Max Cameron, Amanda Delaney, Kathy Diamantopoulou, Bella Lough - June 2003 - Report No. 202

This project was funded by a special grant from the Victoria Police and by the Centre’s Baseline Research Program for which grants have been received from:

• Department of Justice Roads Corporation (VicRoads)

• Transport Accident Commission

The objective of this project was to provide a scientific base for the development of a safety camera strategy that will:

(1)     maximise the road safety benefit of the safety camera program, and

(2)     continue to build on the positive outcomes achieved by enforcement programs in Victoria over the last ten years.

A review of previous evaluation research concerning Victorian, interstate and international automated enforcement programs was conducted. The review concentrated on the way in which this research can inform the future use of new and existing safety camera technologies in Victoria. Strategic principles relating to the maximisation of available intelligence and technology were formulated. The deterrence mechanisms behind each of the enforcement programs are discussed and world’s best practice is identified where possible.

The deterrence value of the Victorian safety camera program is assessed in relation to the

principal road trauma problems addressed (speeding and red-light running), the number of

serious casualty crashes targeted by each offence detection technology, information on

likely effects on these crashes, the influence of the timing and severity of penalties, and the

supporting role of mass media publicity. Community acceptance and support for the

program is also discussed in detail. These elements of a traffic law enforcement program

play a key role in determining the effectiveness of the program in achieving reductions in

road trauma.

This report provides a valuable scientific base for developing a strategy for the future

directions of the safety camera program, but the report is not that strategy. The limited

information available about the effects of the new technologies, and recent changes to the

mobile speed camera operations, precludes that step from a scientific point of view.

Speed, speed camera, traffic signals, red-light camera, enforcement, publicity

Reproduction of this page is authorised.

1 Author may be contacted by internet e-mail at

max.cameron@general.monash.edu.au

Monash University Accident Research Centre,

Building 70, Monash University, Victoria, 3800, Australia.

Telephone: +61 3 9905 4371, Fax: +61 3 9905 4363

The objective of this project was to provide a scientific base for the development of a

safety camera strategy that will:

1) maximise the road safety benefit of the safety camera program, and

2) continue to build on the positive outcomes achieved by enforcement programs in

Victoria over the last ten years.

A review of previous evaluation research concerning Victorian, interstate and international

automated enforcement programs was conducted. The review concentrated on the way in

which this research can inform the future use of new and existing safety camera

technologies in Victoria. Strategic principles relating to the maximisation of available

intelligence and technology were formulated. The deterrence mechanisms behind each of

the enforcement programs are discussed and world’s best practice is identified where

possible.

The deterrence value of the Victorian safety camera program is assessed in relation to the

principal road trauma problems addressed (speeding and red-light running), the number of

serious casualty crashes targeted by each offence detection technology, information on

likely effects on these crashes, the influence of the timing and severity of penalties, and the

supporting role of mass media publicity. Community acceptance and support for the

program is also discussed in detail. These elements of a traffic law enforcement program

play a key role in determining the effectiveness of the program in achieving reductions in

road trauma.

This report provides a valuable scientific base for developing a strategy for the future

directions of the safety camera program, but the report is not that strategy. The limited

information available about the effects of the new technologies, and recent changes to the

mobile speed camera operations, precludes that step from a scientific point of view.

From the point of view of the first specific objective, there is scope to expand the planned

operations of the new technologies to a sufficient extent so that:

(a) a general effect of the technology is achieved across the road environment on which it

is applied (i.e., all freeways and highways in the case of fixed and point-to-point speed

cameras; all signalised intersections in the case of red-light and red-light/speed

cameras)

(b) the marginal economic benefits of the road trauma savings achieved by the general

effect are just greater than the marginal cost of each increase in the technology

operation (i.e., the cost of each extra camera installation, and necessary offence

processing capacity).

So far as the mobile speed camera program is concerned, it has been found that the

increase in camera hours from 4000 to 6000 hours per month is likely to be economically

worthwhile. The influence of the operational changes to make the enforcement more covert

and unpredictable, and to reduce the speeding offence detection threshold, on the marginal

economic benefits is unknown. However, it is expected that these latter changes have made

the program more efficient. The general effect of the program on crashes across the broad

road environment is expected to continue to operate.

Thus, from the point of view of the second specific objective of this project, it can be

concluded that the current mobile speed camera program should continue. There may be a

case for the program to be expanded further, with economic justification. However, a

decision to reduce the mobile speed camera program, in order to provide resources to

implement or expand other safety camera technologies, should be viewed with caution.

This may result in an erosion in the overall positive benefits achieved by traffic

enforcement programs in Victoria over the last ten years.

The widespread use of safety cameras in Victoria, commencing in the late 1980s, marked

the beginning of significant reductions in road trauma in this State. It also represented a

shift away from traditional labour intensive methods of traffic law enforcement towards the

use of semi-automated techniques. Scientific evaluations of these new techniques have

demonstrated the positive contribution of automated enforcement technologies in reducing

road trauma. Similarly, interstate and international research has evaluated the effects of

similar enforcement operations in other states and countries. In considering the future

directions of automated traffic law enforcement, the existing body of research may provide

critical insights into the way in which existing and emerging technologies can be used to

improve road safety outcomes in Victoria.

In view of the above, it is the objective of this project is to provide a scientific base for the

development of a safety camera strategy that will maximise the road safety benefit and

continue to build on the positive outcomes achieved by enforcement programs over the last

ten years. In completing this task, a review of previous evaluation research concerning

Victorian, interstate and international automated enforcement programs is conducted. This

review concentrates on the way in which this research can inform the future use of new and

existing safety camera technologies in Victoria. In particular, attention is given to the

formulation of strategic principles relating to the maximisation of available intelligence

and technology. In doing so, the deterrence mechanisms behind each of the enforcement

programs are discussed and world’s best practice is identified where possible.

The deterrence value of the Victorian safety camera program is assessed in relation to the

principal road trauma problems addressed (speeding and red-light running), the number of

serious casualty crashes targeted by each offence detection technology, information on

likely effects on these crashes, the influence of the timing and severity of penalties, and the

supporting role of mass media publicity. Community acceptance and support for the

program is also discussed in detail. These elements of a traffic law enforcement program

play a key role in determining the effectiveness of the program in achieving reductions in

road trauma.

The remainder of this report is structured as follows. Section two provides necessary

background information relating to the deterrence mechanisms that operate in traffic law

enforcement operations and common features of such programs. An examination of

existing and emerging technologies and their effectiveness is conducted in section three

and strategic principles for their future use are developed. Section four focuses on the

scheduling of enforcement operations. The deterrence value of the Victorian safety camera

program is discussed in section five and section six summarises survey information about

recent trends in community opinion about the program.

The conclusion emphasises that this report provides a scientific basis for developing a

strategy for the future directions of the safety camera program, but the report is not that

strategy. The limited information available about the effects of the new technologies, and

recent changes to the mobile speed camera operations, precludes that step from a scientific

point of view. However the report provides valuable information for policy-makers in the

area.

The primary objective of traffic law enforcement is to ‘contribute to the safe operations of

the road traffic system’ in the context of the criminal justice system (Cameron &

Sanderson, 1982). The two primary mechanisms through which automated enforcement

achieves this objective are general and specific deterrence. The key reasoning behind these

processes relies on utility theory as described by Ross (1981). In general, this assumes that

road users will decide whether on not to commit a traffic offence based on a rational

analysis of the benefits and risks associated with committing the offence. It is noted, that it

the action. Therefore, where the perceived benefit of committing an offence outweighs the

perceived disbenefit, an individual will elect to commit the offence. Similarly, where the

perceived risks of committing an offence are greater than the perceived benefits, a rational

individual will elect not to commit the offence.

Although both the general and specific deterrence mechanisms are based on an assumption

of rational behaviour, there are considerable differences in the operation of the two

mechanisms.

fear of detection and the consequences, to avoid offending (Cameron & Sanderson, 1982).

Therefore, operations employing general deterrence mechanisms necessarily target all road

users irrespective of whether they have previously offended. It follows that general

deterrence programs have the potential to influence the behaviour of all road users.

There are thought to be three key elements that influence the effectiveness of a general

deterrence program; the perceived risk of detection, the severity of punishment and the

immediacy of punishment. The higher the perceived risk of detection the less likely a road

user is to commit an offence. The actual risk of detection is less relevant given that it is

most often unknown by the driver. Indeed, the perceived risk of detection may be higher

than the actual risk. Nevertheless, over time the perceived risk of detection will be

informed by the road user’s own experience and that of their acquaintances (Shinar &

McKnight, 1985). Therefore, some relationship between the perceived and actual risk of

detection is expected, although the precise form of the relationship is unknown.

The severity of punishment is also relevant although it is it is not the primary mechanism

of general deterrence. Past research has concluded that where the perceived risk of

detection associated with an activity is low, severe punishment of the offence will have

little impact (Ross, 1990). Similarly, two studies examining the effect of increases in

penalties for speeding found no associated changes in driver behaviour (Arberg et al, 1989,

and Andersson, 1989). It has therefore been suggested that it is the existence of a penalty

rather than the size of the penalty that provides the general deterrence (Bjørnskau & Elvik,

1990).

