An Improved Multi-Mode Two-Step Floating Catchment Area Method for Measuring Accessibility of Urban Park in Tianjin, China

Abstract

Parks, as a major infrastructure that provide public service for urban residents, play a vital role in promoting urban livability and public health. Under the framework of spatial equity, more sophisticated accessibility methods were used on measuring urban park accessibility such as multi-mode 2SFCA. However, the accessibility of residential areas near parks was seriously underestimated by using the multi-mode 2SFCA method. Thus, this study aimed to propose an improved multi-mode 2SFCA method to measure urban park accessibility with a more appropriate approach, by taking residential areas of Tianjin central city as the spatial unit. The results indicate that all residential areas can obtain urban park accessibility, but the spatial distribution of urban park accessibility is heterogeneous. The numerical value of urban park accessibility decreases as the travel time from residential areas to urban parks increases; it is shown that the proposed method can provide a more realistic evaluation compared to the traditional multi-mode 2SFCA method. This study provides a comprehensive and realistic insight into acquainting with urban park accessibility and helps urban planners formulate effective policies and strategies to ease spatial imbalance.

1. Introduction

Urban parks are the most unrestricted green infrastructures in modern cities, which play a vital role in not only improving residents’ physical and mental health, but also in promoting urban health and sustainable development [1,2,3]. In the supply process of urban parks, its spatial configuration will directly or indirectly affect the fairness and effectiveness of residents’ enjoyment of urban park resources. Therefore, optimization on the spatial configuration of urban parks plays an important role in further promoting the happiness of residents, increasing the inclusiveness of space, and maintaining social harmony and stability [4,5,6].

The original indicators used to describe the configuration of urban parks were mainly statistical, such as park area per capita and park area percentage. Though these indicators could describe the quantitative characteristics of urban parks to some extent, they could not reflect the spatial pattern and the service configuration fairness of urban parks [7,8], which can be better indicated by accessibility. Accessibility refers to the difficulty of accessing a destination from any point in space, reflecting the space resistance people overcome in the process of accessing the destination, and is usually measured by distance, time, and cost [9,10,11]. Accessibility has been widely used in research on distribution rationality and the service equity of public facilities such as urban parks, healthcare services, public transportation, and shopping centers [12,13,14,15,16]. Commonly, accessibility methods mainly include the statistical index method [17], travel distance or travel cost method [18], minimum distance method [19], gravity model method [20], and the two-step floating catchment area (2SFCA) method [21]. Of all the above methods, the 2SFCA method is the most extensive method for evaluating accessibility. It reflects the actual situation of residents choosing spatially close facilities across administrative boundaries by setting the limit travel distance (time), which can better measure the spatial accessibility of public service facilities. Moreover, it is easy to understand and apply. Consequently, this method received more attention and development. In the subsequent research, various extension forms have emerged one after another into a huge 2SFCA method model family, which provide a feasible basic framework for various possible expansion forms [22].

One of the most important extensions of the 2SFCA method integrates multiple trip modes into the accessibility method. The idea of the multi-mode 2SFCA method is to evaluate accessibility by integrating multiple trip modes at the same time, which provides more accurate and objective results of accessibility. This was the first time that Mao and Nekorchuk [23] tried to integrate multiple trip modes in accessibility evaluation. They divided the entire population group within the study area into multi-population subgroups, corresponding to different trip modes, and applied it to a healthcare accessibility evaluation in Florida, USA. Langford, Higgs, and Fry [24] later put forward an improved version of the multi-mode 2SFCA method, considering distance decay function, which reflected its accessibility characteristic better. Tao, Yao, Kong, Duan, and Li [25] proposed a general form multi-mode 2SFCA method based on previous research and applied it to the evaluation of healthcare accessibility in Shenzhen, China. In recent years, the multi-mode 2SFCA method has been increasingly applied to measure park accessibility [26,27,28]. In these studies, scholars had integrated trip modes of bicycling, driving, public transit, and walking, into the 2SFCA model, and claimed that the multi-mode 2SFCA method could give a better accessibility measurement.

