In his report, “Road Safety and Tree-Lined Avenues – The Experience from West-Berlin and Eastern Germany,” Vollpracht, a former road director in West Berlin, discusses tree-lined avenues in both urban and rural contexts [34]. Vollpracht identified environmental and ecological benefits as well as improved aesthetics because of roadside trees, and stated that driving behaviors could be indirectly affected by roadside trees. Nonetheless, trees were identified as a significant risk for crashes and fatalities. New guidelines in Germany call for a specified distance between trees and carriageways of certain speed limits, and where this is not possible, crash barriers must be installed. The author concluded that whenever feasible and not historically sensitive, trees should be removed from every roadside; if removal is not possible, shielding is preferred. In some locations with historical merit, alternative safety treatments may be required to accommodate the roadside trees.
State DOT Tree Removal Marketing Examples
Some state DOTs engaged in marketing campaigns to promote the positive benefits of roadside tree removal. An example of Oregon’s public campaign to utilize roadside trees to improve the states ecology is shown in Figure 13 [79]. A public safety announcement about roadside tree clearing produced by Caltrans is shown in Figure 14 [80].
Figure 13. Oregon DOT (ODOT) Tree Management Program Advertisement [79]
5 ANALYSIS OF STATE DOT TREE AND UTILITY POLE CRASH DATA Motivation
The literature review of crash data for trees and utility poles indicated a significant, widespread concern regarding trees and utility pole crashes. Because both trees and utility poles can be nearly rigid and are frequently found adjacent to the roadway, crashes may be both harmful and relatively frequent. However, most of the tree crash analysis studies available in literature focused on relatively small datasets; were conducted under varying economic, social, and political climates; and not all datasets were complete. The researchers desired to estimate the national average crash cost of tree and utility pole crashes based on average crash severity costs using crash data collected from many state DOTs through a similar time period. These parameters would provide a more robust, complete perspective of tree and utility pole crashes to ensure that crash cost estimates are representative of state and national averages. In addition, a large dataset would lead to more statistically significant conclusions regarding crash frequency and annual crash rates. This research would supplement the findings of FARS [4], which collects data for fatal crashes, and IIHS [2], which collects topographical data related to ROR crashes.
Methods and Procedures
Researchers at the Midwest Roadside Safety Facility (MwRSF) contacted state DOTs and requested information on crashes that involved a tree or utility pole over a five-year span between 2009 and 2014 (e.g., 2009-2013 or 2010-2014). Twelve state DOTs provided crash data for a total of more than 400,000 tree and utility pole crashes. Database fields provided by state DOTs are summarized in Tables 13 and 14. State DOT databases contained various parameters depending on data availability and safety interests. The parameters provided by each state varied, and no state database contained every field tracked in this study.
Table 13. Crash Data Parameters and Definitions
Crash Data Parameter Definition
Crash ID Unique case ID used to differentiate crashes; unique per state
Severity Crash severity. Converted to KABCO whenever possible. For some states, injury noted as “I” for non-fatal injury and was not coded to KABCO.
Date Crash date
Time Crash time
County County where crash occurred
City (includes nearby) City name recorded if crash occurred within or in proximity to city
Longitude & Latitude GPS coordinates of crash
Sequence of Events Series of events which occurred prior to or during crash
Ambient Light Ambient light at time of crash (e.g., daylight, dark/lighted, dark/not lighted)
Road Conditions Road surface conditions at time of crash (e.g., dry, wet, icy)
Weather Conditions Weather conditions at time of crash (e.g., clear, mist, rain, snow)
Road Material Road material at crash location (e.g., asphalt, concrete, gravel)
Road Alignment (Curve or Grade) Roadway alignment and elevation (e.g., curve left, sag, incline)
Road Classification Roadway use (e.g., urban/municipal, rural, state highway)
Speed Limit Speed limit at crash location
Vehicle Year, Make, Model Vehicle data
VIN Unique code used to assist with vehicle identification
Vehicle Class (Type) Type of vehicle involved in crash (motorcycle, car, light truck, large truck)
Selt Belt Used Seat belt use indication (per occupant)
Est Crash Cause Police-reported estimate of major factors contributing to crash
Table 14. Summary of Data Types Provided in Crash Summary
Crash Data
States which Provided Data
IL IN KS NH NJ NC OH OR SD UT WA WI Crash ID Severity Date Time County
City (includes nearby) Longitude & Latitude Sequence of Events Ambient Light Road Conditions Weather Conditions Road Material Road Alignment (Curve or Grade) Road Classification Speed Limit Vehicle Year, Make, Model
VIN
Vehicle Class (Type) Selt Belt Used Factors Contributing to Crash (Estimated)
Crash data was sorted and organized into a table for analysis and comparison. Due to the large number of crashes collected, analysis of individual crash records was not possible. During data analysis, a number of observations were made:
Some states provided redundant crash records for each occupant of the crashed vehicle. Only one crash record containing the maximum injury severity in a given vehicle was retained and analyzed.
Injury data was often provided using a KABCO+U format:
o K = killed or died within the reporting period of a crash report at a hospital; o A = severe injury resulting in loss of consciousness, incapacitation, permanent
injury, extended hospitalization, or chronic pain;
o B = moderate injury resulting in temporary incapacitation or loss of work which is not prolonged;
o C = minor (sometimes denoted “possible”) injury which may be treated on scene or in which an occupant is transported to a hospital and released, or in which treatment is refused;
o O = property damage only (PDO), no major injuries reported which require treatment or hospitalization; and
o U = unknown injury.
For analysis purposes, it was assumed that crashes with “U” injury code were entirely PDO crashes. Thus, crash cost and severity results may understate actual injury contributions.
Injury severities in crashes may be subjective; it is up to the responding officer to determine if injuries are A, B, or C severity. Furthermore, some “K” fatalities may be miscoded if the injured occupant remains in medical care for an extended duration. Fatality can result from medical complications, brain or spinal damage, prolonged loss of consciousness (i.e., non-responsive), or patient or caregiver (e.g., family) decisions to remove life support. Thus, actual fatal crash results may be
underreported.
For some state DOT databases, all non-fatal injuries (A, B, and C severity) were coded as “I.” Data from state DOTs using “I”-injury data were considered
independently from state DOT data which contained a complete KABCO distribution. Sometimes data was not available for every crash in a state. Reasons for data
omission include: crash reports filled out later and not at the scene of a crash; data was not available or could not be measured; errors in data entry/coding; and data was accidentally omitted from a form.
Causality could not be determined for crashes. If crashes involved trees in a series of events, researchers could not determine if the tree was the most harmful event (MHE) unless the state provided data to indicate MHE. In addition, it is not guaranteed that the MHE resulted in the most severe injury if multiple harmful events each resulted in injury. Because not every state provided the sequence of events and few states
indicated which event was MHE, all crash data provided to researchers which were related to trees or utility poles were included in this analysis.