Finally, the swiftness of punishment impacts on the effectiveness of enforcement

operations relying on the general deterrence mechanism. Unfortunately, there is little

conclusive research evidence detailing the optimal timing of punishment (Zaal, 1994).

Nevertheless, reductions in casualty crashes were found in Victoria at a time when

infringement notices resulting from automated enforcement operations took approximately

two weeks to be received (Rogerson et al, 1994). This indicates that a two-week delay

from detection to punishment does not entirely negate the deterrent effect of an

enforcement program. The effect of this delay is however unknown.

apprehended offender, through his actual experience of detection and the consequences, to

avoid re-offending (Cameron & Sanderson, 1982). Therefore, the potential impact of a

specific deterrence program is more limited than that of a program relying on the general

deterrence mechanism. Enforcement programs relying solely on the mechanism of specific

deterrence have the potential to influence only those offenders who have previously been

detected and punished for committing offences. It follows that the magnitude of the

penalty, especially that applying if subsequent offences are committed, is of particular

importance. The choice of penalty, whether it be a warning letter, a fine, demerit points on

a licence or some combination of these, is likely to affect the recurrence of offending

behaviour.

In order to achieve reductions in road trauma, speed enforcement programs may invoke

one or both of the two deterrence mechanisms discussed above. The exact mechanism

used will depend on the nature of the program and its operations. Therefore, it is necessary

to understand the specific characteristics of an enforcement program in order to evaluate its

effects.

Enforcement programs are generally classified as either overt or covert. It is the intention

of overt operations to be highly visible to road users. In doing so, these types of operations

are thought to increase the perceived risk of detection and thus alter the behaviour of road

users immediately in time and space. On the other hand, covert operations are not intended

to be seen by road users and road users should be unaware of the location and timing of

such enforcement operations. Effective covert operations will create a perception that

detection may occur at any location and at any time.

The effects of an enforcement program can either be either localised or general across a

broader road network than the specific locations at which the enforcement operates. Overt

operations are likely to have localised effects, but can also have general effects if the

density of operations is sufficient. Enforcement programs relying on the general deterrence

mechanism commonly have general effects, but the two “general” concepts are different.

Programs based on the specific deterrence mechanism can have both general and localised

effects on crashes.

The type of enforcement program that can be implemented will be influenced by the type

of technology available. In general, speed enforcement technology can be either fixed or

mobile. Fixed devices, such as the safety cameras located in the Burnley and Domain

tunnels in Victoria, are located permanently at one site. In contrast, technologies such as

slant radar speed cameras, are portable and tend to operate at one site for only a short

period of time. This technology, along with others that can be moved from site to site, is

referred to as mobile technology.

In some circumstances, the location of safety cameras, whether fixed or mobile, may be

chosen to affect a known problem of high crash risk or the risk of particularly severe

crashes in a defined area. Such treatments are referred to as black spot treatments. Where

the increased risk relates to a particular route or area the treatment can be spread across this

black route or area. In general, black spot or black route programs are intended to have the

greatest effect at the black spot site or along the black spot route and are rarely aimed at

treating speed across the road network.

As described above, a wide range of enforcement technologies are available for use in a

variety of settings. Table 1 below summarises the characteristics of the available

enforcement technologies and the purposes for which they can be used.

(ANPR = Automatic Number Plate Recognition)

The majority of the technologies in the above table are relevant to automated enforcement

in Victoria. Therefore, the remainder of this section examines each of the relevant

enforcement technologies in terms of their demonstrated effectiveness both in Victoria and

elsewhere.

Slant radar speed cameras have been operating in Victoria since the initial trial period of

this technology in 1985. Since that time, the speed camera program has undergone

significant change and has expanded to its current form. There are currently 54 slant radar

speed cameras used to achieve the target of 6,000 enforcement hours per month for the

speed camera program.

The camera technology consists of two components; a slant radar and a camera control

unit. The camera unit is capable of photographing 2 speeding vehicles per second and can

monitor vehicles travelling towards or away from the camera. The radar unit can be

mounted on a tripod on the roadside or less conspicuously within a vehicle. In general,

these devices have been mounted inside unmarked police vehicles to reduce the visibility

of operations. In addition, in 2001, flashless units were introduced to the speed camera

program for use in clear weather conditions, thereby further reducing the ability of road

users to detect the use of speed cameras.

The mobile cameras use conventional wet-film technology similar to that used in a

standard 35mm camera. Current digital technology is not sufficiently developed to enable

the introduction of mobile, digital safety cameras to the enforcement program. Trials of

these devices conducted by LMT found that the resolution of the images produced by

digital cameras is not sufficiently high to warrant their introduction. The clarity of images

captured by the existing wet-film, slant radar cameras is superior to that of the images

produced by current digital technology. Further, the processing of the images captured by

digital cameras in not currently cost efficient. It is expected that over the next two to five

years digital camera technology for use in a mobile setting will become available and

produce images of sufficient quality to enable their use in an enforcement setting.

In considering the future operations of the Victorian speed camera program it is necessary

to consider four key issues. The first of these is the effectiveness of the existing program

in terms of its impact on the frequency of casualty crashes in Victoria. The following

section highlights the key results of evaluations of the speed camera program as it operated

up to the year 2000/01 in Victoria. International experience in the use of mobile speed

cameras is also considered to assist in the development of strategic principles relating to

future operations.

The initial trial of speed cameras in 1985 attempted to significantly increase the number of

vehicles that could be detecting speeding per hour in comparison to traditional enforcement

methods. This objective was to be achieved through the use of a small number of cameras

operating at high crash frequency sites. The operations were highly visible and all camera

sites were clearly identified with appropriate signage. The overt nature of the program was

intended to invoke the general deterrence mechanism by raising the perceived risk of

detection for speeding offences. Those offenders actually detected were also expected to

be deterred by the penalties imposed as a result of the offence.

The results of an evaluation of the program’s operation demonstrated that the effect of the

speed camera trial was minimal. No statistically significant reductions in casualty crashes

in the areas surrounding the camera sites were found. In addition, the effect on speed was

limited to distances of approximately one to two kilometres from the camera sites (Portans,

1988). This suggests the following strategic principle.

The poor results of the initial speed camera trial led to a re-evaluation of the program and

the introduction of the covert speed camera program detailed in section 3.1 in 1989. This

program significantly increased the average number of infringement notices issued each

month with a target of 4,000 enforcement hours per month being set. During 2001, this

target was raised to 6,000 hours per month.

When evaluated from the period between December 1989 and December 1991 this

program had a significant effect on casualty crash frequency and severity (Cameron et al.,

1992). In particular, from December 1989 to March 1990, there was a statistically

coincided with low levels of both speed camera enforcement and speed related publicity.

During the period April 1990 to June 1990, when the publicity campaign was launched but

prior to extensive enforcement operations, low alcohol hour crashes were reduced by 34%

on Melbourne arterial roads and 21% in country towns. Reductions in the severity of

injuries sustained in these crashes were also found in Melbourne during this period.

Following the high levels of both publicity and enforcement experienced from July 1990,

low alcohol hour casualty crashes were reduced on arterial roads in Melbourne, country

towns and on rural highways by 32%, 23% and 14% respectively. The injury severity of

these crashes was also found to have decreased, principally in Melbourne. The effect of

the speed camera enforcement program on high alcohol hour crashes is less clear.

In addition to the estimation of casualty crash reductions, the impact of the mobile speed

camera program on speeds was also examined (Rogerson et al., 1994). From November

1989 to March 1990 no significant change in mean speeds was found. However, the

proportion of vehicles detected exceeding the speed limit by more than 15 km/h had

decreased in 60 and 70 km/h speed zones during this period. In particular, in 60 km/h

speed zones the proportion of vehicles exceeding the speed limit by more than 15 km/h had

decreased from approximately 11% to 5.5%.

Further analysis of the period from July 1990 to December 1991 indicated that there is an

additional effect of the mobile speed camera program which is localised in space to the

immediate area surrounding the enforcement sites (Rogerson et al., 1994). During the two

weeks following the receipt of TINs by offending motorists, a statistically significant 10%

one kilometre of the camera site. However, there was no reliable evidence of crash

reductions within one kilometre of the camera site during the week immediately following

a speed camera enforcement session. Further, no localised reductions in low alcohol hour

crashes or the severity of crashes were found during this period.

high-alcohol hours of the week are those periods when alcohol related crashes are more likely to occur.

The above results were updated in a more recent study examining the localised effects of

the speed camera program during the period from July 1990 to December 1993 (Newstead

et al. 1995). It was found that the speed camera program had no statistically significant

additional effect on casualty crashes following enforcement operations or the receipt of

TINs in rural towns. In contrast, statistically significant casualty crash reductions, in

addition to the general effect, in metropolitan Melbourne were identified and linked to both

camera operations and the receipt of TINs. The influence of TINs was evident during the

three weeks following their receipt and was greatest on all roads during high alcohol hours.

