Study on landscape evaluation and optimization strategy of Central Park in Qingkou Town

Comprehensive parks in small towns generally serve the urban residents within a few kilometers of the park. Parks are generally in an area with a large flow of people in a small town15. However, the geographical blockage indicates that the users of parks in small towns are generally of limited educational level, so the investigation process is more complicated. This study adopts a multimethod design with DPOE (diagnostic post-use evaluation) as the main research method, supplemented by the analytic hierarchy process (AHP) and GIS technology, to quantitatively evaluate the current use of comprehensive parks in small cities and towns, and make a decision based on the evaluation results regarding the corresponding optimization and promotion strategy. The research was divided into the following three stages.


Basis of evaluation

Based on the perspective of “human body—movement—space—place—environment”21, stimulation theory and control theory in environmental psychology are used as the main directions for setting up and investigating recreational behaviors. Field investigations were conducted on the current environment, region, culture and other relevant factors of Central Park in Qingkou Town from April to May 2020. The survey covered weekdays, weekends and holidays, sunny/rainy days and mornings and evenings. The main task of the survey was to supervise the development and use of parks in small towns, the number of users, the types of facilities, and the appearance and maintenance of the park22 and to list the problems related to the environment and its ecology, the benefits provided by groups of parks, or the benefits provided to the surrounding enterprises and schools. Especially after the epidemic, residents have had a more active and urgent need for the participation of green space. To cover a wider range of weather and time conditions, data were collected in sequence during the observation period according to a preset scheme for four periods of the day, alternating between two working days and one weekend each week23. Two team members (interviewers) were in contact with tourists at different times in the park from 8 am to 8 pm. In order to minimize selection errors, each respondent was invited to participate in the survey (the targets included adults, children, and adolescents). Elderly people who agreed to participate in the survey were asked about their visit activities and usage behaviors in the park, such as the frequency of visits, distance, and satisfaction with respect to park maintenance, safety, and infrastructure construction24. Finally, the survey requested relevant social demographic information, such as age, highest education level, and marital status (with or without children).

Construction of a performance evaluation index system for comprehensive park landscapes in small towns

The evaluation index of the general applicability of small-town parks was established using a field investigation and the combination of the American landscape performance series (LPS)25. The index system was divided into three levels. The first level was the target level, that is, the evaluation index system of the comprehensive landscape performance of small-town parks. The second level was the criterion level, including environmental performance, health performance and economic performance. As the embodiment of the second level, the three-level index layer mainly includes park construction, infrastructure setting, landscape quality, garden atmosphere, and tourist behavior. Considering the accuracy of comprehensive park evaluation in small towns, the three-level index layer was used to obtain 19 related indexes after expert advice and screening20.

Questionnaire design

Since this study is aimed at parks in small towns and the surrounding residents generally have a low level of education, to obtain more effective data, the questionnaire was in the form of ticking. In terms of content, the questionnaire was divided into two parts. The first part collected basic information about tourists, such as the mode of transportation to the park and visit frequency. Second, there were 19 evaluation index factors. The Likert method was used to evaluate the index26, and the answers each had five levels: very satisfied, satisfied, average, dissatisfied and very dissatisfied.

Combination weight analysis

Some papers in the past ten years have discussed the limitations of the AHP in dealing with the complexity and uncertainty of evaluation indicators and used fuzzy comprehensive evaluation methods to deal with the problem of uncertainty27,28,29,30. However, the fuzzy evaluation method has gradually been eliminated due to its inability to quickly determine the evaluation content31. Therefore, this study used indicator weights for analysis and evaluation. On the basis of the AHP analytic hierarchy process, the coefficient of variation (CV) is added to the weight of each indicator32. The main purpose of adopting this method is to establish a landscape performance evaluation system for Qingkou Central Park, refer to the satisfaction evaluation of tourists through the DPOE, and determine the content that the park needs to be optimized.


The places where tourists gather or where tourists are the most often have a certain reference significance for the planning and design of parks. Therefore, collecting data on tourist gathering places is needed in the research process33. An increasing number of studies have pointed to the use of social media to examine users’ daily life behaviors and spatial distribution relationships. Wood et al. attempted to evaluate the access rate of entertainment venues based on the location of photos posted on Flickr34. Hamstead et al. use geolocation data from Flickr and Twitter to assess changes in the use of all parks in New York city35. Because the identification system for comprehensive parks in most small towns is not clear, tourists cannot clearly identify a place to visit or a location they often visit. Therefore, the following attempts were made: (1) A total of 182 points of interest of tourists in Qingkou Central Park were collected through Octopus Aata Collector 8, Six-Feet and other software. (2) The distribution of interest points and on-site observations were used to identify five gathering points that cover most of the park landscape. They were named A, B, C, D and E (Table 1) for fixed video recording. The average weekday traffic and weekend traffic information was obtained. (3) The collected data were imported into Excel for sorting, and the utilization of parks was visually expressed in different time periods through GIS. (4) From May to August 2020, 300 copies of paper questionnaires were distributed at nodes A, B, C, D and E. (5) In addition, the questionnaire content was imported into Excel for simple processing, classification and deletion of duplicate data. (6) Finally, the spatial and temporal distribution results and comprehensive evaluation results of tourists in comprehensive parks in small towns were obtained, and optimization suggestions were summarized.

