a fatality, thus
reinforcing the need to examine the placement of vegetation within the clear zone,
defined herein as a flat unobstructed area for errant vehicles to recover. This study
explores the relationship between the size of the clear zone and the presence of roadside
vegetation on selected driver attributes, including both driver speed and lateral
positioning. To evaluate the effect on the driver speed selection process, a static
evaluation was employed. Completed by more than 100 drivers, the static evaluation was
utilized to gather speed selections on both real and virtual roads containing four
combinations of clear zone size and roadside vegetation density. Additionally, field data
was collected to validate the findings of the static evaluation and to determine the extent
to which roadside vegetation impacts driving attributes. When presented with a large
iii
clear zone, drivers positioned the vehicle further from the edge of the road as the
vegetation density increased. Furthermore, the speeds observed in the field correlated
with the speeds that participants selected when watching a video of the same road.
Finally, the UMassSafe Traffic Safety Data Warehouse was utilized to link crash and
roadway data, allowing for an in-depth analysis of run-off-the-road (ROR) crash severity.
The results of this study further demonstrate the nature of the relationship between clear
zone design and driver behavior. Additionally, the results provide information regarding
improved clear zone design practices that may translate into improved roadway safety.
iv
TABLE OF CONTENTS
Page
ABSTRACT....................................................................................................................... iii
LIST OF TABLES ............................................................................................................ vii
LIST OF FIGURES ......................................................................................................... viii
CHAPTER
1 INTRODUCTION ........................................................................................................... 1
1.1 Problem Statement .................................................................................................... 1
1.2 Research Hypotheses and Objectives ........................................................................ 2
1.3 Scope ................................................................................................................. 3
2 BACKGROUND ............................................................................................................. 5
2.1 Psychological Effect of Trees ................................................................................... 5
2.2 Crash Data Analyses ................................................................................................. 7
2.3 Behavioral Studies................................................................................................... 12
3 METHODOLOGY ........................................................................................................ 16
3.1 Literature Review .................................................................................................... 16
3.2 Static Evaluation ..................................................................................................... 16
3.2.1 Static Evaluation Development........................................................................ 16
3.2.2 Development of Simulated Environments ....................................................... 17
3.2.3 Field Video Capture ......................................................................................... 18
3.3 Real World Validation of Static Evaluation Results ............................................... 20
3.3.1 Field Validation Observation Spots ................................................................. 20
3.3.2 Standard Data Collection Method .................................................................... 20
3.3.3 Linked Site Data Collection ............................................................................. 21
3.3.4 Data Analysis Technique ................................................................................. 21
3.4 Crash Data Analysis ................................................................................................ 22
3.4.1 Filtering of Run-off-the-road Crashes from All Crashes ................................. 23
3.4.2 Software used for Analysis and Plotting of Crash Data .................................. 23
4 FINDINGS AND RESULTS ......................................................................................... 24

4.1.1 Factors Influencing Speed................................................................................ 25
4.1.2 Speed Decisions ............................................................................................... 27
4.2 Validation of Static Evaluation Findings ................................................................ 31
4.2.1 Speed Results ................................................................................................... 32
4.2.2 Lateral Positioning ........................................................................................... 32
4.2.3 Similar Location as Static Evaluation .............................................................. 33
4.2.4 Linked Locations ............................................................................................. 34
4.3.1 Gender and Time of day .................................................................................. 36
4.3.2 Driver Age ....................................................................................................... 38
v
4.3.3 Most Harmful Roadside Objects ...................................................................... 39
4.3.4 Guardrails ......................................................................................................... 40
4.3.5 Speed Limit and Shoulder Width..................................................................... 40
4.3.6 County .............................................................................................................. 42
5 DISCUSSION ................................................................................................................ 45
5.1.1 Simulation Videos............................................................................................ 45
5.1.2 Real World Videos ........................................................................................... 46
5.1.3 Limitations of Static Evaluation ...................................................................... 46
5.2.1 Observed Speeds .............................................................................................. 47
5.2.2 Lateral Positioning and Linked Locations ....................................................... 47
5.2.3 Similar Location as Static Evaluation .............................................................. 48
5.2.4 Limitations of Field Validation ........................................................................ 49
5.3.1 Gender and Time of Day.................................................................................. 49
5.3.2 Driver Age ....................................................................................................... 50
5.3.3 Most Harmful Roadside Objects ...................................................................... 50
5.3.4 Guardrails ......................................................................................................... 51
5.3.3 Speed Limit and Shoulder Width..................................................................... 51
5.3.4 County .............................................................................................................. 52
6 CONCLUSIONS............................................................................................................ 53
6.4 Future Work ............................................................................................................ 55
6.4.1 Validation of Field Data Collection Methodology .......................................... 55
6.4.2 Full Scale Driving Simulator Study Using Developed Scenarios ................... 56
6.4.3 Investigate the Efficacy of Guardrails at Preventing ROR Crashes ................ 56
REFERENCES ................................................................................................................. 57
vi
LIST OF TABLES
Table Page
1 Design Characteristics of Livable and Comparison Sections of Colonial Drive from

Safe Streets, Livable Streets. (10) ....................................................................................... 9
2 Factors Included within Assessment of Driver Speed Selection Process ...................... 17
3 Clear Zone Size and Vegetation Densities Associated with Figure 5. .......................... 19
4 Locations of the Field Validation Observation Spots .................................................... 20
5 Objectives and Hypotheses ............................................................................................ 24
6 Ranking of Static Evaluation Factors Influencing Speed .............................................. 25
7 Standard Deviations from the Static Evaluation Ranked from Most

Variability to Least ....................................................................................................... 26
8 Comparison of Responses for Density of Roadside Vegetation versus Other Speed

Influencing Factors ........................................................................................................... 27
9 Within Subjects Speed Comparison Measuring the Effect of Utility Poles on Speed
Choice ............................................................................................................................... 31
10 P-Values From Chi-squared Test Associated with Data Displayed in Figure 18 ........ 37
11 P-Values from Chi-squared Test Associated with Data Displayed in Figure 19. ........ 38
12 P-Values from Chi-squared Test Associated with Data Displayed in Figure 20. ........ 39
vii
LIST OF FIGURES
Figure Page
1 The three videos shown to subjects showing the varying levels of vegetation. Top
Left Least Vegetation, Top Right Medium, Bottom Left Most. (4) .......................... 6
2 90% of fatal ROR accidents in Australia occurred from collisions with fixed objects
less than 20 feet away from the roadway. (1) ................................................................... 12
3 Four different scenarios drivers encountered in The Street Tree Effect. (19) ................ 14
4 Driving simulator scenarios depicting different clear zone size/vegetation density

