DEER-VEHICLE CRASH PATTERNS AND DEER CROSSING SIGN PLACEMENT

These results indicate that this research considered sign locations with higher than average DVC frequencies and rates in the state and county.The significance of the differences of the

DEER-VEHICLE CRASH PATTERNS AND DEER CROSSING SIGN PLACEMENT

by

Xin Yi

A thesis submitted in partial fulfillment of

the requirement for the degree of

I would like to express my sincere appreciation to my advisor, Dr Keith K Knapp for his advice, encouragement, patience and support throughout my entire study and research activity at University of Wisconsin-Madison I would also like to thank Robert E Rolley from the Wisconsin Department of Natural Resources, and Matt Rauch, Richard Lange, Andrew Schilling, Timothy McClain, and Mary Kunkel from Wisconsin Department of Transportation for their valuable assistance with my data collection I would also like to thank Drs Bin Ran and David Noyce for their valuable suggestions and considerate comments Most importantly, I would like to express my appreciation to my father, my mother, and my friends who have encouraged and supported my study and made my work possible

i

evaluated the effectiveness of an ordinary DC sign The MUTCD provides some quantitativeguidance for DC sign installation and a few jurisdictions have DVC related criteria for installation, but no studies were found that supported the basis of these criteria The purpose

of this research is to help better define the DVC problem in Wisconsin, and investigate the DVC patterns within the selected study segments with DC signs Procedure guideline for theinstallation of DVC signs is also recommended

Twelve deer carcass removal (DCR) and DVC frequencies and rates, along with DCR to DVC ratio, were calculated for each county in Wisconsin The magnitude of the DVC problems in Wisconsin was discussed by county First, the DCR to DVC ratios were

analyzed by county Fifty-eight out of 71 counties that have available data had the DCR to DVC ratio between 0.93 and 4.90, while the others ranged from 6.53 to 49.33 Then the 12 DCR and DVC measures were ranked for the 58 counties The top ten counties within the 12

ii

were presented, and five counties were selected from the sixteen counties for this research, which included Adams, Dane, Sauk, Waupaca and Shawano County.

Seventy-six DC sign installation locations in the five counties were selected for this study The 1996 to 1998 DVC data within two miles of the DC sign pairs were then collected and analyzed The DC sign pairs were first grouped into 30 crash analysis sites (CASs), which included 38 sign pairs Twenty-two of the CASs had single DC sign pairs and eight includedtwo DC sign pairs for each CAS The average overall length of the 30 CASs was 7.9 miles with an average roadway segment length between the DC signs of 3.5 miles, and an average length outside the DC signs of 4.4 miles

Twenty-eight of the 30 CASs had higher DVC rates (DVCs per HMVMT) between the DC signs than the state average Twenty-five CASs had higher than state average DVC rates outside the DC signs The average DVC rate between the DC signs was more than five timesthe state average And the average DVC rate outside the DC signs was about four times the state average Similar traits were found when the CAS averages were compared to the county rate averages Twenty-two of the 30 sites had an average DVC rate between the DC signs that was higher than the county average, and 18 had an average DVC rate outside the

DC signs higher than the county average However, DVC frequency (DVCs per mile per year) average between the DC signs was 14 times the state average, and the outside

iii

frequency average was 10 times the state average The average DVC frequencies between and outside the DC signs in all the CASs were also higher than the county averages These results indicate that this research considered sign locations with higher than average DVC frequencies and rates in the state and county.

The significance of the differences of the DVC frequency and rate between the CAS DC signs and these measures outside these signs was evaluated with a paired T-test The T-test results indicated that the DVC frequencies and rates between the DC signs of the CASs were significantly greater than those outside the DC signs A comparison of these differences for the single and multiple DC pair CASs showed that the average DVC frequency and rate differences for the multiple pair CASs, were 120 and 75 percent higher, respectively, than theaverage DVC comparable differences in the single DC sign pair CASs In other words, a greater reduction in DVCs from between the DC signs to outside the DC signs occurred on multiple DC sign pair roadway segments than on segments within a single DC sign pair Due

to the small sample size, a non-parametric statistic analysis was also conducted on its

significance, however no significant differences were found Additional research is needed

to determine if multiple signs truly impact the DVC patterns between and near the signs