An 8.9% reduction in casualty crashes was experienced in high alcohol hours, on all roads,

during the week following the receipt of TINs.

It is worth noting the differences in the results of the two studies detailed above. First, the

estimated additional effect of the receipt of TINs is slightly less in the 1995 study; 6.2%

compared to 8.4%. Second, the 1995 study identified additional crash reductions during

the week immediately following the enforcement presence during both high and low

alcohol hours. The earlier study found no such evidence of crash reductions following the

enforcement presence. There are a number of possible explanations for these changes.

The additional two years of data included in the 1995 analysis may be affected by changes

in the behaviour of motorists during this period. Drivers may have become more aware of

speed camera sites and adjusted their driving behaviour at these sites. In turn, this may

have resulted in the reduced crash risk observed at the sites. However, further study would

be required to examine this possibility.

The mechanisms that drive reductions in casualty crashes have also been identified. Based

on 1990-91 data, relationships between the monthly level of low alcohol hour casualty

crashes and the inputs of the enforcement program have been established. Crash frequency

was related to the number of speeding TINs issued (generally 2-3 weeks after the offence

occurred) and the publicity levels in the same month. Also, crash severity was related to

camera operating hours and the number of speeding TINs issued (Cameron et al. 1992).

These results imply that actual detection of speeding drivers, as evidenced by the number

of TINs issued, is a key driver of the frequency of casualty crashes. TINs issued as a result

of speed camera operations were estimated to contribute reductions in serious casualty

crashes of 8-9% during the 1990-1993 period (Newstead et al., 1995).

An update of the primary analysis using data from January 1994 to December 1996

produced some different results (Gelb et al., 2000). In particular reductions in casualty

crash frequency were no longer attributable to TIN issuance. In addition, the number of

hours of enforcement no longer had a positive effect on crash severity. Nevertheless,

casualty crash severity was still affected by the number of TINs issued. This suggests that

there has been a change over time in the mechanisms behind the speed camera enforcement

program.

A more recent study has confirmed the key role of the number of speeding TINs detected

having an influence on crashes in subsequent periods (Cameron et al., 2003a, b). During

1999, the Victoria Police varied the levels of speed camera activity substantially in four

Melbourne police districts according to a systematic plan. Camera hours were increased or

reduced by 50% or 100% in respective districts for a month at a time, during two separate

months when speed-related publicity was present and during two months when it was

absent. Monthly casualty crashes in the ten Melbourne police districts during 1996-2000

were analysed to test the effects of the enforcement, publicity and their interaction.

Monthly levels of speeding offences detected by cameras varied substantially over time in

all districts, but the most extreme variations occurred in the four districts as planned.

Changes in crash frequency were found to be inversely associated with changes in the

levels of speeding TINs detected in the same district during the previous month (Figure 1).

The risk of fatal outcome of the casualty crashes was also reduced by more than 40% when

the level of speeding TINs detected during the previous month was at relatively high levels

(more than 30% greater than average).

Figure 1: Relationship between crashes and level of speeding TINs detected by speed cameras

-4

-2

0

2

4

6

8

0 1000 2000 3000 4000 5000 6000

From the above discussion the following conclusions and strategic principles related to the

mobile speed camera program in Victoria can be suggested.

1. Significant reductions in low alcohol hour casualty crashes were experienced during

the initial implementation phase of the new, covert speed camera program when

publicity was high but actual enforcement levels were relatively low. This suggests

2. Following the full implementation of the mobile speed camera program (including

supporting mass media publicity), statistically significant reductions in low alcohol

hour casualty crashes were found across arterial roads in Melbourne and country towns

3. Localised additional reductions in high alcohol hour casualty crashes were experienced

4. Later studies have identified additional casualty crash reductions within one kilometre

of enforcement sites in the week following enforcement sessions. This demonstrates

5. From all the research relating crash and speed effects to the receipt of TINs emanating

To ensure that these principles are consistent with national and international experience of

mobile speed cameras, it is useful to examine the results of evaluations of these

technologies in British Columbia, Canada, New Zealand and Queensland, Australia.

The province of British Columbia implemented a province-wide speed camera program

commencing in March 1996. The program made use of photo radar devices transported in

and operated from unmarked vans. No signage was present at the speed camera sites and

the radar unit was visible only after a vehicle had passed the van. Therefore, the camera

program can be considered to have operated covertly. However, there was significant

mass media publicity of the enforcement operation prior to and throughout its operation.

Further, a survey of residents found that approximately 95% of residents were aware of the

program prior to its commencement.

Operations were scheduled to operate primarily during daytime hours at sites with a history

of high crash frequency or perceived speeding behaviour. Within the first year of

operation the cameras achieved approximately 30,000 enforcement hours which resulted in

250,000 infringement notices being generated. The penalty for exceeding the posted speed

limit ranged from $100-150, however, during the first five months of the program warning

letters were issued and no fines imposed.

The program was found to have beneficial outcomes when evaluated with respect to a

number of variables (Chen et al., 2001). First, it was found that the proportion of vehicles

exceeding the speed limit at the camera sites dropped by 50% from May to December

1996. However, reductions of 75% were found for those vehicles exceeding the speed

limit by more than 16 km/h. The single largest monthly reduction was experienced in

August, when fines for exceeding the speed limit were introduced. Reductions in traffic

speed were also identified at the control sites. Prior to the implementation of the program,

an average of 69% of motorists exceeded the speed limit at these sites. Following the

introduction of fines this had decreased to an average of 61% of motorists. The authors

suggest that this is due to a generalised effect of the speed camera program.

The estimated program effect with respect to the frequency of daytime unsafe speed related

crashes, indicates that the program led to a 25% reduction in these crashes after

infringement notices were incorporated into the program. It is noted that these crashes are

classified as crashes in which unsafe speed was identified as a contributing factor by the

Police attending the crash scene. Daytime traffic collision injuries requiring transportation

by ambulance also decreased by an estimated 11% during this period and fatalities

resulting from daytime traffic collisions were estimated to have fallen by 17% as a result of

the speed camera program.

The above results are generally consistent with those found in relation to the operation of

the Victorian mobile speed camera program and confirm that covert operations can

successfully reduce the frequency of casualty crashes. Although the analysis of the

Canadian program does not examine the causal links between the speed cameras and

improvements in road safety, when viewed in conjunction with the Victorian research it

suggests the following conclusions and strategic principles.

1. The British Columbian program demonstrates that the use of covert, mobile speed

camera operations is an effective means of reducing both vehicle speeds and casualty

2. The imposition of monetary penalties will also increase the deterrence effect of

The introduction of mobile speed cameras in New Zealand commenced in late 1993. The

operation of the cameras was restricted to roads classified as ‘speed camera areas’ based on

a record of speed related crashes. Entrances to these roads were clearly sign posted to

ensure that motorists were aware of the potential presence of the speed cameras. Further,

the majority of speed cameras were mounted on police cars and operators were prohibited

from hiding the cameras. In urban areas, limited use was made of fixed position speed

cameras mounted on poles, however, these were subject to the same signage requirements

as the mobile camera operations. In total, 13 fixed and 31 mobile cameras have been

operating in New Zealand since 1993. Prior to July 2000, the enforcement threshold was

Financial penalties (but no demerit points) were imposed where vehicles were detected

travelling at or above the enforcement threshold. However, since 1 July 2000 a flat 10

km/h enforcement threshold has been in operation.

An evaluation of the effect of the speed camera program described above, found that fatal

and serious crashes on roads with speed limits of 70 km/h or less were reduced by an

estimated 13% during low alcohol times of day (Mara et al., 1996). In speed camera areas,

the reduction in fatal and serious low alcohol hour crashes was 23.3%. Less substantial

reductions in all injury crashes were experienced in speed camera areas on roads with

speed limits of 100 km/h. No effect on crashes was identified on these roads when nonspeed

camera areas were included in the analysis.

Mobile speed cameras were first introduced to Queensland in 1997 and a total of 15

camera units were in operation by June 1997. The camera program operates in an overt

manner with marked vans stationed at the enforcement sites. Signs advising of the camera

presence are also displayed when cameras are in operation. In addition, a public education

program was conducted in late 1996 prior to the commencement of operations. The

camera technology used in Queensland is the same as that currently being used in Victoria

and discussed earlier in this report. The selection of speed camera sites is based on speed

related crash history, however the scheduling of the operations is conducted randomly.

The randomised allocation of resources between the approved camera sites is based on the

Random Road Watch technique as detailed in section 4.2.

Preliminary results of the evaluation of the speed camera program indicate that the mobile

speed camera program has reduced crash frequency by between 12 and 17 percent within a

6km radius of the camera sites. The greatest effects were found within a 2km radius of the

camera sites. Further, there is some suggestion that the crash reduction increased over time

particularly for higher severity crashes. No investigation as to the existence of a

generalised effect of the camera program was conducted.

On the basis of the analysis of both the New Zealand and Queensland mobile speed camera

programs, the following conclusion and strategic principle can be defined.

mobile speed camera program has yet been identified.

The second key issue in relation to the operation of mobile speed cameras is the contrast

between overt and covert operations. There has been little research directly comparing the

impact of the mode of operation on the effectiveness of a mobile enforcement program.

