Towards Sustainable Parking

[Full title: Towards Sustainable Parking: Analyzing the Characteristics of Periodic Off-Street Parking Lots and Their Application in Shared Parking]

[Find the meaning and references behind the names: Zhang, Mode, Double, New, Resources, Four, Transport, Doi, Human, January, Feifei, Basel, Road, Yifei, Enough, Long, Xiao, Show, Development, Time, Power, Triple, December, Land, Edu, Tools, China, Large, Bus, Maas, Car, Data, Under, Major, Far, Street, Point, Cars, Open, Energy, Cai, Area, Severe, Hard, Pan, Smart, Market, Due, Civil, Still, Study, Strong, Lei, Peak, Small]

Received: 8 December 2024 Revised: 19 January 2025 Accepted: 20 January 2025 Published: 21 January 2025 Citation: Cai, Y.; Pan, X.; Zhang, L.; Xu, F.; Zhang, S. Towards Sustainable Parking: Analyzing the Characteristics of Periodic Off-Street Parking Lots and Their Application in Shared Parking Sustainability 2025 , 17 , 833. https:// doi.org/10.3390/su 17030833 Copyright: © 2025 by the authors Licensee MDPI, Basel, Switzerland This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https://creativecommons.org/ licenses/by/4.0/). Article Towards Sustainable Parking: Analyzing the Characteristics of Periodic Off-Street Parking Lots and Their Application in Shared Parking Yifei Cai 1,2 , Xiao Pan 3 , Lei Zhang 3 , Feifei Xu 1,2 and Shuichao Zhang 1,2, * 1 School of Civil and Transportation Engineering, Ningbo University of Technology, Ningbo 315211, China; 13675102961@163.com (Y.C.); xufeifei@nbut.edu.cn (F.X.) 2 Zhejiang Engineering Research Center of Digital Road Construction Technology, Ningbo 315211, China 3 Ningbo Municipal Public Investment Co., Ltd., Ningbo 315000, China; panx@ningboparking.com (X.P.); zhangl@ningboparking.com (L.Z.) * Correspondence: zhangshuichao@nbut.edu.cn Abstract: The pollution and congestion caused by the shortage of parking spaces are threatening the sustainable development of cities. Smart parking platforms are one of the major tools to solve the problem by providing the efficient usage of parking resources. However, current platforms can only realize limited functions, and shared parking is far from being implemented on a large scale. Since off-street parking provides the majority of potential shared parking spaces, this paper takes periodic off-street parking lots as the starting point for opening the shared parking market. Based on data from the Ningbo Yongcheng parking platform, power spectral density (PSD) and the autocorrelation function (ACF) are used to identify periodic parking lots. A Density-Based Spatial Clustering of Applications with Noise (DBSCAN)-based method is applied to clustering the occupancy time series. Land use, user type, parking duration, and parking patterns are then analyzed to study shared parking supply characteristics. The results show that (1) 31.3% of off-street parking lots are periodic parking lots, and 90.3% of them have regular users exceeding 50%. (2) Periodic parking lots are classified into four types. Most parking lots show convex flat peak, double peak, or triple peak characteristics. (3) The shared parking spaces demonstrate spatial and temporal imbalances. But in a small area, even considering the concentration of land use and the peak period, there are still enough spaces available. The above research is of significance for the large-scale implementation of shared parking, which can promote the sustainable development of a city Keywords: applications of sustainability; shared parking; off-street parking; periodicity detection; cluster method 1. Introduction Parking space scarcity and the disparity between supply and demand have become major issues in recent years [ 1 , 2 ]. The long time spent cruising for parking not only wastes fuel but also contributes to environmental pollution. If Mobility as a Service (Maas) is the solution to traffic problems in large cities with severe human–land conflicts [ 3 – 5 ], then one ought to consider that bus use is shrinking and the metro has not reached saturation or there may not even be any metro at all in smalland medium-sized cities. Therefore, cars will still be the dominant mode of transport in these cities [ 6 – 8 ]. Meanwhile, considering the increase in private car ownership due to new energy vehicles, as well as the saturation of land development that makes it hard to build more parking spaces, how to effectively Sustainability 2025 , 17 , 833 https://doi.org/10.3390/su 17030833

[Find the meaning and references behind the names: Stage, Change, Town, Less, Pilot, List, Local, Date, Low, Risk, Poor, Fast, Apps, Matter, Pay, Ten, Share, Main, Areas, Join, Profit, Lot, Cost, Sell, Place, Patience, Free, Camera, Image, White, Target, App, Location, Need, Ones]

Sustainability 2025 , 17 , 833 2 of 23 utilize existing parking resources has become increasingly important [ 9 ] in promoting sustainable cities With the development of new technologies such as Bluetooth tags and image recognition, low-cost monitoring methods for parking lot usage have become very mature [ 10 ]. This has also led to the emergence of parking APP software such as SpotHero, J parking, ETCP etc. in recent years, allowing users to check the availability of parking spaces connected to the APP and also enabling fast parking through online payment [ 11 ]. Some cities have even implemented a citywide parking platform, integrating a large number of on-street and off-street parking spaces through government intervention, which also provides effective data support for the use of parking spaces Normally, parking can occur either on the street or in designated off-street areas On-street parking is allowed within specific zones, while off-street parking takes place in enclosed lots or garages. These parking facilities may be owned by local governments, public entities, or private individuals [ 12 ]. The number of off-street parking spaces is usually much larger than that of on-street parking spaces. In China, the ratio of off-street parking spaces ranges from 80% to 90% while on-street parking spaces account for less than 10%. Thus, the main target market for shared parking is off-street parking lots. When some parking spaces are vacant during certain hours, individuals or facilities can then sell their parking spaces for a limited time through a shared platform, which helps to utilize the unused capacity of parking lots [ 13 ]. However, although convincing these parking lots to join the platform for online payment is easy, convincing them to allow social vehicles to enter is another matter, especially for factories, companies, the government, residential communities, and other enclosed parking lots [ 14 ]. Even for commercial parking lots such as those of shopping malls, entertainment facilities, etc., their parking fees are more expensive than those of regular parking, facing the risk of outsider vehicles occupying their users’ space without generating enough profit, which also leads to a low willingness to share their spaces [ 15 ]. Meanwhile, considering that parking platforms need to negotiate one-to-one with different parking lot management parties, this makes parking platforms have a strong regional characteristic. Since the smart parking market is still in the development stage and has not formed a mature business mode, apart from cities with strong intervention by the municipal government that can achieve the objective of one city one platform, most cities still have multiple commercial platforms [ 16 ], which is also a problem for the large-scale implementation of shared parking. In the comments section of these APPs, it is often seen that people write the following: “It’s too expensive, why should the free on-street parking be turned into paid parking. . .. . .”, “The UI is too poor, there are only ten parking spaces to choose from, and I have to input a lot of passwords and camera permissions. . .. . .”, “Vehicles with out-of-town license plates cannot use it, the charging standards cannot be found, and the previous payment has not shown the parking date and location”. Without sufficient parking space supply, users are unlikely to have the patience to download multiple apps In fact, only a few pilot projects have achieved point-to-point shared parking in China, such as shopping malls providing overnight parking monthly cards for nearby residents, and hospitals providing shared parking for nearby users on the approved “white list”. However, point-to-point shared parking is not suitable for a platform-based mode, as potential users are not limited to parking lots in the immediate vicinity but can access ones from all over the city, which would greatly change the characteristics of parking lots Therefore, in order to achieve the widespread adoption of shared parking, it is necessary to pay attention to these off-street parking lots.

[Find the meaning and references behind the names: Every, Park, Ways, Rapid, Better, Day, Zone, Urban, Ayala, Single, Million, Tool, Original, Forward, Put, High, Year, Selling, Guo, Non, Match, Cloud]

Sustainability 2025 , 17 , 833 3 of 23 If off-street parking lots are classified according to their occupancy characteristics, they can be divided into periodic and non-periodic parking lots. Non-periodic parking lots have fluctuating occupancy throughout the day, and the sudden peak of original users at a certain moment will directly affect the effect of shared parking. Meanwhile, for stable periodic parking lots, predictable original user usage characteristics allow managers to avoid worrying about original user spaces and to better develop shared parking strategies. Meanwhile, the data source of this paper, the “Yongcheng Parking” platform, is a citywide smart parking platform. Users who park by querying about available parking spaces cause fluctuations in the occupancy of parking lots. In other words, parking lots that can still maintain a periodic parking pattern in this platform are parking lots that are not open for query and have significant shared spaces. They are the starting point for large-scale shared parking implementation In summary, to achieve citywide shared parking and thereby promote sustainable urban development, this study takes periodic off-street parking lots as the starting point and contributes to the literature in the following ways. First, it provides a feasible framework to find and cluster the periodic parking lots, PSD and the ACF are used to verify parking lot periodicity, and a hybrid DBSCAN-based method is applied to cluster the complex parking patterns. Second, the connection between land use and periodic parking lots, the causes of periodic occupancy, and their parking patterns are analyzed to help understand the actual parking patterns in large cities. Finally, the shared parking supplies are studied in spatial–temporal aspects, which is essential for the large-scale implementation of shared parking in single or multiple parking platforms. The remainder of this paper is structured as follows: a description of off-street parking occupancy data is presented in Section 2 , the developed method is presented in Section 3 , the analysis of the parking data and policy implications are presented in Section 4 , and finally conclusions are drawn in Section 5 . 2. Literature Review Parking space scarcity has emerged as a significant contributor to deteriorating traffic conditions, especially long-time cruising for parking spaces. Shoup [ 6 ] indicates that cruising for parking is responsible for 30% of traffic on average in downtown areas Ayala et al. [ 17 ] concluded that in Chicago, 8.37 million gallons of gasoline would be consumed and 129 thousand tons of CO 2 emitted every year due to cruising for parking This phenomenon has led to severe congestion, increased pollution, and unsustainable transportation Extensive efforts have been undertaken to address the mobility challenges associated with urban parking. Since the first parking regulation put forward in 1910, parking policy and management has developed a lot, which includes unbundling parking costs [ 18 ], minimum parking requirements [ 19 ], dynamic parking prices [ 8 ], and hybrid parking structures [ 20 ]. In recent years, with the rapid development of information and communication technology, private parking sharing is enabled through a smart parking platform to help match the supply with the demand, which is a promising tool in achieving sustainable parking. Scholars have performed various research on the operation mode and parking allocation method of shared parking [ 21 , 22 ]. Among the operation mode research, Guo et al. [ 23 ] were among the earliest to develop a simulation-based approach for the decision making of repurchasing private parking spots and selling them to public users. Cai et al. [ 14 ] discussed creating shared parking zones by grouping multiple adjacent off-street shared parking lots in high-demand areas. Smart parking technologies such as vacant space detection and cloud control processing are integrated into the shared parking zone. Li et al. [ 24 ] analyzed the parking characteristics of residential areas and built a parking sharing service capability evaluation model. Some

[Find the meaning and references behind the names: Gap, Mei, Level, Fill, Mile, Final, Train, Comes, Lack, Days, Fee, Last, Hong, Manual, Cover, Rather]

