Parking difficulty and parking information system technologies and costs.(Technical report)

Journal of Advanced Transportation March 22, 2008 | Teng, Hualiang; Qi, Yi; Martinelli, David R.

Before the implementation of a parking information system, it is necessary to evaluate the parking difficulty, technology choice, and system costs. In this study, the parking problem was quantified by asking parkers to express their parking difficulties in five scaled levels from the least to the most difficult. An ordered Probit model was developed to identify the factors that influence a parker to feel the parking difficulty. The results indicate that the amount of parking information parkers had before their trips was directly related to their parking search time, which in turn, influenced their perceptions of parking difficulty. Parkers’ preferences to parking information technologies were identified based on developing binary and multinomial probit models. The results indicate that personal business trips and older persons would like to use the kiosk, while the more educated and males would not. Trips with shopping and social/recreation purposes and the drivers who had visited the destination areas frequently would like to choose roadside display. Drivers who had planned their parking and had Internet access would use in-vehicle device. The system cost was estimated based on the cost for each component of the system. The results show that providing en-route parking search information through roadside displays is more expensive than providing pre-trip information through a web site.

Introduction In the Central Business Districts of large urban areas, a significant amount of VMT is usually generated by drivers looking for parking spaces. One effective way to reduce parking searches is to provide drivers with information on parking availability in real-time. A system that provides real-time information about parking is called a parking information system, which has been widely deployed in Europe, but not so in the U.S. The research in Rayman and Stannet (1981) noted that the parking information system in Torbay of U.K. is the first system of this kind in the world. This system includes three large car parks, with eight parking guidance signs. Three messages were displayed on the signs: the number of spaces available, almost full, and full. A unique feature of this system is the coordination of displaying these messages with signal control to influence the travel patterns on streets. Similar systems can be found in other places in Europe (Axhausen et al. 1994). Most of them only display the remaining capacity of parking facilities on parking guidance signs. Instead of using parking guidance signs, the research in Polak et al. (1992) described a radio-based parking information system where the information was broadcast to motorists.

The system described in Edwards and Kelcey, Inc. (1997) is a parking information system deployed in St. Paul, Minnesota. Unlike most such systems in Europe, the St. Paul system is event-based. In other words, this system only works for special events, such as ball games, which usually generate sudden increases in parking activities. The researches in Carter (1999) and Kimley-Hom and Associates, Inc. (1999) describe two similar systems. Chicago, Des Moines (The Urban Transportation Monitor, August 20, 1999), San Jose (Spencer and West 2004), San Francisco (Rodier et al. 2005) were the cities in the U.S. who planned to implement such systems. Garber et al. (2005) applied the same technologies to truck stops on the interstate highway systems. The system implemented in the Seattle Center was web-based where the occupancies of selected parking facilities are transmitted to a web site from which travelers can get real-time updated information as well as information that doesn’t change frequently. The parking information system that is described in this paper for the implementation in NYC in 1999 is different from the systems mentioned above because it includes both a web site and roadside displays to provide parking information, one for pretrip parking planning, and the other for en-route parking search.

Before the implementation of a parking information system, it is necessary to evaluate the extent of parking difficulty that is felt by parkers, the technologies that would be preferred by the parkers and the costs for such a system. Specifically, parking difficulty is defined in this study as the psychological measure of inconvenience each individual parker would experience in searching a parking space. It is measured by adopting the Likert scale (1 to 5), a psychometric response scale often used in questionnaires. The population of parkers distributed in different levels (or scales) of the parking difficulty can be identified by developing models such as the ordered Probit model in this study. The evaluation of parking difficulty is necessary because it can be used to justify the implementation of the parking information system in NYC. Implementation of a parking information system can be justified when the reduction of parking difficulty due to the improvement of parking information provision is viewed substantial by the parkers. As far as parking information technologies are concerned, many of them are available for example: web sites, kiosks, roadside displays, in-vehicle devices, radio broadcasts, mobile phone, etc. Obviously, web sites and kiosks are more suitable for pre-trip planning, while roadside displays are better in providing en-route information. In-vehicle devices and radio broadcasts, on the other hand, can be used both for pre-trip planning and as provision of en-route information. Note that mobile phones (Jin and Guo 2006) were not popular in the U.S. in providing parking information when this study was conducted in 1999 and thus not considered in the evaluation in this study. To evaluate the feasibility of a parking information system, it is necessary to identify the characteristics of the parkers who will choose a technology. The cost estimation of the parking information system is also needed because the decision for the implementation of parking information systems relies upon whether the available budget is sufficient to cover the costs of the system. An analysis of the hardware, performance features, and other attributes is necessary to determine which technology is most cost effective for various applications.

The objectives of this study are to provide a quantitative analysis of the parking difficulty with regard to provision of parking information, the choice of technologies that can be adopted in a system to provide parking information, and the estimation of the cost for the system. Parking difficulty was quantified as the degree (five scales) of difficulty that parker expressed for their parking. An ordered probit model was developed to identify the factors such as parking search time, provision of parking information and parkers’ characteristics that influence users’ perceptions of parking difficulty. Through the developed model, the relationship between parking information provision and parking difficulty can be identified. Parkers’ preferences to parking information technologies were identified based on developing binary and multinomial probit models. Even though parking information systems have existed for more than two decades, the research on the identification of the parking information technologies is still limited. The cost of the system was derived based on the estimation of the cost for each component of the system, for which a system architecture was developed. By this system architecture, the system components and connections between them were defined.