However, from the result of these studies [26,27], we found that the accessibility of residential areas close to urban parks calculated by the multi-mode 2SFCA method was lower than the single-mode 2SFCA method. But according to logic, when the trip mode increases, the urban park accessibility should be higher, rather than lower. Thus, we thought it was not appropriate to measure urban park accessibility by the traditional multi-mode 2SFCA method.

Therefore, to address this problem, this study intends to improve the multi-mode 2SFCA method, using the central city of Tianjin, China, as a case study area. In this study, based on Baidu Map Services (https://lbsyun.baidu.com/ accessed on 18 May 2021), one of the main online map service providers in China, the trip time of three trip modes (driving, public transit, and walking) was obtained and integrated into the improved multi-mode 2SFCA method. We further compared the results of the improved multi-mode 2SFCA method, the single-mode 2SFCA method, and the traditional multi-mode 2SFCA method. The main questions to be explored in this study are as follows: What are the spatial characteristics of the urban park accessibility results with the improved multi-mode 2SFCA method? What are the differences in urban park accessibility between the improved multi-mode 2SFCA method, the traditional multi-mode 2SFCA method and the single-mode 2SFCA method? Can the improved multi-mode 2SFCA method provide a better urban park accessibility measurement?

The rest of the study arrangements are as follows: Section 2 reviews the generalized 2SFCA method and the traditional multi-mode 2SFCA method, and proposes an improved multi-mode 2SFCA method. Some important parameters of the proposed method are also introduced. Section 3 includes an introduction to the study area, data sources, and processing. Section 4 describes the urban park accessibility results using the improved multi-mode 2SFCA method, comparing them with the traditional multi-mode 2SFCA method results, followed by a discussion. Section 5 further analyzes the limitations and future work. Section 6 summarizes the study.

2. Methods

2.1. The Generalized 2SFCA Method

The two-step floating catchment area (2SFCA) method was first proposed by Radke et al. [21] and further improved by Luo et al. [29]. The 2SFCA method evaluates accessibility in two steps. In the first step, for each supply point (i.e., service facility), all demand points within its catchment area are searched and the supply-to-demand ratio is calculated. In the second step, for each demand point, all supply points within their catchment areas are searched and the accessibility of each demand point is obtained by summing up the supply-to-demand ratios of all supply points located within the catchment area of each demand point.

The generalized 2SFCA method uses a dichotomous method to deal with distance decay [29], which assumes that the supply points are equally accessible within the catchment area, and completely inaccessible outside the catchment area. This is often considered as a drawback of the traditional 2SFCA method [22]. To overcome this drawback, some extended forms have been developed for distance decay functions, essentially introducing additional distance decay functions within the catchment area of the traditional 2SFCA method, including enhancement of the multistage discrete function, Gaussian function [30], Kernel density function [31], and the inverse power function [32] in the 2SFCA method [33].

For such extensions, Wang [34] proposed a generalized 2SFCA form (generalized 2SFCA method) by adding a distance decay function $f_{\left(d_{ij}\right)}$ to the method, which is used to generalize and represent the form of the distance decay function in different extensions, which is calculated as:

• Step 1: Calculate the supply-to-demand ratio:

$$R_j=\frac{S_j}{{\sum }_{i\in \left\{d_{ij}\le d_0\right\}}{D}_i{f}_{\left(d_{ij}\right)}}$$

(1)

where $R_j$ is the supply-to-demand ratio of supply point $j$, $S_j$ is the supply capacity of supply point $j$. $D_i$ is the quantity of demand size of demand point $i$, which is usually expressed by the population size. $d_{ij}$ is the travel time between demand point $i$ and supply point $j,$ $f_{\left(d_{ij}\right)}$ is a distance decay function between $i$ and $j$.