However, some evidence does exist and is discussed below.

As detailed above the operation of mobile speed cameras in New Zealand is conducted in a

highly visible manner. However, from mid-1997 to mid-2000 a trial of the covert use of

speed cameras was conducted in one of the four police regions in New Zealand on roads

with speed limits of 100 km/h. This involved adding to existing signage an indication to

motorists that hidden cameras may operate in the speed camera areas. In addition to the

extra signage, there were high levels of newspaper and radio publicity relating to the trial

prior to its commencement. It is also noted that in the first year of operation there was a

26% increase in the operational hours of speed cameras in the trial region. In the second

year of operation, the number of operational hours decreased by 13% from the first year

level. There were no changes in the operation of speed cameras during the trial period in

other areas of New Zealand. In particular, on all roads in non-trial speed camera areas,

speed camera operations remained overt. Further, on roads with speed limits of 70 km/h or

less in the trial region, speed cameras were operated overtly.

An evaluation of the hidden camera trial in terms of vehicle speeds and reportable crashes

demonstrated that during the first two years of the trial, improved road safety outcomes

were experienced (Keall et al., 2002). First, average speed in the trial regions decreased by

an estimated 1.3 km/h over the first two years of the trial. The speed below which 85

percent of vehicles travelled in the trial region fell by an estimated 4.3 km/h. In addition,

reportable crashes in the trial region fell by 11% in comparison to reportable crashes in the

control regions. Further, it was found that the number of casualties in the trial region fell

by 19% in comparison to casualties in the control regions. The number of casualties per

crash fell by 9% on open roads in the trial region compared to open roads in the control

regions. It is noted that these results relate to reductions across the treated region and not

only at camera sites. This indicates that the covert mobile operations were able to

generalise the effect of the New Zealand program beyond the speed camera sites.

Despite the above results it is difficult to draw conclusions from this study on the relative

effectiveness of overt and covert automated speed enforcement programs. During the trial

period, enforcement levels in the trial region were higher than in the non-trial regions.

Further, the number of penalties issued in relation to incidents in the trial areas increased

four fold (Keall et al., 2002). Therefore, based on previously established relationships

between speed enforcement and crashes (Cameron et al., 1995), it is not unexpected that

improvements in road trauma would occur as the level of enforcement increases.

Nevertheless, the authors point to three factors which they believe together support the

conclusion that the introduction of covert speed cameras influenced the casualty crash

reductions. First, the fall in the frequency of casualty crashes coincided with the

introduction of the covert program. Second, mean and high percentile speeds fell

significantly during the trial. Finally, the reduction in the number of casualties per crash

also confirms that speeds fell during the trial period.

To clarify the comparitive effect of covert and overt mobile speed enforcement operations,

it is useful to examine some related Victorian research although it is noted that it does not

relate directly to mobile speed cameras.

The effect of mobile (moving mode) radar speed detection devices on road trauma in rural

Victoria has been examined in terms of the type of enforcement operation. That is, the

effect of covert (unmarked car), overt (marked car) and mixed (marked and unmarked cars)

mobile radar operations has been examined to identify any differences between the

outcomes of different types of enforcement activity (Diamantopoulou and Cameron, 2001).

The analysis was conducted using crash data from July 1995 to June 1997 which was

divided into two periods. These periods were July 1995 to June 1996 and July 1996 to

June 1997 and corresponded with the use of 48 and 73 mobile radar devices respectively.

Analysis was also conducted on the two periods combined when up to 73 mobile radar

device were in operation.

enforcement presence was identified during the period from July 1995 to June 1996. The

days after enforcement however, the effect was less pronounced. During the period from

July 1996 to June 1997, the largest reductions in casualty crashes occurred following

This effect was greatest on the day on which the enforcement activity took place (40.2%

reduction).

The results of the combined period in which up to 73 mobile radar devices were in

operation found that the most significant reductions in casualty crashes occurring one to

mix of overt and covert enforcement was also found to be effective in reducing casualty

crashes during this period.

It is noted that the crash reductions presented above are not statistically significant.

Nevertheless the results are indicative of the likely relationships between overt, covert and

mixed mobile radar enforcement and casualty crashes in rural Victoria.

In developing strategic principles in relation to the mode of operation of the mobile speed

camera program, the limitations of the above research must be noted. First, the Victorian

research concerns the operation of mobile radar operations and not mobile speed cameras.

Second, the research relates to rural areas of Victoria only.

Despite these limitations the following conclusion and strategic principle is suggested on

the basis of the New Zealand and Victorian research.

camera presence, and marked cars in parallel in the case of mobile radar, may help to

remind drivers of the unseen threat of the covert operations, thus increasing general

deterrence.

The intensity of operation is relevant to the consideration of the mobile speed camera

program. A cost benefit analysis of the Victorian mobile speed camera program as it

operated up to 2000, has been conducted (Gelb et al., 2000). The study aimed to evaluate

whether the existing speed camera operations involving approximately 4000 hours of

operation per month was best practice. This was done using a marginal cost/benefit

analysis in terms of both hours of operation and TIN issuance.

The economic analysis was limited to speed camera enforcement in Melbourne during low

alcohol hours because most speed camera activities occur in Melbourne during these hours.

In addition, crash data for low alcohol hours is less likely to include accidents associated

with drink driving. Crash data from 1987 to 1998 was used along with data relating to the

actual costs of running the speed camera program and the costs of casualty crashes to

society in general. Further, the results presented below are drawn from the economic

analysis based on TIN issuance. These results are likely to provide a better estimate than

those based on hours of operation due to the greater variability of the TINs issued per

month.

It was determined that, in order to reduce the social costs (camera operations plus casualty

crash costs), the number of TINs issued per month should fall within the range of 37,000 to

66,000. This corresponds to a range of 3,592 to 6,408 enforcement hours per month and an

optimal average investment per month of 5,146 enforcement hours. This would be

expected to result in a reduction in monthly levels of low alcohol hour casualty crashes of

13%. Further, the marginal benefit cost ratio was determined to be 6.3. That is, by

investing in an average of 5,146 operational hours per month the benefits obtained in

reduced social costs per casualty crash would be 6.3 times the cost of investment.

These results are limited to benefits of speed camera enforcement under the program

existing up to 1998. The results would change if any of the elements of the program such

as camera technology or the operational principles underlying the timing and location of

enforcement operations were to change. Any technological advance that would increase

the number of speeders detected per hour of camera operation would increase the

benefit/cost ratio of the program. Further, if it became possible to issue the same number

of TINs with fewer hours of enforcement, the cost of the program would decrease and the

benefit/cost ratio would necessarily increase. If such changes did occur, the estimates

provided above could be viewed as the lower bounds of an economic assessment of a

redesigned speed camera program.

During August 2001 to February 2002, the number of hours per month planned for speed

cameras to operate was increased from 4,000 to 6,000 hours. There have been few other

changes in mobile speed camera operations, apart from the use of cameras without flash

assistance during daylight conditions, and reductions in speeding offence detection

thresholds (both changes likely to increase, rather than decrease, the number of speeders

detected per hour, at least in the short-term; see section 5.6.2). Research based on the

increased speed camera hours in some Melbourne police districts during specific months in

1999 provides support for likely reductions in crashes following a 50% increase in camera

hours (Cameron et al., 2003a, b; see Figure 1). The increased number of hours per month is

consistent with the range of estimates for the optimal investment of camera hours per

month, based on the economic analysis described above. This suggests the following

strategic principle.

Finally, the issue of enforcement threshold is a relevant consideration to the future

operation of the mobile speed camera program. An enforcement threshold is defined as the

speed at or above which an infringement notice will be issued. In Victoria, this speed is

not identical to the posted speed limit and has changed over time. The speed tolerance is

the speed at which vehicles may travel without incurring a penalty. When the speed

camera program was first introduced in Victoria, a speed tolerance of 10 percent of the

speed limit plus 3 km/h was set. The Australian Design Rule (ADR) covering

speedometers require that the indicated speed be within 10% of the actual speed. Further,

the speed camera technology may have an error of up to 3 km/h. It is believed that the

original speed tolerance was set in view of the variation in these devices. Therefore, in a

60 km/h speed zone, vehicles could travel up to the tolerance level of 69 km/h without

receiving an infringement notice. In February 1993, 110 km/h speed limits were

reintroduced for high quality freeways. To reduce the speed tolerance in these speed

zones, a flat 10 km/h enforcement threshold was introduced on all roads. In March 2002,

Victoria Police announced further staged reductions of speed enforcement thresholds.

However, the details of these reductions have not been widely publicised.

It is noted that the travel speed alleged on traffic infringement notices is 3 km/h less than

the speed detected by the camera device. Prior to 2000, both the detected and alleged

speeds were recorded on infringement notices. Since at least January 2001 only the

alleged speed has been documented on infringement notices.

As the reduction in the enforcement threshold is a recent occurrence, the effect of this

action on casualty crashes or the speeds of vehicles has not yet been assessed. However,

the effect of reducing enforcement thresholds has been evaluated in the Swedish context.