In the body of this paper, the first section introduces the literature on the quantification of parking difficulty, evaluation of parking technologies, and a cost estimation of parking information systema. The second section is dedicated to the quantification of parking difficulty. The third section presents the factors that influence parkers in choosing one information technology than the others. The fourth section describes the cost estimation of the system, with presentation of the development of the architecture for the parking information system. The fifth section describes life cycle costs estimation. The last section presents the conclusions made for this study and priorities for future work.

Literature Review Even though there are a few studies (Polak et al. 1992, Edwards and Kelcey, Inc. 1997, Falcocchio et al. 2000) on parking search time, which has been viewed as an important indicator of parking difficulty, the studies on directly quantifying parking difficulty are rare.

Most of the literature on parking information systems addresses the evaluation of the impact that such systems had on travelers’ behaviors and system performance (Axhausen et al. 1994, Polak et al. 1992, Polak et al. 1990, and Allen 1993). The literature for the identification of parking information technologies is sparse.

On the other hand, some literature does exist on studying the needs of traffic information technologies (Hoff 1971, Dudeck and Jones 1971, Ng et al. 1995, Spyridakis et al. 1991, and Barfield et al. 1991). The study in Hoff (1971) documents research that was conducted at the time when variable message signs emerged as an effective way of disseminating traffic information. Two approaches were taken in the study to evaluate the choice of six techniques for presenting traffic information. In one approach, respondents were presented all possible pair combinations of the alternative techniques and asked to indicate which one in the pair they felt gave the best information. As a result, a preference scale was developed where each alternative was ranked. In the second approach, the respondents were presented each of the alternatives with different traffic congestion levels and asked to indicate whether to make a diversion. A comparison of the alternatives was made based on the percentage of diversion made by the responses. The research presented in Dudek and Jones (1971) was another work on determining traffic information technologies. Four visual displays for traffic information were evaluated in the study based on the preliminary analysis of rating and ranking data made by survey participants. In the study of Ng et al. (1995), an investigation was made to estimate whether commercial drivers and dispatchers would use in-vehicle traffic information systems, and how important the ATIS information is to them. The studies presented in Spyridakis et al. (1991) and Barfield et al. (1991) on traffic information technologies summarized travelers’ preference of obtaining traffic information from different media, which included TV, electronic message signs, highway advisory radio, commercial radio station, phone, and CB radio.

In summary, the methods employed in the 1970s or 1980s were primarily based on experiments, and their analysis to investigate the relationship between the characteristics of travelers and trips and their preference of information and technologies was weak. Contrarily, the methods employed in 1990s are primarily survey-based in nature and the emphasis was given on establishing a relationship between the characteristics of the travelers and trips and the chosen information and technologies. It is the later approach that was also employed in this study.

The study that fully presented the cost estimation for a parking information system is Edwards and Kelcey, Inc (1997) and HNTB Corporation (1997). Note that life cycle costs were not provided in these two studies. National ITS Architecture provides cost estimates for different types of ITS components. However, these costs need to be updated to reflect the NYC conditions.

Methodology: Parking Difficulty In this study, the quantification of the parking difficulty and its relationship with parking information was based on a mail-back survey where parkers were asked to rate their parking difficulty on a scale from 1 (the least difficulty) to 5 (the most difficulty). The rating is ordered in nature and can be better modeled by ordered probit models. Basically, an ordered probit model is built around a latent regression model:

y* = [beta]’x + [epsilon], (1) where x represents independent variables such as personal and trip characteristics and y* is an unobserved variable. What can be observed is y = 1 if y* [less than or equal to]0, =2 if 0 =3 if [[micro].sub.1] =J if [[micro].sub.J-2][less than ro eqaul to]y*, where y (i.e., 0, 1 ….. J) represents the degree of parking difficulty. The [micro]’s are the thresholds that define y* into integer ordering. They can be estimated with [beta]. If [epsilon] is assumed normally distributed across observations and its mean and variance are normalized to 0 and 1, then we have the following probabilities:

Prob(y = 1)=[PHI](-[beta]’x) Prob(y = 2)=[PHI]([[micro].sub.1]-[beta]’x)-[PHI](-[beta]’x) Prob(y = 3)=[PHI]([[micro].sub.2]-[beta]’x)-[PHI]([[micro].sub.1]-[beta]’x) (3) Prob(y = J)= 1-[PHI]([[micro].sub.J-2] – [beta]’x) The parameters of [beta]’s and [micro]’s can be obtained based on maximum likelihood estimation. In this study, the Statistical Software Tool (SST) was used in the estimation.

Intuitively, parking search time should be one of the independent variables in x of Equation (1) because parkers would have stronger feeling of parking difficulty when they incurred longer parking search times. However, parking search time is also viewed as a variable that is derived from the same parking search process that is endogenoues with regard to the variable of parking difficulty. Thus, parking search time cannot be directly included in x. To deal with the endogeneity of parking search time, an instrumental variable approach was taken, in which an estimated parking search time was used to replace the true value of parking search time. In this condition, Equation (1) can be written as:

y* = [beta]’ x’ + [[beta].sub.1]I + C (4) where x’ represents the set of independent variables that doesn’t contain parking search time. [beta]’ denotes the corresponding coefficients for the variables included in x’. I is the instrumental variable for parking search time. [[beta].sub.1] is the corresponding coefficient.

To obtain the estimated parking search time I, a Probit model is developed for parking search time, which can be expressed as:

I = [[beta].sub.z]’z + [epsilon]’ (5) where z represents the independent variables that influence parking search time. [[beta].sub.z] denotes the corresponding coefficients of the independent variables, [epsilon]’ is the error term. Even though parking search time is a continuous value, it was divided into ordered categories in the survey, which can be better modeled by ordered Probit models.

The Log-likelihood ratio test can be used for indicating the goodness-of-fit for the developed ordered Probit models for parking difficulty and parking search time. It is based on the ratio of the values of two likelihood functions, one derived from the hypothesis being tested and another without the constraints of the hypothesis under test. Basically, the statistic for the log-likelihood ratio test can be expressed as below, which follows a chi-squared distribution with the degrees of freedom equal to the number of restrictions imposed. go to website best parking nyc

-2 1n([L.sub.0])/1n([L.sub.[beta]]) ~ [x.sup.2] where 1n([L.sub.[beta]])is the log-likelihood at convergence; 1n([L.sub.0])is the loglikelihood for the restricted model (with all coefficients equal to 0), and J is the degrees of freedom (equal to the number of estimated parameters in the developed model).

If the significance level of the Log-likelihood ratio test is less than a given critical value (such as 0.05), the null hypothesis that all coefficients equal to 0 can be rejected, which implies that the developed model is statistically significant. The significance level being close to 0 indicates that the model is a good fit.

Methodology: Parking Technology In the survey, parkers were asked to choose an information technology ranging from web site, kiosk, and in-vehicle devices for pretrip planning, and choose one technology among roadside display and an in-vehicle device for en-route planning. The choice set consists of three technologies for the pre-trip planning. Thus, a multinomial Logit model was developed to identify the related market segment. The choice set consists of two technologies for the en-route planning. Thus, the binary Logit model was developed. The same log-likelihood ratio test performed for the ordered Probit models can be applied to the developed multinomial Logit and binary Logit models.

Methodology: Parking Cost Capital costs were calculated for each of the system components in terms of present and annualized costs. The annualized cost was calculated by using the following formula:

[C.sub.a] = m.c.i(1+i)[sup.n]/(1+i)[sup.n] -1 (6) where m is the quantity of a system component, c denotes the unit cost, i represents the interest rate, and n is expressed as life time, which varies for different components of the system. By adding up the costs for each component, the total annualized and present costs for the system can be derived.

General Profile of the Survey Respondents The surveys were distributed at four study areas in NYC: (1) Midtown Manhattan, (2) Lower Manhattan, (3) Downtown Brooklyn, and (4) Downtown Flushing. In an effort to maximize the number of respondents, survey sites were selected based on the size of the parking facilities located within each of the four survey areas. A total of 38 offstreet parking facilities were selected. Each of the sites was surveyed during the hours of 7:00AM-2:00 PM and 4:00PM-8:00 PM.

In total, 191 questionnaires were returned and used for analysis in this study. The demographic information collected for parkers’ in the survey include age, household, education, and gender. Age was classified into six groups: under 25, 25-34, 35-44, 45-54, 55-64, and over 64. The distribution of these six age groups is: 2.67%, 13.37%, 26.20%, 27.81%, 20.32%, and 9.62%. It can be seen that most respondents are between 25 to 64 years. Five categories of household income were surveyed: less than 29,999, 30,000-49,999, 50,000-69,999, 70,000-99,999, and 100,000 or more, and their distribution is: 7.82%, 11.73%, 18.99%, 22.91%, and 38.55%. It shows that the household incomes are mostly above $50,000. The survey classified the education into four levels: some high school, high school graduate, some college, and college graduate, and the distribution of them is: 6.95%, 4.28%, 16.58%, and 72.19%. It indicates that most of them have some college or college education. In terms of gender, as surveyed, about 70% of the respondents are male and 30% are female.

The trip characteristics collected in the survey include: (1) Trip purposes, (2) Trip frequency in a specific area, (3) Known parking information level before starting their trips, (4) Whether or not checking traffic information before their trips, (5) Whether or not having Internet access, (6) Search times that they incurred, and (7) Parking difficulty they have experienced.

Five trip purposes were surveyed in the study: work, shopping, social/recreation, personal business, and other. The distribution of them is: 71.58%, 10.53%, 9.47%, 6.32%, and 2.10%, which indicates that about 70% of the trips were for work. As for trip frequency, there are four levels surveyed: less than once a month, 1-3 days a month, 1-3 days a week, and 4 or more days a week. The distribution of the frequency is: 21.16%, 18.52%, 16.93%, and 43.39%. It shows that more than 40% of the places, where the respondents were given the questionnaire, were visited 4 or more days a week.

The amount of the parking information level known before their trips was obtained from answers to the following statements: (1) I didn’t know where to park, (2) I partially knew where to park, and (3) I knew exactly where to park. Actually, three types of information can be derived from these statements: levels of parking information they had before they started their trip, the strategies they took to search for a parking space, and the extent that their parking destinations was planned before their trips. The parkers making the first statement may be very new to their destination areas and had no parking information available to have a prechoice of an area, location, or facility for searching a space. These parkers may randomly select a direction to search until they find a parking space. In other words, their parking destinations were not well planned before their trip. The second statement suggests the parkers who may have been familiar with their destination areas may have a certain preference for either on-street or off-street parking and are sure that vacant parking spaces are easier to find in one area than in others. With this partial information, the drivers may simply drive to a chosen onstreet or off-street parking facility and start to search until an available space was found. In this case, their parking processes can be viewed as not planned. The parkers who chose the third statement may be those who had paid a monthly rate to a parking facility or those who are somewhat sure that an area would have spaces available when they arrive. Thus, they may just directly go to their familiar area without a searching process. The survey results for these three parking information levels are: I don’t know where to park, 7.37%, I partially knew where to park, 13.68%, and I knew exactly where to park, 78.95%.

It is conceivable that parkers checking traffic information regularly may prefer parking information to be provided together with traffic information. The survey reveals that 56.61% of the respondents checked traffic information before their trips.

Having access to the Internet influences parkers’ preference for information and technologies. It might be reasonable to expect that parkers having Internet access would prefer technologies that required Internet access. In the survey, 84.65% of the respondents had Internet access.

Whether a long or short search time was incurred might also influence parkers’ preference to information and technology. In this study, search time was classified into eight categories, each with 5 minutes increment (except the first two categories). The survey results are: 0-1 55.68%, 2-5 22.16%, 6-10 10.27%, 11-15 5.41%, 16-20 0.54%, 21-25 1.62%, 26-30 1.08%, and Over 30, 3.24%.

Whether parkers experienced difficulty in finding a parking space may affect their choice of parking information items and technologies. To determine the degree of the parking difficulty, the survey asked respondents to indicate from 1 (least difficulty) to 5 (most difficulty) their feelings about their parking difficulties. The survey results are: 1 61.90%, 2 15.34%, 3 12.17%, 4 4.76%, and 5 5.82%.

For the purpose of validation of the data collected from the survey, the background of the parkers was compared with those in Falccochio et al. (1995) and Hayden ] Wegman (1997) which are studies for the same area and also include off-street parking. In Falccochio et al. (1995), the distribution of trip purpose in terms of work, shopping, social/recreation, personal business, and others is 70%, 6%, 9%, 8%, and 6%, which is very close to 71.58%, 10.53%, 9.47%, 6.32%, and 2.10% collected in this study. In the study by Hayden|Wegman (1997), the income distribution is 6%, 22%, 30%, 42% for the categories of Under $30,000, $30,000-$60,000, $60,000-$100,000, which is comparable to the case in this study 7.82%, 11.73%, 18.99%, 22.91%, and 38.55% for the categories of less than 29,999, 30,000-49,999, 50,000-69,999, 70,000-99,999, and 100,000 or more. Note that the categories of trip purposes and income are not the same between the studies, and thus a statistical analysis cannot be performed for the validation purpose. It is also noted that the New York City area is unique in its transit service that consists of a large physical network and makes it possible for a substantial population such as low-income people to travel within the city without owning an automobile. Such a unique transportation service in NYC makes the background of parkers different from those in other cities. In addition, the total of 191 questionnaires returned in this study can be viewed as sufficiently large for analysis, especially when it is compared with some published studies such as Abdel-Aty and Jovanis (1998) where sample size less than 100 was used.

Degree of Parking Difficulty Table 1 lists the results for the ordered Probit model that was developed for the degree of difficulty in finding a parking space. The log-likelihood ratio test indicates the developed model is statistically significant. The variables that are statistically significant include search time, age, and household income. A positive coefficient for search time indicates that the more time drivers spent looking for a parking space, the greater their parking difficulty. The coefficient for age has a negative sign, which suggests that older parkers tended to have a greater capability of tolerating the difficulty they felt in looking for a parking space. A negative coefficient for household income indicates that drivers with higher household income may be less concerned with parking fees and thus more likely experience less difficulty to find a parking space.

From Table 2, it can be seen that the developed model is statistically significant. Shopping trips and parking information levels are statistically significant factors. The shopping trip purpose has a positive coefficient, which indicates that drivers on shopping trips were likely to incur a longer search time. As for the variable of parking information level before starting trips, it has a negative coefficient. This implies that the more parkers knew about the parking location around their destination areas, the less likely they were to end up with a longer search time.

From Table 2 it can be derived that an increase in parking information level will result in a decrease in parking search time, while from Table 1 it can be seen that a decrease in parking search time will make the parkers less difficult when they search for parking spaces. Thus, it can be concluded that providing parking information could successfully lighten parkers’ parking difficulty.

Choice of Parking Information Technologies The multinomial Logit (MNL) model results for the choice of pretrip parking information are presented in Table 3. When developing the MNL model, an in-vehicle device was used as the reference. As presented in the table, the calibrated model is statistically significant. The results imply that respondents who are older or had personal business trips, would choose a kiosk. The respondents who had received higher education or were male, however, were likely to choose the in-vehicle device or web site.

The binary Logit model results for the choice of en-route parking information are presented in Table 4. The log-likelihood ratio test indicates that the calibrated model is statistically significant. The results show that people whose trips were for shopping, social, or recreation, and those who had visited their destination areas more frequently would choose roadside display. Those who had known their parking places very well and had Internet access would likely choose an in-vehicle device. Drivers who had frequently visited their destination areas may have a better knowledge in finding a parking space. Thus, they didn’t have the need to use an in-vehicle device either. In contrast, the person who had well planned his/her parking trip would likely be a parker who was in a work trip that is more time sensitive than shopping and other trip purposes. Thus, they need to use an in-vehicle device to improve the efficiency of finding a parking space. In addition, the way that a web site and an in-vehicle device present parking information may be quite similar, which may make the respondents who are using the Internet to choose in-vehicle device.

Costs of Parking Information Systems In the estimation of the costs for the parking information system in NYC, it was assumed that the system will be piloted in the Downtown Flushing area. For this pilot area, a system architecture was first developed and the total cost of the parking information system was estimated based on a preliminary design of the system in this area.

Parking Information System Architecture In the development of a system architecture for the parking information system, user services were first identified. Among the 31 user services specified in the National ITS Architecture (U.S. Department of Transportation 2000), four are identified as being related to the parking information system: (1) Pre-trip Travel Information, (2) Traveler Services Information, (3) Travel Demand Management, and (4) Traffic Control. For each of these user services, the corresponding possible market packages are chosen. As a result, the following three are most relevant to the parking information system: (1) Interactive Traveler Information, (2) Traffic Information Dissemination, and (3) Parking Facility Management. Based on the identified market packages, a system architecture for parking information, as shown in Figure 1, was developed.

From Figure 1 it can be seen that the users of the parking information system are internet web users and drivers on the road. The web users can access the parking information web site which receives updated information from a parking information management center. Drivers receive updated parking formation from roadside displays which are also connected to the same parking information management center. The updated information on the roadside display is received from each individual off-street parking facility where detectors (surveillance) are supposed to be installed to count in-and-out vehicles.

Parking Management Center A Parking Management Center is a focal location where parking information is received from parking facilities and transmitted to variable message signs and a parking information web server. In the central location, the following equipment as well as an operator is needed: (1) Hardware, (2) Software, (3) Staff, and (4) Web server.

The critical hardware is a personal computer that can be used both as a server for the parking web site and a place for receiving data from parking facilities. The software in the PC should have the capability to perform tasks such as checking the status of variable messages. A multiserial board is required to receive data from each of the parking garages. In addition, modems are required for receiving data from parking garages. For six parking facilities considered in this study, one multi-serial communication board and six modems are needed for receiving data, and another ten modems are needed for transmitting data to roadside displays.

An operator is needed for the operation and maintenance of the systems. His/her responsibility is to monitor the operational status of every segment of the parking information system at all times and provide prompt service such as keying-in of variable messages. In addition, he/she also needs to manage the maintenance of the system when failures of the system components are discovered and update the static parking information periodically.

The web site is a place where travelers can access to get both static and real-time parking information. The operation of this web site relies on the support of a server that obtains dynamic parking information from a central PC. Based on the static web site that was developed during an earlier stage of this study and currently on-line, the cost to be incurred to upgrade the system to a dynamic web site is small and has been included in the design of the system.

Parking Management The following devices are required for each off-street parking facility to collect and disseminate parking information: (1) vehicle detector, (2) controller, and (3) modem. A vehicle detector counts the in-and-out of the vehicles at control sections in each parking facility. A control section could be defined as an access point, or entrance and exit, and is the likely area to be outfitted with vehicle detectors. The detector can be a vehicle detector, ticket-spitter, or cash register. In this study, a loop detector was chosen for this system. In the cost calculation, the number of detectors needed for this system was derived by identifying the location and number of entries and exits at each off-street parking facility. In total, 29 loop detectors are required for the whole system.

A controller calculates the number of vacant spaces available based upon the counts of in-and-out and passes on the information to a parking management center through transformation of data by modems and transmission of data via telecommunications. A modem transforms the count into a format suitable to be transmitted to a central location. It is assumed that each off-street facility only needs one controller and one modem. Thus, the system needs to have 6 controllers and 6 modems installed in 6 parking facilities in the piloted area.

[FIGURE 1 OMITTED] Roadside Displays In this study, four types of parking signs were proposed: dynamic variable message signs, map signs, internal signs, and front signs. On the dynamic message signs which were supposed to be placed in the periphery of the area, the following information is supposed to be displayed: facility’s name, direction to a facility, and the number of spaces available, among which the last one is updated in real-time. The information items of fee structure and hours of operation were also wanted by the parkers as indicated in the survey. Due to the space limitation of the variable message signs, these information items were not to be displayed. The information to be displayed on the later three signs is static and thus these signs are referred to as a static sign. A map sign is used to show the zone code of parking facilities in Downtown Flushing and the locations of those parking facilities participating in the system, and thus should be placed at least one block before the dynamic message signs. Internal signs are the continuation of the dynamic signs, and the parking information on each panel of the board contains names of a parking facility and directions to the facility. As shown in Figure 2, 10 dynamic signs and 49 static signs are required in the pilot area.