combinations shown to participants during the static evaluation. .................................... 18
5 The route taken during field video capture and the locations of the clear
zone/vegetation combinations. .......................................................................................... 19
6 A setup where video could not be obtained from the inside of a vehicle. ..................... 21
7 Computerized portrayal of the lateral positioning analysis. .......................................... 22
8 UMass Safety Data Warehouse. .................................................................................... 22
9 Flow chart of methodology to isolate run-off-the-road crashes..................................... 23
10 Average speed selected after watching the four driving simulator drives. .................. 28
11 Average speed selected after watching the nine videos of real drives. ........................ 29
12 Static evaluation scenario with directional utility pole presence. ................................ 30
13 Average normalized speeds from the field study ......................................................... 32
14 Average normalized lateral positions measured by the distance in feet from the edge
line..................................................................................................................................... 33
15 Average speeds of the four locations studied in both the static evaluation and field
validation........................................................................................................................... 34
16 Linked location on Route 202 Southbound in Belchertown, MA ............................... 35
17 Linked location on Route 9 in Amherst, MA .............................................................. 36
18 ROR injury rates as a function of both gender and time of day. ................................. 37
viii
19 ROR injury rates versus driver age. ............................................................................. 38
20 Crash injury rates as a function of both driver age and gender. .................................. 39
21 Crash injury rates for various run off the road crash indicators present in one of the
four possible sequences of events within a crash report. .................................................. 40
22 The effect of hitting a guard rail on ROR crash severity. ............................................ 40
23 Posted speed limit versus crash injury rate. ................................................................. 41
24 Crash injury rates as a function of the size of the right shoulder and
posted speed limit. ............................................................................................................ 42
25 County map of run off the road serious injury rates as compared to the statewide
average. ............................................................................................................................. 43
26 County map of run off the road minor injury rates as compared to the statewide
average. ............................................................................................................................. 44
ix
CHAPTER 1
INTRODUCTION
Due to increasing levels in the built environment, opportunities to preserve the
natural environment are critical. Studies on climate change have shed light on the critical
role trees play within the environment with regards to mitigating negative impacts.
Specifically, trees have the ability to reduce heat islands by providing shade, leading to
lower pavement temperatures and decreased emissions (1). Positive psychological
implications, such as reduced stress, decreased road rage, alleviated depression and
expedited recovery from injuries have been associated with natural environments.
While trees provide psychological and environmental benefits, they pose a
potential risk to drivers when placed within proximity to the traveled way.
Approximately 1.9 percent of all crashes are with fixed objects such as trees and 46
percent of these crashes are fatal (1). Although there is research that describes both the
positive and negative aspects of natural vegetation along the roadside, there is a need to
further understand the impacts on driver performance. Additionally, there is a need to
investigate the correlation between the built environment and run-off-the-road crashes.
The research conducted within this thesis outlines a methodology that was employed to
evaluate the effects of roadside elements on driver performance as well as an analysis of
run-off-the-road crashes. The corresponding results, included herein provide describe the
nature of the relationship between roadside vegetation and select driver attributes.
1.1 Problem Statement
As described within the American Association of State Highway and
Transportation Officials (AASHTO) A Policy on Geometric Design of Highways and
1
Streets (Green Book), the clear zone is a design element on both local and collector roads
and is intended to provide a recovery area for errant vehicles (2). The AASHTO Green
Book stipulates that the clear zone located on roads with or without a curb should be a
minimum of 7 and 10 ft, respectively (2). Implementation of the clear zone may often
present a challenge as it may require the roadway builder to purchase additional right of
way and to remove trees.
The efficacy of the clear zone remains a bit ambiguous within the literature as
past studies show contradicting results with regard to roadside elements. The natural
environment seems to produce psychological benefits on drivers, yet trees were shown to
increase the severity of run-off-the-road crashes. Analysis on the benefits and hazards of
vegetation placed within the clear zone has not been fully investigated, thus there is a
need for an improved understanding as to how elements placed within the clear zone
impact the related driver behavior. Specifically, there is a need to compare the frequency
and severity of crashes on roads with adequate clear zones versus roads with roadside
trees. Lastly, there is a need to evaluate and understand the specific impacts of trees
within the clear zone on driver behavior.
1.2 Research Hypotheses and Objectives
Based upon the identified problem statement, the overarching goal of this thesis
research was to evaluate the impacts resulting from the presence of trees within close
proximity to the traveled way. Within the framework of this overarching goal, a series of
research objectives and corresponding hypotheses were developed as outlined in the
following section.
2
Objective 1: Understand the effect that roadside vegetation density has on vehicle
speeds. The first hypothesis is that the density of trees along the roadside will have a
measurable effect on operating speeds. When trees are more thickly settled, drivers may
perceive a faster operating speed or feel unsafe, and in turn reduce their speed. The
second hypothesis is that vehicle speeds observed in the real world will be similar to the
speeds surveyed during the static evaluation.
Objective 2: Understand how the size of the clear zone affects driver behavior. It
is hypothesized that a smaller clear zone may cause drivers to position their vehicle
further from the edge of the road or accept lower operating speeds in fear of running off
the road.
Objective 3: Determine the effect that various roadway environmental elements
have on drivers speed selection. It is hypothesized that participants will not rate
vegetation density as influential to their speed selection because it is a roadway element
that is only subconsciously noticed.
Objective 4: Utilize crash data to investigate run-off-the-road crashes and the
patterns associated with their severity. It is hypothesized that a direct correlation will
exist between the severity of run-off-the-road crashes and the type of roadside elements
involved during the crash.
1.3 Scope
While there are many factors that are believed to influence operating speeds, the
scope of this study focused solely upon roadside elements, and most specifically upon
3
roadside vegetation. Other variables such as traffic volumes, road conditions, weather
conditions and functional classification were held constant or not considered as variables.
Additionally, the scope of the crash data analysis component was limited to
Massachusetts run-off-the-road crashes occurring between 2007 and 2010. Other types
of crashes were examined for comparison purposes but the focus was on run-off-road
crashes.
4
CHAPTER 2
BACKGROUND
Concepts relating to both roadside vegetation and run off the road safety have
been the focus of a myriad of research efforts. The published literature was studied to
identify previous research that is related to the topics presented within this thesis. More
specifically, the research review focused upon three primary topics: the psychological
effects that natural environments have on people in general, driver behavioral studies and
crash data analyses. The literature reviewed, while broad in topics, makes a good
portrayal of the benefits and consequences of roadside trees.
2.1 Psychological Effect of Trees
A recent study was conducted by the Kentucky Transportation Center at
University of Kentucky and had participants watch a virtual simulation online of different
roadway types (3). Some roads were in an urban setting while others were rural. They
varied the roadway width, clear zone width, surrounding plant intensity and barrier type.
After the subjects went through a roadway scenario they were asked to rate their
discomfort on a ten point scale. The study found that discomfort increases as vegetation
becomes more intense and increased with narrower roadways. The researchers
hypothesized that an increase in discomfort could correlate to reduced vehicle speeds as
drivers have a lower sense of security. The researchers concluded that it would be useful
to run a direction validation test, eliciting data for each scenario in a five-axis simulator
and comparing these data with response data gathered using the online visualization
method. (3)
5
A paper published in Environment and Behavior tried to address the question of
can highway vegetation mitigate automobile driver frustration? (4) To answer this
question, 106 people watched a video of a drive down one of three highway corridors.
The subjects were asked to solve an unsolvable anagram; the researchers would measure
their level of anger and their frustration tolerance, or how long they kept trying before
giving up.
Figure 1 The three videos shown to

subjects showing the varying levels of
vegetation. Top Left Least Vegetation,
Top Right Medium, Bottom Left Most.
(4)
The researchers found that the videos had no effect on how angry the subjects got
from failing to solve the anagram. However, the researchers did find that the people who
viewed the videos with more vegetation had a higher tolerance for frustration as more
time was spent trying to solve the anagram before giving up. The researchers mention
that this finding is significant because it means vegetation can reduce the chances of road
rage occuring.
6
A study conducted in 2003 attempted to determine what made a stretch of road
more or less preferable for pedestrians (5). To do so, three types of walking behavior
were first identified: walking to commute, walking for health reasons, or walking for
spiritual or relaxation purposes. The researcher then asked people to respond to a
questionaire asking the reasons why they walk, and what they did or did not like about
the road they were walking along. Common responses for positive attributes were
greenery, grass, and road was well maintained. These factors also showed up when they
were not present along the stretch of road. In the end the author summarized that it is
clear that greenery/grass/landscape appears to be fairly significant when walking for
spiritual renewal or stress relief. (5)
A team of researchers from Texas A&M studied the effects that roadside
environments had on stress recovery and immunization (6). They subjected people to a
mild stressor and then had them watch a video of a simulated drive with varying levels of
greenery. Subjects were again subjected to another mild stressor. After that stressor, the
experimenters measured their blood pressure. As expected, the researchers found that the
subjects who viewed the videos of more natural settings had less elevated blood pressures
than the other groups.
2.2 Crash Data Analyses
In Seattle on SR-99, the city converted two-way left turn lanes into landscaped
medians with left turn/U-turn pockets, among other improvements. Researchers from the
Washington State DOT analyzed three years of crash data before and after the change (7).
The study found a decrease in crash rates but an increase in crash severity; however,
neither finding was statistically significant. The authors noted that 32 trees had to be
7
replaced in the three year analysis period and while collisions with trees were generally
property damage only (PDO) crashes, concern does exist that as tree width and strength
grow over time, tree crash severity may increase. (7)
In Texas, Researchers examined ten sites that had recently undergone landscape
improvements (8). These landscape improvements included interchange, roadside and
median landscaping as well as roadside planting. The crash rate significantly decreased at
8 of the 10 sites studied and increased slightly at the other two. The authors identified the
reason for the increase was due to the sites being complex grade separated interchanges
which require a higher number of roadside vertical objects, such as columns and road
signs. A 71% decrease of tree collisions was observed which was attributed largely to one
of the ten sites. The researchers concluded that landscape along the roadside is having a
positive effect on driver behavior and perception. (8)
Five roads were studied in Toronto, Canada that had recently undergone
landscape improvements. Crash frequency and severity were observed for three years
before the improvements and three years after the improvements had been made (9). The
researchers found that there was between a 5 and 20 percent reduction in crash rate and
severity as a result of the changes. The authors say that their research indicates a
possible correlation between greening and reduction in mid-block accident frequency and
severity. (9)
A researcher from Georgia Institute of Technology studied five years of crash
data from Colonial Drive in Orlando, Florida (10). He did so to challenge the assertion
that the wider the clear zone, the safer it will be. A livable section of the road was
compared with a section that adhered to the more conservative design standards. Through
8
analysis of the crash data, he found that livable streets had 10.7 percent fewer crashes and
23.6 percent fewer injurious crashes. A summary of the roadway characteristics can be
found in Table 1.
Table 1 Design Characteristics of Livable and Comparison Sections of Colonial