• Step 2: Calculate the accessibility:

$$A_i={\displaystyle \sum }_{i\in \left\{d_{ij}\le d_0\right\}}^k{R}_j{f}_{\left(d_{ij}\right)}$$

(2)

where $A_i$ is the accessibility of demand point $i,$ $k$ is the number of supply points within a catchment area $d_0$ of demand point $i$.

$f_{\left(d_{ij}\right)}$ is a generalization of distance decay function. As mentioned above, the distance decay function $f$ can take various forms. In this study, the distance decay function is the Gaussian function. The Gaussian 2SFCA method was proposed by Dai [30,31]. This function shows the accessibility decay in an “S” shape, which means the decay speed of accessibility with distance is slower in the near and far stages, but faster in the middle [34]. The distance decay function $\mathrm{f}$ can be written as:

$$f_{\left(d_{ij}\right)}=\{\begin{array}{c}\frac{e^{-\frac{1}{2}\times \frac{d_{ij}}{d_0}}-e^{-\frac{1}{2}}}{1-e^{-\frac{1}{2}}},\\ 0, d_{ij}\ge d_0\end{array}{d}_{ij}<d_0$$

(3)

where $d_0$ is the size of the catchment area.

2.2. The Traditional Multi-Mode 2SFCA Method

The traditional multi-mode 2SFCA method was first applied to health services, and the trip modes were mainly driving and public transit. In the evaluation of urban park accessibility, walking is a key trip mode. Therefore, when the multi-mode 2SFCA method is used to evaluate urban park accessibility, the calculation formulas are as follows:

• Step 1: Calculate the supply-to-demand ratio:

$$R_j=\frac{S_j}{{\sum }_{i\in \left\{d_{ij}\le d_0\left(w\right)\right\}}{D}_{i,w}\ast f(d_{ij,w})+{\sum }_{i\in \left\{d_{ij}\le d_0\left(p\right)\right\}}{D}_{i,p}\ast f(d_{ij,p})+{\sum }_{i\in \left\{d_{ij}\le d_0\left(c\right)\right\}}{D}_{i,c}\ast f(d_{ij,c})}$$

(4)

where $R_j$ is the supply-to-demand ratio of urban park $j$, which represents the potential per capita urban park area. $S_j$ is the supply capacity of urban park $j$, which represents the urban park area in this study. $d_0\left(w\right),d_0\left(p\right),d_0\left(c\right)$ are the predefined threshold travel time from $j$ by walking, public transit, and driving. $D_{i,w}$, $D_{i,p}$, $D_{i,c}$ are the demand scales at residential area $i$ respectively, that is, the number of people who can be served by urban parks within their catchment areas by walking, public transit, and driving. $f(d_{ij,w})$, $f(d_{ij,p})$, and $f\left(d_{ij,c)}\right)$ are the distance decay function, and other variables have the same meaning as the above formula.

• Step 2: Calculate the total urban park accessibility according to the population weight under walking, public transit, and driving respectively, as follows:

$$A_i=\frac{D_{i,w}{\sum }_{i\in \left\{d_{ij}\le d_0\left(w\right)\right\}}^j{R}_{j,w}{f}_{\left(d_{ij,w}\right)}+D_{i,p}{\sum }_{i\in \left\{d_{ij}\le d_0\left(p\right)\right\}}^j{R}_{j,p}{f}_{\left(d_{ij,p}\right)}+D_{i,c}{\sum }_{i\in \left\{d_{ij}\le d_0\left(c\right)\right\}}^j{R}_{j,c}{f}_{\left(d_{ij,c}\right)}}{D_{i,w}+D_{i,p}+D_{i,c}}$$

(5)

where $A_i$ is the total urban park accessibility of residential area $i$ under the traditional multi-mode 2SFCA method, which is the weighted average of the sub-accessibility of residential area $i$ under walking, public transit, and driving. Other variables have the same meaning as the above formula.