The 38 sign pairs identified in this study were also categorized into positive, negative, and conflicting sign locations based on their DVC patterns Four DVC measures, the peak 1/4-mile DVC frequency and rate, and the peak average DVC frequency and rate, were

iv

between the DC signs were selected as positive sign locations (PSLs) Based on these results, it was considered high probability that these locations were in the proper location They were the focus of the further safety measure investigation For example, all the PSLs had both between to outside DVC frequency and rate ratios higher than 1.16, with average of 2.15 and 2.52, respectively The average DVC frequency between the DC signs of the PSLs was 3.62 DVCs per mile per year, and the average DVC rate for the same segments was 244.5 DVCs per HMVMT These results were used in the creation of the DC sign

installation guidance procedure recommended in this research

Recommendations were made on the future research and data collection, such as between to outside sign DVC measure ratios for the entire state Most importantly, the procedure

guidelines for DC sign installations were recommended when a DC sign installation request was received in Wisconsin The procedure limitations were also presented

v

TABLE OF CONTENTS

ACKNOWLEDGEMENT I ABSTRACT II CHAPTER 1 INTRODUCTION 1

PROBLEM STATEMENT

2 RESEARCH OBJECTIVE

3 ORGANIZATION

5 DVC SPATIAL PATTERNS

18 SUMMARY

22

DCR to DVC Ratios 24 County DCR and DVC Frequency and/or Rate Comparison 25 County Selection 31

SIGN INSTALLATION STUDY SITE SELECTION

32 SUMMARY

36 CAS ANALYSIS

Overall Group Pattern Summary 54

vii

LIST OF TABLES

TABLE 1 COUNTY DESCRIPTIVE STATISTICS A 23

TABLE 2 FREQUENCIES, RATES AND DCR/DVC RATIO DESCRIPTIVE STATISTICS A 26

TABLE 3 COUNTIES WITH HIGHEST DCR FREQUENCY AND RATES A 28

TABLE 4 COUNTIES WITH HIGHEST DVC FREQUENCY AND RATES A 28

TABLE 5 SIGN PAIR DISTANCES BY COUNTY 34

TABLE 6 CAS LOCATIONS AND CHARACTERISTICS 39

TABLE 7 CAS LOCATIONS AND CHARACTERISTICS (CONT.) 40

TABLE 8 CAS LOCATIONS AND CHARACTERISTICS (CONT.) 41

TABLE 9 CAS DVC FREQUENCIES AND RATES A 44

TABLE 10 CAS DVC FREQUENCY AND RATES DESCRIPTIVE STATISTICS 48

TABLE 11 STATE AND COUNTY AVERAGE DVC MEASURES (1996 TO 1998) 49

TABLE 12 CAS BETWEEN AND OUTSIDE PAIRED T-TEST DVC COMPARISON RESULTS B 52

TABLE 13 DVC FREQUENCIES, RATES AND BETWEEN TO OUTSIDE SIGN RATIOS FOR PSLS .60

viii

FIGURE 1 DEER CROSSING SIGN (8) 2

FIGURE 2 SUPPLEMENTAL SIGN (8) 2

FIGURE 3 DVC TEMPORAL PATTERNS BY MONTH IN WISCONSIN (4) 7

FIGURE 4 LIGHTED “DEER XING” SIGN (25) 13

FIGURE 5 ANIMATED DC SIGN WITH “DEER XING” SUPPLEMENTAL SIGN (25) 13

FIGURE 6 THE LIGHTED, ANIMATED DEER CROSSING SIGN(7) 16

FIGURE 7 COUNTY DCR TO DVC RATIOS 25

FIGURE 8 ADAMS COUNTY DEER CROSSING SIGN LOCATIONS (32) 36

FIGURE 9 DANE COUNTY DEER CROSSING SIGN LOCATIONS (32) 1

FIGURE 10 SAUK COUNTY DEER CROSSING SIGN LOCATIONS (32) 32

FIGURE 11 SHAWANO COUNTY DEER CROSSING SIGN LOCATIONS (32) 32

FIGURE 12 WAUPACA COUNTY DEER CROSSING SIGN LOCATIONS (32) 34

FIGURE 13 EXAMPLE SINGLE (UPPER) AND MULTIPLE DC SIGN (LOWER) PAIR CASS 38

FIGURE 14 EXAMPLE DVC SIGN PAIR PATTERN (PEAK BETWEEN DC SIGN PAIR) 56

FIGURE 15 EXAMPLE DVC SIGN PAIR PATTERN (CONFLICTING PEAK LOCATIONS) 57

FIGURE 16 EXAMPLE DVC SIGN PAIR PATTERN (PEAK OUTSIDE DC SIGN PAIR) 58

FIGURE 17 DEER CROSSING SIGN INSTALLATION PROCEDURE GUIDANCE 78

ix

CHAPTER 1 INTRODUCTION

Deer-vehicle crashes (DVCs) drew the attention of engineers and researchers more than 30