Drive from Safe Streets, Livable Streets. (10)
Researchers at Caltrans and California Polytechnic State University studied six
years of crash data on 29 different sections of road, of which 19 had trees in the median
(11). They only included crashes that occurred on the side of the road containing the
median as they wanted to see what effect the trees had on frequency and severity. A
significant finding was that large trees in medians were associated with more collisions
and increased severity. However, the authors acknowledged that some of this association
was statistically weak. Lower speeds and larger side clearances were not found to reduce
frequency and severity of crashes with median trees contrary to what might be intuitively
believed.
A study completed in 2003 compared the safety of freeways to parkways, which
are roads which are generally divided by a landscaped median and have medium to high
speeds (12). The study attempted to compare roads that had similar characteristics in
terms of average daily traffic, average speed, and types of drivers. The main performance
indicators that were used were fatal accident rate (FAR) and accident cost (AC). The
9
study found that parkways had a lower FAR as well as a lower AC. Run-off-the-road
crashes were more common on parkways but the authors deduced that this was due to
elements that make up freeways, such as concrete barriers and guardrails which inhibit
run-off-the-road crashes. The authors admit that this current study does not allow
conclusions about any specific landscape elements or settings that may be contributing to
the decrease in collisions and accidents. (12) This paper identifies an area prime for
further researchers, the study of how each individual landscape element affects driver
performance. Such a study could possibly be conducted in a driving simulator.
A 2006 study analyzed national crash data in an attempt to quantify the effect that
urban trees had on traffic safety which was measured by crash incidence and severity
(13). The reasoning for undertaking this project is that circumstances of tree crashes in
urban settings are not well understood. (13) In addition to learning about the patterns of
crashes with trees, the researches wanted to look at the difference in collisions based on if
the area was urban or rural. A relevant finding from the analysis was that collisions with
trees made up 1.9% of all crashes and crashes in rural areas result in a higher rate of
injurious crashes than in urban areas and fixed object collisions were more frequent in
rural areas than urban areas. An interesting observation was made when they were
describing trees as technology, Western European countries have lower crash injury
rates despite less support of tree removal due to aesthetic and environmental reasons.
(13) The article concludes with the statistic that 80 percent of the US population lives in
urban areas and there is a need for better informed road design.
Five years of crash data were studied for a two lane undivided roadway in New
Hampshire (14). The researchers wanted to build a model that identified statistically
10
significant factors predicting the probability of crashes and injury crashes. They wanted
to use this model to perform a risk assessment in the area for which the model was
developed. One relevant finding was that adding sidewalks to a section of the road made
it safer not only for the pedestrians but also for drivers.
Andrew Zeigler analyzed 500 vehicle-tree accidents in Michigan in an attempt to
provide recommendations to state and local road authorities (15). He found that often
times the driver is intoxicated or is not familiar with the road. He observed that vehicle-
tree accidents usually occur along a winding rural road and that no single feature of the
road environment accounts for all the accidents that occur. Like many engineers before
him, Zeigler suggests tree removal as an alternative for addressing environmental and
safety issues.
A literature review in 2007 focused on the benefits that trees have on
communities while balancing the risk they present to motorists (1). Early in the paper,
they talk about how urban trees can be a cost effective approach for urban cities to meet
environmental standards. Part of this effect is that emissions increase with respect to air
temperature, and trees that shade pavement can reduce asphalt temperatures by 36 F
which can reduce air temperatures by 7 F. The authors also cited the statistic that 1.9
percent of all accidents were collisions with trees but they mention that 46 percent of
these collisions were fatal. Due to the way police reports are written, it is hard to study
the effect that vegetation has on crash rates as this detail is rarely noted by the officer(s)
on the scene. Among many suggestions, the authors call for improved data collection
concerning vegetation and before-and-after studies to assess consequences of installing
or removing street and median trees. (1)
11
100
90
Cumulative Frequency (%)

80
70
60
50
40
30
20
10
0
0 5 10 15 20 25 30 35 40 45 50 55
Distance from roadway of fixed object (feet)
Figure 2 90% of fatal ROR accidents in Australia occurred from collisions with
fixed objects less than 20 feet away from the roadway. (1)
2.3 Behavioral Studies
Several studies that have focused upon driver behavior are of relative importance
to the work conducted within this thesis. This section outlines several of these studies
which have focused on driver behavior.
Researchers from China, collaborating with the Texas Department of
Transportation examined the relationship between roadside stimuli and driver fatigue
(16). In the experiment, 12 subjects drove in a driving simulator down a two-way two-
lane road. Along the road they varied the amount of roadside stimuli. For the majority
of the drive, there would be no significant roadside elements; then every 1, 5, 10 or 40
kilometers there would be a section of roadside trees. The researchers used average heart
rate as a measure of driver fatigue as prior research indicated that heart rate was a good
indicator of judging driving load. The study found that drivers were least affected by
12
dense stimuli as they became acclimated to it, the optimal spacing of stimuli to keep
drivers alert was found to be between 5 and 10 kilometers. The study demonstrates that
roadside elements do affect drivers physiologically and demonstrates a need for more
research to determine if this physiological change affects driver performance.
Researchers in Italy used a driving simulator to test six different conditions
involving guard rails and shoulders along the road (17). They had their subjects drive
through one of the six different conditions and they measured their speeds and lateral
position along different parts of the drive which included a left and right curve as well as
a straight section. They found that when a shoulder was present drivers went significantly
faster at 17 of the 42 points at which they were measuring. There was no significant
change at the other measurement points. The subjects also drove nearer to the edge at 33
of 42 points when a shoulder was present. The guardrail had no significant effect on
speed. An interesting finding from this study was that when no guard rail was present
near trees, drivers did not change their position in the travel lane. The researchers
concluded that this implied that the drivers did not perceive the nearby trees as a risk, but
they also acknowledged that this may change in real life when actual physical harm is
possible.
A project from the Texas Transportation Institute used a driving simulator to test
the effect that roadside trees had on drivers perception of safety as well as sense of edge;
or how well they can distinguish the road from surrounding land uses (18) (19). In the
experiment, 31 participants drove through four different scenarios, an urban setting with
and without roadway trees and a rural setting with and without roadway trees. The
participants were then asked to rate their sense of safety and sense of edge. Throughout
13
the experiment, the vehicle speed was also captured. The researchers found that suburban
streets with trees were perceived as safest and urban streets without trees were perceived
as least safe. A mean speed reduction of 3.02 miles per hour was observed when trees
were present along the suburban landscape. The presence of trees also helped drivers with
sensing the edge of the road. The authors summarized by saying that street trees may
provide positive safety benefits and further study with a larger sample size is
warranted. (18) (19)
Figure 3 Four different scenarios drivers encountered in The Street Tree Effect. (19)
Researchers from the University of Minnesota used a driving simulator to test
various changes to the roadway, such as the installation of lighting poles, lighter colored
pavements and shrubs on medians and shoulders (20). They had 32 participants (16 male,
16 female) drive through 8 different combinations of their variables. The researchers
collected the speeds of the drivers as well as their position in the lane. They concluded
14
that landscape treatments indeed had an effect on driver behavior, but the effect was
markedly inconsistent. (20)
Ewing and Dumbaugh reviewed literature regarding which factors cause crashes
in general (21). Their objective was to identify when, where and why traffic collisions
occurred. After their review they were able to make two interesting statements; the
traffic environments of dense urban areas appear to be safer than the lower-volume
environments of the suburbs and less-forgiving design treatments - such as narrow
lanes, traffic-calming measures, and street trees close to the roadway - appear to enhance
a roadways safety performance. (21) The reasoning given for this second observation is
that less-forgiving designs provide drivers with a distinct idea of where the edge of the
roadway is while providing them with a clear sense of what a safe travel speed is. The
authors conclude their review by stating that there has been little research done on
understanding why pre-crash behaviors occur based on the environment in which they
occur.
Researchers from University of Massachusetts Amherst used a driving
simulator, real drivers and a static evaluation to study the difference in perceived speed
versus actual speed. (22-24) The key findings were that speeds perceived by drivers in
the field were similar to the speeds perceived while driving in a replicated driving
simulator environment. The static evaluation, while only consisting of images and not
videos, examined the extent to which pavement markings, functional classifications and
barriers had on drivers speed selection.
15
CHAPTER 3
METHODOLOGY
A series of research tasks were developed based upon the existing literature and
the documented needs for further research. An experimental design was created to
evaluate the effects of clear zone size and roadside vegetation on driver behavior and run
of the road crash severity. The following section outlines the research tasks that were
employed to address the objectives and evaluate the established hypotheses.
3.1 Literature Review
The initial task of the thesis research was a literature review. This task was
initialized at the onset of the study and continued through the thesis process. Any
research regarding roadside elements in the clear zone or the clear zone in general was
studied. An added emphasis of this task was to identify research which has been
conducted using a driving simulator and identify its strengths and weaknesses in order to
build off of it. The results of this literature review were presented previously in Chapter 2
and are integrated into the remaining sections as appropriate.
3.2 Static Evaluation
A static evaluation was developed using Adobe Captivate 6. The purpose was to
evaluate how roadside vegetation and clear zone size affect drivers speed selection.
3.2.1 Static Evaluation Development

The static evaluation consisted of two main elements. The first involved showing
participants a video of either a real or virtual drive and asking them to input the speed
they would select on the given road. The participants were asked to not input how fast the
surrounding traffic was travelling at, what they thought the speed limit should be or the
16
speed at which they thought the video was captured. The second element asked
participants to rate, using a Likert scale, how strongly various factors, which are
displayed in Table 2, affect their speed selection.
Table 2 Factors Included within Assessment of Driver Speed Selection Process

Presence of
Presence of Guard
Lane Width Vehicles Pedestrian Activity
Rails/Barriers
Following You
Internal Distractors
Posted Speed Known Presence of
Weather (Radio, Cell phone,
Limit Police Enforcement
GPS, etc.)
Shoulder Density of Roadside Intersection
Time of Day
Width/Type Vegetation Frequency
Presence of
Presence of Proximity of Roadside
Oncoming Pavement Quality
Passengers Objects
Vehicles
3.2.2 Development of Simulated Environments
Four virtual environments were developed using Realtime Technologies Inc.
(RTI) driving simulation platform software. The virtual environments were designed for
the driver simulator located in the Arbella Insurance Human Performance Lab at
University of Massachusetts Amherst. The environments consist of varying levels of
roadside stimuli and clear zone size, Figure 4.
17
Figure 4 Driving
riving simulator scenarios depicting different clear zone size/vegetation
density combinations shown to participants during the static evaluation.
3.2.3 Field Video Capture
Field video was captured in Amherst, Pelham and Belchertown on a clear Sunday
morning when the surrounding traffic was extremely sparse.. A camera was mounted on a
tripod and held out of a moon roof at an angle that mimicked the viewpoint of the driving
simulator videos. While the speed limit of the selected roads varied, a concerted effort
was made to drive at a constant speed, around 40 mph, during video capture so that the
static evaluation results were not skewed. Figure 5 depicts the route taken and labels the
points of video capture and Table 3 outlines the specific clear zone/roadside vegetation
configuration of each location.
18
Figure 5 The route taken during field video capture and the locations of the clear
zone/vegetation combinations.
Table 3 Clear Zone Size and Vegetation Densities Associated with Figure 5.
Point Clear Zone Size Vegetation Density
A Medium Dense
B Large Dense
C Small Dense
D Medium Dense
E Large Sparse
F Small Dense
G Large Sparse
H Large Dense
19
3.3 Real World Validation of Static Evaluation Results
Eight segments of roads were identified as having similar configurations to the
videos used in the static evaluation and selected for field data collection. Along these
roads, at least 39 vehicle speeds were captured. Vehicle positioning was observed as well
using a video camera. It was hypothesized that the results of this field analysis would
mimic the results found in the static evaluation.
3.3.1 Field Validation Observation Spots