Communications Communications are the means for transmitting data from parking facilities to a parking management center and then to roadside displays. In this study, wireless technology was unaffordable because the data transmission needs to be operated 24 hour a day and thus it was not considered in this study for the data transmission between parking facilities, central location, and roadside displays.

Similar to the parking information system described in (6), we recommend using telephone lines for the communication between the Parking Management Center and the roadside displays. For each dynamic message sign, the required equipment includes a modem and a regular voice telephone line. Correspondingly, 10 connections are needed from the PMC to roadside displays.

Table 5 lists all the costs for the all the system components identified above. Among the maintenance costs, those for staff support in the parking management center and the dynamic roadside displays take up more than 75 % of the total maintenance cost. The primary difference of costs for web and roadside displays lies in the web server and the dynamic signs. Obviously, the cost for the web site ($50,000 for equipment, $5,000 for maintenance) is less than that for roadside display ($500,000 for equipment, $10,185 for construction, and $60,000 for maintenance). In other words, it is more expensive to provide en-route parking information than pre-trip parking information.

[FIGURE 2 OMITTED] Conclusions Based on the investigation in this study, it can be seen that the level of parking information parkers possessed before starting their trips is first significantly related to the parking search time, which in turn influences parking difficulty they felt. This observation justifies a need for providing parking information. It can also be seen that the parking difficulty felt by parkers is highly related to three factors: (1) how long they had spent in searching for a parking space, (2) age, and (3) household incomes. Therefore the time parkers spend looking for a parking space is not the only factor that influences a drivers perceived parking difficulty. bestparkingnycnow.net best parking nyc

As far as the information technology is concerned, when the web site, kiosk, and in-vehicle device were evaluated comparatively for providing pretrip information, the web site and the in-vehicle device were preferred to the kiosk. Personal business trips and older persons would like to use the kiosk, while the more educated and males would not. If the cost for purchasing an in-vehicle device were considered (in-vehicle navigation devices are expected to be standard in most vehicles in the next 3-5 years), it can be conceivable that a web site would be preferred to in-vehicle device. When roadside displays and in-vehicle devices were evaluated comparatively for providing en-route information, roadside displays are preferred. Trips with shopping and social/recreation purposes and the drivers who had visited the destination areas frequently, preferroadside displays. Drivers who had planned their parking and had Internet access prefer an in-vehicle device.

Based on the costs estimated in this study, it can be determined that it is more expensive to provide en-route parking search information through roadside displays than to provide pre-trip information through a web site. Roadside displays are the most expensive component in the system.

It should be noted that there are factors such as parking fee that may also influence the perception of parking difficulty. Advanced models may need to be developed to incorporate these additional factors. Kiosks were considered in this study because they had been installed in several locations in New York City at the time this study was conducted. It is one reason for not considering the cost of kiosks in the cost analysis. Also, cell phones have become more viable since the study was conducted, and should be considered in future studies. The costs for in-vehicle devices would incur only to the users, not the public agency who builds the system eventually. That is why the costs for in-vehicle devices were not considered in this study.

Acknowledgements The authors would like to thank John C. Falcocchio, Raman Patel, Jose M. Ulerio, All Afshar-Ghotli and Allen Huang for their contributions to various parts of the study.

References Rayman N. and A. Stannet. Compact Urban Traffic Control Car Parking Information System, Traffic Engineering + Control, July 1981, pp. 411-417.

Axhausen K.W., J.W. Polak, M. Boltze, and J. Puzicha. Effectiveness of the Parking Guidance Information System in Frankfurt am Main. Traffic Engineering + Control, Vol. 35, No. 5, pp. 304-309, 1994.

Polak J. P. Vythoulkas, and I. Chatfield. Broadcast Parking Information: Behavioral Impacts and Design Requirements. Transportation Research Record 1403, 1992.

Edwards and Kelcey, Inc. St. Paul Advanced Parking Information System Operational Test, Summary Report Detailed System Design, March 1997.

Carter R. J.; Parking and Routing Information System–Phase I Operational Test, Conference Proceedings of Ninth Annual Meeting and Exposition of Intelligent Transportation Society of America, April 19-22,1999.

Kimley-Horn and Associates, Inc., Functional Specifications for the Parking and Traffic Guidance System, Prepared for Metropolitan Government of Nashville and Davidson County, May 27 1999 Teng H., Y. Qi, and P. Yi , Parking Search Time and Parking Information Identification for Off-Street Spaces in Central Business District, Accepted for Publication at the 81St Annual Meeting of the Transportation Research Board, Washington D.C.

Falcocchio J.C., H. Teng, J. Ulerio, A. Afshar, and A. Huang, Evaluation and Feasibility Study for a City-Wide Deployment of Motorist Parking Information. Final report prepared for the New York City Department of City Planning, 2000.

Polak J.W., I.C. Hilton, K.W. Axhausen, and W. Young. Parking Guidance and Information Systems: Performance and Capability. Traffic Engineering + Control, Vol. 31, No. 10, pp. 519-524, 1990.