In 1987 reduced tolerance levels for speeding offences were implemented in the two

Swedish cities of Halmstad and Jönköping. At the same time, increased penalties for

speeding offences were introduced in July 1987 and a campaign aimed at reducing speeds

was implemented. An evaluation study was conducted to determine the impact of these

programs on speeds in the treatment areas (Andersson, 1990). The study used four distinct

urban areas as control sites and measured speeds during 1986 and 1987.

Speeds in the treatment areas fell by approximately 0.8 to 1.2 km/h from 1986 to 1987. In

contrast, in the control areas where the enforcement threshold did not change, there was a

slight increase in speeds over this period. It is suggested by the authors of the study that

the speed reduction was most likely due to the increased risk of detection caused by the

lower enforcement thresholds rather than the increased penalties associated with speeding

offences or the campaign targeted at reducing speeds. This conclusion is reached on the

basis of surveys of drivers travelling through the treatment and control sites. One in three

drivers reported driving slower than previously and the majority of motorists reported that

this was due to the increased levels of police activity. Few drivers were aware of the new

penalties and only ten percent of drivers who reported driving more slowly did so as a

result of the campaign targeted at reducing speeds.

In view of this research the following conclusion and strategic principle is proposed.

This conclusion is further supported by the impact of a reduction in the enforcement

threshold in New Zealand. Prior to July 2000, the enforcement threshold was determined

the introduction of a flat 10km/h enforcement threshold applying across the road network

in July 2000, the number of vehicles exceeding the speed limit at speed camera sites

declined dramatically. Specifically, there was a 50% reduction in the proportion of

vehicles detected exceeding the 10km/h tolerance at camera sites in the six weeks

following the introduction of the reduced tolerance (Robinson, 2001). This reduction has

been sustained over time. The translation of this reduction in offence rates into

improvements in road trauma has not been evaluated. However, the proportion of drivers

travelling at over 110km/h on rural roads (typically with 100km/h speed limits) fell from

24-26% during 1997-1999 to 20% in 2000, 15% in 2001 and 10% in 2002 (Land Transport

Safety Authority, 2003). Whether this reduction in speeding was solely due to the reduced

enforcement tolerance is unclear.

In comparison to other forms of speed enforcement technology, fixed speed cameras are a

relatively new method of automated speed enforcement in Victoria. The first fixed speed

cameras used in Victoria were located on the Monash Freeway (CityLink) in the Domain

tunnel and commenced operation in April 2000 when that tunnel was first opened for

public use. Since that time, fixed speed cameras have also been installed in the Burnley

tunnel and on the Monash Freeway some distance before the entrance to the Domain

tunnel.

The operation of these fixed cameras is intended to be as covert as possible. However,

over time, although the exact location of the cameras may remain unclear, the public have

become aware of the existence of the cameras on the Monash Freeway particularly in the

two tunnels. This has been achieved through extensive media comment on their presence.

Nevertheless, there is no signage indicating the use of speed cameras at the sites and where

necessary blanking plates are used to hide the cameras from view.

The fixed cameras operating in these positions are analog video cameras which are capable

of operating continuously. Within each of the tunnels two pairs of camera banks are

installed facing opposite directions to enable both the front and rear registrations plates to

be captured. However, it is most common for only one pair of these camera banks to be in

operation at one time. Each bank of cameras contains 2 cameras for each lane of traffic to

be viewed. The first of these takes a wide view of the traffic and the second captures close

up images of offending vehicles to enable registration details to be collected. Although the

cameras operate continuously they retain images only when a speeding offence is detected.

Once an offence has been detected the data is relayed for processing remotely and there is

no need to access the site for this purpose. The images captured by the analog video

camera must be converted to digital format to enable the processing of the offences.

In addition to the fixed camera installations described above, new banks of cameras have

been installed along the Western Ring Road and are proposed for the Melbourne-Geelong

Freeway. These roads are thought to have been selected on the basis of high proportions of

speeding drivers and severe injury crashes. The cameras operate in a similar manner to the

existing cameras, however the technology used differs. Digital cameras have been chosen

for these routes following trials to ensure the viability of this technology.

The effectiveness of the fixed-position speed cameras was evaluated in terms of the impact

on vehicle speeds in the tunnel (Diamantopoulou and Corben, 2001). Analysis of the

effect on the enforcement was conducted with respect to each lane in the tunnel, the day of

the week of travel and the hour of the day of travel.

The overall effect of the fixed-position speed cameras was to reduce the proportion of

those drivers exceeding the speed limit and to reduce the average speed of vehicles in the

tunnel. Average vehicle speeds fell from 75.05 km/h to 72.50 km/h. The proportion of

drivers exceeding the 80 km/h speed limit fell by 66%. In addition, the proportion of

drivers exceeding speeds of 90 and 110 km/h were also significantly reduced by 79% and

76% respectively.

A lane-by-lane analysis of the effects of the speed enforcement initiative suggests that the

cameras were effective in reducing average vehicle speeds and the proportion of drivers

exceeding speeds of 80, 90 and 110 km/h in all lanes of the tunnel. However, the effect

was greater in the left lane than the right.

The day-by-day analysis also indicated that reductions in these measures of effect were

experienced on both weekdays and weekends. In particular, average vehicle speeds were

reduced by 2.9% during the week and 5.4% on weekends. Also, the proportion of drivers

exceeding the speed limit was reduced by 65% during the week and 68% on weekends.

For the time of day analysis, it was found that the fixed-position speed cameras were

effective in reducing both average vehicle speeds and the proportion of drivers exceeding

speeds of 80, 90 and 110 km/h for most time periods on weekdays. The most significant

reduction was experienced in the afternoon peak period. However, during the morning

peak period no reductions in either average vehicle speeds or the proportion of vehicles

exceeding 110 km/h were experienced. In addition, there was no reduction in the

proportion of vehicles exceeding 110 km/h during the non-peak daytime period.

The cameras were also shown to be effective in reducing average vehicle speeds during all

time periods on weekends. Reductions in the proportion of vehicles exceeding speeds of

80, 90 and 110 km/h were also generally found in all time periods.

Speed cameras were first introduced into the United Kingdom in 1992 and by 1994 there

were thirty speed cameras in use. Since that time the number of speed cameras available

for use has increased significantly. However, the exact number of fixed speed cameras

currently in use is unclear. A 1996 study conducted by the Police Research Group (Hooke

et al., 1996) indicated that in 10 out of the 43 Police forces in England and Wales there

were 102 cameras in use which rotated through 475 speed camera sites. It is noted,

however, that some of the 102 cameras were used as red-light rather than speed cameras

and both fixed and mobile wet-film radar cameras are included. The number of cameras in

use by the remaining Police forces is unknown.

Both fixed and mobile speed camera operations operate overtly in the UK. Indeed, new

camera visibility rules were introduced in June 2002 to further increase the visibility of

speed camera operations. These rules required that camera housing be yellow and visible

from specified distances from the camera sites. There was a clear intention that drivers be

aware of the location of the cameras. In March 2003, there was a High Court challenge to

this requirement, on the grounds that the increased visibility reduced the ability of the

cameras to have an effect on speeds beyond the camera sites because the element of

unpredictability was reduced. This challenge was withdrawn in Court and the rules remain

in place.

The minimum enforcement threshold for the UK speed camera program is 10% of the

speed limit plus 2 mph (3 km/h), i.e. the camera can detect speeding at 35 mph (56.4 km/h)

and above in the 30 mph (48 km/h) speed limit zones commonly used in urban areas. All

speeding offences attract a fixed fine penalty, which is currently £60, and three demerit

points (accumulation of 12 points in a three year period leads to licence loss, as in

Victoria). There is a proposal to increase the fine for speeding offences 15 mph or greater

in excess of the limit, but currently there is no escalation of the fine or points for higher

levels of speeding, unlike the situation in Victoria (see section 5.13).

A cost benefit analysis of the speed camera program in ten police force areas conducted in

1996 (Hooke et al.) determined that accidents at the speed camera sites fell by 28%

following the installation of the cameras. The speed of vehicles at camera sites also

decreased by an estimated average of 4.2 mph (6.8 km/h) at each site. In financial terms

the cost incurred in installing the cameras were returned five-fold after one year of

operation. After five years, the speed cameras had generated a return 25 times the initial

investment. The benefits were predominantly reductions in crash costs. Further, the fine

revenue from the cameras in nine of the ten Police Force areas studied, was sufficient to

cover the recurrent costs associated with the speed cameras. This demonstrated that the

speed camera program operating in this form led to reductions in road trauma and was cost

beneficial. It was noted, however, that the full benefits of speed cameras were not being

achieved due to budgetary constraints. It was also noted that the fine revenue could cover

the costs of an expansion in the program.