Sustainability 2025 , 17 , 833 4 of 23 other research focuses on “last-mile” problems. Hong et al. [ 25 ] explore travelers’ preference for shared micromobility in the context of last-mile travel from shared parking lots to their final destination in the city center The allocation method is the other stem of shared parking research. Mouskos et al. [ 26 ] first modeled the parking reservation process as a resource allocation problem, with the objective of minimizing the total travel cost, to form a mixed integer linear programming (MILP) problem. Tasseron et al. [ 27 ] made an in-depth study of the parking allocation approach based on a parking reservation. Mei et al. [ 28 ] introduced the agent simulation system and built the simulation framework of parking fee and reservation. With the support of digital technology, shared transportation has developed greatly in recent years [ 29 – 31 ]. However, since shared parking has not yet been put into large-scale use, most of the studies above are still limited to hypothetical data or small-scale survey data. Previous studies have mainly concentrated on on-street parking lots [ 32 ], paying scant attention to off-street parking circumstances [ 33 , 34 ]. Meanwhile, current research on off-street parking lots mostly focuses on specific areas, such as airports, train stations, or hospitals [ 35 , 36 ]. For example, in the fifth edition of the Parking Generation Manual [ 37 ], it contains parking generation rates for over 70 types of land uses, and it also provides information and guidelines about the site selection, permissions, procedures, background, and independent variables of different types of off-street parking lots. But in order to achieve accurate forecasts, the parking lots are selected with strict limitations [ 38 ]: (1) the site should be mature (i.e., at least two years old); (2) occupancy (i.e., at least 85%); (3) sites should be clear for the purpose of controlling parking counts on it; (4) no abnormal condition next to the selected site such as constructions [ 39 ]. These restrictions make the survey-based data difficult to apply in platform-based shared parking [ 40 ]. Since the idle time of individual parking spaces mainly concentrates on working days, the main source of shared parking spaces still comes from off-street parking lots that are currently not open to the public or have relatively high charges [ 41 ]. Furthermore, the platform must engage in one-to-one communication with parking lot managers. Consequently, it is crucial to consider the actual usage of each parking lot rather than using one type of land use to represent all of them [ 42 ]. Meanwhile, since the current research finds it difficult to cover long-term survey data, which also makes it hard to effectively identify the periodicity of off-street parking lots, periodic and non-periodic parking lots are often mixed with each other [ 43 ], as discussed above, and the sudden peak of the original users at a certain moment will directly affect the effect of shared parking Thus, to achieve sustainable parking, the characteristics of off-road parking lots at the city level and their application in shared parking still requires a further investigation, which includes the specific classification of off-road parking lots, the composition of users, and a special phenomenon that cannot be explained by conventional models. Meanwhile, the research on the shared parking allocation method is limited to hypothetical data, while there is a lack of understanding of the spatial and temporal distribution of the shared parking spaces provided by different types of parking facilities. Therefore, to fill the research gap, this paper conducts research on the characteristics and sharing features of off-street periodic parking lots 3. Study Area and Data Collection The data in this paper are provided by the Ningbo “Yongcheng parking” APP v 3.2.3, which is the first parking APP in China to realize “one city one platform”. It has been registered by more than 3 million users, and 1739 off-road parking lots, including more than 380,000 off-road parking spaces. Since the off-road parking lots accessed by the platform cover the urban area of Ningbo and the county-level cities under its jurisdiction,

[Find the meaning and references behind the names: Hotel, Square, Commerce, Residence, Cant, Peer, Set, Hospital, Cut, Cap, Nally, Sharp, Saturday, Ers, Holiday, Coe, Plate, Monday, Table, Num, Lea, Rst, See, April, Parts, Weeks, Min, Sunday]

Sustainability 2025 , 17 , 833 5 of 23 the coverage rate of the platform outside the main urban area is still low. Therefore, this paper selects the main urban area of Ningbo, including Haishu, Yinzhou, and Jiangbei districts, as well as some areas of Zhenhai, Beilun, and Fenghua districts. The study area has a total area of 3689 square kilometers and a population of 5.16 million. A total of 995 off-road parking lots are included (see Figure 1 ). Sustainability 2025 , 17 , x FOR PEER REVIEW 5 of 24 coverage rate of the platform outside the main urban area is still low. Therefore, this paper selects the main urban area of Ningbo, including Haishu, Yinzhou, and Jiangbei districts, as well as some areas of Zhenhai, Beilun, and Fenghua districts. The study area has a total area of 3689 square kilometers and a population of 5.16 million. A total of 995 o ff -road parking lots are included (see Figure 1). Figure 1. The study area. The data set includes two parts: 1: parking lot information, which contains the name, location (longitude and latitude), and number of parking spaces; 2: user information, which involves the license plate, entry time, and departure time, etc. In total, 35 days of o ff -street parking data were collected from 1 April to 5 May 2023. Occupancy in this paper is parking accumulation divided by the capacity, where parking accumulation is the number of vehicles entering minus the number of vehicles leaving. The data were calculated every 15 min. See Equation (1): it it it it i num ent lea occ cap + − = (1) where it occ is the occupancy of parking lot i at time t ; it num , it ent , and it lea are the current number of vehicles, the number of entering vehicles, and the number of leaving vehicles of parking lot i at time t ; and i cap is the capacity of parking lot i . The study period included the Qingming Festival and Labor Day, which disrupts the identi fi cation of parking lot periodicity in a week. For example, 23 April should be Saturday, but it was shifted to a working day due to the 5 days of the May day holiday. Thus, to be tt er identify the parking period, this paper reorders the data from Monday to Sunday and adds a holiday, making 8 days a week. Then, 35 days are reduced to 4 weeks for a total of 32 days. See Figure 2, where day 1 to 7 is Monday to Sunday and day 8 is a holiday. The 3 days cut o ff are the Qingming holiday (only 1 day, unrepresentative), the last day of the May day holiday (5 days for 4 weeks), and the fi rst day of the data set (the residual vehicles are not recorded, making the occupancy inaccurate). Figure 1. The study area The data set includes two parts: 1: parking lot information, which contains the name, location (longitude and latitude), and number of parking spaces; 2: user information, which involves the license plate, entry time, and departure time, etc. In total, 35 days of off-street parking data were collected from 1 April to 5 May 2023. Occupancy in this paper is parking accumulation divided by the capacity, where parking accumulation is the number of vehicles entering minus the number of vehicles leaving. The data were calculated every 15 min. See Equation (1): occ it = num it + ent it − lea it cap i (1) where occ it is the occupancy of parking lot i at time t ; num it , ent it , and lea it are the current number of vehicles, the number of entering vehicles, and the number of leaving vehicles of parking lot i at time t ; and cap i is the capacity of parking lot i The study period included the Qingming Festival and Labor Day, which disrupts the identification of parking lot periodicity in a week. For example, 23 April should be Saturday, but it was shifted to a working day due to the 5 days of the May day holiday Thus, to better identify the parking period, this paper reorders the data from Monday to Sunday and adds a holiday, making 8 days a week. Then, 35 days are reduced to 4 weeks for a total of 32 days. See Figure 2 , where day 1 to 7 is Monday to Sunday and day 8 is a holiday. The 3 days cut off are the Qingming holiday (only 1 day, unrepresentative), the last day of the May day holiday (5 days for 4 weeks), and the first day of the data set (the residual vehicles are not recorded, making the occupancy inaccurate) Sustainability 2025 , 17 , x FOR PEER REVIEW 6 of 24 Figure 2. Revised occupancy rate of Kechuang o ffi ce building. Meanwhile, in order to distinguish the types of parking lots (residence, hospital, etc.), although each parking lot has its corresponding name in the data set, many parking lots are still hard to be identi fi ed only by their name. Considering that most o ff -street parking lots serve certain buildings, the land use data are collected in ARCGIS, and 100 m bu ff ers are constructed around the target areas of each park to fi nd its most relevant land use (see Figure 1). The parking lots are fi nally classi fi ed into 10 categories (see Table 1). Table 1. The types of parking lots. No. Category No. Category 1 Residence 6 Transportation 2 Hospital 7 Hotel and amusement 3 O ffi ce building 8 Industry 4 Commerce 9 Education 5 Public sector 10 Public parking lot 4. Methodology To have a be tt er understanding of the characteristics of periodic parking lots, it is necessary to identify the periodicity of the occupancy time series and classify them into a certain number of categories. Then, the periodicity detection method and cluster method are introduced below. 4.1. Periodicity Detection There are usually two fundamental methods in periodicity detection algorithms: (1) frequency domain methods relying on a periodogram after Fourier transform [44] and (2) time domain methods relying on the ACF [45]. However, a periodogram is not accurate when the period length is long or the time series is with sharp edges. Meanwhile, the estimation of the ACF and the discovery of its maximum values can be a ff ected by outliers and noises easily, leading to many false alarms in practice [46]. Some methods have been proposed in the joint frequency–time domain to combine the advantages of both methods [47]. Finding signi fi cant periods on the ACF is more di ffi cult than using a periodogram while the ACF provides a more fi ne-grained estimation of potential periodicities. Referring to [48], a hybrid method which combines the advantages of PSD and the ACF is used in this paper, and this method uses a periodogram to extract candidate periods and the ACF to verify them. If the number of periods obtained from a periodogram is on a peak of the ACF, it can be considered a valid period; otherwise, it is considered a false alarm. (1) Calculate the periodogram and fi nd the peaks Suppose that X is the DFT of a time series x ( n ), n = 0, 1, …, N − 1. The periodogram P is the squared length of each Fourier coe ffi cient: Figure 2. Revised occupancy rate of Kechuang office building.

[Find the meaning and references behind the names: Max, Powers, Fine]

Sustainability 2025 , 17 , 833 6 of 23 Meanwhile, in order to distinguish the types of parking lots (residence, hospital, etc.), although each parking lot has its corresponding name in the data set, many parking lots are still hard to be identified only by their name. Considering that most off-street parking lots serve certain buildings, the land use data are collected in ARCGIS, and 100 m buffers are constructed around the target areas of each park to find its most relevant land use (see Figure 1 ). The parking lots are finally classified into 10 categories (see Table 1 ). Table 1. The types of parking lots No. Category No. Category 1 Residence 6 Transportation 2 Hospital 7 Hotel and amusement 3 Office building 8 Industry 4 Commerce 9 Education 5 Public sector 10 Public parking lot 4. Methodology To have a better understanding of the characteristics of periodic parking lots, it is necessary to identify the periodicity of the occupancy time series and classify them into a certain number of categories. Then, the periodicity detection method and cluster method are introduced below 4.1. Periodicity Detection There are usually two fundamental methods in periodicity detection algorithms: (1) frequency domain methods relying on a periodogram after Fourier transform [ 44 ] and (2) time domain methods relying on the ACF [ 45 ]. However, a periodogram is not accurate when the period length is long or the time series is with sharp edges. Meanwhile, the estimation of the ACF and the discovery of its maximum values can be affected by outliers and noises easily, leading to many false alarms in practice [ 46 ]. Some methods have been proposed in the joint frequency–time domain to combine the advantages of both methods [ 47 ]. Finding significant periods on the ACF is more difficult than using a periodogram while the ACF provides a more fine-grained estimation of potential periodicities. Referring to [ 48 ], a hybrid method which combines the advantages of PSD and the ACF is used in this paper, and this method uses a periodogram to extract candidate periods and the ACF to verify them. If the number of periods obtained from a periodogram is on a peak of the ACF, it can be considered a valid period; otherwise, it is considered a false alarm (1) Calculate the periodogram and find the peaks Suppose that X is the DFT of a time series x ( n ), n = 0, 1, . . ., N − 1. The periodogram P is the squared length of each Fourier coefficient: P ( f k / N ) = ∥ X ( f k / N ) ∥ 2 k = 0, 1, . . . . . N − 1 2 (2) where k / N indicates the frequency or, equivalently, at period N / k (2) Determine the threshold The peaks of the periodogram are then extracted as candidate periods, but the threshold should be determined to distinguish the period with appropriate power. The permutation of the time series x ( n ) is denoted as x ( n ) ; then, the maximum power of the random time series x ( n ) is recorded as P max , and the threshold is the 99 th largest value of these maximum powers (3) ACF verification

[Find the meaning and references behind the names: Add, Unique, Valley, Left, Ern, Veri, Hour, Lies, Friday, Hill, Schools, Lie, Satis]