Two locations for each clear zone/vegetation density combination were selected
in the surrounding area resulting in observations at eight different sites. Table 4 presents
the eight locations that were selected for the field validation of the driving simulator
study. At two locations, multiple observations took place concurrently at different spots
on the road where the roadside conditions changed.
Table 4 Locations of the Field Validation Observation Spots

Speed Lane
Clear Vegetation Sample
Location Limit Width
Zone Size Density Size
(mph) (ft)
Amherst Rd - Pelham Small Dense 51 40 10
Pomeroy Ln Amherst Small Dense 40 35 12
Feeding Hills Rd Westfield Medium Dense 82 40 11
Bay Road Amherst Medium Dense 39 40 12
Route 202 SB - Belchertown Large Dense 47 50 12
Route 9 SB Amherst Large Dense 53 50 12
West St Amherst Large Sparse 57 40 12
Route 9 SB Amherst Large Sparse 53 50 12
3.3.2 Standard Data Collection Method
At each location, a tripod was set up to collect video of the oncoming direction of
traffic for the purpose of obtaining lateral positioning. A LiDAR speed gun, graciously
on loan from the Massachusetts State Police, was used to collect vehicle speeds. When
possible, video and speeds were captured from the inside of a vehicle for the purpose of
20
being inconspicuous to the drivers. In other locations, such placement of a vehicle was
not possible, Figure 6. While at each location, the posted speed limit and lane widths
were noted for data analysis.
Figure 6 A setup where video could not be obtained from the inside of a vehicle.
3.3.3 Linked Site Data Collection
At two locations, two observers were utilized to capture data at two different
points along the road, each with a different clear zone/vegetation density. The purpose of
a linked observation was to eliminate variables such as speed limit, lane width, shoulder
type and driver demographics.
3.3.4 Data Analysis Technique

Vehicle speeds were spoken into the camera so they could be transcribed later.
Lateral positions were obtained by using a transparency and VLC media player to gauge
where a vehicle was positioned on the road. Figure 7 displays how lateral positions were
obtained from the collected video data.
21
Figure 7 Computerized portrayal of the lateral positioning analysis.
3.4 Crash Data Analysis
ysis
In accordance with object

objective four, the UMass Safety Data Warehouse was
w utilized
to investigate
nvestigate the patterns and trends associated with run-off-the-road crashes,
crashes Figure 8.
The Data Warehouse inn combination with the MassDOT roadway inventory provided
information on the details of all crashes that occurred within the Commonwealth of
Massachusetts between 2007 and 2010

2010.
Figure
igure 8 UMass Safety Data Warehouse.
22
3.4.1 Filtering of Run-off-the-road Crashes from All Crashes
To analyze the severity and trends associated with run-off-the-road crashes, a
methodology for defining these crashes had to first be developed. Figure 9 displays a
flow chart of how a run-off-the-road crash was defined from the crash data.
Figure 9 Flow chart of methodology to isolate run-off-the-road crashes.

Crashes without valid x and y coordinates were removed for the purpose of
ensuring data quality and that all crashes could be geolocated. A total of 29,651 crashes
from the years 2007 to 2010 were obtained, ensuring an adequate sample size for multi-
variable comparisons.
3.4.2 Software used for Analysis and Plotting of Crash Data

Microsoft Access was utilized to link each individual crash with the MassDOT
roadway inventory using the unique identifier for the roadway ID. Then the pivot table
feature in Microsoft Excel was used to investigate the data. Next, Minitab 16 was used to
perform the statistical testing. Finally, graphs were generated using OriginPro 8.6.
23
CHAPTER 4
FINDINGS AND RESULTS
The objectives and hypotheses are presented in Table 5 and are addressed in the
following chapter. Vehicle speed and lateral positioning were the primary metrics used to
evaluated drivers response within the static evaluation and in the field. This chapter
presents the crash data analysis and the results from both phases of the experiment.
Table 5 Objectives and Hypotheses

Objective Hypotheses
(A) The density of trees along the roadside will
Understand the effect that have a measurable effect on operating speeds
1 roadside vegetation density has (B) Speeds observed in the real world will be
on vehicle speeds similar to the speeds surveyed during the static
evaluation.
(A) A smaller clear zone may cause drivers to
position their vehicle further from the edge of the
Understand how the size of the road
2 clear zone affects driver
behavior. (B) A smaller clear zone may cause drivers to
accept lower operating speeds in fear of running off
the road.
Determine the effect that

various roadway environment (A) Participants will not rate vegetation density as
3
elements have on drivers influential to their speed selection
speed selection.
(A) A direct correlation will exist between the

Utilize crash data to investigate
4 severity of run-off-the-road crashes and the type of
run-off-the-road crashes.
roadside elements involved during the crash.
The first section of this chapter provides the results of the static validation and the
following sections describe the results of the field data validation and the crash data
analysis, respectively. The main focus of the analysis was to evaluate driver behavior in
the four different clear zone/vegetation density combinations selected.
24
100 participants participated in an online static evaluation of their speed selection
based on roadway characteristics. While demographic data was not collected, the sample
did consist of males and females, of varying ages and geographic locations in the United
States. Participants were asked to choose the speed they would drive after watching either
1 of 9 real world videos, or 1 of 4 driving simulator videos. After the videos, they were
asked how strongly 16 roadway environment factors affect their speed choice.
4.1.1 Factors Influencing Speed

Participants were asked to rate 16 aspects of the driving environment on the
magnitude that each one affects their speed choice while driving. The participants were
given five options: Very Little, Little, Neutral, Strongly and Very Strongly. For analysis
purposes, the response options were given a numerical value of 1 to 5 (lowest to highest).
Table 6 presents an ordered list of the results with factors listed from most influential to
least influential.
Table 6 Ranking of Static Evaluation Factors Influencing Speed

Factor Influencing Speed Average Rating
Known Presence of Police Enforcement 4.28
Pedestrian Activity 4.25
Weather 4.25
Pavement Quality 3.91
Intersection Frequency 3.91
Proximity of Roadside Objects 3.83
Lane Width 3.78
Posted Speed Limit 3.76
Shoulder Width/Type 3.68
Time of Day 3.51
Presence of Oncoming Vehicles 3.45
Presence of Guard Rails/Barriers 3.27
25
Presence of Vehicles Following You 3.27
Presence of Passengers 3.22
Density of Roadside Vegetation 3.16
Internal Distractors (Cell Phone, Radio, etc.) 2.98
Known police enforcement was narrowly rated as the most influential factor
influencing speed followed by pedestrian activity and weather. Participants selected
internal distractors, such as cell phones and radios, as the least influential on their speed
selection.
The standard deviation of each factor as calculated to investigate the extent to
which variability existed within the responses. Table 7 displays an ordered list of the 16
factors from most variability to least. Density of roadside vegetation, one of the primary
foci of this thesis research, displayed the most variability while rating very low as an
influencing factor on speed choice. This supported the prediction of Hypothesis 3A that
drivers would not consciously think that roadside vegetation affects their operating speed.
Weather and pedestrian activity had the least variability and both were near the top of the
list of most influential. Proximity of roadside objects, the other primary foci, had a low
variability and scored in the upper middle of the pack with a score correlating with
Strongly.
Table 7 Standard Deviations from the Static Evaluation Ranked from Most
Variability to Least
Factor Influencing Speed Standard Deviation
Density of Roadside Vegetation 1.13
Presence of Passengers 1.06
Internal Distractors 1.01
Time of Day 1.00
Presence of Oncoming Vehicles 1.00
Presence of Vehicles Following You 0.94
Posted Speed Limit 0.93
Known Presence of Police Enforcement 0.93
26
Presence of Guard Rails/Barriers 0.93
Shoulder Width/Type 0.91
Pavement Quality 0.83
Proximity of Roadside Objects 0.82
Lane Width 0.77
Intersection Frequency 0.73
Pedestrian Activity 0.69
Weather 0.64
The influence of Density of Roadside Vegetation on speed choice was compared
to the influence of the other 15 choices, Table 8. Despite being ranked as the second least
influential of the 16 factors, roadside vegetation density was still ranked as more
influential than each other factor by at least 6 percent of respondents. 51 percent of
respondents indicated that the proximity of roadside objects was more influential to their
speed choice while 41 percent said that they had equal effects.
Table 8 Comparison of Responses for Density of Roadside Vegetation versus Other

Speed Influencing Factors
Density of Roadside Vegetation
Factor Influencing Speed > =
Internal Distractors 39 30 69
Presence of Guard Rails/Barriers 34 33 67
Presence of Passengers 33 32 65
Presence of Vehicles Following You 28 35 63
Shoulder Width/Type 19 38 57
Time of Day 24 31 55
Presence of Oncoming Vehicles 29 24 53
Lane Width 17 34 51
Proximity of Roadside Objects 8 41 49
Posted Speed Limit 21 25 46
Pavement Quality 13 31 44
Intersection Frequency 13 30 43
Known Presence of Police Enforcement 8 22 30
Pedestrian Activity 6 23 29
Weather 7 21 28
4.1.2 Speed Decisions
27
1300 speed choices were collected in response to nine real world videos and four
driving simulator videos. The maximum speed selected was 90 mph, which was omitted
from the data analysis because it was an extreme outlier. When omitting that value, the
next highest speed was 75 mph. The lowest speed selected was 20 mph and 45 mph was
the most commonly selected speed. Lastly, the median speed that participants chose was
also 45 mph.
4.1.2.1 Speed Decisions from Simulator Videos