years ago (1) More recently, however, this issue has become an increasing concern along the

roadways in Wisconsin and the United States In 2000, for example, there were 258,000

reported animal-vehicle crashes in the United States (2) These crashes included 143

fatalities (2) However, Conover has estimated that up to 50 percent of this type of crashes are unreported (3) In Wisconsin alone, there were more than 20,000 DVCs reported in 2000, and this represented 15 percent of the total crashes reported (4) However, deer carcass

removal (DCR) numbers reported by Wisconsin Department of Natural Resources (DNR) areapproximately twice the number of DVCs reported by the Department of Transportation

(DOT) (5) The vehicular damage cost estimates for the reported DVCs in Wisconsin State are more than $30 million per year (4)

Numerous methods have been developed to reduce DVCs (1, 6, 7) Some of the

countermeasures considered have included fencing, bridge structures, roadside reflectors, vehicle whistles, highway lighting, right-of-way plantings and intercept feeding, salt

alternatives, deer crossing (DC) signs, mirrors, chemical repellants, herd management, and

speed limit reduction(6) DC signs (See Figure 1) with and without supplemental signs (See

Figure 2) have been installed at a large number of locations The primary objective of the installation is to reduce DVCs by warning drivers that they may be in an area with prevalent

FIGURE 1 Deer Crossing Sign (8)

FIGURE 2 Supplemental Sign (8)

deer crossings and/or crashes The existence of these signs suggests caution and reduced vehicle speeds

PROBLEM STATEMENT

DC signs are used throughout the United States However, little research has focused on the placement of DC signs Quantitative guidance on the installation of these signs is almost non-existent The overuse or misuse of DC signs, however, can be costly and also reduce

3their potential effectiveness Most DC signs are typically placed at locations where the need for caution is clear and obvious (e.g., curve warning signs), but this type of situation is not true for DC sign locations The proper locations of these signs (and their supplementary distance message), therefore, become very important At a minimum, the driver of a vehicle expects DC signs to designate roadway segments that have more than a typical number of deer crossings or DVCs The actual magnitude of the DVC problem in Wisconsin is

discussed This research evaluates the DVC patterns surrounding a number of existing DC sign installations in 5 counties to determine if this selection is appropriate for these locations

In addition, no quantitative guidance on the installation of these signs is currently available inWisconsin, and this research has taken an initial step in that direction The results of this research should assist with more effective and efficient installations of DC signs

RESEARCH OBJECTIVE

The objective of this project was to investigate the reported DVC patterns surrounding the

DC sign installations along a sample set of roadways in Wisconsin First, the number and location of reported DVCs near and between DC sign installations were summarized Then, three years of DVC data were used to determine a typical or expected county and statewide DVC frequencies or rates The pattern at each individual sign site was investigated

Wisconsin county DVC and DCR data and their interrelationships were also evaluated to help describe and define Wisconsin DVC issue

presented Chapter five includes the conclusions and specific recommendations reached from this research.

CHAPTER 2 LITERATURE REVIEW INTRODUCTION

This chapter includes discussion and documentation of DVC temporal and spatial pattern studies and the factors that contribute to DVCs Studies related to the potential effectiveness

of DC signs and the installation of these signs were also described and discussed No studies were found that focused on the evaluation of DVCs for DC sign placement However, summaries of several safety measures and statistical analysis methods for traffic safety studies were included

DVC TEMPORAL PATTERNS

The factors that impact the DVC temporal patterns include: traffic volume, deer behavior,

and human behavior (e.g., hunting) (9, 10, 11, 12) Deer behavior is seasonal, and impacts the level of their movement (9) Those movements appear to have an impact on the number

of DVCs that occur (9) In the fall and spring when woodland forage material is scarce, deer tend to search for food and are sometimes attracted to the roadside vegetation (10) Deer

rutting occurs during the fall along with hunting season, which also leads to increased deer

movement (10) During the winter, deer activities decrease due to their involuntary

curtailment of food intake (10) DVC patterns appear to be related to increases in deer

movement For example, they are more prevalent in the spring (e.g., April and May) and fall

and/or early winter (e.g., October and November) (See Figure 3) Figure 3 shows that the largest number of DVCs usually occurs in November, and a minor peak in DVCs occurs

FIGURE 3 DVC Temporal Patterns by Month in Wisconsin (4).

during May Similar DVC patterns have been documented in a number of states (9, 11, 12).

It has also been found that most DVCs occur near sunrise, sunset and at night (9, 10, 13)

Deer spend daylight hours in the wooded parts of their home range, but begin to forage near

sunset (14) They tend to move back into the woods just before dawn (14) Increases in

DVCs near sunset and sunrise may be attributable to a combination of the deer movement and increased traffic volumes During these times, white-tailed deer also have increased

breeding behaviors during dusk and dawn (9, 10)

Some DVC patterns have also been found on a daily basis (9) Data analyzed by Allen and

McCullough indicated that the number of DVCs was highest on Friday evenings but lowest

on Monday mornings (9) They hypothesized that traffic volume was higher on Friday

evenings

DVC SPATIAL PATTERNS

There are a large number of factors that are believed to impact the locations of DVCs (9, 11,

12, 15, 17-21) These factors include, but are not limited to, the following: adjacent traffic

volume, vehicle speeds, land type (e.g., forestland, rural area), adjacent land use (e.g.,

residence, buildings, and parks), human population, deer population, mileage of roadways that have DVCs, roadway features (e.g., number of lanes), roadside features (e.g., bridges, gullies, and rivers), roadside visibility (e.g., sight distance), and vegetation cover

For example, in Ohio, positive correlations were found between the number of county DVCs and county size, amount of forestland, rural land within counties, number of people, and

length of major roadway mileage with DVCs (18) And negative correlations were found with the amount of cropland (18) A study in Illinois State found the positive correlations

between number of county DVCs and county deer density, along with average daily vehicle

miles of travel (VMT) by county (20) The factors with the strongest linear relationships to

the number of county DVCs include amount of forestland, length of major roadway with

DVCs and VMT (18, 20).

Allen and McCullough also indicated that the DVCs occurred most commonly in areas that

were primarily forestland with large deer population (9) DVCs also appeared to occur

approximately in same proportion as the amount of crop, unimproved field, and forest habitat

adjacent to the studied segments (9) They also found that almost three times as many

crashes occurred on two-lane paved roads than on divided paved highways, and very few

DVCs occurred on unpaved roads (9) These results did not take traffic volume into account (9)

Hubbard has investigated the relationships between some of the factors believed to influence

the location of a “high” DVC location (11) The database they evaluated had deer harvest

numbers, traffic volume estimates, and the distance from the DVC location to the nearest town or city and the nearest city with a population greater than 2,000 The number of bridgeswithin the selected segments and roadway lanes are also recorded A set of randomly

selected one-mile segments of roadway was studied, along with the landscape and roadway characteristics within one mile of the segment sites The input variables in the model

developed included the size of different adjacent vegetation patches and the standard

deviation of all patch areas, the number of bridges within the study segments, and the number

of roadway lanes (11) The number of bridges, lanes of traffic, average size of grass patches,

and the amount of woody patches were positively related to the possibility that the segment would be a “high” DVC location and the number of bridges was found to be a strong

predictor Other factors had negative relationships with this probability It was suggested

that mitigation of DVCs be focused on roadway segments with a high number of bridges

(11)

DC SIGN INSTALLATIONS

An assumption by drivers is that DC signs are located in areas that either have experienced orare expected to have an unusual number of DVCs The only guidance on the installation of these signs from the Manual on Uniform Traffic Control Devices (MUTCD), however, is that

DC signs may be installed at appropriate locations as determined by engineering study or

judgment (8) The MUTCD does indicate that a DC sign should be used at locations where

unexpected crossing activities might occur, but judgment is required in the determination of

these locations that might be considered “unexpected” locations for deer crossings (8).