Sustainability 2025 , 17 , 833 7 of 23 The ACF examines how similar a sequence is to its previous values for different τ lags, and it is calculated as follows: ACF ( τ ) = 1 N N − 1 ∑ n = 0 x ( τ ) · x ( n + τ ) (3) The candidate periods selected by the first two steps are verified by the ACF results, and if the period resides on a hill of the ACF, then it is adjusted to the closest peak of the ACF; otherwise, if it lies on the valley, then it can be considered a false alarm Considering that parking occupancy varies on weekdays, weekends, and holidays, the time series are organized as 3 types. If the following criteria are satisfied, then the parking lots are considered periodic (1) Only weekdays: The time series are reordered from Monday to Friday, 4 weeks. The parking lot is considered periodic on weekdays if the period calculated is measured in days (2) Only weekdays and weekends: The time series are reordered from Monday to Sunday, 4 weeks. The parking lot is considered periodic in a week if the period calculated is 1 day or 7 days (3) One week add one holiday: The time series are reordered as in Figure 2 . The parking lot is considered periodic both in non-holidays and holidays if the period calculated is 1 day or 8 days See Figures 3 and 4 where candidate periods are extracted using a periodogram, which are P 1, P 2, P 3, and P 4, and their periods are 764, 128, 93, and 61 with the unit of the number of 15 min, equivalently, comprising 7.95, 1.33, 0.97, and 0.64 days. Meanwhile, in Figure 4 , only P 1 and P 3 lie on the hill of the ACF results, P 2 and P 4 should be removed, and P 1 and P 3 should be adjusted to their closest peaks in the ACF, which are 96 and 768, or 1 day and 8 days. The results show the parking lot is periodic both in non-holidays and holidays Sustainability 2025 , 17 , x FOR PEER REVIEW 7 of 24 2 / / 1 ( ) ( ) 0,1,..... 2 k N k N N P f X f k − = = (2) where k / N indicates the frequency or, equivalently, at period N / k . (2) Determine the threshold The peaks of the periodogram are then extracted as candidate periods, but the threshold should be determined to distinguish the period with appropriate power. The permutation of the time series x ( n ) is denoted as {?}̅ {?} ; then, the maximum power of the random time series {?}̅ {?} is recorded as max P , and the threshold is the 99 th largest value of these maximum powers. (3) ACF veri fi cation The ACF examines how similar a sequence is to its previous values for di ff erent τ lags, and it is calculated as follows: 1 0 1 ( ) ( ) ( ) N n ACF x x n N τ τ τ − = = ⋅ +  (3) The candidate periods selected by the fi rst two steps are veri fi ed by the ACF results, and if the period resides on a hill of the ACF, then it is adjusted to the closest peak of the ACF; otherwise, if it lies on the valley, then it can be considered a false alarm. Considering that parking occupancy varies on weekdays, weekends, and holidays, the time series are organized as 3 types. If the following criteria are satis fi ed, then the parking lots are considered periodic. (1) Only weekdays: The time series are reordered from Monday to Friday, 4 weeks. The parking lot is considered periodic on weekdays if the period calculated is measured in days. (2) Only weekdays and weekends: The time series are reordered from Monday to Sunday, 4 weeks. The parking lot is considered periodic in a week if the period calculated is 1 day or 7 days. (3) One week add one holiday: The time series are reordered as in Figure 2. The parking lot is considered periodic both in non-holidays and holidays if the period calculated is 1 day or 8 days. See Figures 3 and 4 where candidate periods are extracted using a periodogram, which are P 1, P 2, P 3, and P 4, and their periods are 764, 128, 93, and 61 with the unit of the number of 15 min, equivalently, comprising 7.95, 1.33, 0.97, and 0.64 days. Meanwhile, in Figure 4, only P 1 and P 3 lie on the hill of the ACF results, P 2 and P 4 should be removed, and P 1 and P 3 should be adjusted to their closest peaks in the ACF, which are 96 and 768, or 1 day and 8 days. The results show the parking lot is periodic both in non-holidays and holidays. Figure 3. Results of the periodogram. Figure 3. Results of the periodogram Sustainability 2025 , 17 , x FOR PEER REVIEW 8 of 24 Figure 4. Results of the autocorrelation function. 4.2. Cluster Method 4.2.1. Problems Various types of o ff -street parking lots are involved in this paper while each occupancy time series has more than 3000 data points to fully capture its behavior within periods. Compared to the idealized time series, parking lot occupancy varies in di ff erent aspects, which are as follows: (1) Noises: Vehicles looking for parking spaces or that have not yet left cause the fl uctuation in the time series. See Figure 5 a, where an obvious fl uctuation is observed when the parking lot is already saturated. (2) Amplitude di ff erences: It is obvious that di ff erent parking lots have di ff erent number of users and parking spaces, which lead to the amplitude di ff erences in time series. Figure 5 b illustrates that two commercial parking lots share the same peak hour but have a di ff erent peak volume. (3) Phase shifts: A phase shift refers to a global horizontal shift between two time series; one or two hours of a peak hour shift are commonly observed in parking lots. See Figure 5 c. (4) Special pa tt ern: The pa tt ern of the time series may be similar when the parking lot has only one type of user. But when parking lots have two or more types of regular users, even for primary schools, the pa tt erns are completely unlike the combinations of teachers, parents, and the di ff erent grades of students who form a unique parking lot pa tt en. See Figure 5 d. ( a ) Noises ( b ) Amplitude di ff erences Figure 4. Results of the autocorrelation function.

[Find the meaning and references behind the names: Maps, Smooth, Patten, Self, Focus]

Sustainability 2025 , 17 , 833 8 of 23 4.2. Cluster Method 4.2.1. Problems Various types of off-street parking lots are involved in this paper while each occupancy time series has more than 3000 data points to fully capture its behavior within periods. Compared to the idealized time series, parking lot occupancy varies in different aspects, which are as follows: (1) Noises: Vehicles looking for parking spaces or that have not yet left cause the fluctuation in the time series. See Figure 5 a, where an obvious fluctuation is observed when the parking lot is already saturated (2) Amplitude differences: It is obvious that different parking lots have different number of users and parking spaces, which lead to the amplitude differences in time series. Figure 5 b illustrates that two commercial parking lots share the same peak hour but have a different peak volume (3) Phase shifts: A phase shift refers to a global horizontal shift between two time series; one or two hours of a peak hour shift are commonly observed in parking lots. See Figure 5 c. (4) Special pattern: The pattern of the time series may be similar when the parking lot has only one type of user. But when parking lots have two or more types of regular users, even for primary schools, the patterns are completely unlike the combinations of teachers, parents, and the different grades of students who form a unique parking lot patten. See Figure 5 d. Sustainability 2025 , 17 , x FOR PEER REVIEW 8 of 24 Figure 4. Results of the autocorrelation function. 4.2. Cluster Method 4.2.1. Problems Various types of o ff -street parking lots are involved in this paper while each occupancy time series has more than 3000 data points to fully capture its behavior within periods. Compared to the idealized time series, parking lot occupancy varies in di ff erent aspects, which are as follows: (1) Noises: Vehicles looking for parking spaces or that have not yet left cause the fl uctuation in the time series. See Figure 5 a, where an obvious fl uctuation is observed when the parking lot is already saturated. (2) Amplitude di ff erences: It is obvious that di ff erent parking lots have di ff erent number of users and parking spaces, which lead to the amplitude di ff erences in time series. Figure 5 b illustrates that two commercial parking lots share the same peak hour but ( a ) Noises ( b ) Amplitude di ff erences ( c ) Phase shift ( d ) Special pa tt ern Figure 5. Problems with the parking occupancy data. 4.2.2. The DBSCAN-Based Algorithm The problems above make traditional cluster methods such as K-means, K-shapes, or Self-Organizing Maps have limited abilities in identifying clusters. These methods focus on clustering methods and similarity measures with an assumption of idealized time series data. These idealized time series are often low-dimensional (usually less than 1000 Figure 5. Problems with the parking occupancy data 4.2.2. The DBSCAN-Based Algorithm The problems above make traditional cluster methods such as K-means, K-shapes, or Self-Organizing Maps have limited abilities in identifying clusters. These methods focus on clustering methods and similarity measures with an assumption of idealized time series data. These idealized time series are often low-dimensional (usually less than 1000 data points), and the curves are smooth without abnormal patterns [ 49 ]. Thus, referring to [ 50 ], a DBSCAN-based algorithm is used in this paper. The steps are introduced below.

[Find the meaning and references behind the names: Top, Key, Arg, Missing, Cross, Mean, Core]

Sustainability 2025 , 17 , 833 9 of 23 (1) Processing Use linear interpolation to fill the missing values based on their adjacent data points, and adopt z-score standardization to remove differences in amplitude (2) Baseline Extraction (1) Smoothing Extreme Value: remove the top 5% data which deviate the most from the mean value, and then use linear interpolation to fill them (2) Extract Baseline: Apply the moving average with a small sliding window to separate the occupancy rate into two parts: baseline and residuals. The baseline removes most of the noises while preserving its underlying shape and is used as the input of the clustering method (3) Density-Based Clustering (1) Shape-based similarity measure: Use the shape-based distance (SBD) based on cross-correlation to measure the similarity of two baselines [ 47 ]. For two time series, → x ( x 1 , x 2 , . . . . . . , x m ) and → y ( y 1 , y 2 , . . . . . . , y m ) , where x i and y i are the occupancy rate at time i , i = 1, 2, . . . , m For all phase shifts s ∈ [ − m + 1, m − 1 ] , the inner product CC s ( → x , → y ) with shift s is calculated as follows: CC s → x , → y =        m − s ∑ i = 1 x s + i · y i s ≥ 0 m + s ∑ i = 1 x i · y i − s s < 0 (4) And SBD is defined as follows: SBD = 1 − max s ( CC s ( ⇀ x , ⇀ y ) ⇀ x 2 · ⇀ y 2 ) (5) SBD ranges from 0 to 2, where 0 means two time series have exactly the same shape A smaller SBD means higher shape similarity (2) Density-Based Clustering with Parameter Estimation Use DBSCAN to obtain the clusters, which is to find the cores’ p and expand the core that has at least minPts objects within a distance of radius ε from it (excluding p ). The similarity of two time series is valued with the SBD while two key parameters are in the DBSCAN algorithm [ 50 ]. (4) Assignment For time series considered outliers, the centroid of each cluster and the SBD of each outliner to the centroid are calculated, and if an unlabeled outliner whose SBD to its nearest cluster centroid is smaller than 0.2, it can be assigned to this cluster. The centroid is calculated as follows: centroid = arg min ⇀ x ∈ cluster i ∑ ⇀ y ∈ cluster SBD ( ⇀ x , ⇀ y ) 2 (6) 5. Results In this section, we use the above methods to study the characteristics of the off-street parking lots. The abbreviations used in this section can be seen in Table 2 .

[Find the meaning and references behind the names: Sca, Shop, Rules, Red, Rus, Bar, Lower, Blue, Orange, Mid, Short]

Sustainability 2025 , 17 , 833 10 of 23 Table 2. Abbreviations in Section 5 . Abbreviation Explanation Regular user (RU) Users who park a single parking lot more than five times a month Irregular user (IU) Users who park a single parking lot less than five times a month Regular long-time user (RLTU) Regular users who park for more than 6 h Regular mid-time user (RMTU) Regular users who park for less than 6 h and more than 2 h Regular short-time user (RSTU) Regular users who park for less than 2 h Irregular long-time user (ILTU) Similar as an RLTU Irregular mid-time user (IMTU) Similar as an RMTU Irregular short-time user (IMTU) Similar as an RSTU Type 1 parking lots Periodic parking lot on weekdays Type 2 parking lots Periodic parking lot on weekdays and weekends Type 3 parking lots Periodic parking lot on weekdays, weekends, and holidays 5.1. Regular User and Irregular User After using the periodicity detection to examine the 995 off-street parking lots, 312 periodic parking lots are found where 156 are type 1, 80 are type 2, and 76 are type 3 parking lots. To find the effect of an RU on parking lot characteristics, the RU occupancy divided by the total occupancy is calculated. See Figure 6 , where among the 312 periodic parking lots, there are 208 parking lots with an RU ratio larger than 0.7, and only 30 parking lots lower than 0.5, which shows that most of the periodic parking lots are formed due to the high ratio of RUs Sustainability 2025 , 17 , x FOR PEER REVIEW 11 of 24 Figure 6. Ratio of regular users. Figure 7. Relationships between land use and RU ratio. Occupancy curves with a high RU proportion are shown in Figure 8. It is obvious that the red curves (RU) almost overlap with the blue curves (total occupancy), while the yellow curves (IR) only fl uctuate between 0 and 0.2 at the peak hour and contribute to most of the noises of the total occupancy. This conclusion may be incomplete, because fl uctuations are also observed in parking lots with an RU composition and are complicated. As in Figure 8(4), the RU of the large auto parts market is composed of di ff erent retail shop owners, workers, and regular customers, and the sca tt ered opening and closing time leads to variations in peak hours. However, since the total amount remains basically unchanged, it still shows a periodic parking pa tt ern. Figure 8. Occupancy of high RU parking lots. Figure 6. Ratio of regular users For all 995 off-street parking lots, see Figure 7 , where on weekdays, more than 50% of hospital, office building, commerce, and education parking lots show periodicity (the ratio between the red bar and orange bar). This result is a little counterintuitive because it is understandable that if one parking lot has stable service time and users, this parking lot will be periodic, and such as for office buildings, schools, and factories, the high RU ratio of them also verifies this (ratio between blue bar and red bar). On the contrary, residence, hotel, and public parking lots with a complex composition of users are hard to be periodic. It seems that the hospital and commercial market parking lots should also follow the rules; this problem will be discussed in the next section Sustainability 2025 , 17 , x FOR PEER REVIEW 11 of 24 Figure 6. Ratio of regular users. Figure 7. Relationships between land use and RU ratio. Occupancy curves with a high RU proportion are shown in Figure 8. It is obvious that the red curves (RU) almost overlap with the blue curves (total occupancy), while the yellow curves (IR) only fl uctuate between 0 and 0.2 at the peak hour and contribute to most of the noises of the total occupancy. This conclusion may be incomplete, because fl uctuations are also observed in parking lots with an RU composition and are complicated. As in Figure 8(4), the RU of the large auto parts market is composed of di ff erent retail shop owners, workers, and regular customers, and the sca tt ered opening and closing time leads to variations in peak hours. However, since the total amount remains basically unchanged, it still shows a periodic parking pa tt ern. Figure 8. Occupancy of high RU parking lots. Figure 7. Relationships between land use and RU ratio.