Four different driving simulator videos were shown to participants throughout the
static evaluation. Drives with dense vegetation and small, medium and large clear zones
were presented along with a drive with sparse vegetation and a large clear zone. Figure
10 presents the average speeds that participants selected after watching the four videos of
the simulated drive. While the drives with large clear zones presented higher averages
than the medium and small clear zones, these differences were not statistically
significant.
Figure 10 Average speed selected after watching the four driving simulator drives.
4.1.2.2 Speed Decisions from Real World Videos
28
Nine different videos captured from real roads were presented to participants.
Participants responded with an average speed of 41.9 mph for roads with small clear
zones and dense vegetation, Figure 11. The videos of a medium clear zone and dense
vegetation elicited an average speed of 44.4 mph from participants. Within the large clear
zone videos, the dense and sparse vegetation elicited responses of 52.9 mph and 48.7
mph, respectively. The difference in average speed was statistically significant between
each combination. The results from the static evaluation supported Hypothesis 2B as
participants did, in fact, select a lower speed than for the small and medium clear zones.
Hypothesis 1A was not supported as participants were not affected by roadside vegetation
as higher speeds were selected on roadways with dense vegetation than roadways with
sparse vegetation.
Figure 11 Average speed selected after watching the nine videos of real drives.
29
4.1.2.3 The Effect of Utility Poles within Close Proximity of the Roadway
One location used within the evaluation presented an interesting opportunity to
investigate the effect that utility poles within close proximity of roadway has on speed
selection. Figure 12 depicts the scenario that participants were presented with during the
static evaluation.
Figure 12 Static evaluation

valuation scenario with directional utility pole
ole presence.
In one direction there is an open field with absolutely nothing posing a hazard to
vehicles running off the road. In the other direction, utility poles are approximately four
feet from the edge of the road spaced appro

approximately
ximately 150 feet apart. When presented with
the video of the road with no utility poles, participants responded with an average speed
of 50.0 mph. When utility poles were in close proximity of the roadway, participants
selected a speed two miles per hour lower, a statistically significant difference (paired t-
t
test, p=0.003). The individual responses of the 100 responses were examined for the
utility pole scenario, Table 9.. 36 percent of respondents indicated that they would select a
lower speed on the road with the utility poles. 49 percent selected the same speed for both
30
scenarios and only 15 percent of respondents selected a higher speed when utility poles
were present.
Table 9 Within Subjects Speed Comparison Measuring the Effect of Utility Poles on
Speed Choice
Average Speed Average Speed
Percentage of Difference
Response Type on Road w/o on Road w/
Respondents (mph)
Poles (mph) Poles (mph)
Higher Speed
with no Utility 36 54.6 45.6 9
Poles
Equal Speed 49 48.9 48.9 0
Lower Speed
with no Utility 15 43.0 50.9 -7.9
Poles
4.2 Validation of Static Evaluation Findings
Eight locations were selected for analysis based on their clear zone size and
adjacent vegetation density. Due to the differences in lane widths and posted speed limits,
the lateral positions and speeds were normalized to compare different locations. 12 foot
lanes and 40 mph speed limits were used as the baseline as they were the most commonly
found at the observation sites. The equations below detail the normalization process:
12

40

The percent of drivers violating the speed limit was an additional measure that
was used to compare speeds across different locations.
31
4.2.1 Speed Results
To negate the effect that higher speed limits had on vehicle speeds, the results
were normalized as previously described. Figure 13 displays the averaged normalized
speeds across the four combinations of clear zone and vegetation density. Small clear
zone/dense vegetation roads had similar normalized speeds to roads with a large/sparse
configuration. Medium/dense and large/dense road configurations both had significantly
lower normalized average speeds. The percent of drivers violating the speed limit was
also examined as another measure to evaluate the speeding tendencies of the road. The
percentage of observed drivers violating the speed limit was strongly correlated with the
average speed observed. The findings of the field data collection did not support
Hypothesis 1A or 2B as normalized speeds were extremely similar for the four clear
zone/vegetation density combinations.
Figure 13 Average normalized speeds from the field study

4.2.2 Lateral Positioning
Lateral positioning of vehicles was captured using video data and is presented in
Figure 14. With the exception of small clear zones, drivers positioned their vehicle closer
32
to the edge of the road as the proximity of roadside objects decreased which supports
Hypothesis 2A.
Figure 14 Average normalized lateral positions measured by the distance in feet

from the edge line.
4.2.3 Similar Location as Static Evaluation
Four locations which were featured in the static evaluation were studied in the
field as well. Each of the four clear zone/vegetation density combinations were
represented during the field validation, Figure 15. Two of the locations Bay Road and
West St exhibited speeds statistically similar to what participants indicated during the
static evaluation. Pomeroy Lane, a road with a small clear zone and dense vegetation, had
lower speeds than in the static evaluation. Route 202, a densely vegetated road with a
large clear zone, had higher speeds in the field than respondents indicated during the
static evaluation. Hypothesis 1B, which states that speeds observed in the real world will
be similar to the speeds surveyed during the static evaluation, was supported as two of
the locations exhibited statistically similar speeds and the other two locations had average
speeds in the same range as were selected during the static evaluation.
33
Figure 15 Average speeds of the four locations studied in both the static evaluation
and field validation.
4.2.4 Linked Locations
Two locations were observed as a pair for the purpose of removing all external
variables such as differences in speed limits, shoulder types, lane widths and driver
demographics. In the case of one scenario the road transitioned from a medium clear zone
to a large clear zone, all the while having dense vegetation density, Figure 16. While in
the medium clear zone, drivers positioned their vehicle 4.4 feet from the edge of the road
and while in the large clear zone drivers were a half foot closer to the edge of the road.
Drivers reduced their speed by 2.7 miles per hour when transitioning from the medium to
large clear zone. Both differences were statistically significant.
34
Figure 16 Linked
inked location on Route 202 Southbound in Belchertown, MA
Route 9 in Amherst, MA was another location which was studied as a linked
observation. The road transitioned from sparse vegetation to dense vegetation all the
while having a large clear zone, Figure 17.. When vegetation was present, drivers
positioned their vehicle statistically significantly further from the edge of the road. The
difference in speed between the two vegetation densities was not statistically significant.
35
Figure 17 Linked
inked location on Route 9 in Amherst, MA
The UMass Safety Data Warehouse was utilized to investigate the patterns and
trends associated with run

run-off-the-road
road crashes. The crashes were grouped into three
categories for the purpose of analysis. Fatal Injury and Non

Non-fatal
fatal injury - Incapacitating
crashes
shes were grouped as Serious. Non
Non-fatal injury Non-Incapacitating
Incapacitating and Non-fatal
Non
injury Possible were grouped as Minor. No Injury crashes were labeled as Property
Damage Only. Crashes with injury statuses of: Not Applicable, Not Reported,
Reported
orted by Invalid, and Unknown were not included in the analyses as no
conclusion could be drawn regarding the crash severity.
4.3.1 Gender and Time of day

The differences between crash injury rates for males and females
es during both night and
day were quantified and are shown in Figure 18. Run-off-the-crashes,

crashes, which occurred
during the night, had a higher chance of serious than those that
hat occurred during the day,
day
36
though the difference is narrowly insignificant (p=0.06). Male drivers were found to have
a higher rate of Property Damage Only crashes than female drivers for both day and night
driving.
Figure 18 ROR injury rates as a function of both gender and time of day.
A chi-squared test was used to example the statistical significance for the data
corresponding to Figure 18. There were no statistical differences when comparing injury
severities of crashes involving females, Table 10. The difference in serious injury crashes
for night versus day for both males and females was narrowly statistically insignificant
(p=0.06).
Table 10 P-Values from Chi-squared Test Associated with Data Displayed in Figure 18
Serious Minor No Injury

Female: Night vs. Day 0.06 0.31 0.06
Male: Night vs. Day 0.06 0.00 0.00
Day: Male vs. Female 0.02 0.00 0.00
Night: Male vs. Female 0.52 0.00 0.01
*Statistical significance indicated by a P-value 0.05 and a shaded box.
Men experience a higher serious injury rate than females during the day (p=0.02).
Men likely experience a serious injury rate due to their riskier driving behavior and
excessive speeding. Crashes at night probably result in more fatalities because drivers do
not have the sight distance to avoid hazards in comparison to driving during the day.
Also, alcohol is more likely to play a factor in the occurrence of these nighttime crashes.
37
4.3.2 Driver Age
The severity of run-off-the-road crashes was also investigated with respect to the
age of the driver, Figure 19. It was found that crashes involving older drivers (60+ years
of age) were more likely to result in a serious injury than crashes with younger (<22 years
of age) or middle age (23-59 years of age) drivers. Older drivers were also more likely to
sustain minor injuries than young and middle age drivers.
Figure 19 ROR injury rates versus driver age.