Because quantitative guidelines for DC sign installation do not exist, a few jurisdictions haveattempted to create their own guidelines For example, the Washtenaw County Michigan Road Commission gave suggestions to the Michigan Manual of Uniform Traffic Control Devices about the placement of DC signs, which included considerations of the DVC history

along the road segment of interest and one mile in each direction (22) The installation of a

DC sign in Michigan is warranted if five DVCs have occurred within a twelve-month period

(22) It is also suggested that the placement of the sign be reviewed every third year and adjustments made if appropriate (22) For example, if the crash study shows that no DVCs

have occurred in any twelve months within the three-year period, the sign may be removed

(22) In addition, Iowa recommends that DC signs be installed where posted speed is in

excess of 45 miles per hour or obstruction and topography occasionally limit sight distance

(23) States such as Iowa and Minnesota also have some crash related criteria for the

installation of DC signs, but they are very general in nature Both states recommend that

historical DVC data be reviewed to determine the appropriate location of warning signs (23, 24)

DC SIGN EFFECTIVENESS

Warning signs are most effective (i.e., reduce vehicle speeds) when a condition or hazard thatneeds a reaction is clear and obvious to the driver No studies were found that attempted toevaluate the speed reduction effectiveness of ordinary DC signs In fact, it is generallyconsidered that these signs do not effectively reduce vehicle speeds in their vicinity Several

DC new technologies (with deer detection and/or recording system) were applied to improvesign effectiveness, but only two research studies were found that documented an analysis ofhow they improved the DC sign effectiveness, as measured by the reduction in vehicularspeeds, and/or ratio of deer crossings to deer killed during the test periods

The first DC sign enhancement study was initiated in 1968 and included two phases (7, 25).

In the first phase, DC sign with just the lighted words “DEER XING” and a redesigned DCsign with deer animation and “DEER XING” supplemental sign was considered (See Figure

4 and 5) (25) The signs were installed south of Glenwood Springs, Colorado along a lane divided highway with a posted speed limit of 60 mph (25) The vehicle speeds were

four-recorded for 16 days while each sign was turned away from approaching vehicles, and then

for additional 28 days with the lighted crossing signs facing the traffic (25) The speed collection station was located 800 feet past the evaluated DC sign (25) The data indicate

FIGURE 4 Lighted “DEER XING” Sign (25)

FIGURE 5 Animated DC Sign with “DEER XING” Supplemental Sign (25)

that the mean vehicle speed was 54.52 miles per hour (mph), 53.03 mph, and 51.59 mphwhen the sign was turned away from the traffic, with the lighted “DEER XING” sign, and

with the animated sign and “DEER XING” supplemental sign, respectively (25) It was

concluded that the differences between these three mean speeds were small but statisticallysignificant, and that no significant relationship was found between the number of days when

the sign was activated and the average speed of the traffic (25) In other words, there did not

appear to be any habituation for the 28 days to these technologies when they were active

(25)

During the second phase of the Colorado study the effectiveness of two animated DC signs

on DVCs were studied A six foot by six foot DC sign was updated with four silhouettes ofdeer that lighted in sequence from right to left The supplemental warning sign was also

changed from “DEER XING” to “DEER XING NEXT MILE” (See Figure 6) (7) The DC

signs were activated and deactivated during alternate weeks for 15 weeks Vehicle speedswere collected at each sign location, and 0.15, 0.65, and 1.5 miles past the signs Overall,eighty vehicle speeds were randomly chosen for analysis from the 1,800 to 2,200 collected

In addition, nightly spotlight counts were used to estimate the number of deer that crossed theroadways For analysis purposes it was assumed that each deer observed crossed theroadway twice each night

The data collected as part of this project showed a number of patterns (7) The mean vehicle

speed with the signs deactivated was more, but no more than 3.0 mph greater than with the

FIGURE 6 The Lighted, Animated Deer Crossing Sign(7)

signs on at all three collection locations The deer crossing to deer road kill ratio with thesigns deactivated (i.e., 56.5:1), was also almost equal to the ratio with the signs activated

(i.e., 56.9:1) (7) No statistically significant difference was found between these ratios (7)

A short pilot study was also conducted on the speed impacts of placing three deer carcasses

on the shoulder 150, 320, and 350 feet past the sign This was only done one night eachweek (i.e., 15 days) during the study The mean vehicle speed significantly decreased by

7.85 mph with the carcasses and the signs deactivated (7) When the signs were activated and the carcasses added, the mean speed also significantly decreased by 6.24 mph (7).

However, No significant difference was found between these two average speed reductions

with signs on and off (7)

17The Colorado researchers concluded that the lighted and animated DC signs were not

effective at reducing the number of deer killed (7) The change in wording of the supplemental signs was not valuable (7) The driver response in terms of speed reduction was not enough to have an effect on the crossings per road kill ratio (7) Vehicle speeds were reduced significantly when deer carcasses appeared (7)

A second and more recent DC sign impact study investigated the impacts of using technology

to improve the effectiveness of a typical deer crossings sign (26) This study considered the

Flashing Light Animal Sensing Host (FLASH) system, which consisted of infrared sensors (with a backup system, a geophone system, which detects ground vibration) for roadside deer

detection and data collection (26) Roadside fencing forced deer into a 300-foot wide gap to cross the roadway, and the data collection equipment was installed in that gap (26) If a mule

deer was detected on the roadside, signs about 1,000 feet from the detection zone would activate a “Attention: Migratory Deer Crossing” message and a “Deer on Road when Lights

are Flashing” message (26)

Vehicle speed and classification data were collected when the signs were not activated and

also they were activated (26) This information was collected in both directions of traffic before and within the segment with the roadside detectors (26) Deer activity was also recorded by the FLASH system, the geophone system, and a video camera (26) Overall, it was concluded that the geophone detectors were most reliable of those considered (26)

When the infrared and geophone devices activated the signs, a mean operating vehicle speed

reduction of 3.6 mph was observed (26) Also, the addition of a roadside deer decoy and

flashing lights resulted in average speed reductions of 12.32 and 6.62 mph for passenger

vehicles and tractor trailers, respectively (26) Overall, the presence of the decoy appeared to

have more impact on the average vehicle speed than the active flashing light sign with and

without the appearance of an actual deer (26)

MEASURE OF SAFETY

This research focuses on the measurement of DVC safety near DC sign locations Crash data

and statistics are needed to quantify and describe crashes (27) Crashes are generally

described by the types of crashes Two measures of the magnitude of the DVC problem include crash frequency (e.g., crashes per unit of time) and rate (e.g., crashes per area

population, number of registered vehicles, roadway mileage, or vehicle miles of travel) (27)

Involvement statistics often focus on the category of the vehicles and drivers involved in the

crashes (27) Crash severity is another measure of safety generally expressed in fatalities and injuries (27)

In many cases, including this research, it is also necessary to compare the level of safety at different locations Those locations with more significant safety problems may need to be considered before those with less significant crash issues One method of identifying crash-

prone locations is to rank locations by crash frequency or crash rate (28) A classic statistical

method to identify “high” crash frequency and rate locations is shown in Equation 1 Those

19that satisfy the inequality in Equation 1 are then considered to have more crash than are usually expected

))(

(K S

XA

where OBi = Crash frequency or rate at location i

XA = Mean frequency or rate for all locations under consideration

K = Constant corresponding to a level of confidence in the finding

S = Sample standard deviation for all locations

Equation 1 requires the standard deviation of safety data that is not often available The rate quality control method in Equation 2 (limited to crash rate) assumes the crash rate follows a Poisson distribution A location would be considered hazardous if its crash rate satisfies the inequality:

i i

i

V V

XS K XS

OBR

2

1)

( 0 5 ++

where OBRi = Crash rate observed at location i

XS = Mean crash rate for locations with characteristics similar to those of location i

Vi = Volume of traffic at location i, in the same units as the crash rates

K = Constant corresponding to a level of confidence in the finding

Both of these measures of “high” crash location identification have been used in the past and continue to be used

SUMMARY

DVCs occur most often in spring, fall and early winter It is believed that this pattern of DVCs is the result of deer movement due to scarcity of food, rutting, and/or hunting season During the days, DVCs also occur more frequently near sunrise and sunset This is believed

to be the combined result of deer movement and increased traffic flow During the week, DVCs appear to peak on Friday evenings Apparently, this is the result of an increase in deermovement and a peak in traffic flow at this time Factors that contribute to DVCs appear to

be deer behavior, traffic volume characteristics, and human activities (e.g., hunting) (9, 10,

11, 12)

A number of studies have shown relationships between several factors and numbers of DVCs

at the proximity of a location These factors include adjacent traffic volume, vehicle speeds, land type (e.g., forestland, rural area), adjacent land use (e.g., residence, buildings, parks), human population, deer population, mileage of roadways that have DVCs, roadway features (e.g., number of lanes), roadside features (e.g., bridges, gullies, rivers), roadside visibility

(e.g., sight distance), and vegetation cover (9, 11, 12, 15, 17-21).