In response to a recommendation of the 1996 study concerning the constraints being placed

on expanding the camera enforcement activity by the costs of cameras and their operation,

the relevant authorities agreed to allow a two-year trial in eight areas of Great Britain in

which the costs of camera enforcement and prosecution could be recovered from fine

revenue. The trial commenced in April 2000 and as expected this resulted in increased

enforcement in the trial areas. These areas were chosen to achieve a balance between

geography, crash history and different enforcement strategies and technologies. In each

area an operational partnership was formed to run the “safety camera scheme” comprising

the police force for the area, the highway authorities and the courts. The term “safety

cameras” is used as a generic term to include both speed and traffic light cameras used in

the partnership areas, but the majority are speed cameras.

The impact of the increased funding for speed camera operations was examined in terms of

both reductions in speed and casualty crashes during the trial. The results from the first

year were so encouraging that the U.K. government took the decision to extend the system

nationally before the pilot phase was completed (Gains et al., 2003).

The results from the first two years in the pilot areas allowed the effects of the fixed and

mobile cameras to be compared (Gains et al., 2003). The fixed cameras being permanent

could be expected to affect speeds and crashes at all times, but the mobile cameras may

have an effect only at the time they are present and for a period thereafter. (However the

study examined effects throughout the year at mobile camera locations; it is not known

how frequently each mobile site was enforced.)

Over the full two years, average speed at camera sites fell by 10% or 3.7 mph (6 km/h).

The decrease in average speed was slightly greater at fixed camera sites, but there was a

much greater fall in the proportion of vehicles speeding at fixed camera sites (67%) than at

mobile camera sites (37%). When excessive speeding was examined (exceeding the speed

limit by more than 15 mph), the proportion of vehicles fell by 96% at fixed camera sites

and by 55% at mobile camera sites.

Serious casualty reductions were also somewhat smaller over the full two years (35%

reduction in fatal and serious injuries at camera sites), but the results showed that the

reduction was greater at fixed camera sites (65%) than at mobile camera sites (28%), as

could be expected given the relative magnitudes of the speed behaviour changes. The crash

effects were similar in urban and rural areas, with pedestrians being particular beneficiaries

of the program (56% reduction in fatal and serious injuries at camera sites). It should be

noted that at mobile camera sites crash data for the whole two year period was used and

not just those crashes occurring at a specified time after mobile camera activity at a site.

The effects of the overt cameras appeared to generalise across the whole of the trial areas,

with the average number of fatal and serious injuries in each area being 4% below the

long-term trend in serious road trauma in the rest of Great Britain. While the camera sites

are located in speed-related “accident hot spots”, the density of their locations and/or their

threat to speeding motorists appears to be sufficient to produce a general effect which

extends beyond the camera sites.

The study also found that public support for the use of speed cameras was consistently

high throughout the period, with 80% agreeing that “cameras are meant to encourage

drivers to keep to the limits not punish them”. It also found that the system was successful

in redirecting £20 million of fine revenue to local areas to fund the camera operations, and

that there were benefits to society, in terms of the value of road casualties saved, of £112

million during the first two years of the program in the trial areas.

Trials of fixed position speed cameras were conducted in Sweden from 1990-1992. Eight

roads with speed limits of 90 km/h and eight roads with speed limits of 50 km/h were each

fitted with two fixed position speed cameras. Signs advising drivers of the use of speed

cameras were also used on these road sections.

The average speed of vehicles travelling on the treatment roads was reduced by

approximately 2.3 km/h. The speed reduction at the camera sites themselves was

estimated to be between 5 and 10 km/h. In addition to these reductions in speed, casualty

crashes were estimated to have fallen by 5 percent and the number of casualties involved in

collisions fell by an estimated 9 percent. However, it is noted that these results are not

statistically significant.

In the above three studies the evaluation of fixed speed camera technologies has been

conducted primarily in terms of reductions in average speeds and casualty crashes. The

UK study also conducted a comprehensive cost-benefit analysis of the program.

Prior to drawing conclusions based on the above research it is necessary to note the

differences in the programs discussed. First, the operation of fixed speed cameras in

Victoria remains semi-covert. In contrast, the existence of fixed position speed cameras in

the UK and Sweden are heavily sign-posted. In addition, the road environment in which

the different programs operated did vary somewhat. The use of fixed speed cameras in

Victoria is currently restricted to use in tunnels and urban freeways (with rural freeways to

be included soon). The UK and Swedish operations were not so restricted in their

operation.

Despite these differences a number of conclusions and strategic principles can be

proposed.

camera sites themselves.

3. In general, fixed speed camera programs result in reductions in both average speeds

4. Fixed position speed cameras can operate effectively in a number of environments

including tunnels and low and higher speed roads.

Point-to-point speed measurement devices do not currently operate in Victoria. However,

the Hume Highway has been identified as a potential route on which to use this

technology. The technology uses a number of cameras mounted at staged intervals along a

particular route. The cameras are able to measure the average speed between two points or

the spot speed at an individual camera site. In order to measure the average speed between

two points the cameras must be linked to one another and the time clocks on both machines

must be synchronised. The average speed is then determined by dividing the distance

travelled by the time taken to travel between the two points. The distance between two

camera sites may vary from as low as 300 meters to up to tens of kilometres. An

enforcement threshold may also be implemented to allow for acceptable variations in

driver speed along the route. However, the speed limit and enforcement threshold set for

these roads need to be considered in light of the accuracy of the available technology and

the type of speed being measured. Potentially, however, a lower enforcement threshold

could be considered for the average speed measured by this technology than the spot

speeds measured by mobile and fixed speed cameras.

Although point-to-point speed camera technology has not yet been introduced for use on

Australian roads, the equipment has been trialled in the Netherlands. The trial involved the

installation of three cameras along a major motorway with high traffic volumes. Vehicles

passing each of the camera sites were photographed and an electronic image processing

device was used to match records of the same vehicle. The cameras were located 750 m

and 2.25 km apart. If the average speed of vehicles between the camera sites (determined

by the distance/time relationship) was in excess of the speed limit, infringement notices

were issued. No study of the effectiveness of the Dutch trials in terms of speed or crash

effects is currently available.

In the U.K., the first implementation of point-to-point camera technology, using digital

imaging, was installed on Nottingham’s main link road from the M1 Motorway in July

2000, as part of the trial program of additional speed cameras in eight Police areas (see

section 3.2.2). The evaluation of the trial found that fatal and serious injuries fell by 31% at

camera sites in the Nottingham area, and that the results from the point-to-point camera

site were not significantly different from the general effect (Gains et al., 2003). In a

comparison with traditional wet-film spot-speed fixed cameras, Keenan (2002) found that

reported casualty crashes at the Nottingham digital camera site fell from 33 during the year

before installation to 21 during the year after, a reduction of 36%. In contrast, crashes at

the spot-speed camera sites studied appeared to increase, but not statistically significantly

so.

The Nottingham point-to-point camera system was manufactured by Speed Check Services

(SPECS) and consisted of two digital cameras located approximately 0.5 km apart on a 40

mph (64 km/h) limit urban ring road. SPECS systems have been used at other locations in

Great Britain, sometimes using more than two linked cameras, including other urban links

and rural main road sites.. The SPECS web-site claims that, at the original Nottingham site,

there was 5-6 mph (8-10 km/h) reduction in speeds, 40% reduction in serious injuries, and

30% reduction in slight casualties during the two years after installation.

Commenting on the relative merits of the new technology, Keenan (2002) noted that the

spot-speed fixed cameras have a site-specific effect whereas the point-to-point camera

system has a link-long influence on drivers and their speeds. The new technology achieves

speed enforcement along a whole length of road by calculating a driver’s average speed

over the link. The overt nature of the operations in the U.K. would result in drivers

knowing this.

Keenan (2002) also noted from his study that “around the [spot-speed camera] sites a

significant proportion of the drivers observed manipulated their behaviour in close vicinity

to the installations, suddenly applying their brakes 50 metres before the camera and then

promptly accelerating away from it. Most alarming was the fact that the accident statistics

at some of the [spot-speed camera] sites had worsened since the camera installation”.

While the crash data were probably too few for Keenan to claim that the situation had

worsened, it is possible that any speed and crash reduction benefits at the overt fixed spotspeed

camera sites were eroded by some drivers behaving in the way Keenan suggests.

However, given the policy in the U.K. of making fixed camera sites conspicuous and the

placing of advance camera warning signs a requirement of the scheme, there should be less

likelihood of drivers being taken by surprise. This effect may be even less likely to be a

significant consequence of the point-to-point camera systems.

However, the relative cost of the two types of fixed camera systems is a disadvantage.

Keenan (2002) estimated that a typical spot-speed wet-film speed camera cost £45,000 to

install, whereas the single point-to-point system in Nottingham cost £136,400 to install, a

three times factor. The SPECS web-site estimated that a traditional camera costs about

£40,000, whereas a pair of SPECS digital cameras costs £70,000 and requires at least

another £100,000 for the computer network to support it.

Red light cameras were first introduced to Victoria in a six-month trial of the technology

conducted in 1981. In August 1983, ten new red light cameras were purchased and

commenced operations throughout Victoria. There are currently 35 red light cameras

operating throughout metropolitan Melbourne and these cameras rotate through 132 red

light camera sites. The red light cameras operate by taking two photographs of a vehicle as

it passes over the detection loops imbedded in the road surface. The first captures the rear

of the vehicle and will enable it to be identified. The second is used to determine whether

the vehicle continued into the intersection after passing over the detection mechanism. It is

noted that the camera only operates when a vehicle passes over the detection loop 0.9 or

more seconds after the traffic lights have turned red.