Table 11 P-Values from Chi-squared Test Associated with Data Displayed in Figure 19.
Serious Minor No Injury
Young vs. Old 0.00 0.00 0.00
Young vs. Middle 0.02 0.01 0.18
Middle vs. Old 0.09 0.00 0.00
While it can be concluded that the severity of these crashes is higher when it
involves an older driver, no conclusions can be drawn regarding the frequency of these
crashes as there is no reliable data source which can describe the vehicle miles travelled
(VMT) for each age group. A similar trend is seen when comparing the fatality rate with
both gender and age of the driver, Figure 20. Middle aged male drivers had statistically
higher serious crash rates when compared to females from the same age group. Young
males and females both were statistically less likely to sustain serious injuries from a
ROR crash than older drivers. All statistical comparisons can be seen in Table 12.
38
Figure 20 Crash injury rates as a function of both driver age and gender.
Table 12 P-Values from Chi-squared Test Associated with Data Displayed in Figure 20.
Serious Minor PDO
Female: Young vs. Old 0.01 0.20 0.01
Female: Young vs. Middle 0.26 0.01 0.03
Female: Middle vs. Old 0.05 0.00 0.00
Male: Young vs. Old 0.03 0.00 0.00
Male: Young vs. Middle 0.03 0.32 0.89
Male: Middle vs. Old 0.34 0.00 0.00
Young: Male vs. Female 0.14 0.00 0.00
Middle: Male vs. Female 0.02 0.00 0.00
Old: Male vs. Female 0.94 0.09 0.11
4.3.3 Most Harmful Roadside Objects
For the purpose of influencing roadside design, it is important to investigate the level of harm each
level of harm each roadside object presents. To determine the harmfulness of each roadside object, the crash
roadside object, the crash injury rates for eight ROR indicators were determined by their presence within one of
a
presence within one of the four possible sequences of events, A statistically significant difference (*)
when compared to all run of the road crashes (grey) was achieved for P<0.05.
Figure 21. These fatality rates were then compared with the fatal accident rate for all
ROR crashes. When a collision with a tree occurs, the chance of a serious injury is
statistically higher (p = 0.00) supporting Hypothesis 4A. It was also found that collisions
with light poles, ditches and mailboxes have a statistically lower chance of a serious
injury when compared with the serious injury rate of all ROR crashes.
39
a
A statistically significant difference (*) when compared to all run of the road crashes (grey) was
achieved for P<0.05.
Figure 21 Crash injury rates for various run off the road crash indicators present in
one of the four possible sequences of events within a crash report.
4.3.4 Guardrails
Collisions with guard rails, a common barrier placed alongside roadways to
prevent ROR crashes, were examined to evaluate their effectiveness. ROR guard rail
crashes (Figure 22) demonstrated a higher serious injury rate in comparison with ROR
crashes not involving a guard rail. However, this comparison is not significant (p=0.14).
Figure 22 The effect of hitting a guard rail on ROR crash severity.

4.3.5 Speed Limit and Shoulder Width
The severity of ROR crashes was analyzed as a function of the posted speed limit
as speeding is one of the factors that increase the severity of ROR crashes, Figure 23.
For posted speed limits ranging from 20 to 65 mph, an upward trend was observed
reaching serious injury rates of approximately 9 percent. A similar was trend was seen for
40
minor injuries, as the speed limit increased, so did the chance of a minor injury. The
chance of a PDO crash occurring decreased as the speed limit increased.
Figure 23 Posted speed limit versus crash injury rate.

The width of the right shoulder for three speed groups was compared to the crash
severity to determine if the extra room allowed drivers to regain control and lower the
severity of the crash, Figure 24. However, very few statistical differences were observed.
A large right shoulder (10+ feet) results in a higher serious injury rate than a 0-4 foot
shoulder for roads with a speed limit of 40-50 mph.
41
Figure 24 Crash injury rates as a function of the size of the right shoulder and
posted speed limit.
4.3.6 County
The serious and minor injury rates were also compared at the county level versus
the state average to see if different regions experienced different patterns regarding run-
off-the-road crashes. 9 of the 14 counties experience serious injury rates that are not
statistically different from the overall serious injury rate. However, five counties have
statistically significant differences. Middlesex County, home of approximately 25 percent
of Massachusetts population and the 23rd most populated county in the nation, has a
ROR serious injury rate of only 6.1 percent. Suffolk County however, the county which
houses Boston, has a ROR fatal accident rate of 8.6 percent. The other three differences
are displayed in Figure 25.
42
Figure 25 County map of run off the road serious injury rates as compared to the
statewide average.
Minor injury rates were also compared for the 14 Massachusetts counties, Figure
26. Again, Middlesex County experienced injury rates lower than the state injury rate.
For minor injuries, Suffolk County does not show any significant difference from the
state rate. Worcester, Plymouth, Bristol and Barnstable counties all have minor injury
rates higher than the state average.
43
Figure 26 County map of run off the road minor injury rates as compared to the
statewide average.
44
CHAPTER 5
DISCUSSION
A total of 100 participants, all of whom were licensed drivers, participated in the
static evaluation. This type of evaluation allowed for insight into the drivers mindset as
they determined their operating speed for each video of a real road or driving simulator
scenario. The benefits of an evaluation distributed via internet were a large sample size
and a wide geographic distribution. The following section discusses the results and
findings documented in the previous chapter.
5.1.1 Simulation Videos

Drivers were presented with four driving simulator videos where the size of clear
zone and density of roadside vegetation varied. It was hypothesized that the small clear
zone and dense vegetation would cause drivers to select a slower speed than when
presented with trees sparse and a large clear zone. However, the average speeds selected
by participants did not vary as expected between the four driving simulator videos.
Statistically, the same speed was chosen for each simulator video. The results are not
supported by field validation concluding that either participants did not have enough time
to assess the conditions from the 10 second clip or they did not feel fully immersed
because of the low quality video. Thus, this should be considered as a preliminary
investigation into speed-related driving simulator environments and should be explored
further when better software is available to capture more realistic driving simulator
drives.
45
5.1.2 Real World Videos
Drivers were shown nine videos of roadways with the same clear zone/vegetation
density scenarios as the four driving simulator drives. When subjects were presented with
dense trees close to the road, they selected statistically lower speeds than when the trees
were further from the road. An abnormally high average speed was selected in the large
clear zone/dense vegetation video combination. This spike in speed can be attributed to at
least 12 foot wide lanes and large, paved shoulders on the road whereas in the large clear
zone/sparse vegetation video combination, shoulders were either gravely or non-existent.
Since participants rated lane width and shoulder width/type as Strongly affect their speed
choice, the abnormally high average speed in large clear zone/dense vegetation video
combination was likely influenced by shoulder width/type. Overall, the results
determined from real world videos captured during the static evaluation were as
hypothesized.
The case study involving utility poles in close proximity to the road was
enlightening as it confirmed that drivers change their speed when roadside objects are in
close proximity to the roadway. Utility poles pose a large hazard in the case of run-off-
the-road crashes so it is comforting that drivers are aware and do take caution.
5.1.3 Limitations of Static Evaluation

While an online static evaluation has many advantages such as large sample size
and wide geographic distribution, it lacks the level of immersion that driving creates and
many variables cannot be held constant as could be in a driving simulator. Best efforts
were made to choose similar roads but often times the speed limit, lane width, shoulder
type and shoulder width varied.
46
Ideally, locations could have been found where there was a constant speed limit,
constant lane widths and similar shoulder widths/types. However, this was not possible so
the speeds and lateral positions had to be normalized for comparison.
5.2.1 Observed Speeds

The locations with small clear zones and dense vegetation had the highest
normalized speeds but also had the highest percentage of drivers violating the posted
speed limit, nearly 75 percent. For roads with medium and large clear zones containing
dense roadside vegetation, only ~60 percent and ~50 percent of drivers were speeding,
respectively. The roads with large clear zones and sparse vegetation also had high
normalized speeds and a high percentage of speeders. This may be explained by the fact
that speed limits are placed abnormally low on densely vegetated roads with small clear
zones for the purpose of safety.
5.2.2 Lateral Positioning and Linked Locations

The lateral positioning of vehicles was captured for vehicles traveling along roads
where either the vegetation density changes for a constant clear zone or where the clear
zone changes but the vegetation density is held constant. As the vegetation density
increased for a large clear zone, driver positioned their vehicles further from the edge of
the road. With the exception of small clear zones, drivers positioned their vehicle closer
to the edge of the road as the proximity of roadside objects decreased. The exception of
small clear zones with dense vegetation could be due to oncoming vehicles. Drivers
might have felt trapped and positioned their vehicle closer to the edge to gain extra space.
Additionally, one of the small clear zone roads, Amherst Rd in Pelham, has 10 foot lanes
47
which could cause drivers to be extra cautious with regard to oncoming vehicles. While
best efforts were made to ensure uniformity, shoulders varied from location to location
and could have influenced lateral positioning. Drivers might have changed their
positioning based on the quality and size of the shoulder in addition to the size of clear
zone and density of roadside vegetation.
5.2.3 Similar Location as Static Evaluation