Allen P.A. Driver Response to Parking Guidance and Information Systems. Traffic Engineering + Control, Vol. 34, No. 6, pp. 302-307, 1993.

Hoff G.C. A Comparison of Selected Traffic Information Devices. Highway Research Record 366, pp. 116-129, 1971.

Dudek C.L. and H.B. Jones. Evaluation of Real-Time Visual Information Displays for Urban Freeways. Highway Research Record 366, pp. 64-78, 1971.

Ng L., R.L. Wessels, D. Do, F. Mannering, and W. Barfield. Statistical Analysis of Commercial Driver and Dispatcher Requirements for Advanced Traveler Information Systems. Transportation Research, Part C., Vol. 3, No. 6, pp. 353-369, 1995.

Spyridakis J., Barfield W., L. Conquest, M. Haselkorn, and C. Isakson. Surveying Commuter Behavior: Designing Motorist Information Systems. Transportation Research, Vol. 25A, No. 1, pp. 17-30, 1991.

Barfield W., M. Kaselkorn, J. Spridakis, and L. Conquest. Integrating Commuter Information Needs in the Design of a Motorist Information System. Transportation Research, 25A, No. 2/3, pp. 7178, 1991.

HNTB Corporation, Saint Paul Advanced Parking Information System Operational Test, Operational Test Evaluation Report, July 1997.

Abdel-Aty M.A. and P.P. Jovanis. Modelling Incident-Related Routing Decisions Including the Effect of Traffic Information Using A Nested Logit Structure. Paper No. 980003, 77th Annual Meeting of the Transportation Research Board, Washington D.C., January 1998.

Falcocchio J.C., G. Darsin, and E. Prassas. An Inquiry on the Traffic Congestion Impacts of Parking and Pricing Policies in the Manhattan CBD. Polytechnic University, Transportation Training and Research center, Draft final report for the New York City Department of Transportation, Division of Parking, and the University Transportation Research Center, Region II, February 1995.

Hayden | Wegman, Consulting Engineers. Parking Facilities Users’ Survey and Parking Need Survey in Community Districts 1-8 in Manhattan. Final report to the New York City Department of City Planning, March 1997.

U.S. Department of Transportation, The National ITS Architecture, A Framework for Integrated Transportation into the 21st Century, Version 3.0, 2000.

Jin, M. and W. Guo, EZ M-parking System, Transportation Research Record No. 1944, 2006, pp 67-71.

Rodier C. J., S.A. Shaheen, and A. M. Eaken, Transit-Based Smart Parking in the San Francisco Bay Area, California–Assessment of User Demand and Behavioral Effects, Transportation Research Record No. 1927, 2005, pp. 167-173.

Spencer, M.E. and J. West, Parking Guidance System–San Jose, California, Transportation Research Record No. 1886, 2004, pp. 34-39.

Garberl, N. J., Y. Lul, H. Teng. A proposed methodology for implementing and evaluating a real-time truck parking information system, 12th World Congress on ITS, 6-10 November 2005, San Francisco Paper 4796.

Hualiang (Harry) Teng Yi (Grace) Qi David R. Martinelli Hualiang (Harry) Teng, Department of Civil and Environmental Engineering, Howard R. Hughes College of Engineering, University of Nevada, Las Vegas, NV, USA Yi (Grace) Qi, Texas Southern University, Houston, Texas, USA David R. Martinelli, Department of Civil and Environmental Engineering, West Virginia University, Morgantown, WV, USA Received. October 20006 Accepted. December 2006

Table 1. Ordered Probit Model Estimation Results for Parking Difficulty

Variable (a) Estimated t- Stat. P-value Coeff. (P>[t])

Constant 2.011 4.412 1.9E-05

Search Time (Instrumental variable) 0.137 1.666 0.09765 (1 0-1 min, …, 8 Over 30 min)

Age -0.021 -2.684 0.00803 (1 Under 25, …, 6 Over 64)

Household Income -0.370 4.797 3.6E-06 (1$100,000

[micro.sub.1] 0.561 5.638 7.5E-08

[micro.sub.2] 1.147 7.965 2.8E-13

[micro.sub.3] 1.470 8.043 1.8E-13

Log-likelihood at zero = -195.0358

Log-likelihood at convergence = -173.5180 Log-Likelihood ratio test (significance level) = 43.0356 (3.28324E-07)

Number of observation = 169 (b) Percent correctly predicted = 62.13

(a) Dependent Variable: 1 least difficulty, …, 5 most difficulty

(b) Only 169 respondents answered the question on parking difficulty

Table 2. Ordered Probit Model Estimation Results for Search Time

Variable (a) Estimated t-Stat. P-value Coeff. (P>[t])

Constant 1.618 3.530 5E-04

Trips with shopping purpose 0.750 2.579 0.011 (1 shopping trip, 0 otherwise)

Parking information level 0.660 4.126 6E-05 (1 didn’t know where to park, 2 knew partially where to park, 3 know exactly where to park)

[micro.sub.1] 0.775 7.006 6E-11

[micro.sub.2] 1.257 9.294 9E-17

[micro.sub.3] 1.626 10.089 7E-19

Log-likelihood at zero = -209.0767

Log-likelihood at convergence = -192.7877 Log-Likelihood ratio test (significance level) = 32.578 (1.26397E-05)

Number of observation = 169 (b) Percent correctly predicted (%) = 55.03

(a) Dependent variable: 1 0-1 minute, 2 2-5, 3 6-10, 4 11-15, 5 over 15 minutes.