[Find the meaning and references behind the names: Own, Garden, Normal, Sleep, Moon, Fixed, Station, Line, Lake]

Sustainability 2025 , 17 , 833 11 of 23 Occupancy curves with a high RU proportion are shown in Figure 8 . It is obvious that the red curves (RU) almost overlap with the blue curves (total occupancy), while the yellow curves (IR) only fluctuate between 0 and 0.2 at the peak hour and contribute to most of the noises of the total occupancy. This conclusion may be incomplete, because fluctuations are also observed in parking lots with an RU composition and are complicated As in Figure 8 (4), the RU of the large auto parts market is composed of different retail shop owners, workers, and regular customers, and the scattered opening and closing time leads to variations in peak hours. However, since the total amount remains basically unchanged, it still shows a periodic parking pattern Sustainability 2025 , 17 , x FOR PEER REVIEW 11 of 24 Figure 6. Ratio of regular users. Figure 7. Relationships between land use and RU ratio. Occupancy curves with a high RU proportion are shown in Figure 8. It is obvious that the red curves (RU) almost overlap with the blue curves (total occupancy), while the yellow curves (IR) only fl uctuate between 0 and 0.2 at the peak hour and contribute to most of the noises of the total occupancy. This conclusion may be incomplete, because fl uctuations are also observed in parking lots with an RU composition and are complicated. As in Figure 8(4), the RU of the large auto parts market is composed of di ff erent retail shop owners, workers, and regular customers, and the sca tt ered opening and closing time leads to variations in peak hours. However, since the total amount remains basically unchanged, it still shows a periodic parking pa tt ern. Figure 8. Occupancy of high RU parking lots. Figure 8. Occupancy of high RU parking lots For low RU periodic parking lots, see Figure 9 . It can be observed that the red RU patterns still remain stable, while the orange IU ones with higher occupancy also show periodicity. Considering there are only 30 periodic parking lots with an RU proportion lower than 0.5, each parking lot thus has its own unique characteristics to some extent. For example, in Figure 9 (7), the moon lake garden is a famous tourist attraction for its night market, and an evening peak can be observed in the IR curve. Since the parking lot is already saturated on normal days, the operators have to squeeze more parking spaces when the evening volume is high (blue curve in day 2). Thus, the periodicity is mainly caused by the saturation and the fixed peak hour It should be noted that not all parking lots with a high RU proportion are periodic; there are still 215 aperiodic parking lots with a RU proportion higher than 0.7, examples of which are shown in Figure 10 , where blue line is the total occupancy rate and red line is the RU occupancy. Most of them are residence, hotel, and public parking lots (orange bar in Figure 7 ), and as discussed before, once these parking lots have mixed users and whole-day service time form periodic occupancy, there is a high possibility that they have a special location or attract particular users, such as a P&R parking lot near a metro station, a suburb “sleep town” residence community, or a tourist hotel during off seasons.

[Find the meaning and references behind the names: Connect, Closed, Six, Green]

Sustainability 2025 , 17 , 833 12 of 23 Sustainability 2025 , 17 , x FOR PEER REVIEW 12 of 24 For low RU periodic parking lots, see Figure 9. It can be observed that the red RU pa tt erns still remain stable, while the orange IU ones with higher occupancy also show periodicity. Considering there are only 30 periodic parking lots with an RU proportion lower than 0.5, each parking lot thus has its own unique characteristics to some extent. For example, in Figure 9(7), the moon lake garden is a famous tourist a tt raction for its night market, and an evening peak can be observed in the IR curve. Since the parking lot is already saturated on normal days, the operators have to squeeze more parking spaces when the evening volume is high (blue curve in day 2). Thus, the periodicity is mainly caused by the saturation and the fi xed peak hour. Figure 9. Occupancy of low RU parking lots. It should be noted that not all parking lots with a high RU proportion are periodic; there are still 215 aperiodic parking lots with a RU proportion higher than 0.7, examples of which are shown in Figure 10, where blue line is the total occupancy rate and red line is the RU occupancy. Most of them are residence, hotel, and public parking lots (orange bar in Figure 7), and as discussed before, once these parking lots have mixed users and whole-day service time form periodic occupancy, there is a high possibility that they have a special location or a tt ract particular users, such as a P&R parking lot near a metro station, a suburb “sleep town” residence community, or a tourist hotel during o ff seasons. Figure 10. High RU non-periodic parking lots. 5.2. Parking Time To further study the characteristics of periodic parking lots, RUs and IUs are classifi ed into three groups separately according to their parking time. Among the 312 periodic Figure 9. Occupancy of low RU parking lots Sustainability 2025 , 17 , x FOR PEER REVIEW 12 of 24 For low RU periodic parking lots, see Figure 9. It can be observed that the red RU pa tt erns still remain stable, while the orange IU ones with higher occupancy also show periodicity. Considering there are only 30 periodic parking lots with an RU proportion lower than 0.5, each parking lot thus has its own unique characteristics to some extent. For example, in Figure 9(7), the moon lake garden is a famous tourist a tt raction for its night market, and an evening peak can be observed in the IR curve. Since the parking lot is already saturated on normal days, the operators have to squeeze more parking spaces when the evening volume is high (blue curve in day 2). Thus, the periodicity is mainly caused by the saturation and the fi xed peak hour. Figure 9. Occupancy of low RU parking lots. It should be noted that not all parking lots with a high RU proportion are periodic; there are still 215 aperiodic parking lots with a RU proportion higher than 0.7, examples of which are shown in Figure 10, where blue line is the total occupancy rate and red line is the RU occupancy. Most of them are residence, hotel, and public parking lots (orange bar in Figure 7), and as discussed before, once these parking lots have mixed users and whole-day service time form periodic occupancy, there is a high possibility that they have a special location or a tt ract particular users, such as a P&R parking lot near a metro station, a suburb “sleep town” residence community, or a tourist hotel during o ff seasons. Figure 10. High RU non-periodic parking lots. 5.2. Parking Time To further study the characteristics of periodic parking lots, RUs and IUs are classifi ed into three groups separately according to their parking time. Among the 312 periodic Figure 10. High RU non-periodic parking lots 5.2. Parking Time To further study the characteristics of periodic parking lots, RUs and IUs are classified into three groups separately according to their parking time. Among the 312 periodic parking lots, for the six types of users, i.e., RLTUs, RMTUs, RSTUs, ILTUs, IMTUs, and ISTUs, we connect the six types of occupancy curves one by one and use the cluster method in Section 4.2.2 to classify these parking lots into different categories. Four major categories and some outliners are exhibited according to their patterns (1) Enclosed parking lot: In total, 21 parking lots among 312 periodic parking lots belong to this category. These are usually parking lots of closed industrial zones or office buildings, with most of the parking time being occupied by RUs. There are only a few visitors, and their entry requires registration. This means that these parking lots’ patterns are closely related to the usage of the main building. The following parking pattern is shown in Figure 11 , where RLTUs (red solid line) form the main framework, with some RMTUs (green solid line) leaving during lunchtime and returning in the afternoon. IUs or short-term parking users can be ignored.

[Find the meaning and references behind the names: Lines, Middle, Record, Company, Black, Parks, Worth, Seven, Flow]

Sustainability 2025 , 17 , 833 13 of 23 Sustainability 2025 , 17 , x FOR PEER REVIEW 13 of 24 parking lots, for the six types of users, i.e., RLTUs, RMTUs, RSTUs, ILTUs, IMTUs, and ISTUs, we connect the six types of occupancy curves one by one and use the cluster method in Section 4.2.2 to classify these parking lots into di ff erent categories. Four major categories and some outliners are exhibited according to their pa tt erns. (1) Enclosed parking lot: In total, 21 parking lots among 312 periodic parking lots belong to this category. These are usually parking lots of closed industrial zones or o ffi ce buildings, with most of the parking time being occupied by RUs. There are only a few visitors, and their entry requires registration. This means that these parking lots’ pa tt erns are closely related to the usage of the main building. The following parking pa tt ern is shown in Figure 11, where RLTUs (red solid line) form the main framework, with some RMTUs (green solid line) leaving during lunchtime and returning in the afternoon. IUs or short-term parking users can be ignored. Figure 11. Enclosed parking lot. (2) Parking lots mainly serve internal users: In total, 190 parking lots belong to this category. See Figure 12, including universities, non-enclosed o ffi ce buildings (where the entire building is not owned by the same company or there are some open facilities), industrial parks, or wholesale markets (with a large number of employees, regular customers, and fewer irregular users). The main parking characteristics and types of these parking lots are similar to enclosed parking lots, but there are some short-term users. This causes certain fl uctuations in the overall parking curve. It is worth noting that the red dashed line (ILTU) occupies about 10% of the parking time and hardly changes overtime. These vehicles represent “zombie cars” because there is only one entry record. These vehicles that have been parked for more than one day will be identi fi ed as irregular longtime parking. Figure 12. Parking lot mainly for internal users. (3) Parking lots mainly serve external users: There are 62 parking lots in this cluster, such as those of hospitals, shopping malls, primary and middle schools, hotels, and government service buildings. See Figure 13, unlike category 1 and 2, these parking lots have a lower ratio of employees or regular customers. Their main composition, in addition to RUs (employees and regular customers), also includes a large number of ISTUs and IMTUs (blue and green dashed lines). The opening hours and number of customers of the main buildings of these parking lots are basically the same on a weekday, which leads to Figure 11. Enclosed parking lot (2) Parking lots mainly serve internal users: In total, 190 parking lots belong to this category. See Figure 12 , including universities, non-enclosed office buildings (where the entire building is not owned by the same company or there are some open facilities), industrial parks, or wholesale markets (with a large number of employees, regular customers, and fewer irregular users). The main parking characteristics and types of these parking lots are similar to enclosed parking lots, but there are some short-term users. This causes certain fluctuations in the overall parking curve. It is worth noting that the red dashed line (ILTU) occupies about 10% of the parking time and hardly changes overtime. These vehicles represent “zombie cars” because there is only one entry record. These vehicles that have been parked for more than one day will be identified as irregular long-time parking Sustainability 2025 , 17 , x FOR PEER REVIEW 13 of 24 parking lots, for the six types of users, i.e., RLTUs, RMTUs, RSTUs, ILTUs, IMTUs, and ISTUs, we connect the six types of occupancy curves one by one and use the cluster method in Section 4.2.2 to classify these parking lots into di ff erent categories. Four major categories and some outliners are exhibited according to their pa tt erns. (1) Enclosed parking lot: In total, 21 parking lots among 312 periodic parking lots belong to this category. These are usually parking lots of closed industrial zones or o ffi ce buildings, with most of the parking time being occupied by RUs. There are only a few visitors, and their entry requires registration. This means that these parking lots’ pa tt erns are closely related to the usage of the main building. The following parking pa tt ern is shown in Figure 11, where RLTUs (red solid line) form the main framework, with some RMTUs (green solid line) leaving during lunchtime and returning in the afternoon. IUs or short-term parking users can be ignored. Figure 11. Enclosed parking lot. (2) Parking lots mainly serve internal users: In total, 190 parking lots belong to this category. See Figure 12, including universities, non-enclosed o ffi ce buildings (where the entire building is not owned by the same company or there are some open facilities), industrial parks, or wholesale markets (with a large number of employees, regular customers, and fewer irregular users). The main parking characteristics and types of these parking lots are similar to enclosed parking lots, but there are some short-term users. This causes certain fl uctuations in the overall parking curve. It is worth noting that the red dashed line (ILTU) occupies about 10% of the parking time and hardly changes overtime. These vehicles represent “zombie cars” because there is only one entry record. These vehicles that have been parked for more than one day will be identi fi ed as irregular longtime parking. Figure 12. Parking lot mainly for internal users. (3) Parking lots mainly serve external users: There are 62 parking lots in this cluster, such as those of hospitals, shopping malls, primary and middle schools, hotels, and government service buildings. See Figure 13, unlike category 1 and 2, these parking lots have a lower ratio of employees or regular customers. Their main composition, in addition to RUs (employees and regular customers), also includes a large number of ISTUs and IMTUs (blue and green dashed lines). The opening hours and number of customers of the main buildings of these parking lots are basically the same on a weekday, which leads to Figure 12. Parking lot mainly for internal users (3) Parking lots mainly serve external users: There are 62 parking lots in this cluster, such as those of hospitals, shopping malls, primary and middle schools, hotels, and government service buildings. See Figure 13 , unlike category 1 and 2, these parking lots have a lower ratio of employees or regular customers. Their main composition, in addition to RUs (employees and regular customers), also includes a large number of ISTUs and IMTUs (blue and green dashed lines). The opening hours and number of customers of the main buildings of these parking lots are basically the same on a weekday, which leads to a fixed parking cycle for IUs. Most RUs with a parking ratio of more than 40% but less than 70% have this parking characteristic Sustainability 2025 , 17 , x FOR PEER REVIEW 14 of 24 a fi xed parking cycle for IUs. Most RUs with a parking ratio of more than 40% but less than 70% have this parking characteristic. Figure 13. Parking lot mainly for external users. (4) Pseudo-periodicity parking lots: In total, 32 parking lots belong to this type, such as those of tourist a tt ractions, hotels, commercial apartments, and entertainment facilities. See Figure 14, unlike the fi rst three categories of parking lots, these parking lots have a complex fl ow of customers and do not have a majority of stable periodic users. However, these parking lots have fi xed peak periods, and during peak hours, they are fully occupied, resulting in a saturated occupancy rate (black solid line) that forms a periodic situation due to the limitation of the parking lot capacity. Therefore, it is called a pseudo-periodicity parking lot in this paper. Figure 14. Pseudo-periodicity parking lots. (5) Outsiders: There are seven outsiders. Unlike the fi rst four main categories, these parking lots have no signi fi cant characteristics for both RUs and IUs, and their occupancy rate is not saturated during peak hours, but they exhibit a periodic pa tt ern. In addition to parking lots covered by categories 3 and 4, outsiders also include some public and park parking lots. The periodic pa tt erns of these parking lots all have certain speci fi cities, such as the public parking lot shown in Figure 15 a. It is located next to a large comprehensive hospital, as shown in Figure 15 c, which mainly serves as a spillover parking lot for the hospital and also exhibits the same three peak periods as the hospital, but with di ff erent characteristics. Figure 15 a is mainly composed of temporary users marked by dashed lines, while Figure 15 c includes more employees and regular patients. Figure 13. Parking lot mainly for external users (4) Pseudo-periodicity parking lots: In total, 32 parking lots belong to this type, such as those of tourist attractions, hotels, commercial apartments, and entertainment facilities. See Figure 14 , unlike the first three categories of parking lots, these parking lots have a complex flow of customers and do not have a majority of stable periodic users. However,