Speeds were captured on four of the same roads presented in the static evaluation.
For medium clear zone/dense vegetation and large clear zone/spare vegetation
combinations, static evaluation participants chose a speed statistically the same as
observed in the field.
However, drivers who encountered small clear zones with dense vegetation drove
at statistically lower speeds than static evaluation participants chose. Drivers immersed
on a road with a high density of trees may make an unconscious decision to drive slower
due to the high rate of visual flow in the surrounding environment. This level of
immersion while driving cannot be replicated by viewing video clips.
A statistical difference was seen between drivers and static evaluation participants
for the road with a large clear zone and dense vegetation. Static evaluation participants
stated slower speeds than observed field speeds. The ability to gauge the size of a clear
zone from a video versus while driving could explain this difference. These differences
could be furthered explored by replicating the road scenarios in a driving simulator in
order to hold all other variables constant.
48
5.2.4 Limitations of Field Validation
The original intent was to separate the vegetation density into dense, moderate
and sparse designations. While in the field, defining a site as having dense vegetation or
moderate vegetation is very subjective. Defining the clear zone size is much easier but
still presents a challenge as the clear zone often changes along the same segment of
roadway. While there were many locations with sparse or dense trees, locations with
consistent moderate vegetation density were challenging to identify. For improved
accuracy, the final analysis did not include sites with moderate vegetation density.
29,651 run-off-the-road crashes occurring in Massachusetts from 2007 to 2010
were analyzed. While it was possible to investigate the outcome and locations of each
crash, it was not possible to determine a cause of each crash. Many conclusions were
made regarding the injury rates of ROR crashes. Unfortunately, the frequency at which
these crashes occur within demographic subset as accurate Vehicles Mile Traveled
(VMT) data is not available. Conclusions could have been drawn using Registry of Motor
Vehicle (RMV) data on the number of licensed drivers. However, those conclusions
would be skewed as not all people with a drivers license drive the same amount. For
example, somebody living in urban Boston may only drive once a week whereas
somebody living in northwestern Massachusetts may have to drive 25+ miles per day.
5.3.1 Gender and Time of Day

While there were no significant differences for serious and minor injury rates
between night and day for women, men more were likely to sustain a minor injury when
they were involved in a run-off-the-road at night. Men have been proven to engage in
49
riskier driving behavior such as speeding or drunk driving, these risky behaviors become
more frequent at night and when there are fewer drivers on the road. It was expected that
there would be a significant rise in serious injury rates at night, but the difference was
narrowly not statistically significant for both males and females (p=0.06).
Males also were likely to experience no injury and only property damage whereas
females were more likely to sustain minor injuries as a result of ROR crashes. Societal
pressures may play a role as in males may be more likely to refuse medical treatment for
minor injuries whereas females would not feel this societal pressure.
5.3.2 Driver Age

Younger drivers (16-22) and middle aged drivers (23-59) had similar serious and
minor injury rates when involved in a ROR crash. This was unsurprising as the injury rate
does not take into account the frequency of these types of crashes but rather the outcome
when they occur.
Older drivers were more likely to sustain serious injuries than younger drivers and
more likely than younger and middle aged drivers to sustain minor injuries. The ability of
the body to recover from injuries is the likely the cause of this difference. As the body
ages, it loses its ability to overcome traumatic events. Additionally, reaction times
increase as you get older. When a ROR crash, a younger or middle aged driver may be
able to steer or hit the brakes a little sooner lessening the severity of the crash.
5.3.3 Most Harmful Roadside Objects

The injury rates of every roadside object which may be struck during a run-off-
the-road crash was compared to the overall ROR crash injury rate. The results from this
analysis confirmed the initial hypothesis that roadside trees were extremely dangerous to
50
drivers. 10 percent of crashes involving a tree resulted in a serious injury. Conversely,
only 55 percent of crashes involving a tree only involved property damage. It is safe to
conclude that trees are the most harmful roadside object. The second most harmful
roadside would most likely be utility poles. While the chance of a serious injury is not
significantly higher for utility pole crashes, the chance of a minor injury is significantly
greater.
The common factor for utility poles and trees is that they are both immovable
objects. Most utility poles are at least 18 inches in diameter and roadside trees can vary
from anywhere from 4 to 24+ inches in diameter. While the data was not available within
the UMass Safety Data Warehouse, it is likely that the severity of the crash increases as
the diameter of the fixed object increases. This speculation is supported by the fact that
mailboxes and light poles, each having smaller diameters and constructed of more
frangible materials, have statistically lower serious and minor injury rates.
5.3.4 Guardrails
It was difficult to evaluate the efficacy of guardrails in lessening the severity of
run-off-the-road crashes as it is rare that a ROR crash still occurs after a collision with a
guard rail. The primary of purpose of rigid roadside barriers is to keep drivers from
leaving the roadway and guard rails seem to do just that. While not within the scope of
this thesis research, an in depth analysis of guardrails could determine how well they
prevent ROR crashes from occurring, this is discussed further in the Future Work section.
5.3.3 Speed Limit and Shoulder Width

Expectedly, the serious and minor injury rates increased as the posted speed limit
increased. While it would be very informative to know the speed at which every ROR
51
crash occurred, a thorough and costly investigation at the site of each crash would be
required. It is likely that the severity of the crash increases as the drivers speed increases.
An interesting comparison would be whether the absolute speed at which the crash
occurred or the amount over the speed limit the crash occurred more greatly affected the
severity. An example of this type of study would involve a comparison of a crash at 70
mph on an interstate and a crash at 55 mph on a windy rural road to see which would be
more dangerous.
Injury rates were also investigated with respect to the shoulder width at constant
speeds. Findings demonstrated that shoulder width was more likely to have an effect on
the frequency of these crashes rather than influencing the severity.
5.3.4 County
Finally, ROR injury rates were compared across the 14 counties in Massachusetts.
While counties are not an ideal way to geographically group crashes, they can provide a
generalized idea as to the crash patterns in the area. Suffolk County, the most urban
county in Massachusetts had a serious injury rate of 8.5%, which was significantly higher
than the state rate. A comparison between crash severity and seat belt usage demonstrated
that Worcester County, one of the highest observed seat belt usage rates during the
annual Massachusetts Safety Belt Usage Observation Study (22), had the highest serious
and minor injury rates in the state. Bristol County, which had one of the lowest belt usage
rates, had a serious injury rate around the state average and a minor injury rate
statistically lower than the state average. Therefore, it appears that belt usage does not
appear to be a strong factor in determining the severity of ROR crashes.
52
CHAPTER 6
CONCLUSIONS
This research investigated the effect of clear zone size and surrounding vegetation
on driver behavior using an online static evaluation, a field validation of the static
evaluation findings and a crash data analysis, which focused on run-off-the-road crashes
in Massachusetts. The results and discussion presented in the previous chapters resulted
in a number of conclusions, which are outlined within this chapter.
The online static evaluation included 100 participants who selected speeds based
on roadway characteristics. After watching 13 short videos and selecting an operating
speed for each road, the participants were asked to rate how strongly roadway
environment factors affected their speed choice. The results are as follows:
o Respondents selected speeds that were statistically similar for the four driving
simulation video clips despite the differences in clear zone size and vegetation
density.
o Responses for real world videos showed as the size of the clear zone increased, so
did the speed at which respondents said they would travel.
o When utility poles were in close proximity of the roadway, the average speed
selected was 2 miles per hour lower than the exact same road without utility
poles.
53
o The known presence of police enforcement was rated as the most influential on
respondents speed choice. Proximity of roadside objects was rated as having a
strong influence whereas roadside vegetation density was rated as having a
neutral effect.
Eight locations containing similar roadside configurations as depicted in the static
evaluation were studied in the field to see if real drivers exhibited similar behavior as
respondents indicated. Video data was captured to see if the presence of trees affected the
lateral positioning of vehicles on the roadway. The data collected in the field indicated
the following:
o Two of the four locations studied in the field and featured in the static evaluation
had statistically similar speeds as what respondents indicated. One location had
lower speeds and the last location had higher speeds.
o With the exception of roads with small clear zones, vehicles drove closer to the
edge as the clear zone size increased. It was hypothesized that narrower lanes on
the roads with a small clear zone may have caused the exception.
o At the two linked locations with multiple observers, drivers positioned their
vehicle farther from the edge line as vegetation density increased for a constant
clear zone and decreased their distance from the edge line as clear zone size
increased for constant vegetation density.
54
The UMass Safety Data Warehouse was utilized to investigate the patterns and
trends associated with 29,651 run-off-the-road crashes occurring in Massachusetts from
2007 to 2010. The key crash data analysis findings are as follows:
o While the difference in the severity of crashes occurring at night vs. day was not
significant, the data would suggest that nighttime ROR crashes are more serious
in nature.
o Men are more likely to escape injury during ROR crashes but this difference may
be due to a reporting bias.
o Older drivers are more likely than younger and middle aged drivers to sustain
serious or minor injuries during a ROR crash.
o The injury severity of ROR crashes does not correlate on a regional level to the
seat belt usage rate observed by UMass Safe during Summer 2012.
6.4 Future Work
Despite the findings and conclusion of this research, additional research questions
remain. Some of the future research that is recommended include a validation of the field
data collected, a full scale driving simulator study and a thorough investigation of the
efficacy of guardrails at preventing ROR crashing. Each of these are briefly expanded
upon in the following sections.
6.4.1 Validation of Field Data Collection Methodology

In order to ensure that the presence of a camera and observer had minimal effects
on driver behavior, one or two locations should be observed a second time. In the future,
a camera could be mounted on a utility pole and the lane could be chalked 100 feet apart.
55
Speeds would be obtained not by direct observation, but by analyzing the video frame-
by-frame.
6.4.2 Full Scale Driving Simulator Study Using Developed Scenarios

A full scale driving simulator study could investigate the effect of subconscious
decisions on speed selection. Scenarios similar to those in the static evaluation could be
used to see if drivers selected speeds similar to those reported in the evaluation. This
study would garner more insight on which factors affect subconscious speed related
decisions. As discussed in section 5.2.1, a driving simulator would allow for the
classification of roadside vegetation density into dense, moderate and sparse rather than
just dense and sparse. Lastly, a driving simulator study could also examine the effect that
roadside vegetation has on lateral positioning and these data could be compared to the
data collected during the field validation.
6.4.3 Investigate the Efficacy of Guardrails at Preventing ROR Crashes

While ROR collisions with guardrails were examined in terms of injury severity,
a separate analysis could investigate how effective guardrails are at preventing ROR from
occurring. A secondary analysis would be to determine whether prevention of run-off-
the-road crashes would always results in less serious injuries.
56
REFERENCES
1. Benefits and Risks of Urban Roadside Landscape: Finding a Livable, Balanced
Response. Dixon, Karen K and Wolf, Kathleen L. Seattle, Washington : s.n., 2007. 3rd
Urban Street Symposium.
2. AASHTO. A Policy on Geometric Design of Highways and Streets, 5th Edition.
Washington, DC : s.n., 2004.
3. Evaluation of Highway Design Parameters on Influencing Operator Speeds Through
Casewise Visual Evaluation. Stamatiadis, Nikiforos, et al., et al. 2010, Transportation
Research Record, pp. 143-149.
4. The Restorative Effects of Roadside Vegetation: Implications for Automobile Driver
Anger and Frustration. Cackowski, Jean Marie and Nasar, Jack L. 2003, Environment
and Behavior, pp. 736-751.
5. Landscape Design in the Clear Zone: The Effect of Landscape Variables on Pedestrian
Health and Driver Safety. Naderi, Jody Rosenblatt. 2003, Transportation Research
Board.
6. The View from the Road: Implications for Stress Recovery and Immunization. Parsons,
Russ, et al., et al. 1998, Journal of Environmental Psychology, pp. 113-139.
7. The Safety Effects of Urban Principle Arterial Streetscape Redevelopment Projects
Including Street Trees: A Context-Sensitive Case Study. St. Martin, Anna J, Milton,
John C and Hallenbeck, Mark E. 2007, Transportation Research Board.
8. Landscape Improvement Impacts on Roadside Safety in Texas. Mok, Jeong-Hun,
Landphair, Harlow and Naderi, Jody. 2006, Landscape and Urban Planning, pp. 263-
274.
9. An Integrated Approach to Environmental Impact Mitigation and Safety Management:
Case Studies in the Municipality of Metropolitan Toronto. Naderi, Jody Rosenblatt and
Bahar, G B. Toronto, Canada : s.n., 1997. 13th World Meeting of the International Road
Federation.
10. Safe Streets, Livable Streets. Dumbaugh, Eric. 2005, Journal of the American
Planning Association, pp. 283-300.
11. Investigation of Median Trees and Collisions on Urban and Suburban Highways in
California. Sullivan, Edward C and Daly, James C. 2005, Transportation Research
Record 1908, pp. 114-120.
12. Parkways or Freeways: Safety Performance Linked to Corridor Landscape Types.
Mok, Jeonghun and Landphair, Harlow C. 2003, Transportation Research Board.
13. Urban Trees and Traffic Safety: Considering US Roadside Policy and Crash Data.
Wolf, Kathleen L and Bratton, Nicholas. 2006, Internation Society of Arboriculture,
pp. 170-178.
14. Roadway Safety in Rural and Small Urbanized Areas. Ossenbruggen, Paul J,
Pendharkar, Jyothi and Ivan, John. 2001, Accident Analysis and Prevention, pp. 485-
498.
57
15. Risk of Vehicle-Tree Accidents and Management of Roadside Trees. Zeigler, Andrew
J. 1987, Transportation Research Record 1127, pp. 37-43.
16. A Study of the Relationship between Driver Fatigue and Monotonous Road. Zhao,
Xiaohua, et al., et al. 2012, TRB Annual Meeting.
17. Effect of Roadside Configuration and Cross-Section on Driver's Behavior. Bella,
Francesco and Tulini, Gabriele. 2010, 89th Annual Meeting Transportation Research
Board.
18. Simulating Impacts of Curbside Street Trees on Driver Performance and Perception.
Naderi, Jody Rosenblatt, Kweon, Byoung-Suk and Maghelal, Praveen. 2006,
Transportation Research Board.
19. The Street Tree Effect and Driver Safety. Naderi, Jody Rosenblatt, Kweon, Byoung
Suk and Meghelal, Praveen. 2008, ITE Journal on the Web, pp. 69-73.
20. Scallen, Stephen and Carmody, John. Investigating the Effects of Roadway Design
on Driver Behavior: Applications for Minnesota Highway Design. s.l. : Minnesota
Department of Transportation, 1999.
21. The Built Environment and Traffic Safety: A Review of Empirical Evidence. Ewing,
Reid and Dumbaugh, Eric. 2009, Journal of Planning Literature, pp. 347-367.
22. University of Massachusetts Traffic Safety Research Program. Massachusetts
Safety Belt Usage Observation Study. 2012.
23. Office of Degree Requirements. Guidelines for Master's Theses and Doctoral
Dissertations. [Online] Graduate School - University of Massachusetts Amherst, March
2010. http://www.umass.edu/gradschool/thesis/pdf_files/Electronic_Guidelines.pdf.
24. Stanley, Budd. Auto Journal: Adventure through transportation. Blogspot. [Online]
2010. [Cited: June 19, 2012.] http://buddstanley.blogspot.com/.
25. Renn, Aaron. Urbanophile. Carmel: Leadership in Action, Part One. [Online]
February 11, 2007. [Cited: June 19, 2012.]
http://www.urbanophile.com/2007/02/11/carmel-leadership-in-action-part-one/.
26. Rahm, Eddie. Biker's Dozen. The Loops. [Online] 2012. [Cited: June 19, 2012.]
http://www.bikersdozen.com/loops/loop.asp?loop=5.
27. Hurwitz, David S. Static and Dynamic Evaluation of the Driver Speed Perception
and Selection Process. M.S. Thesis. 2006.
28. An Initial Evaluation of the Driver Speed Selection Process on Posted and Unposted
Roadways Varying in Functional Classification. Knodler, Michael A and Dulaski,
Daniel M. Las Vegas, NV : Institute of Transportation Engineers Spring Conference,
2005.
29. Speed Perception Fidelity in a Driving Simulator Environment. Hurwitz, David S,
Knodler, Michael A and Dulaski, Daniel M. Orlando, FL : Driving Simulation
Conference, 2005.
58

The Effect of Roadside Elements On Driver Behavior and Run-Off-the-Road Crash Severity

Uploaded by

Copyright:

Available Formats

The Effect of Roadside Elements On Driver Behavior and Run-Off-the-Road Crash Severity

Uploaded by

Document Information

Original Description:

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

The Effect of Roadside Elements On Driver Behavior and Run-Off-the-Road Crash Severity

Uploaded by

Copyright:

Available Formats

University of Massachusetts Amherst

The Effect of Roadside Elements on Driver

Follow this and additional works at: http://scholarworks.umass.edu/theses

Submitted to the Graduate School of the

University of Massachusetts Amherst in partial fulfillment

of the requirements of the degree of

MASTER OF SCIENCE IN CIVIL ENGINEERING

Approved as to style and content by:

THE EFFECT OF ROADSIDE ELEMENTS ON DRIVER BEHAVIOR AND RUN-

OFF-THE-ROAD CRASH SEVERITY

B.S.C.E OREGON STATE UNIVERSITY

M.S.C.E UNIVERSITY OF MASSACHUSETTS - AMHERST

Directed by: Michael A. Knodler Jr. Ph.D.

Roadside vegetation provides numerous environmental and psychological benefits to

4 FINDINGS AND RESULTS ......................................................................................... 24

1 Design Characteristics of Livable and Comparison Sections of Colonial Drive from

2 Factors Included within Assessment of Driver Speed Selection Process ...................... 17

4 Locations of the Field Validation Observation Spots .................................................... 20

5 Objectives and Hypotheses ............................................................................................ 24

6 Ranking of Static Evaluation Factors Influencing Speed .............................................. 25

7 Standard Deviations from the Static Evaluation Ranked from Most

8 Comparison of Responses for Density of Roadside Vegetation versus Other Speed

4 Driving simulator scenarios depicting different clear zone size/vegetation density

7 Computerized portrayal of the lateral positioning analysis. .......................................... 22

8 UMass Safety Data Warehouse. .................................................................................... 22

9 Flow chart of methodology to isolate run-off-the-road crashes..................................... 23

12 Static evaluation scenario with directional utility pole presence. ................................ 30

13 Average normalized speeds from the field study ......................................................... 32

16 Linked location on Route 202 Southbound in Belchertown, MA ............................... 35

17 Linked location on Route 9 in Amherst, MA .............................................................. 36

22 The effect of hitting a guard rail on ROR crash severity. ............................................ 40

23 Posted speed limit versus crash injury rate. ................................................................. 41

Due to increasing levels in the built environment, opportunities to preserve the

lower pavement temperatures and decreased emissions (1). Positive psychological

While trees provide psychological and environmental benefits, they pose a

further understand the impacts on driver performance. Additionally, there is a need to

evaluate the effects of roadside elements on driver performance as well as an analysis of

1.1 Problem Statement

As described within the American Association of State Highway and

Transportation Officials (AASHTO) A Policy on Geometric Design of Highways and

way and to remove trees.

within the clear zone on driver behavior.

1.2 Research Hypotheses and Objectives

research objectives and corresponding hypotheses were developed as outlined in the

speeds surveyed during the static evaluation.

Objective 3: Determine the effect that various roadway environmental elements

vegetation density as influential to their speed selection because it is a roadway element

that is only subconsciously noticed.

Objective 4: Utilize crash data to investigate run-off-the-road crashes and the

involved during the crash.

portrayal of the benefits and consequences of roadside trees.

2.1 Psychological Effect of Trees

A recent study was conducted by the Kentucky Transportation Center at

hypothesized that an increase in discomfort could correlate to reduced vehicle speeds as

Figure 1 The three videos shown to

spiritual or relaxation purposes. The researcher then asked people to respond to a

clear that greenery/grass/landscape appears to be fairly significant when walking for

spiritual renewal or stress relief. (5)