Red-light camera sites are selected on the basis of the crash history at individual sites.

Three or more ‘cross traffic type’ collisions must have occurred at an intersection over the

preceding five years in order for a red-light camera to be installed at that intersection. The

rotation of cameras through sites is determined on the basis of crash history, offence

frequency, equipment and site availability, and operator knowledge of intersections.

Further, the operation of the red light cameras is no longer entirely overt in nature. In the

past, each red light camera sight was sign posted on all legs of the intersection regardless

of whether a camera was in operation at the time. Such signage is no longer required.

There has been only limited research conducted on the effectiveness of red-light cameras

both in Victoria and internationally. Below the results of evaluations of the Victorian,

U.K. and Californian red-light camera programs are presented.

An evaluation of the effectiveness of the red-light camera program using crash data from

1981 to 1986 has been conducted examining the impact of the program on various crash

types (South et al., 1988). During this time, 46 sites were in operation and cameras rotated

between these sites. The results of this study indicate that the proportion of right-angle

crashes occurring at treatment sites decreased by an estimated 32% over this period.

(Right-angle crashes are defined as those crashes in which vehicles approaching from

adjacent arms of an intersection collide at right angles.) No statistically significant

reductions in other crash types were identified in this study. Although non-significant, the

results suggests that rear-end collisions also decreased after the introduction of the redlight

camera program. This surprising result is explained by the authors as a result of

vehicles approaching treatment intersections at lower speeds than previously and thus

reducing the likelihood of a crash. Overall, the number of casualties involved in collisions

at the treatment sites decreased by an estimated 10.4%. Further, accident cost savings of

13.8% were estimated to have been generated by the red-light program.

The use of red-light cameras in the UK in now widespread. In 1996, 254 red-light camera

sites were in operation in10 of the 43 Police forces in England and Wales. A total of 102

cameras were available for use, however, these were distributed between both speed and

red-light camera sites. The use of the red-light cameras were associated with an 18%

reduction in casualty crashes at the camera sites (Hooke et al., 1996). The cost benefit

analysis of this technology also produced positive results. Within a year of implementing

the program the majority of sites had more than recovered the cost of the initial investment.

Within five years the program had returned twelve times the initial investment.

The red-light camera program introduced in July 1997 in Oxnard, California has also been

the subject of an evaluation study (Retting et al., 1999). This program involved the

installation of nine red-light cameras and numerous signs throughout the city warning

motorists of automated enforcement at signalised intersections. The penalty for failing to

obey the traffic signals was a financial penalty of $104 and one demerit point on the

driver’s licence. A one month warning period operated prior to full enforcement during

which no penalties were issued.

The results of the study indicate that the proportion of vehicles offending at both camera

and non-camera sites decreased by approximately 42% in the four months following the

introduction of the program. The proportion of vehicles offending at control sites, located

outside the city of Oxnard, did not change significantly during the evaluation period. This

suggests that the effect of the speed camera program extended beyond the treatment sites to

other nearby intersections in the city of Oxnard. The reductions in offence rates within

Oxnard were associated with a 29% reduction in casualty crashes at signalised

intersections. In addition, the frequency of rear-end collisions did not increase

significantly. These results are consistent with those of the Victorian study, however, it is

noted that this study is limited to the effects of the camera program in the four months

following the implementation of the speed camera program. Therefore, no conclusions can

be drawn from this study about the long term impact of the program on crashes associated

with red-light running.

In considering the above evaluations, it is noted that the red-light camera programs

discussed above, were conducted at a time when signage always accompanied the

automated enforcement at signalised intersections. In Victoria such signage is no longer

required and the use of red-light cameras has become more covert in nature. Therefore, the

following conclusions must be read with this in mind.

leading to reductions in casualty crashes at nearby intersections.

As detailed previously, the use of speed and red-light cameras in Victoria has been wide

spread over the past decade. However, during 2002, a trial of new red-light/speed camera

technology commenced. Combined red-light/speed cameras are capable of operating

simultaneously as both red-light and speed cameras. The devices are installed in fixed

housings at signalised intersections and it is expected that they will be capable of rotation

through a number of designated sites.

During the green and amber phases of traffic signals the cameras operate as fixed position

speed cameras only. That is, they are able to assess the speed of vehicles passing through

the intersection and capture images of those vehicles detected exceeding the speed limit.

During the red-light phase, the cameras continue to operate as speed detection devices.

However, in addition, they are able to capture vehicles entering the intersection against the

red light. It therefore becomes possible to generate two infringement notices related to the

same event in the case of vehicle entering the intersection against the signal at a speed in

excess of the posted speed limit.

Trials of this technology are currently being conducted in Victoria using a mixture of wetfilm

and digital technology. The digital cameras are capable of operating without a flash

and ensure that the operations are conducted as covertly as possible. However, it is

believed that further refinements of the digital technology are required before the use of

wet-film cameras can be eliminated.

As the use of combined red-light/speed cameras is only just commencing in Victoria, no

evaluation of the effectiveness of these devices in reducing offence rates has been

conducted. However, it is assumed that the mechanisms which drive the individual speed

and red-light camera programs will also apply to the operation of the combined

technology.

In Victoria, there are currently approximately 4,300 authorised speed camera sites. Of

these approximately half are actively in use. The scheduling of mobile safety camera

operations at these sites is of importance to the overall effectiveness of the program. To

achieve the target of 6,000 enforcement hours per month across Victoria, the enforcement

resources must be allocated across time and space.

The target number of enforcement hours to be conducted on each day of the week is

determined centrally by the Traffic Intelligence Unit (TIU) of Victoria Police. The

proportion of enforcement operations to be conducted on each day of the week corresponds

to the proportion of casualty crashes occurring on that day over a one-year period ending

three months prior to the planned enforcement operations. The time of day at which the

enforcement is to take place is also determined by previous crash history at that time of

day. The recommendations of the TIU are distributed to individual Traffic Management

Units (TMUs) to enable them to schedule operations at individual locations in line with the

recommendations.

At the TMU level no definitive guidelines exist to determine the placement of the speed

cameras. However, the speed camera locations must be chosen from the pool of active

sites. This pool of sites has developed over time and each of the active sites must meet the

criteria set down in the Victoria Police Speed Camera Policy & Operations Manual. In

brief, this requires that a site have a recent history of speed-related collisions or a written

complaint concerning excessive speeds. Further, the physical characteristics of the site

must be such as to enable efficient operation of the speed camera technology and minimise

public complaints. It is noted that these guidelines apply only to the operation of mobile

speed cameras. If the Senior Sergeant of a TMU believes that these criteria are met, the

site can be submitted to the TIU for approval. In general, this process takes approximately

six weeks and the site cannot become active until this process is completed.

In addition to contributing to the selection of new speed camera sites, Senior Sergeants or

Sergeants of individual TMUs must schedule speed camera operations by the location and

time of day. This must be completed in the context of the advice from the TIU regarding

the intensity of operations on each day of the week. The general guidelines governing site

selection are set out in the Victoria Police Speed Camera Policy & Operations Manual.

This states that TMU Senior Sergeants and Sergeants ‘must still target high-risk behaviour,

accident locations and subsequently allocate resources in such a manner as to ensure the

maximum possible benefit in crash reduction whilst increasing the deterrent effect’.

Nevertheless, it does not appear that more specific guidelines than these exist.

In general, it is believed that TMUs rely primarily on crash statistics to determine the

geographic location of the speed cameras. In particular, it appears that sufficient data is

available to provide each TMU with the locations in their area involving the highest

number of crashes over the previous 12-month period. Further, some TMUs have

implemented a system to provide feedback on the number of vehicles assessed and

offences detected during each camera session. This information is then used to ensure that

the maximum deterrent effect is achieved.

Once a proposed schedule of operation is complete, it is submitted to the Regional Traffic

the schedule applies. This allows sufficient time for the schedule to be forwarded to LMT

who are responsible for the allocation of operators and vehicles to each camera session.

Using this method of resource allocation, the mobile speed camera program has achieved

the positive results detailed in section 3.1.1. A general effect on serious crashes has

apparently been achieved across the broader road network, although the camera locations

and times have been chosen to reflect recent crash patterns and evidence of excessive

speeds in time and space.

Zaidel (2002), in a comprehensive review of the impact of enforcement on crashes, has

noted that most police traffic agencies plan their deployment according to a “black spot

approach” and/or “black time approach”. Most experts advise the police to do this, as it

appears consistent with good management principles. Zaidel notes that the approach is

responsive to public concerns and can have, sometimes, immediate benefits at the specific

locations. He suggests, however, that it is not necessarily a good guide for routine and

sustained traffic enforcement aimed to raise compliance over the entire network, at all

times.