(b) Only 169 respondents answered the question on parking difficulty and search time

Table 3. Multinomial Logit Model Estimation Results for the Choice of Pretrip Parking Information Technologies

Variable (a) Estimated t- Stat. P-value Coeff. (P>[t])

Constant 0.006 0.036 0.97

Personal Business (specific to Kiosk) 1.733 1.935 0.05 (1 Personal business, 0 Otherwise)

Age (specific to Kiosk) 0.321 1.713 0.09 (1 Under 25, 2 25-34, 3 35-44, 4 45-54, 5 55-64, 6 Over 64)

Education (specific to Kiosk) -0.605 -2.785 0.01 (1 Some high school, 2 High school graduate, 3 Some college/technical school, 4 College graduate or beyond)

Male (specific to Kiosk) -1.447 -2.396 0.02 (1 Male, 0 Otherwise)

Log-likelihood at zero = -161.5

Log-likelihood at convergence = -130.87 Log-Likelihood ratio test (significance level) = 61.26 (6.67166E-12)

Number of observation = 147 (b) Percent correctly predicted = 48.98

(a) Dependent variable: 1 Web site 2 Kiosk 3 In-vehicle device

(b) There are only 147 valid data for this model

Table 4. Binary Logit Model Estimation Results for the Choice of En-Route Parking Information Technologies

Variable (a) Estimated t- Stat. P-value Coeff. (P>[t])

Constant 1.079 1.007 0.32

Shopping 1.939 1.724 0.09 (1 Shopping, 0 Otherwise)

Social/Recreation 1.889 1.995 0.05 (1 Social/recreation, 0 Otherwise)

Parking Information Level Before -0.6000 -1.473 0.14 Trip (1 know nothing about parking, 2 know partially about parking, 3 know exact parking locations)

Visit Frequency 0.387 1.790 0.08 (1 Less than once a month, 2 1-3 days a month, 3 1-3 days a week, 4 or more days a week)

Internet Access -1.314 -1.761 0.08 (1 Yes 0 No)

Log-likelihood at zero = -74.86 Log-likelihood at convergence = -67.37 Log-Likelihood ratio test (significance level) = 14.98 (0.02)

Number of observation = 108 (b) Percent correctly predicted = 64.815

(a) Dependent variable: 1 Chose roadside display, 0 Otherwise.

(b)There are only 108 valid data for this model

Table 5. Cost Estimates Equipment

Unit Life Unit Total Annua- time Cost lized (year)

Parking Management Center

Hardware * PC & monitor 1 5 $3,000 $3,000 $751 * Multi serial 1 20 $50 $40 $5 board * Modem 16 20 $125 $2,000 $204 PC System 20 $5,050 $960 Subtotal

Software 1 20 $50,000 $50,000 $5,093 Staff support 1 1 $0 $0 $0 Web 1 10 $50,000 $50,000 $7,451 Subtotal $105,050 $13,504

Parking Management

Loop detector 29 5 $500 $14,500 $3,632 Controller 10 10 $5,000 $50,000 $7,451 Modem-card 10 20 $150 $1,500 $153 Subtotal $66,000 $11,236

Roadside Display

Dynamic sign 10 20 $50,000 $500,000 $50,926 Static signs with 49 20 $400 $19,600 $52,922 pole Subtotal $519,600 $0

Comm: Parking->PMC

Phone circuit 6 [infinity] $0 $0 $0 activation Annual comm 6 $0 $0 $0 Subtotal $0 $0

Comm: Parking->PMC

Phone circuit 10 [infinity] $0 $0 $0 activation Annual comm 10 $0 $0 $0 Subtotal $0 $0

Total $690,650 $77,663

Construction Maintenance Per Year

Unit Total Annua- Units Total Cost lized

Parking Management Center

Hardware * PC & monitor * Multi serial board * Modem PC System $1,000 $1,000 $102 $1,000 $1,000 Subtotal

Software $5,000 $5,000 Staff support $40,000 $40,000 Web Subtotal $1,000 $102 $46,000

Parking Management

Loop detector $800 $23,000 $5,811 $100 $2,900 Controller $1,000 $10,000 $1,490 $600 $6,000 Modem-card $10 $100 $10 $0 $0 Subtotal $33,000 $7,311 $8,900

Roadside Display

Dynamic sign $10,000 $100,000 $10,185 $6,000 $60,000 Static signs with $200 $9,800 $998 $100 $4,900 pole Subtotal $109,800 $11,183 $64,900

Comm: Parking->PMC

Phone circuit $20 $120 $10 activation Annual comm $450 $2,700 Subtotal $120 $10

Comm: Parking->PMC

Phone circuit $20 $200 $16 activation Annual comm $450 $4,500 Subtotal $200 $16 $4,500

Total $144,420 $18,622 $127,000 Teng, Hualiang; Qi, Yi; Martinelli, David R.