Table 1 Landscape node and application of the Central Park.

Spatial distribution and experience of respondents

The behavior track, gathering area and spatial distribution of tourists in the park affect the evaluation of the park. However, due to the different behavioral habits of tourists, some areas of the park will have a crowd gathering effect36. As a result, the use of environmental resources in the park is uneven, resulting in the waste of environmental resources and ecological damage.

During the study, to make the data more accurate, a total of six samples were randomly selected from three working days and three weekends from May to July 2020 for data collection. The collection method consisted of five people at five important nodes in the park that basically cover the popular areas (including A. Northeast Main Entrance Square, B. Southwest Main Entrance Square, C. West Secondary Entrance Square, D. Sports Theme Square, and E. Waterfront wooden platform); the sites were filmed from 06:00 to 20:00 (fifteen minutes were taken from 06:00–08:00, 08:00–11:00, 11:00–13:00, 13:00–17:00 and 17:00–20:00 for each time period) to record the activity track of tourists in a day. Then, the average weekday samples and weekend samples were averaged, and the data of the flow of people and the length of stay were processed. The flow of people in the different periods of each node of the two samples was sorted using Excel, and coordinate marks were made on the construction drawings according to the image data to screen out repeated and unreasonable data37,38. Then, ArcGIS was used to perform data visualization (scenic spot heat), and the following conclusions could be drawn: there was a significant difference in the stay time of tourists at different nodes (Fig. 2).

Figure 2

Schematic diagram of the flow of people in different times and spaces.

Through node analysis, it was found that tourists stayed in the park for at least 30 min. Compared with the research process, it took approximately 1 h and 10 min to complete the whole track through the Six-Foot app, which showed that tourists stayed in the park for a longer time. Among them, the passenger flow peaks were during 06:00–08:00, 11:00–13:00 and 17:00–20:00, while the passenger flow was lower at other times. In addition, according to observation and survey data summary, passenger flow was not evenly concentrated among the five node areas6. Overall, node A and node B had a large flow of tourists and a longer stay time. Node D was ranked next, while node C and node E had less traffic and shorter stay times.

Overview of respondents

According to the POE field survey and questionnaire survey, 292 valid questionnaires (recovery rate 97.3%) were obtained, among which 58.2% (169 persons) were female, slightly higher than the 41.8% (123 persons) that were male. Among them, 40.5% (118 persons) were carrying children under 8 years old. In terms of age, young people between 30 and 39 years old accounted for 43.7% (128 people), most of whom carried children, followed by younger people between 19 and 29 years old and young people between 40 and 49 years old, which fully showed that the urban center where the park is located was dominated by young people. Among the visitors, 71.9% (210 people) were local residents, 18.5% (54 people) were temporary residents, and 9.6% (28 people) were foreign tourists. In terms of how people arrived at the park, most of them came by walking (116 people), accounting for 38.9%, followed by electric vehicles and private cars (61 people each), accounting for 21.6%. Among them, 53% (154 people) chose to come to the park when they were free, and 24% (70 people) came to the park three or four times a week. In the park system of this small town, the public green space served the local residents to a large extent and gradually became a recreational place for the real-time entertainment of nearby residents (Table 2).

AHP-CV comprehensive weight analysis

The development of the AHP-CV combined weights has led to a change in the method of determining indicator weights from a single subjectiveness to a comprehensive objectivity39. In order to avoid the evolution of the AHP to a single weighting method, the AHP method is combined with the SW and CV methods here to calculate objective weight, based on the principle of minimum information entropy combining two kinds of weighted information40.

  1. 1.

    To ensure the reliability of the data, first, check the consistency of the paired comparison matrix:

    $$textRC = text IC/textIR$$

    If RC < 0.1, the result is acceptable and has the significance of in-depth research.

  2. 2.

    The CV method of calculating objective weights is a statistical measure of the degree of dispersion of the probability distribution or frequency distribution32, that is, the ratio of the standard deviation to the evaluation mean, expressed as a certain degree of change. This study uses the CV method to calculate objective weights.

    Coefficient of variation calculation model:

    $$C_i = fracsigma_i mu_i $$

    Weight calculation model:

    $$W_2i = fracC_i mathop sum nolimits_i = 1^n C_i $$

  3. 3.

    Combine the obtained weight i with the weight obtained by CV, and the combined weight of each evaluation index adopts the calculation principle of minimum information entropy. The combination of methods makes full use of the AHP and CV methods and solves the uncertainty problem caused by the fuzzy analysis method.

Method model:

$$W_i = fracsqrt W_1i W_2i mathop sum nolimits_i = 1^n sqrt W_1i W_2i $$

Experiments statement

All test methods were performed in accordance with relevant guidelines and regulations. All pilot schemes have been approved by the ethics committee of Fuzhou Municipal government, including any relevant details. Informed consent has been obtained from all participants in the questionnaire and/or their legal guardians (Tables 2, 3).

Table 2 Characteristics of visitors.
Table 3 Qingkou central park evaluation index weight and weight ranking.