2.3. The Improved Multi-Mode 2SFCA Method

By comparing the urban park accessibility results of Hu et al. [27], we found that the results calculated by the multi-mode 2SFCA method showed that the accessibility of the residential areas located nearby the urban parks decreased, and was even severely lower than the results under a single walking mode. Furthermore, this phenomenon existed in other multi-mode 2SFCA applications if the walking mode was included. In order to explain this unreasonable result, we analyzed the formula of the multi-mode 2SFCA. A further investigation showed that the multi-mode 2SFCA method utilized a universal supply-to-demand ratio in evaluating accessibility by multiple trip modes. In addition to the walking population, the denominator of the supply-to-demand ratio of the urban park by multiple trip modes incorporated a large population in public transit and driving. Within the unique urban park, the population at the same time threshold increased on a large scale, thus the denominator in the supply-to-demand ratio increased exponentially, which resulted in a conspicuous underestimate of the supply-to-demand ratio. Even in the process of accessibility summation calculation, the number of urban parks increased as the catchment area of urban parks increased. Nevertheless, because of the extremely low supply-to-demand ratio for each urban park, the overall accessibility reduced correspondingly.

Therefore, this study improved the traditional multi-mode 2SFCA method and calculated the supply-to-demand ratios for different trip modes, respectively.

The calculation process mainly includes three steps:

• Step 1: Calculate the supply-to-demand ratios of the urban park services for the three trip modes, as follows:

$$R_{j,w}=\frac{S_j}{{\sum }_{i\in \left\{d_{ij,w}\le d_w\right\}}{D}_{i,w}\ast f\left(d_{ij,w}\right)}$$

(6)

$$R_{j,p}=\frac{S_j}{{\sum }_{i\in \left\{d_w<d_{ij,p}\le d_p\right\}}{D}_{i,p}\ast f\left(d_{ij,p}\right)}$$

(7)

$$R_{j,c}=\frac{S_j}{{\sum }_{i\in \left\{d_w<d_{ij,c}\le d_c\right\}}{D}_{i,c}\ast f\left(d_{ij,c}\right)}$$

(8)

where $R_{j,w}$,$R_{j,p}$, $R_{j,c}$ are the supply-to-demand ratios of the residents by walking, public transit, and driving, respectively. $D_{i,w}$, $D_{i,p}$, $D_{i,c}$ are the demand scales at residential area $i$, that is, the number of people who can obtain the services of urban parks within their respective catchment areas by walking, public transit, and driving. The hypothesis based on this study is that within the range that can be reached by walking, all residents choose walking as the trip mode; outside the range that cannot be reached by walking, residents can use public transit or driving as the trip mode. $d_{ij,w}$, $d_{ij,p}$, $d_{ij,c}$ are the travel time of walking, public transit, and driving, respectively, $d_w,d_p,d_c$ are the catchment areas of walking, public transit, and driving, respectively. $f(d_{ij,w})$, $f(d_{ij,p})$, $f\left(d_{ij,c}\right)$ are the distance decay function.

• Step 2: Calculate the accessibility of the three trip modes according to the supply-to-demand ratio, as follows:

$$A_{i,w}={{\displaystyle \sum }}_{k\in \left\{d_{ij,w}\le d_w\right\}}{R}_{j,w}\ast f\left(d_{ij,w}\right)$$

(9)

$$A_{i,p}={{\displaystyle \sum }}_{k\in \left\{d_w<d_{ij,p}\le d_p\right\}}{R}_{j,p}\ast f\left(d_{ij,p}\right)$$

(10)

$$A_{i,c}={{\displaystyle \sum }}_{k\in \left\{d_w<d_{ij,c}\le d_c\right\}}{R}_{j,c}\ast f\left(d_{ij,c}\right)$$

(11)

• Step 3: Calculate the total accessibility, as follows:

$$A_i=A_{i,w}+A_{i,p}+A_{i,c}$$

(12)

where $A_{i,w}$, $A_{i,p}$, $A_{i,c}$ are the accessibility of residential areas by walking, public transit, and driving, respectively. $A_i$ is the urban park accessibility of residential area $i$ by the improved multi-mode 2SFCA method, and other variables have the same meaning as the above formula.

3. Case Study: Access to Urban Park in the Central City of Tianjin

3.1. Study Area

Tianjin is in the Bohai Rim region of China, which is one of the four municipalities directly under the Central Government of China. Located within the outer ring expressway, the central city of Tianjin is the political, economic, and cultural center of the city and the most densely populated area. The study area covers six central districts (Heping District, Hexi District, Nankai District, Hedong District, and Hongqiao District) and parts of four surrounding districts (Xiqing District, Dongli District, Jinnan District, and Beichen District), as shown in Figure 1, cover an area of 434 km². By the end of 2018, there were 84 streets in the study area, with a residential population of 6.582 million.

3.2. Data Sources and Processing

3.2.1. Urban Parks

Because the tourists of comprehensive parks may select different trip modes, such as walking, driving, or public transit, this study selected comprehensive parks as the study target to measure accessibility. Referring to the “Tianjin Urban Parks List”, the distribution of 21 urban parks was vectorized from high-precision satellite images, as shown in Figure 1. Information about the parks includes name, size, and location. For ensuring calculation accuracy, the entrances of the urban parks were also collected from Baidu map and street view (https://map.baidu.com accessed on 10 May 2021).

3.2.2. Population Data

In order to compare the accessibility results of different methods at micro-scale, residential communities were taken as the minimum spatial units for the study of urban park accessibility. In the study area, we scratched 3120 residential communities from Baidu Map API, the information about the residential areas include name, area, and boundary coordinates. The population data was obtained from the statistical yearbook of each administrative region of Tianjin in 2018. The population of each residential area was obtained by following formula:

$$RP=\frac{RA}{DA}\times DP$$

(13)

where $RP$ is the population of each residential area (person), $RA$ is the area of residential area (km²), $DA$ is the area of the street where the residential area is located (km²), and $DP$ is the population of the street (person).

By combining the population data with locations of residential areas, a population distribution map was made, as seen in Figure 2. The population of each residential area is used to represent the degree of its demand in calculating the accessibility by the improved multi-mode 2SFCA method.

3.2.3. Travel Time

Travel time restricts people’s urban park accessibility, and it depends on the actual local road network conditions and actual trip modes [25]. Most commonly, the travel time is measured by modeled road network and assumed travel speeds, neglecting the effects of temporal impedance of the transportation system. To obtain a more accurate travel time, some studies used travel time generated by online Maps to obtain the accessibility [35]. Travel time is generated by Maps service providers, combining transport network and real-time traffic conditions; therefore, this study used the path planning function provided by Baidu Maps Web API. Taking the center of the residential area as the origin, and the entrance point of the urban park as the destination, travel time and travel distance can be retrieved and extracted via Python by setting three travel modes (driving, public transit, and walking). If a park has several entrances, the shortest travel time will be taken as the travel time from residential areas to the urban park. To remove uncertainty in travel time data by spatial traffic conditions, the average travel time departing at 10 a.m. for four weekdays (from 18 May to 21 May 2021) was used as the travel time.

Since this study focuses on exploring the differences between single and multiple trip modes, with the need for consistency of the other variables, we finally determined 20 min as the catchment area for residents to reach the urban park under the three trip modes of walking, public transit, and driving, according to field research and previous related studies [36]. The main reason was the consideration of the maximum acceptable travel time for residents in walking, public transit, and driving.

4. Result Analysis

The improved multi-mode 2SFCA method was used to calculate the urban park accessibility in Tianjin central city, as shown in Figure 3. In order to facilitate comparative analysis, the traditional multi-mode 2SFCA method and the single-mode 2SFCA method were used to calculate urban park accessibility, respectively, and walking was considered as a trip mode by the single-mode 2SFCA method, as shown in Figure 4.

From the spatial distribution of urban park accessibility in the study area, the calculation results of the improved multi-mode 2SFCA method (Figure 3) show that all residential areas can obtain urban park accessibility, but the accessibility varies greatly. On the whole, urban park accessibility in the southern and western regions is higher than that in the central and eastern regions. The high-value areas are clustered around the urban parks as their centers. Furthermore, the spatial distribution of urban park accessibility obtained by the traditional multi-mode 2SFCA method (Figure 4a) is similar to that obtained by the improved multi-mode 2SFCA method. These results show that all residential areas have urban park accessibility, and there is significant spatial disparity. However, the spatial distribution of urban park accessibility obtained by the single-mode 2SFCA method (Figure 4b) is quite different from the results obtained by the multi-mode methods. These results show that only residential areas near urban parks are accessible, which accounts for 46.25% of the whole study area. Meanwhile, 53.75% of residential areas cannot obtain urban park accessibility by a single mode. These results show that as the trip mode option increases, the number of residential areas with urban park accessibility increases simultaneously from the accessibility spatial distribution.

From the numerical value of urban park accessibility (

Table 1. The comparison of improved multi-mode 2SFCA, traditional multi-mode 2SFCA, and single-mode 2SFCA.

	Maximum	Minimum	Mean	SD
Improved Multi-mode	90.586	0.301	3.861	4.596
Traditional Multi-mode	1.068	0.030	0.253	0.147
Single-mode	88.301	0	1.182	4.238

), both the improved multi-mode 2SFCA method and the traditional multi-mode 2SFCA method can increase the number of residential areas which can obtain urban park accessibility. However, there are obvious differences between the two methods. According to the result of the improved multi-mode 2SFCA method, the range of urban park accessibility is 0.301–90.586, with a mean of 3.861. Compared with the results of the traditional multi-mode 2SFCA method, the minimum value increases by 0.271, the maximum value increases by 89.518, and the overall mean value increases by 3.608, which indicates a significant improvement in the overall urban park accessibility. In addition, it can also be found from Table 1 that, compared with the results of the single-mode 2SFCA method, the maximum, minimum, and average values of accessibility of the improved multi-mode 2SFCA method are optimized. Meanwhile, compared with the single-mode 2SFCA method, the minimum value of the traditional multi-mode 2SFCA method result increases by 0.030, the maximum value decreases by 87.233, and the overall mean value decreases by 0.929. This indicates that the traditional multi-mode 2SFCA method reduces urban park accessibility for residential areas by a high value, compared with the single-mode 2SFCA method. Furthermore, we can see from the standard deviation of the three methods that the improved multi-mode 2SFCA method makes a more significant difference in urban park accessibility than the single-mode 2SFCA method. By contrast, the traditional multi-mode 2SFCA method severely reduces more urban park accessibility values of residential areas than the single-mode 2SFCA method. Therefore, it is unreasonable to use the traditional multi-mode 2SFCA method to evaluate urban park accessibility.