The major alternative to the method of scheduling mobile speed camera operations

described previously, is the randomisation of the allocation of resources. This method is

based primarily on the application of general deterrence theory. That is, this method aims

to ensure that the perceived risk of detection amongst road users is high at all locations and

times. In turn it is expected that any speed and casualty crash reduction will occur across

the whole road network.

Queensland has successfully implemented this method of scheduling enforcement

operations and the details and outcomes of the program are discussed below.

The Random Road Watch program (RRW) of traffic policing was introduced during

December 1991 in the rural areas of the Southern Police Region of Queensland. Since that

time the program has been extended and it now operates throughout the state. The

program aims to allocate enforcement resources in a random way so as to maximise road

safety benefits. In addition, the randomisation of the timing and location of visible road

safety enforcement enables police to cover larger parts of the road network than would be

the case with conventional policing and road users are less able to predict the location and

timing of enforcement activities.

This approach is implemented by using the existing Police structure of regions and districts

to select a number of road segments (approximately 40) that will be the subject of

enforcement. These road segments are chosen to ensure that roads covering over 50

percent of all road crashes are included in the program. The next stage of implementation

involves dividing each day into 2-hour segments for enforcement between 6 am and 12

midnight. No enforcement takes place between midnight and 6 am each day. A random

selection of sites and times is then selected for enforcement activities. The number of

hours per week required by each Division is tailored to match Police resources available

within each Division. Actual enforcement is conducted conspicuously from marked Police

vehicles.

The RRW program was evaluated in terms of the effect of its implementation on crash

frequency over the period of December 1991 to July 1996 (Newstead et al., 1999).

The analysis indicated that for all non-metropolitan areas of Queensland the RRW program

resulted in statistically significant crash reductions at all severity levels. The crash

reductions increased as the severity level of the crash increased. Examining crash

reductions for rural and urban areas separately produced some interesting results. In rural

areas, there was a statistically significant 34% reduction in fatal crashes but reductions in

other crash categories were not statistically significant. On the other hand, urban areas

experienced crash reductions for all categories except fatal crashes. However, the failure

to identify statistically significant reductions in fatal crashes may be due to insufficient

data.

In addition to the variation between metropolitan and rural areas, the outputs and crash

effects of the program differed across Police regions. The relationship between the outputs

of the program, such as the number of hours of enforcement, and the crash effects of the

program in each region was investigated with the aim of determining the mechanisms that

drive the program. Significant variations in the offences detected per crash treated and

enforcement hours per crash treated were identified across regions. Treated crashes are

defined as crashes in the year prior to the introduction of RRW on routes and in time bands

enforced by RRW. It was found that the crash coverage of the program, (i.e. the

percentage of previous crashes in the region covered by the program) was positively

related to both the total number of crashes saved and the percentage of crashes saved in the

region. The analysis also indicated that total crashes saved and the percentage crash

savings are positively related to offences detected and hours enforced, however, these

associations were not statistically significant.

The effects of the program over time have also been analysed. The results show that the

effect of the RRW program on all crash types except those involving fatalities has

increased over time. The effect of the program on fatalities appears to be fairly consistent

across the three years immediately following the implementation of the scheme.

It is noted that similar programs have been conducted in other jurisdictions and although

the outcomes of these are not conclusive they indicate that reductions in crash frequency

can be achieved by implementing randomly scheduled police enforcement.

1. Random scheduling of enforcement operations has successfully produced statistically

significant crash reductions at all severity levels in non-metropolitan Queensland. The

effect of the program in metropolitan areas has been more difficult to assess. Further,

this effect has increased over time for all crash types (except those involving fatalities,

for which the effect was approximately constant).

3. The positive effect of the RRW program on crashes of various severities in both rural

4. The statistically significant association between program coverage and crash effects

indicates that in operational terms, wide spread but perhaps less intense, randomised

enforcement will result in greater crash effects than more intense but less diverse

To the extent that available information allows, the deterrence value of the Victorian safety

camera program is assessed in the following sections by consideration of:

running),

The overall effect of the safety camera program is also discussed in relation to the

objectives of achieving maximum road safety benefit and continuing to build on the

positive outcomes achieved by enforcement programs in Victoria over the last ten years.

The problem due to speed of vehicles on Victoria’s roads can be categorised as follows:

the regulated or sign-posted speed limit of the road zone)

It is not generally possible for police to focus enforcement resources on inappropriate

speeds. However, safety cameras could be used to enforce speed limits which have been

varied to lower levels due, for example, to weather conditions, the presence of numerous

pedestrians, or high traffic densities. The principal responsibility for variable speed limit

technology lies with VicRoads. When used, police can then focus on increased crash risk

due to illegal speed, in much the same way as enforcement focuses on speeds above fixed

speed limits.

The risk of crash involvement for vehicles travelling at free (unimpeded) speed has been

calibrated for each speed level in urban and rural South Australia. The initial study

estimated the relative risks of casualty crash involvement, relative to a speed of 60 km/h, in

60 km/h speed zones in urban Adelaide (Kloeden et al., 1997). The casualty crashes were

those resulting in an injury severe enough to require ambulance transport, and were thus

more severe than those crashes requiring a Police report because someone was injured to

any degree. Five percent resulted in a fatality, 28% in hospital admission, and the

remainder involved treatment at hospital. This well-known study found that the risk of a

(relatively severe) casualty crash approximately doubled for each 5 km/h increase in free

speed above 60 km/h, but found inconsistent estimates for the risk associated with free

speeds below 60 km/h.

A recent re-analysis of this study has fitted smooth risk relationships to the data, resulting

in more reliable risk estimates, especially at the lower speeds (Kloeden et al., 2002). The

relationship fitted to the absolute speeds in urban 60 km/h speed zones is shown in Table 2.

The relationship of the same type fitted to the difference from average speeds (which was

58.8 km/h overall in 60 km/h zones) produced similar results.

Speed in 60 km/h

speed zone (average

speed = 58.8 km/h)

Difference in speed

from average speed

in rural speed zone

45 0.27 n.a. n.a.

50 0.39 -10 0.54

55 0.60 -5 0.72

65 1.82 5 1.45

70 3.57 10 2.20

75 7.63 15 3.49

80 17.66 20 5.77

85 44.36 25 9.96

90 120.82 30 17.94

The second study calibrated the risk relationship for vehicles travelling at free speed in

rural speed zones of 80 km/h or greater (Kloeden et al., 2001). The majority of the crashes

(52%) occurred in 100 km/h zones, with 25% occurring in 110 km/h zones and 21% in 80

km/h zones. Because a variety of speed zones were covered in this study, the risk

estimates were calculated for the difference in speed from the average speed in the speed

zone. (It could be expected that the average speed is about the same as the speed zone

limit in each environment.) The crashes considered were those in which someone was

treated at hospital or killed, and thus were more severe than those crashes requiring a

Police report because someone was injured to any degree. Twenty-three percent resulted

in a fatality, 46% in hospital admission, and the remainder involved treatment at hospital.

The relative risks of involvement in these (relatively severe) casualty crashes are also

shown in Table 2. The study did not provide an estimate of the risk for vehicles travelling

15 km/h slower than average.

Table 2 indicates that the risk of casualty crash involvement rises more rapidly for speeds

in excess of the speed limit in urban areas than it does in rural areas. However the

probability of the crash resulting in death or hospital admission (serious injury) is likely to

be greater in rural areas because of the higher travelling speeds and hence impact speeds.

The relationship of injury severity in crashes to travelling speed has not been fully covered

by the risk estimates in Table 2 and will be covered in the following section.

A MUARC study has matched the crashed vehicles in the initial South Australian urban

study (Kloeden et al., 1997) with Police accident reports on the same crashes

(Diamantopoulou et al., 2002). This has allowed examination of the injury severity of

persons occupying or hit by the vehicle whose free travelling speed before crashing was

estimated. The injury severity of these casualties generally increased with the travelling

speed, as expected (Figure 2).

The injury severity distribution in Figure 2 was used to adjust the urban casualty crash

risks from Table 2 to estimate the risks of a serious casualty crash (ie. a crash resulting in

fatality or hospital admission) given in Table 3. The risks were estimated for vehicle

speeds in illegal speed categories related to the levels at which monetary and demerit point

penalties change in Victoria, the previous speed offence threshold level (10 km/h in

excess), and a lower level (5 km/h in excess) considered to be the tolerance level likely to

be most acceptable by the community on urban roads (see Section 6.4).

There were too few fatal crashes (8) to reliably estimate fatal crash risk, and too few cases

to provide reliable estimates of injury severity for speeds in the 61-65, 66-70 and 71-75

km/h ranges separately. The injury severity in crashes involving vehicles travelling in each

illegal speed category is shown in the third column of Table 3 and the relative injury

severity in the fourth column. The serious casualty crash risk (relative to that at 60 km/h)

is the product of the casualty crash risk (second column: ex Table 2) and the relative injury

severity.

Figure 2: Injury severity related to travel speed before impact

􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀

􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀

􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀

􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀

􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀

􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀

􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀

􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀

􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀

􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀

􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀

􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀

􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀

􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀

􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀

􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀

􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀

􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀

􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀

􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀

􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀

􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀

􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀

􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀

􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀

􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀􀀀