Fuzhou is located in the eastern part of Fujian Province, China, within the southeastern coastal region (Figure 1). It borders the East China Sea to the east, covering a total area of 11968 square kilometers and a coastline of 1137 km. The Min River flows through the city for 150 km, with over 30 tributaries. The city’s landscape is primarily characterized by mountains and hills, while tidal flats, plains, and basins make up approximately 30% of the total land area. The forest coverage rate in Fuzhou is approximately 45%, and its rich landscape pattern provides a complex and diverse research scenario for GSSP. However, Fuzhou, as a rapidly urbanizing coastal capital city, faces typical problems such as green space fragmentation and impaired ecological connectivity, so it is urgent to optimize its green space system.

Primary data for the analysis were sourced from relevant platforms, including the Geospatial Data Cloud and the Geographical Information Monitoring Cloud Platform (https://www.gscloud.cn/sources/?cdataid=302&pdataid=10, http://www.dsac.cn/DataDownLoad/Search?dataID=301400). Information related to the Fuzhou Master Plan was provided by the Fuzhou Planning and Design Institute.

The analysis began with GIS preprocessing of the necessary data. Fragstats 4.4 was then employed to conduct a landscape index analysis to assess current land use patterns. Following this, Conefor 2.6 was used to evaluate connectivity and classify ecological protection areas (GPAs). The minimum cumulative resistance model (MCR) was applied to identify ecological corridors, while the gravity model and network analysis methods were utilized to define the first-level corridors. Finally, green strategic nodes were identified through the intersection of the minimum and minimum-maximum resistance corridors, establishing the “area-corridor-node” structure within the city’s GSSP (Figure 2).

GIS was used to correct and reclassify Fuzhou’s land use status data (provided by the Fuzhou Planning and Design Institute), satellite image data, and the Master Planning. There are a total of five land use categories, including arable land, wood land, grass, waters and construction land. And then data were converted into Tiff files as the base map for the remaining analysis work, and resampling the resolution to 100 m, as shown in Figure 3.

Landscape pattern index analysis is a critical method in spatial pattern analysis and is considered essential for GSSP (Qiao et al., 2024). Fragstats, one of the most widely used landscape pattern analysis software packages, operates at three analytical scales—patch, class, and landscape—and can analyze over 60 landscape indicators (Bhattacharya et al., 2024). These pattern indices reflect properties such as the type, diversity, complexity, and connectivity of landscape patches. The results from the quantitative analysis of landscape patterns can significantly assist in GSSP. In this study, land use types were categorized into five classes: woodland, grassland, arable land, water, and construction land. Additionally, 11 landscape indices closely related to GSSP were selected for analysis, including class area (CA), percent of landscape (PLAND), and number of patches (NP).

Based on scenic areas, nature reserves, and geologically vulnerable zones identified in the Fuzhou Master Plan, ecological land types such as woodland, grassland, and water areas were re-evaluated to delineate 18 GPAs (Figure 4).

Many studies have pointed out that the connection between ecological areas plays a vital role for ecological functions, the maintenance of biodiversity, and ecological vitality (Pietsch, 2018; Montis et al., 2016). We used the probability of connectivity metric ( P C ) to analyze the relationship between GPA and took the importance level ( d P C ) as the primary basis for the classification of GPA. For detailed information on the P C and d P C , please see Saura et al. (2011) and Saura and Torné (2009). The calculation formulas used were as follows (Equations 1, 2): P C = ∑ i = 1 n ∑ j = 1 n p i j a i a j A 2 (1) d P C = P C − P C r e m o v e P C × 100 % (2) where P C is a graph-based availability metric that quantifies functional connectivity, 0 ≤ PC ≤ 1, the larger the P C value is, the higher the connectivity degree of the GPA is; n represents the total number of GPA in the city area; p i j is the maximum product of all path probabilities between GPA i and GPA j ; a i and a j are the areas of GPA i and j ; A is the total area of the city area; d P C is the change (in %) of the connectivity index after removing one GPA, it represents the importance level of one GPA; P C r e m o v e is the overall index value of the remaining GPA after removing a single GPA.

Minimum cumulative resistance model (MCR) is a model for calculating the minimum cumulative resistance from a grid-based map on which estimated dispersal resistances basic on landscape types (Guo et al., 2025; Sun et al., 2024). The calculation formula used was as follows (Equation 3): V M C R = f ⁡ min ∑ j = n i = m D i j R i (3) where V M C R is the value of the minimum cumulative resistance; f is a positive correlation function between the minimum cumulative resistance and the ecological process; min denotes the minimum value of cumulative resistance produced in different processes of patch i transforming into a different patch j ; D i j is the spatial distance between patch i and patch j ; and R i represents the resistance value that exists in the ecological transition. The system of the resistance value will have a significant impact on the results of V M C R . The same patch may have different resistance value for different ecological processes.

Land use type was the main factor for the resistance surface of Fuzhou GSSP. Referencing related studies (Fu et al., 2020) and considering Fuzhou’s actual situation, we built the resistance value system, as shown in Table 1. According to the resistance value, GIS (Arc Toolbox - Spatial Analyst - Raster Reclass - Reclassify) was used to develop the resistance surface graphics (Figure 5).

Scholars have used the MCR in the field of ecological planning, with some success (Zhu et al., 2020; Xu et al., 2018). In the Fuzhou GSSP, MCR was used to construct ecological corridors with minimal resistance and green strategic nodes.