[Find the meaning and references behind the names: Range, Minor, Pink]

Sustainability 2025 , 17 , 833 14 of 23 these parking lots have fixed peak periods, and during peak hours, they are fully occupied, resulting in a saturated occupancy rate (black solid line) that forms a periodic situation due to the limitation of the parking lot capacity. Therefore, it is called a pseudo-periodicity parking lot in this paper Sustainability 2025 , 17 , x FOR PEER REVIEW 14 of 24 a fi xed parking cycle for IUs. Most RUs with a parking ratio of more than 40% but less than 70% have this parking characteristic. Figure 13. Parking lot mainly for external users. (4) Pseudo-periodicity parking lots: In total, 32 parking lots belong to this type, such as those of tourist a tt ractions, hotels, commercial apartments, and entertainment facilities. See Figure 14, unlike the fi rst three categories of parking lots, these parking lots have a complex fl ow of customers and do not have a majority of stable periodic users. However, these parking lots have fi xed peak periods, and during peak hours, they are fully occupied, resulting in a saturated occupancy rate (black solid line) that forms a periodic situation due to the limitation of the parking lot capacity. Therefore, it is called a pseudo-periodicity parking lot in this paper. Figure 14. Pseudo-periodicity parking lots. (5) Outsiders: There are seven outsiders. Unlike the fi rst four main categories, these parking lots have no signi fi cant characteristics for both RUs and IUs, and their occupancy rate is not saturated during peak hours, but they exhibit a periodic pa tt ern. In addition to parking lots covered by categories 3 and 4, outsiders also include some public and park parking lots. The periodic pa tt erns of these parking lots all have certain speci fi cities, such as the public parking lot shown in Figure 15 a. It is located next to a large comprehensive hospital, as shown in Figure 15 c, which mainly serves as a spillover parking lot for the hospital and also exhibits the same three peak periods as the hospital, but with di ff erent characteristics. Figure 15 a is mainly composed of temporary users marked by dashed lines, while Figure 15 c includes more employees and regular patients. Figure 14. Pseudo-periodicity parking lots (5) Outsiders: There are seven outsiders. Unlike the first four main categories, these parking lots have no significant characteristics for both RUs and IUs, and their occupancy rate is not saturated during peak hours, but they exhibit a periodic pattern. In addition to parking lots covered by categories 3 and 4, outsiders also include some public and park parking lots. The periodic patterns of these parking lots all have certain specificities, such as the public parking lot shown in Figure 15 a. It is located next to a large comprehensive hospital, as shown in Figure 15 c, which mainly serves as a spillover parking lot for the hospital and also exhibits the same three peak periods as the hospital, but with different characteristics. Figure 15 a is mainly composed of temporary users marked by dashed lines, while Figure 15 c includes more employees and regular patients Sustainability 2025 , 17 , x FOR PEER REVIEW 15 of 24 Figure 15. Outsiders. 5.3. The Occupancy Pa tt ern In Section 5.2, we analyzed the main reasons and di ff erent user a tt ributes that contribute to the formation of periodic parking lots. In this section, the curve shapes of parking occupancy rates in di ff erent parking lots on weekdays, weekends, and holidays are analyzed. Through clustering algorithms, we obtained four types of parking occupancy curve pa tt erns from a total of 312 parking lots that only have periodic pa tt erns on weekdays. The pink range represents the highest and lowest parking occupancy rate curves for this type of parking lot. In the 312 periodic parking lots, the occupancy curve shape can be divided into four categories, see Figure 16 1 Cluster 1: There are 165 parking lots, which show a convex peak shape from 8 am to 5 pm, with peak hours being consistent and there being pseudo-periodic situations. Meanwhile, some parking spaces will show double peak shapes at 9 am and 3 pm, but it is not obvious. 2 Cluster 2: a total of 54 parking lots, which have signi fi cant double peaks at 9 am and 3 pm, and some parking spaces with three peaks at 9 pm. 3 Cluster 3: a total of 65 parking lots, which are between cluster 1 and 2, with minor peaks at 9 am, 3 pm, and 9 pm. 4 Cluster 4: a total of 17 parking lots, which show concave peak shapes from 8 am to 8 pm, mainly composed of residential, some parks, and public parking spaces. Figure 16. Parking pa tt erns of type 1 parking lots. Figure 15. Outsiders 5.3. The Occupancy Pattern In Section 5.2 , we analyzed the main reasons and different user attributes that contribute to the formation of periodic parking lots. In this section, the curve shapes of parking occupancy rates in different parking lots on weekdays, weekends, and holidays are analyzed. Through clustering algorithms, we obtained four types of parking occupancy curve patterns from a total of 312 parking lots that only have periodic patterns on weekdays. The pink range represents the highest and lowest parking occupancy rate curves for this type of parking lot.

[Find the meaning and references behind the names: Multi, Pool, Fit]

Sustainability 2025 , 17 , 833 15 of 23 In the 312 periodic parking lots, the occupancy curve shape can be divided into four categories, see Figure 16 Sustainability 2025 , 17 , x FOR PEER REVIEW 15 of 24 Figure 15. Outsiders. 5.3. The Occupancy Pa tt ern In Section 5.2, we analyzed the main reasons and di ff erent user a tt ributes that contribute to the formation of periodic parking lots. In this section, the curve shapes of parking occupancy rates in di ff erent parking lots on weekdays, weekends, and holidays are analyzed. Through clustering algorithms, we obtained four types of parking occupancy curve pa tt erns from a total of 312 parking lots that only have periodic pa tt erns on weekdays. The pink range represents the highest and lowest parking occupancy rate curves for this type of parking lot. In the 312 periodic parking lots, the occupancy curve shape can be divided into four categories, see Figure 16 1 Cluster 1: There are 165 parking lots, which show a convex peak shape from 8 am to 5 pm, with peak hours being consistent and there being pseudo-periodic situations. Meanwhile, some parking spaces will show double peak shapes at 9 am and 3 pm, but it is not obvious. 2 Cluster 2: a total of 54 parking lots, which have signi fi cant double peaks at 9 am and 3 pm, and some parking spaces with three peaks at 9 pm. 3 Cluster 3: a total of 65 parking lots, which are between cluster 1 and 2, with minor peaks at 9 am, 3 pm, and 9 pm. 4 Cluster 4: a total of 17 parking lots, which show concave peak shapes from 8 am to 8 pm, mainly composed of residential, some parks, and public parking spaces. Figure 16. Parking pa tt erns of type 1 parking lots. Figure 16. Parking patterns of type 1 parking lots 1 Cluster 1: There are 165 parking lots, which show a convex peak shape from 8 am to 5 pm, with peak hours being consistent and there being pseudo-periodic situations. Meanwhile, some parking spaces will show double peak shapes at 9 am and 3 pm, but it is not obvious 2 Cluster 2: a total of 54 parking lots, which have significant double peaks at 9 am and 3 pm, and some parking spaces with three peaks at 9 pm 3 Cluster 3: a total of 65 parking lots, which are between cluster 1 and 2, with minor peaks at 9 am, 3 pm, and 9 pm 4 Cluster 4: a total of 17 parking lots, which show concave peak shapes from 8 am to 8 pm, mainly composed of residential, some parks, and public parking spaces Meanwhile, on workdays, there are 11 parking lots that do not fit into any of the four categories, exhibiting characteristics of single peaks, multi-peaks, or even convex–concave change peaks. This is greatly related to the complexity of the building’s characteristics and the user composition. As shown below, in Figure 17 (1), the company has a two-shift working hour, resulting in a sharp decline in occupancy rate at 15 pm when the first shift workers leave and an increase in occupancy at night. In Figure 17 (2), the vegetable market, which is only open in the morning, shows a single peak from 5:00 AM to 11:00 AM. In Figure 17 (3), the elementary school shows two peak periods for parents picking up and dropping off their children, and a sharp decline in occupancy during lunchtime when employees eat outside. In Figure 17 (4), the swimming pool only has a peak after working hours. In summary, these special parking lots do not represent the typical usage patterns of single land use parking lots. However, if the actual usage is not taken into account, it will be difficult to achieve shared parking in these parking lots Sustainability 2025 , 17 , x FOR PEER REVIEW 16 of 24 Meanwhile, on workdays, there are 11 parking lots that do not fi t into any of the four categories, exhibiting characteristics of single peaks, multi-peaks, or even convex–concave change peaks. This is greatly related to the complexity of the building’s characteristics and the user composition. As shown below, in Figure 17(1), the company has a two-shift working hour, resulting in a sharp decline in occupancy rate at 15 pm when the fi rst shift workers leave and an increase in occupancy at night. In Figure 17(2), the vegetable market, which is only open in the morning, shows a single peak from 5:00 AM to 11:00 AM. In Figure 17(3), the elementary school shows two peak periods for parents picking up and dropping o ff their children, and a sharp decline in occupancy during lunchtime when employees eat outside. In Figure 17(4), the swimming pool only has a peak after working hours. In summary, these special parking lots do not represent the typical usage pa tt erns of single land use parking lots. However, if the actual usage is not taken into account, it will be di ffi cult to achieve shared parking in these parking lots. Figure 17. The type 1 parking lots in the outlier cluster. For 156 parking lots exhibiting periodic pa tt erns on working days and weekends, the occupancy curve takes on a similar shape to that of a workday, see Figure 18, i.e., convex single, double, and triple peaks, and a concave valley. The di ff erence lies in the total volume of parking during workdays and weekends, i.e., cluster 1 with more parking on workdays than weekends, cluster 2 with similar parking on workdays and weekends, and cluster 3 with less parking on workdays than weekends. Figure 17. The type 1 parking lots in the outlier cluster.

[Find the meaning and references behind the names: Plus, Size, Sample, Common]

Sustainability 2025 , 17 , 833 16 of 23 For 156 parking lots exhibiting periodic patterns on working days and weekends, the occupancy curve takes on a similar shape to that of a workday, see Figure 18 , i.e., convex single, double, and triple peaks, and a concave valley. The difference lies in the total volume of parking during workdays and weekends, i.e., cluster 1 with more parking on workdays than weekends, cluster 2 with similar parking on workdays and weekends, and cluster 3 with less parking on workdays than weekends Sustainability 2025 , 17 , x FOR PEER REVIEW 16 of 24 Meanwhile, on workdays, there are 11 parking lots that do not fi t into any of the four categories, exhibiting characteristics of single peaks, multi-peaks, or even convex–concave change peaks. This is greatly related to the complexity of the building’s characteristics and the user composition. As shown below, in Figure 17(1), the company has a two-shift working hour, resulting in a sharp decline in occupancy rate at 15 pm when the fi rst shift workers leave and an increase in occupancy at night. In Figure 17(2), the vegetable market, which is only open in the morning, shows a single peak from 5:00 AM to 11:00 AM. In Figure 17(3), the elementary school shows two peak periods for parents picking up and dropping o ff their children, and a sharp decline in occupancy during lunchtime when employees eat outside. In Figure 17(4), the swimming pool only has a peak after working hours. In summary, these special parking lots do not represent the typical usage pa tt erns of single land use parking lots. However, if the actual usage is not taken into account, it will be di ffi cult to achieve shared parking in these parking lots. Figure 17. The type 1 parking lots in the outlier cluster. For 156 parking lots exhibiting periodic pa tt erns on working days and weekends, the occupancy curve takes on a similar shape to that of a workday, see Figure 18, i.e., convex single, double, and triple peaks, and a concave valley. The di ff erence lies in the total volume of parking during workdays and weekends, i.e., cluster 1 with more parking on workdays than weekends, cluster 2 with similar parking on workdays and weekends, and cluster 3 with less parking on workdays than weekends. Figure 18. Parking patterns of type 2 parking lots Among the five outsiders, there are also obvious deviations from the clustering results due to the uniqueness of the parking lots, such as those of companies and office buildings in Figure 19 a,b which deviate from the usual clustering 1 pattern and show a morning peak on weekdays and a valley on weekends, and the parking lots of Yinzhou park and the Olympic Center in Figure 19 c,d which show a three-peak characteristic on weekends Sustainability 2025 , 17 , x FOR PEER REVIEW 17 of 24 Figure 18. Parking pa tt erns of type 2 parking lots. Among the fi ve outsiders, there are also obvious deviations from the clustering results due to the uniqueness of the parking lots, such as those of companies and o ffi ce buildings in Figure 19 a,b which deviate from the usual clustering 1 pa tt ern and show a morning peak on weekdays and a valley on weekends, and the parking lots of Yinzhou park and the Olympic Center in Figure 19 c,d which show a three-peak characteristic on weekends. Figure 19. The type 2 parking lots in the outlier cluster. In the 76 parking lots with periodic occupancy on workdays, weekends, and holidays, see Figure 20, it is di ffi cult to obtain representative clustering results due to the complexity of the curves and the small sample size. However, on holidays, the overall occupancy curve shape of parking lots is similar to that on normal days, with the di ff erence being the total occupancy and the peak time. Figure 20. Parking pa tt erns of type 3 parking lots. 5.4. Analysis of Shared Parking Space Supply The study of the characteristics of periodic parking lots can support parking lot design, while the other purpose of this paper is to study whether these periodic parking lots can be used for shared parking. Assuming that all of the vacant parking spaces in these periodic parking lots can be shared, and using the three common peak hours found in Section 5.3, i.e., 9 am, 15 pm, 19 pm, plus midnight 24 pm as representatives, the total number of shared parking spaces that periodic parking lots can provide is obtained in Figure 21. Figure 19. The type 2 parking lots in the outlier cluster In the 76 parking lots with periodic occupancy on workdays, weekends, and holidays, see Figure 20 , it is difficult to obtain representative clustering results due to the complexity of the curves and the small sample size. However, on holidays, the overall occupancy curve shape of parking lots is similar to that on normal days, with the difference being the total occupancy and the peak time 5.4. Analysis of Shared Parking Space Supply The study of the characteristics of periodic parking lots can support parking lot design, while the other purpose of this paper is to study whether these periodic parking lots can be used for shared parking. Assuming that all of the vacant parking spaces in these periodic parking lots can be shared, and using the three common peak hours found in Section 5.3 , i.e., 9 am, 15 pm, 19 pm, plus midnight 24 pm as representatives, the total number of shared parking spaces that periodic parking lots can provide is obtained in Figure 21 .

[Find the meaning and references behind the names: Plazas, Southern, Ces, Half, Early]

Sustainability 2025 , 17 , 833 17 of 23 Sustainability 2025 , 17 , x FOR PEER REVIEW 17 of 24 Figure 18. Parking pa tt erns of type 2 parking lots. Among the fi ve outsiders, there are also obvious deviations from the clustering results due to the uniqueness of the parking lots, such as those of companies and o ffi ce buildings in Figure 19 a,b which deviate from the usual clustering 1 pa tt ern and show a morning peak on weekdays and a valley on weekends, and the parking lots of Yinzhou park and the Olympic Center in Figure 19 c,d which show a three-peak characteristic on weekends. Figure 19. The type 2 parking lots in the outlier cluster. In the 76 parking lots with periodic occupancy on workdays, weekends, and holidays, see Figure 20, it is di ffi cult to obtain representative clustering results due to the complexity of the curves and the small sample size. However, on holidays, the overall occupancy curve shape of parking lots is similar to that on normal days, with the di ff erence being the total occupancy and the peak time. Figure 20. Parking pa tt erns of type 3 parking lots. 5.4. Analysis of Shared Parking Space Supply The study of the characteristics of periodic parking lots can support parking lot design, while the other purpose of this paper is to study whether these periodic parking lots can be used for shared parking. Assuming that all of the vacant parking spaces in these periodic parking lots can be shared, and using the three common peak hours found in Section 5.3, i.e., 9 am, 15 pm, 19 pm, plus midnight 24 pm as representatives, the total number of shared parking spaces that periodic parking lots can provide is obtained in Figure 21. Figure 20. Parking patterns of type 3 parking lots Sustainability 2025 , 17 , x FOR PEER REVIEW 18 of 24 Figure 21. Supply of shared parking spaces. For the total parking demand of 1.85 million in Ningbo, these periodic parking lots can provide an average of more than 80,000 shared parking spaces. Considering that these periodic parking lots are the beginning for opening the shared parking market, if these parking spaces can be e ff ectively utilized, it can still greatly alleviate the parking pressure compared to building new parking spaces. Meanwhile, after using the Kriging interpolation method, the spatial distribution of shared parking spaces in the urban area of Ningbo can be obtained. As shown in Figure 22, taking 9 am and 19 pm on workdays and weekends as examples, it can be found that there is a signi fi cant imbalance in the spatial–temporal distribution of these shared parking spaces. 1 Imbalance in time: On weekdays, during 9 am and 15 pm (see Figure 21) there are nearly half fewer shared parking spaces than during non-working hours. While on weekends and holidays, the number of parking spaces at 15 pm and 19 pm is about 20,000 to 30,000 less than that in the early morning. 2 Imbalance in parking lots: Shared parking spaces are concentrated in a few large parking lots (red circles in Figure 22). As shown in Figure 22, among 66,367 shared parking spaces at 9 am on workdays, 37 large parking lots that can provide more than 500 shared parking spaces occupy 37,730 parking spaces. Meanwhile, for 120 parking lots that can provide less than 50 shared parking spaces, only 1654 parking spaces can be provided. 3 Imbalance in spatial distribution: Due to concentrated land use, from the perspective of large areas, the peak periods of parking lot usage overlap with each other. See Figure 22 a, where Area A is the high-tech development zone of Ningbo, and there are su ffi cient parking spaces. The International Hotel, Convention and Exhibition Center, municipal government, and other parking lots contained therein have a large number of unused parking spaces both on workdays and weekends. Area B is the city center, where most of the periodic parking lots are o ffi ce buildings, hospitals, government agencies, and commercial plazas, and these parking lots are all in short supply of parking spaces. Area D is the southern business district, where the parking lots are all large o ffi ce buildings, therefore showing the same shared characteristic of fewer spaces on weekdays and more on weekends. Meanwhile, Area C is a mixed land use of residences, businesses, o ffi ces, and education in the suburbs, and there is a complementary situation of shared parking spaces in this area. Figure 21. Supply of shared parking spaces For the total parking demand of 1.85 million in Ningbo, these periodic parking lots can provide an average of more than 80,000 shared parking spaces. Considering that these periodic parking lots are the beginning for opening the shared parking market, if these parking spaces can be effectively utilized, it can still greatly alleviate the parking pressure compared to building new parking spaces Meanwhile, after using the Kriging interpolation method, the spatial distribution of shared parking spaces in the urban area of Ningbo can be obtained. As shown in Figure 22 , taking 9 am and 19 pm on workdays and weekends as examples, it can be found that there is a significant imbalance in the spatial–temporal distribution of these shared parking spaces.

[Find the meaning and references behind the names: Circle, Mall]

Sustainability 2025 , 17 , 833 18 of 23 Sustainability 2025 , 17 , x FOR PEER REVIEW 19 of 24 ( a ) Weekday 9 am ( b ) Weekday 19 am ( c ) Weekend 9 am ( d ) Weekend 19 am Figure 22. Distributions of shared parking spaces. If large-scale area analysis provides the design basis for city managers, the practical implementation of shared parking still requires one-to-one discussions between parking lots and the platform in small-scale areas. For users, the purpose of choosing shared parking is to see whether there are vacant parking spaces within an acceptable walking distance near their destination. Therefore, the Guangji Primary School parking lot in Area B and in-city shopping mall parking lot in Area C are taken as examples. If the blue circle in Area B is enlarged, it represents a 1000 m bu ff er with Guangji Primary School as the center. Within this area, there are 18 periodic parking lots, which can provide a total of 4715 parking spaces as shown in Figure 23 a. The occupancy of these Figure 22. Distributions of shared parking spaces 1 Imbalance in time: On weekdays, during 9 am and 15 pm (see Figure 21 ) there are nearly half fewer shared parking spaces than during non-working hours. While on weekends and holidays, the number of parking spaces at 15 pm and 19 pm is about 20,000 to 30,000 less than that in the early morning 2 Imbalance in parking lots: Shared parking spaces are concentrated in a few large parking lots (red circles in Figure 22 ). As shown in Figure 22 , among 66,367 shared parking spaces at 9 am on workdays, 37 large parking lots that can provide more than 500 shared parking spaces occupy 37,730 parking spaces. Meanwhile, for 120 parking

[Find the meaning and references behind the names: Slow, Nine]

Sustainability 2025 , 17 , 833 19 of 23 lots that can provide less than 50 shared parking spaces, only 1654 parking spaces can be provided 3 Imbalance in spatial distribution: Due to concentrated land use, from the perspective of large areas, the peak periods of parking lot usage overlap with each other. See Figure 22 a, where Area A is the high-tech development zone of Ningbo, and there are sufficient parking spaces. The International Hotel, Convention and Exhibition Center, municipal government, and other parking lots contained therein have a large number of unused parking spaces both on workdays and weekends. Area B is the city center, where most of the periodic parking lots are office buildings, hospitals, government agencies, and commercial plazas, and these parking lots are all in short supply of parking spaces. Area D is the southern business district, where the parking lots are all large office buildings, therefore showing the same shared characteristic of fewer spaces on weekdays and more on weekends. Meanwhile, Area C is a mixed land use of residences, businesses, offices, and education in the suburbs, and there is a complementary situation of shared parking spaces in this area If large-scale area analysis provides the design basis for city managers, the practical implementation of shared parking still requires one-to-one discussions between parking lots and the platform in small-scale areas. For users, the purpose of choosing shared parking is to see whether there are vacant parking spaces within an acceptable walking distance near their destination. Therefore, the Guangji Primary School parking lot in Area B and in-city shopping mall parking lot in Area C are taken as examples If the blue circle in Area B is enlarged, it represents a 1000 m buffer with Guangji Primary School as the center. Within this area, there are 18 periodic parking lots, which can provide a total of 4715 parking spaces as shown in Figure 23 a. The occupancy of these parking lots on weekdays is shown in Figure 23 b. It can be seen that all parking lots exhibit a convex peak from 9 am to 17 pm, Meanwhile, the parking lot of the Hong’an office building with 1085 parking spaces (green curve) directly affects the supply characteristics of the entire area. Only this parking lot has a small peak from 17:00 to 20:00, and if all parking lots are considered as one parking lot, as shown in Figure 23 c, we can still see a slow decline from 17:00 to 20:00. Besides this interruption, the total occupancy curve is stable, with the maximum occupancy appearing at 9 am to 15 pm, with around 3850 vehicles, which means that even at the peak, there are nearly 1000 vacant parking spaces in the area Sustainability 2025 , 17 , x FOR PEER REVIEW 20 of 24 parking lots on weekdays is shown in Figure 23 b. It can be seen that all parking lots exhibit a convex peak from 9 am to 17 pm, Meanwhile, the parking lot of the Hong’an o ffi ce building with 1085 parking spaces (green curve) directly a ff ects the supply characteristics of the entire area. Only this parking lot has a small peak from 17:00 to 20:00, and if all parking lots are considered as one parking lot, as shown in Figure 23 c, we can still see a slow decline from 17:00 to 20:00. Besides this interruption, the total occupancy curve is stable, with the maximum occupancy appearing at 9 am to 15 pm, with around 3850 vehicles, which means that even at the peak, there are nearly 1000 vacant parking spaces in the area. ( a ) 100 m bu ff er ( b ) Occupancy of nearby parking lots ( c ) Total parking occupancy Figure 23. Shared parking supply around Guangji primary school. There are nine periodic parking lots in the vicinity of the in-city shopping mall, with a total of 4336 parking spaces, which are mainly shopping mall, cultural, entertainment, and school parking lots. This shows mixed land use and may provide complementary shared parking supply. The occupancy for each parking lot is shown in Figure 24 b, and although there is a similar peak from 9 am to 20 pm, it is clear that the peaks between the parking lots are o ff set by a few hours. Similarly, if all parking lots are considered as one parking lot, the total occupancy curve can also be obtained in Figure 24 c, and except for a sudden peak in the evening of Friday, the occupancy in this area mainly presents a double peak at 10 am and 19 pm, and about 1000 parking spaces can still be provided at 19 pm. ( a ) 100 bu ff er ( b ) Occupancy of nearby parking lots ( c ) Total parking occupancy Figure 24. Shared parking supply around in-city shopping mall. 6. Conclusions In order to achieve sustainable parking, this paper focuses on the promotion of largescale shared parking in the parking platform, periodic o ff -street parking lots are identi fi ed through a hybrid PSD and ACF method, and the parking lot occupancy is clustered using a DBSCAN-based algorithm. Among the 995 o ff -street parking lots, 312 are periodic parking lots. Figure 23. Shared parking supply around Guangji primary school There are nine periodic parking lots in the vicinity of the in-city shopping mall, with a total of 4336 parking spaces, which are mainly shopping mall, cultural, entertainment, and school parking lots. This shows mixed land use and may provide complementary shared parking supply. The occupancy for each parking lot is shown in Figure 24 b, and although there is a similar peak from 9 am to 20 pm, it is clear that the peaks between the parking

[Find the meaning and references behind the names: Given, Reason]

Sustainability 2025 , 17 , 833 20 of 23 lots are offset by a few hours. Similarly, if all parking lots are considered as one parking lot, the total occupancy curve can also be obtained in Figure 24 c, and except for a sudden peak in the evening of Friday, the occupancy in this area mainly presents a double peak at 10 am and 19 pm, and about 1000 parking spaces can still be provided at 19 pm Sustainability 2025 , 17 , x FOR PEER REVIEW 20 of 24 parking lots on weekdays is shown in Figure 23 b. It can be seen that all parking lots exhibit a convex peak from 9 am to 17 pm, Meanwhile, the parking lot of the Hong’an o ffi ce building with 1085 parking spaces (green curve) directly a ff ects the supply characteristics of the entire area. Only this parking lot has a small peak from 17:00 to 20:00, and if all parking lots are considered as one parking lot, as shown in Figure 23 c, we can still see a slow decline from 17:00 to 20:00. Besides this interruption, the total occupancy curve is stable, with the maximum occupancy appearing at 9 am to 15 pm, with around 3850 vehicles, which means that even at the peak, there are nearly 1000 vacant parking spaces in the area. ( a ) 100 m bu ff er ( b ) Occupancy of nearby parking lots ( c ) Total parking occupancy Figure 23. Shared parking supply around Guangji primary school. There are nine periodic parking lots in the vicinity of the in-city shopping mall, with a total of 4336 parking spaces, which are mainly shopping mall, cultural, entertainment, and school parking lots. This shows mixed land use and may provide complementary shared parking supply. The occupancy for each parking lot is shown in Figure 24 b, and although there is a similar peak from 9 am to 20 pm, it is clear that the peaks between the parking lots are o ff set by a few hours. Similarly, if all parking lots are considered as one parking lot, the total occupancy curve can also be obtained in Figure 24 c, and except for a sudden peak in the evening of Friday, the occupancy in this area mainly presents a double peak at 10 am and 19 pm, and about 1000 parking spaces can still be provided at 19 pm. ( a ) 100 bu ff er ( b ) Occupancy of nearby parking lots ( c ) Total parking occupancy Figure 24. Shared parking supply around in-city shopping mall. 6. Conclusions In order to achieve sustainable parking, this paper focuses on the promotion of largescale shared parking in the parking platform, periodic o ff -street parking lots are identi fi ed through a hybrid PSD and ACF method, and the parking lot occupancy is clustered using a DBSCAN-based algorithm. Among the 995 o ff -street parking lots, 312 are periodic parking lots. Figure 24. Shared parking supply around in-city shopping mall 6. Conclusions In order to achieve sustainable parking, this paper focuses on the promotion of largescale shared parking in the parking platform, periodic off-street parking lots are identified through a hybrid PSD and ACF method, and the parking lot occupancy is clustered using a DBSCAN-based algorithm. Among the 995 off-street parking lots, 312 are periodic parking lots Among the periodic parking lots, it is easier to form periodic parking occupancy in hospital, office building, commerce, and education parking lots during weekdays, because regular users are relatively stable, while the opposite is true for residence, hotel, and public parking lots. Additionally, by analyzing the proportion of RUs and IUs, it is found that the main reason for the periodic off-street parking lots is due to the regular users, who form the backbone of parking lots, while a small number of short-term users cause certain fluctuations Therefore, periodic parking lots are divided into four major categories: enclosed parking lots, parking lots serving internal users, parking lots serving external users, and pseudo-periodic parking lots. Subsequently, the time distribution characteristics of periodic parking lots are analyzed, and it is found that most parking lots exhibit a flat peak from 8 to 19 h or a double or triple peak, while only a few parking lots exhibit a concave pattern. Finally, based on the distribution of periodic parking spaces, the total supply of shared parking and its spatial distribution are studied, and the following conclusions are drawn: (1) Shared parking should prioritize large parking lots with more than 500 parking spaces, as these few parking lots provide the majority of shared parking spaces. In small areas, shared parking should be built around these parking lots, making them the core supply of the area because the occupancy characteristics of these parking lots directly affect the pattern of shared parking supply (2) From the perspective of a large area, the total number of parking spaces provided by periodic parking lots is not very large, but from the perspective of a small area, even during peak hours, it can provide more than 1000 shared parking spaces. For the situation of one city with multiple platforms, if small areas can be reasonably divided, they can also attract enough registered users (3) Should the matching of peak hours be given special consideration? Mixed use parking can provide shared parking spaces at different times, which is effective for point-topoint shared parking, and as such residential areas and shopping malls can complement each other. However, from the platform perspective, this may still be insufficient.

[Find the meaning and references behind the names: Plan, Hammond, Board, Jia, Kong, Lyons, Fang, Brain, Read, Great, Fight, Part, Mackay, Zheng, Graham, Evolution, Thank, Case, Author]

Sustainability 2025 , 17 , 833 21 of 23 Users of the entire city may use these shared spaces, which will greatly interfere with the original characteristics of the parking lot. At the same time, urban land use is usually concentrated, and it is easy for an entire district to have the same peak in parking demand. Therefore, the shared parking platform should pay more attention to the total number of parking spaces rather than the matching of parking spaces It can be seen that after the platform-based operation, due to the variety of parking lots and their unique parking occupancy due to usage pattens, there are still great difficulties in the large-scale implementation of shared parking. For a certain type of parking lot such as a hospital parking lot, it can be further divided into large comprehensive hospitals, specialized hospitals (women’s and children’s hospitals, brain hospitals), clinics, and rural hospitals. In addition, some special parking phenomena, such as spillover, zombie cars, pseudo-peak, convex and concave changes, etc., lead to the ineffective utilization of shared parking if the actual parking characteristics are not considered. Meanwhile, if a periodic parking lot is opened for sharing, it also has an interaction with non-periodic parking lots and on-street parking lots, and the periodic occupancy formed by original regular users also changes. Therefore, it is important to consider the actual parking patterns of a parking lot and the interaction of on-street parking lots in future research on shared parking allocation methods Author Contributions: Conceptualization, Y.C. and S.Z.; methodology, Y.C.; software, Y.C.; validation, Y.C., S.Z. and F.X.; formal analysis, X.P. and L.Z.; investigation, X.P. and L.Z.; resources, X.P. and L.Z.; data curation, X.P. and L.Z.; writing—original draft preparation, Y.C.; writing—review and editing, Y.C.; visualization, Y.C.; supervision, S.Z.; project administration, X.P. and L.Z.; funding acquisition, S.Z. All authors have read and agreed to the published version of the manuscript Funding: This research was supported by Ningbo key Research and Development Plan Project (2023 Z 230), “Innovation Yongjiang 2035” Major Application Demonstration Programme (2024 Z 003) Institutional Review Board Statement: Not applicable Informed Consent Statement: Not applicable Data Availability Statement: The data can be available by contacting the authors Acknowledgments: The authors thank the Ningbo Yongcheng parking platform for providing the parking data. The contents of this paper reflect the views of the authors, who are responsible for the facts and the accuracy of the information presented herein Conflicts of Interest: Authors Pan Xiao and Zhang Lei were employed by the company Ningbo Municipal Public Investment Co., Ltd. The remaining authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest References 1 Wu, F.; Ma, M. Clustering Analysis of the Spatio-Temporal On-Street Parking Occupancy Data: A Case Study in Hong Kong Sustainability 2022 , 14 , 7957. [ CrossRef ] 2 Jia, S.; Li, Y.; Fang, T. Can driving-restriction policies alleviate traffic congestion? A case study in Beijing. China Clean Technol. Environ 2022 , 24 , 2931–2946. [ CrossRef ] 3 Lyons, G.; Hammond, P.; Mackay, K. The importance of user perspective in the evolution of MaaS Transp. Res. Part A Policy Pract 2019 , 121 , 22–36. [ CrossRef ] 4 Polydoropoulou, A.; Pagoni, I.; Tsirimpa, A.; Roumboutsos, A.; Kamargianni, M.; Tsouros, I. Prototype business models for Mobility-as-a-Service Transp. Res. Part A Policy Pract 2020 , 131 , 149–162. [ CrossRef ] 5 Hrcher, D.; Graham, D.J. MaaS economics: Should we fight car ownership with subscriptions to alternative modes? Econ. Transp 2020 , 22 , 100167. [ CrossRef ] 6 Shoup, D.C. Cruising for parking Transp. Policy 2006 , 13 , 479–486. [ CrossRef ] 7 Zheng, N.; Geroliminis, N. Modeling and optimization of multimodal urban networks with limited parking and dynamic pricing Transp. Res. Part B Methodol 2016 , 83 , 36–58. [ CrossRef ]

[Find the meaning and references behind the names: Le Mouel, Liu, Lee, Saberi, Real, Najmi, Nissan, Xie, Arch, Yuan, Bikes, Soc, Dasgupta, Path, Zhao, Gao, Palombo, Hsu, Wang, Putra, Jiang, Int, Sci, Ash, Rashidi, Chu, Inf, Kuo, Garrick, Redondo, Sens, Lucky, Wei, Price, Zhu, Jang, Atkinson, Tsai, Big, November, Lett, Beach, Feng, Tao, Cats, Rec, Sign, Martens, Meng, Laporte, Chen, Rich, Cyprus, Prs, Lin, Yan, Levine, Yang, Wolfson, Bernstein, Chou]