No studies were found in the literature review that investigated and evaluated the placement

of ordinary DC signs, or quantitatively defined criteria to locate the signs Previous studies

21related to DC signs have typically assumed that they are not adequate, but that they are installed in the most appropriate location For example, two studies completed in 1970s investigated the impact of enhancements to typical DC signs at locations believed to be appropriate Both studies found that these improvements reduced the average speed of traffic

by no more than 3.0 mph A study of a system that detected deer and activated flashing lights on a DC sign was also summarized This study found a reduction on vehicle speed, butthe average of the reduction was only 1.4 mph Speeds were reduced more significantly when a deer carcass or a target was placed on the roadside

In the last section of this chapter, the reader were introduced to safety measures and ranking methods to identify crash-prone locations A ranking method is used in Chapter 3 to select study counties

CHAPTER 3 STUDY COUNTY AND SITE SELECTION INTRODUCTION

This chapter describes the methodology used in selection of study site counties as well as specific sign installation sites The deer carcass removal (DCR) to DVC ratios were

calculated for each county It was assumed that DCRs were the majority of all the vehicle hits Therefore it was considered that the data from the counties with large DCR to DVC ratios were questionable, and were not considered Selection of a subset of counties from the rest of the counties examined in this study was based on a comparison of the DCR and DVC frequencies and rates Those counties that ranked high on the list of frequencies and rates were considered The magnitude of the DVC problem in Wisconsin was also described through investigating the top ten counties of DCR and DVC frequency and rate lists Individual sign installation sites were selected based on data availability and apparent installation year of the signs The data collected is calculated for each of these sites and discussed in this chapter

deer-SELECTION OF STUDY COUNTIES

Wisconsin has 72 counties, however, the data from Menominee County is questionable because DNR does not collect data in this county (i.e Indian reservation) and DOT only reported no more than 10 DVCs each year Therefore it is not included in the analysis and selection process The Wisconsin DOT has approximately 1,100 DC sign installations

23described within its current sign management system (SMS) Therefore, the first step in this DVC pattern research had to be the selection of the counties within which sign installation sites would be located From the 71 counties, the selection was based on the magnitude of ranking of several DCR and DVC frequency and rate measures, which were calculated with average hundred million vehicle miles of travel (HMVMT), roadway mileage, county land area, human population, and pre-hunt deer population estimates Table 1 indicates the range, mean, and standard deviation of the data used to compare counties and calculate DCR and DVC rates

TABLE 1 County Descriptive Statistics a

County

Standard Deviation

a The data were collected from 2000 (4, 29, 30).

b Deer population estimates data were from 1997 (31).

The total number of reported DVCs ranged from 19 to 1,177 The average for the 71

counties was 280 with a standard deviation of 235 The total DCRs, on the other hand, ranged from 25 to 1,902 with average of 670 and standard deviation of 448 Clearly, not all

the DVCs were reported or there were a number of collisions with more than one deer

involved As shown in Table 1, there is a wide range of county VMT and human population

in Wisconsin and it has a large amount of variation, as their standard deviation values were large and the differences between their mean and median were significant Length of

roadways, county land area, and deer population were less variable relative to their own mean value, although their data have large ranges in terms of the magnitude The mean HMVMT is only 8.1, but it ranges from 0.8 to 78.2 The average county population is about 75,000 The average area and length of roadways in a county were 760 mile2 and 1,576 miles, respectively Deer population estimates ranged from 1,189 to 39,716 with an average

of 17,295 per county The data summarized in Table 1 is shown in Appendix A These data were used to calculate five different DCR and DVC rates for each county (i.e., DVCs or DCRs per HMVMT, DVCs or DCRs per county land area) in addition to their total DVCs and DCRs

DCR to DVC Ratios

The ratio of DCRs to DVCs was calculated for each county to see how closely these two numbers match within each county Only those counties with an acceptable ratio were considered in the analysis The ratios calculated are plotted as shown in Figure 7, and the complete ratio data by county is given in Appendix A However, the researchers only

wanted to consider those counties with what might be considered a typical ratio for

Wisconsin As shown in Figure 7, first, the two counties with ratios larger than 14.85 were viewed as outliers in Wisconsin and eliminated from further consideration The remaining

FIGURE 7 County DCR to DVC Ratios

county ratios between 0.93 and 14.85 had an averaged 3.43 and the standard deviation was 3.56 However, the county ratios between 0.93 and 4.90, which covered includes 58 out of

the 71 counties, averaged 2.01 and had a standard deviation of only 0.98 Therefore, this range of data (See Figure 7) appears to represent the typical Wisconsin county Those counties included in the 0 to 4.90 DCR to DVC ratio range were evaluated further