To further compare differences between the improved multi-mode 2SFCA method and the traditional multi-mode 2SFCA method, this study also calculated the difference of accessibility obtained by them, as shown in Figure 5. It can be seen that urban park accessibility of all residential areas has been improved in varying degrees by using the improved multi-mode 2SFCA method. The urban park accessibility of residential areas near the urban parks has been particularly improved, with a maximum increase of 89.810. We also compared the results of two multi-mode 2SFCA methods, respectively, with the result of the single-mode 2SFCA method, as shown in Figure 6. We can see that by the traditional multi-mode 2SFCA method, the catchment area is expanded and the number of residential areas which can obtain urban park accessibility is increased, but the residential areas with high-value accessibility are reduced; that is, the residential areas near the urban parks are underestimated. In contrast, the results calculated by the improved multi-mode 2SFCA method improve not only the accessibility evaluated by the single-mode 2SFCA method, but also the accessibility of residential areas near urban parks. By the Section 2.2 and the formula (4) in it, we evaluated the supply-to-demand ratio of multiple trip modes by the traditional multi-mode 2SFCA method, but the total number of people walking, using public transit, and driving, which we took as the total population, is larger than the actual number of people served by the three kinds of trip modes. As a result, the supply-to-demand ratio decreased. Although the number of residential areas increased, urban park accessibility of residential areas with high-value accessibility reduced. On the contrary, by the improved multi-mode 2SFCA method, we set up three supply-to-demand ratios for three kinds of population of different trip modes, respectively. The denominators of three kinds of supply-to-demand ratios are the population of different trip modes, respectively, rather than all populations. Therefore, the calculation results not only increase the number of residential areas, but also improves the overall urban park accessibility, as shown in Figure 7. The improved multi-mode 2SFCA method avoids this problem deftly, and after taking account of multiple trip modes, it achieves a more reasonable evaluation of urban park accessibility.

5. Discussion

The traditional multi-mode 2SFCA method underestimates urban park accessibility and the improved multi-mode 2SFCA method, reintegrating multiple trip modes, provides a more rational evaluation of urban park accessibility. The results of this study show that under the calculation based on the improved multi-mode 2SFCA method, the catchment area is expanded and the number of residential areas gaining urban park accessibility increases. Compared with the traditional multi-mode 2SFCA method, the improved multi-mode 2SFCA method ameliorates the overall underestimation of urban park accessibility, especially for residential areas near urban parks, and this method is closer to reality. Generally, the improved multi-mode 2SFCA method can achieve a more reasonable evaluation of urban park accessibility to reflect spatial equity (Figure 7). However, there are still limitations in this study on the topic of urban park accessibility. First, in obtaining the service’s ability of parks, this study mainly used the parks’ service scale (area) as the only factor for measurement of the serviceability of parks. Serviceability is also affected by other attributes, such as function, type, landscape quality, facilities, park maintenance, and public psychological perception [27,37,38,39]. Hence measuring urban park accessibility based on their scale alone is limited. Second, the study does not consider different needs for parks of various groups of the population. For example, vulnerable groups including children [40], youth [41], the elderly [42,43], and the disabled [44], may lead to deviations in the actual needs of residents for parks. However, our current data lacks statistical information of relevant population attributes, which may affect the accuracy of urban park accessibility. Third, there are still boundaries in the accessibility calculations line effect, that is, residents living close to the city boundary may choose parks across the city for services, which leads to deviations in the final result [45]. Finally, in the choice of transportation, residents may have other options for traveling, such as bike-sharing [26]. Therefore, this study will explore these issues further in the future.

6. Conclusions

Based on the framework of the multi-mode 2SFCA method, we proposed a more innovative and rational method to incorporate trip modes into the evaluation of urban park accessibility. Taking Tianjin central city as the study area, this study illustrated the implementation of the improved multi-mode 2SFCA method and compared its accessibility evaluation with that of the traditional multi-mode 2SFCA method.

The comparison analysis suggested that the traditional multi-mode 2SFCA method tended to underestimate accessibility in residential areas around urban parks. These miscalculations led to population variation in different trip modes being ignored. Considering populations within various trip modes, the improved multi-mode 2SFCA method provides a more realistic calculation.

In conclusion, the improved multi-mode 2SFCA method is more suitable for actual urban park accessibility services and residents’ travel conditions. It also reflects the spatial layout characteristics of urban parks in Tianjin central city more accurately. Overall, compared with previous applications of the multi-mode 2SFCA method, this study introduces a more rational and objective method for the evaluation of urban park accessibility, which is beneficial to correctly evaluate the spatial fairness of urban park accessibility, and provide decision-making guidance for fair-oriented urban park planning. This method can also be used for the evaluation of other public facilities.