Sustainability 2025 , 17 , 833 22 of 23 8 Gu, Z.; Najmi, A.; Saberi, M.; Liu, W.; Rashidi, T.H. Macroscopic parking dynamics modeling and optimal real-time pricing considering cruising-for-parking Transp. Res. Part C Emerg. Technol 2020 , 118 , 102714. [ CrossRef ] 9 Jiang, Z.; Mondschein, A. Analyzing parking sentiment and its relationship to parking supply and the built environment using online reviews J. Big Data 2021 , 3 , 61–79. [ CrossRef ] 10 Al-Turjman, F.; Malekloo, A. Smart parking in IoT-enabled cities: A survey Sustain. Cities Soc 2019 , 49 , 101608. [ CrossRef ] 11 Lin, T.; Rivano, H.; Le Mouel, F. A survey of smart parking solutions IEEE Transp. Intell. Transp. Syst 2017 , 12 , 3229–3253 [ CrossRef ] 12 Kuo, P.F.; Hsu, W.T.; Putra IG, B.; Sulistyah, U.D. The proposed model for analyzing off-street parking Dynamics: A case study of Taipei City Transp. Res. Part A Policy Pract 2024 , 180 , 103965. [ CrossRef ] 13 Yang, Y.; Yang, B.; Yuan, Z.; Meng, R.; Wang, Y. Modelling and comparing two modes of sharing parking spots at residential area: Real-time and fixed-time allocation IET Intell. Transp. Syst 2024 , 18 , 599–618. [ CrossRef ] 14 Cai, Y.; Chen, J.; Zhang, C.; Wang, B. A parking space allocation method to make a shared parking strategy for appertaining parking lots of public buildings Sustainability 2018 , 11 , 120. [ CrossRef ] 15 Mladenovi´c, M.; Delot, T.; Laporte, G.; Wilbaut, C. A scalable dynamic parking allocation framework Comput. Oper. Res 2021 , 125 , 105080. [ CrossRef ] 16 Gao, J.; Li, S.; Yang, H. Shared parking for ride-sourcing platforms to reduce cruising traffic Transp. Res. Part C Emerg. Technol 2022 , 137 , 103562. [ CrossRef ] 17 Ayala, D.; Wolfson, O.; Xu, B.; DasGupta, B.; Lin, J. Pricing of parking for congestion reduction. In Proceedings of the 20 th International Conference on Advances in Geographic Information Systems, Redondo Beach, CA, USA, 6–9 November 2012 18 Yan, X.; Levine, J.; Marans, R. The effectiveness of parking policies to reduce parking demand pressure and car use Transp. Policy 2019 , 73 , 41–50. [ CrossRef ] 19 McCahill, C.; Haerter-Ratchford, J.; Garrick, N.; Atkinson-Palombo, C. Parking in urban centers Transp. Res. Rec. J. Transp Res. Board 2014 , 2469 , 49–56. [ CrossRef ] 20 Yan, P.; Cai, X.; Ni, D.; Chu, F.; He, H. Two-stage matching-and-scheduling algorithm for real-time private parking-sharing programs Comput. Oper. Res 2021 , 125 , 105083. [ CrossRef ] 21 Wang, S.; Li, Z.; Xie, N. A reservation and allocation model for shared-parking addressing the uncertainty in drivers’ arrival/departure time Transp. Res. Part C Emerg. Technol 2022 , 135 , 103484. [ CrossRef ] 22 Xie, M.; Zhang, X.; Wu, Z.; Wei, S.; Gao, Y.; Wang, Y. A shared parking optimization framework based on dynamic resource allocation and path planning Phys. A 2023 , 616 , 128649. [ CrossRef ] 23 Guo, W.; Zhang, Y.; Xu, M.; Zhang, Z.; Li, L. Parking spaces repurchase strategy design via simulation optimization J. Intell Transp. Syst 2016 , 20 , 255–269. [ CrossRef ] 24 Li, C.; Tao, Y.; Liu, S. A shared parking space optimization model to alleviate China’s parking problem considering travelers’ tiered credit risk Transp. Lett 2021 , 13 , 45–52. [ CrossRef ] 25 Hong, D.; Jang, S.; Lee, C. Investigation of shared micromobility preference for last-mile travel on shared parking lots in city center Travel Behav. Soc 2023 , 30 , 163–177. [ CrossRef ] 26 Mouskos, K.C.; Tvantzis, J.; Bernstein, D.; Sansil, A. Mathematical formulation of a deterministic parking reservation system (PRS) with fixed costs. In Proceedings of the 10 th Mediterranean Electrotechnical Conference. Information Technology and Electrotechnology for the Mediterranean Countries, Lemesos, Cyprus, 29–31 May 2000; Volume 2, pp. 648–651 27 Tasseron, G.; Martens, K. Urban parking space reservation through bottom-up information provision: An agent-based analysis Comput. Environ. Urban Syst 2017 , 64 , 30–41. [ CrossRef ] 28 Mei, Z.; Feng, C.; Ding, W.; Zhang, L.; Wang, D. Better lucky than rich? Comparative analysis of parking reservation and parking charge Transp. Policy 2019 , 75 , 47–56. [ CrossRef ] 29 Chou, S.Y.; Dewabharata, A.; Zulvia, F.E. Dynamic space allocation based on internal demand for optimizing release of shared parking Sensors 2021 , 22 , 235. [ CrossRef ] 30 Wu, X.; Lin, J.; Yang, Y.; Guo, J. A digital decision approach for scheduling process planning of shared bikes under Internet of Things environment Appl. Soft Comput 2023 , 133 , 109934. [ CrossRef ] 31 Ni, X.Y.; Lu, W.; Zhang, C.; Liu, Y.; Zhao, J. A comparative analysis of information provision strategies for parking variable message sign display problems J. Intell. Fuzzy Syst 2020 , 39 , 2725–2735. [ CrossRef ] 32 Cats, O.; Zhang, C.; Nissan, A. Survey methodology for measuring parking occupancy: Impacts of an on-street parking pricing scheme in an urban center Transp. Policy 2016 , 47 , 55–63. [ CrossRef ] 33 Pu, Z.; Li, Z.; Ash, J.; Zhu, W.; Wang, Y. Evaluation of spatial heterogeneity in the sensitivity of on-street parking occupancy to price change Transp. Res. Part C Emerg. Technol 2017 , 77 , 67–79. [ CrossRef ] 34 Tsai, M.C.; van Gasselt, S.; Lin, T.C. Adaptation of parking behaviour to price-policy adjustments in taipei city: Patterns and spatial analysis Int. Arch. Photogramm. Remote Sens. Spat. Inf. Sci 2020 , 43 , 205–212. [ CrossRef ]

[Find the meaning and references behind the names: Eng, June, Bauer, Ben, Acm, Dave, Parmar, Ajeng, Olariu, Castelli, Hipp, York, Qin, Field, Dallas, Huang, Rosenblum, Deep, Melbourne, Rep, Pei, Rawat, Chun, Alberico, Ling, Ideas, Floyd, Hamadneh, Joseph, Mag, Med, Newport, Fry, Dang, Herbst, Xin, Ming, Sahili, Hudson, Washington, Quality, Canada, Property, Siam, Liao, Gim]

Sustainability 2025 , 17 , 833 23 of 23 35 Qin, H.; Gao, J.; Zhang, G.; Chen, Y.; Wu, S. Nested logit model formation to analyze airport parking behavior based on stated preference survey studies J. Air Transp. Manag 2017 , 58 , 164–175. [ CrossRef ] 36 Yan-ling, W.; Xin, W.; Ming-chun, Z. Current situation and analysis of parking problem in Beijing Procedia Eng 2016 , 137 , 777–785 [ CrossRef ] 37 ITE Parking Generation Manual , 5 th ed.; ITE: Washington, DC, USA, 2019 38 Al-Sahili, K.; Hamadneh, J. Establishing parking generation rates/models of selected land uses for palestinian cities Transp. Res Part A Policy Pract 2016 , 91 , 213–222. [ CrossRef ] 39 Ajeng, C.; Gim, T.H.T. Analyzing on-street parking duration and demand in a Metropolitan City of a developing country: A case study of Yogyakarta City, Indonesia Sustainability 2018 , 10 , 591. [ CrossRef ] 40 Yan, G.; Yang, W.; Rawat, D.B.; Olariu, S. Smart parking: A secure and intelligent parking system IEEE Intell. Transp. Syst. Mag 2011 , 3 , 18–30 41 Parmar, J.; Das, P.; Dave, S. Study on demand and characteristics of parking system in urban areas: A review J. Traffic Transp. Eng 2019 , 7 , 111–124. [ CrossRef ] 42 Rosenblum, J.; Hudson, A.W.; Ben-Joseph, E. Parking futures: An international review of trends and speculation Land Use Policy 2020 , 91 , 104054. [ CrossRef ] 43 Fry, D.; Hipp, J.A.; Alberico, C.; Huang, J.-H.; Lovasi, G.S.; Floyd, M.F. Land use diversity and park use in New York City Prev. Med. Rep 2021 , 22 , 101321. [ CrossRef ] 44 Mantouka, E.G.; Fafoutellis, P.; Vlahogianni, E.I. Deep survival analysis of searching for on-street parking in urban areas Transp. Res Part C Emerg. Technol 2021 , 128 , 103173. [ CrossRef ] 45 Bauer, A.; Züfle, M.; Herbst, N.; Kounev, S.; Curtef, V. Telescope: An automatic feature extraction and transformation approach for time series forecasting on a level-playing field. In Proceedings of the 2020 IEEE 36 th International Conference on Data Engineering (ICDE), Dallas, TX, USA, 20–24 April 2020; IEEE: New York, NY, USA, 2020; pp. 1902–1905 46 Li, Z.; Zhao, Y.; Liu, R.; Pei, D. Robust and rapid clustering of KPIs for large-scale anomaly detection. In Proceedings of the 2018 IEEE/ACM 26 th International Symposium on Quality of Service, Banff, AB, Canada, 4–6 June 2018; IEEE: New York, NY, USA, 2018 47 Paparrizos, J.; Gravano, L. k-shape: Efficient and accurate clustering of time series. In Proceedings of the 2015 ACM SIGMOD International Conference on Management of Data, Melbourne, Australia, 31 May–4 June 2015; ACM: New York, NY, USA, 2015; pp. 1855–1870 48 Vlachos, M.; Yu, P.; Castelli, V. On periodicity detection and structural periodic similarity. In Proceedings of the 2005 SIAM International Conference On Data Mining, Newport Beach, CA, USA, 21–23 April 2005; Society for Industrial and Applied Mathematics: Philadelphia, PA, USA, 2005; pp. 449–460 49 Liao, T.W. Clustering of time series data—A survey Pattern Recogn 2005 , 38 , 1857–1874. [ CrossRef ] 50 Ding, R.; Wang, Q.; Dang, Y.; Fu, Q.; Zhang, H.; Zhang, D. Yading: Fast clustering of large-scale time series data Proc. VLDB Endow 2015 , 8 , 473–484. [ CrossRef ] Disclaimer/Publisher’s Note: The statements, opinions and data contained in all publications are solely those of the individual author(s) and contributor(s) and not of MDPI and/or the editor(s). MDPI and/or the editor(s) disclaim responsibility for any injury to people or property resulting from any ideas, methods, instructions or products referred to in the content.

Other Environmental Sciences Concepts:

[back to top]

Discover the significance of concepts within the article: ‘Towards Sustainable Parking’. Further sources in the context of Environmental Sciences might help you critically compare this page with similair documents:

Noise, Outsider, Triple Peak, Vegetable market, Spatial distribution, Land use, Fluctuation, False alarm, Holiday, Sustainable development, Digital technology, Peak hour, Amplitude difference, Hypothetical data, Elementary school, Peak time, Linear interpolation, Time series, Double peak, Phase shift, Occupancy rate, Conventional model, Parameter estimation, Temporal distribution, Moving average, Mixed land use, Urban Land Use, Frequency-domain method, Clustering algorithm, User information, Single peak, Regular patients, Autocorrelation function, Parking lot, Morning peak, Traffic condition, Power spectral density, Hybrid method, Sliding window, Irregular user.