County DCR and DVC Frequency and/or Rate Comparison

The DCR and DVC frequencies and rates (i.e., based on HMVMT, miles of roadways, county land area, human population, and deer population) were calculated for the 58 countieswith DCR to DVC ratios smaller or equal to 4.90 These frequencies and rates are

calculated, along with the county DCR to DVC ratios, and are shown in Appendix A with a summary in Table 2 The counties with the highest magnitudes of DCR and DVC

TABLE 2 Frequencies, Rates and DCR/DVC Ratio Descriptive Statistics a

County Characteristics Range Mean Median Standard

b DCRs or DVCs per mile = DCRs or DVCs per miles of roadways by county

c DCRs or DVCs per mile 2 = DCRs or DVCs per county land area

d DCR per DVC ratios were calculated from the data from July 1999 to June 2000.

Frequencies and/or rates are listed in Table 3 and Table 4 The top ten counties from each list are discussed and the magnitude of the DVC problems in Wisconsin is described

TABLE 3 Counties with Highest DCR Frequency and Rates a

Total

DCRs

DCRs per HMVMT

DCRs per Mile of Roadways

DCRs per County Land Area (mile 2 )

DCRs per Thousand Human Population

DCRs per Thousand Deer Population b

3 Dane Florence Green Lake Winnebago Waushara Milwaukee

4 Waukesha Waupaca Waushara Shawano Shawano Washington

5 Columbia Green Lake Columbia Ozaukee Marquette Brown

6 Marathon Waushara Oneida Green Lake Waupaca Walworth

7 Oneida Oneida Marquette Waushara Oneida Winnebago

8 Waushara Marquette Adams Columbia Jackson Sheboygan

9 Dodge Taylor Florence Brown Green Lake Fond du Lac

10 Adams Clark Winnebago Washington Columbia Dane

a The data is mostly from 2000 (29).

b The rates of DCRs per thousand deer population were calculated from data in 1997 (31).

TABLE 4 Counties with Highest DVC Frequency and Rates a

Total

DVCs

DVCs per HMVMT

DVCs per Mile of Roadways

DVCs per County Land Area

DVCs per Thousand Human Population

DVCs per Thousand Deer Population b

2 Marathon Shawano Shawano Winnebago Adams Milwaukee

3 Shawano Taylor Green Lake Waukesha Shawano Ozaukee

4 Waupaca Green Lake Columbia Ozaukee Marquette Brown

5 Columbia Waupaca Sheboygan Dane Jackson Sheboygan

6 Portage Forest Portage Waupaca Green Lake Kenosha

7 Waukesha Florence Winnebago Columbia Taylor Washington

8 Sheboygan Marquette Sauk Washington Forest Dane

10 Winnebago Portage Marquette Green Lake Columbia Racine

a DVC data is from 2000.

b The rates of DVCs per thousand deer population were calculated from the data in 1997 (31).

The 12 ranking lists in Tables 3 and 4 show a significant amount of overlap, and cover 30 out

of the 58 remaining counties included No individual county exists in all 12 lists, but

Waupaca and Shawano County are both present in 10 of the lists, respectively In addition, these two counties and most of the others in the lists appear in the equal (or close to equal) number of DCR and DVC lists This results from the fact that their DCR and DVC numbers are close enough in magnitude to produce similar lists In fact, 29 out of the 31 counties considered have the DCR to DVC ratios between 0.93 and 3.00 The other two counties, Oneida and Waushara, with the largest DCR to DVC ratios in the range considered (i.e., 4.65and 4.02) only appear four and five times, respectively, in the DCR lists They do not appear

in any of the DVC lists This fact is the result of the large DCR to DVC ratios The DVC reporting might be questioned in this case Portage County, on the other hand, which has one

of the smallest DCR to DVC ratios of the group considered (i.e., 1.06), appears three times inthe DVC lists and none in the DCR lists This appears to be the result of its relatively small DCR to DVC ratio among the counties considered Therefore, the DCR data might be questioned for Portage County Dane and Sheboygan County appear four times in DVC lists,and twice and once in the DCR lists, respectively They both have low DCR to DVC ratios (i.e., 1.06 and 1.11),

As indicated in the previous paragraph, 31 of the 58 counties evaluated did appear in the 12 top ten lists created Seven counties only appear in the DCR lists, and six counties only appear in the DVC lists Most of these ten counties only appeared once or twice in the 12