Usually, a gravity model provides an estimate of the volume of flows of, for example, goods, services, or people between two or more locations. In recent years, some scholars have used the model in the field of ecological planning and corrected the formula to better adapt to the actual requirements of the planning (Yang et al., 2017a). The formula as shown below (Equation 4): G a b = L max 2 ⁡ ln S a S b L a b 2 P a P b (4) where G a b is the interaction force between GPA a and b, P a and P b are the resistance values of GPA a and b, S a and S b are the areas of GPA a and b, L a b is the cumulative resistance value of the corridor between GPA a and b, and L max is the maximum resistance of all the corridors in the study area. L a b and L max can be calculated by the Spatial Analyst Tools-Distance-Cost Distance command in GIS (Figure 9).

Combining with the d P C and G a b of GPA, we simulated five scenarios of the first-level corridors network, which within all the potential ecological corridors generated (Figure 10). The network analysis method (Dalton et al., 1973; Haggett and Chorley, 1972) was introduced to evaluate the five scenarios. Taking the loop index α , the average connection index β , the network connection degree λ , and the cost ratio index C R represents the network relationship between GPA and first-level corridors. We compared the indexes of the scenarios and selected a better one. The indexes calculation formula as follows (Equations 5–8): α = l − v + 1 2 v − 5 (5) where α is the loop index, the number of loops present divided by the maximum number of loops possible; l is the number of corridors, and v is the number of GPA. β = l v (6) where β is the average connection index, if β < 1, there is a dendrogram that occurs; if β = 1, there is a single circuit; and if β > 1, it means more complex levels of connectivity exist. γ = l l max = l 3 v − 2 (7) where λ is the network connectivity index, the ratio of the number of links in a network to the maximum number of links possible. C R = 1 − l d (8) where C R is the cost ration and reflects the network’s effectiveness, d is the accumulative resistance of the corridors calculated according to resistance value by using ArcGIS.

The analysis results of landscape pattern indexes in Fragstats 4.2 are shown in Table 3. The woodland, which was the main component of the GPP, had apparent advantages compared with other types of land in terms of CA, CONNECT, AI, PLAND, and TE. The analysis showed the landscape base of Fuzhou city as well. For the GSSP, the main task was to protect the green quantity and maintain a green space structure.

Conefor is a software package that allows quantifying the importance of habitat areas and links for the maintenance or improvement of connectivity. We took the d P C , a significant index of Conefor, to evaluate the critical level of the GPA, setting a distance index of 1000 m and corresponding to a probability of 0.5. To facilitate the comparison between GPA, we normalized the results to 100, as shown in Table 4.

According to the P C and d P C formulas, the area of GPA has a positive relationship with its d P C . The results (Table 4) showed that for GPA 10, the Min River, although the area was small, but because it crossed the city and d P C was high. GPA 6, 7, 8, 9, and 11 were located in the urban area and also in the center of the entire city, a smaller area with a high d P C value, which meant an essential position in the overall structure of the GSS.

As Figure 11 shows, mountains surround the north and southwest sides of the Fuzhou urban area, and the minimum resistance corridors between GPAs intersected in these areas, playing an essential role in the connectivity between GPAs and thus representing essential parts of Fuzhou’s GSS.

We normalized the gravity model’s calculation results to 100,helping to identify the optimal corridors for the first-level corridors (Table 5). According to the G a b values, the gravity of GPA17 and other GPAs were minimal. However, as it was a coastal wetland with a unique landscape and a specific ecological function, we connected it to the first-level corridors. Five scenario networks were simulated for comparison, as shown in Figure 10. Although the indexes α , β , and λ of scenario 5 were relatively better, we took scenario 1 as the final first-level network after careful consideration of C R and d , as shown in Figure 12.

The ecological corridor between GPA 9 and 11 was the only ecological corridor that spanned the entire urban area. It plays a crucial role in the green connection of the northern and southern urban areas, while also linking urban and suburban areas of Fuzhou. More attention should be paid to protection and construction of this vital corridor.

As shown in Figure 13, the intersection of ecological corridors were the green strategic nodes. It was found that GPA 6, 7, 8, 9, and 11 were located at multiple intersections of ecological corridors, and played an essential role in the connectivity between GPAs. They could be regarded as GPAs of strategic importance.

Since low-resistance woodland occupied a substantial part of Fuzhou City, the number of maximum resistance corridors generated was small, and the area where they intersected with the minimum resistance corridor is the minimum-maximum resistance strategic nodes. As shown in Figure 14, strategic nodes 1 and 2 were located in the urban area and the corridor between GPA 9 and 11, which were of strategic importance.

This study uses the Conefor connectivity indicator to classify GPAs, which fits the international trend of prioritizing the protection of critical urban green spaces, while targeting the particular challenges of coastal urbanization in Fuzhou. Similar to the research of Verma et al. (2020), Kefalas et al. (2019), and Xu et al. (2018), individual diversities of GPA type and area were identified in Fuzhou. In particular, there was a significant difference between the area around the urban and suburban areas, as well as differences in functions, as also shown by the studies of García et al. (2020), and Li Collins et al., 2020. Conefor 2.6, which takes connectivity as the primary standard and characteristic, can quickly identify the importance value of a GPA (Qian et al., 2023). However, compared to the common single modeling strategy of MSPA and connectivity analysis in traditional GSSP (Xu et al., 2023), this study proposes the Fragstats- Conefor framework in the source site identification stage. Ecological sources with higher conservation priority are identified through this dual-model extraction framework.

In the analysis process for Fuzhou City, it was found that some GPAs were small in size, but their location in a prominent position resulted in high d P C values. The recognition effect of Conefor has been confirmed by many studies (Jin et al., 2025; Luo et al., 2024), and the simplicity of the required necessary data and the ease of software operation make it a better choice for the city GSSP.

In the process of constructing ecological corridors using the minimum resistance model (Guo et al., 2025), we found that the assignment of resistance values was somewhat subjective. The range of values used in different studies varied significantly (Fu et al., 2020), which had a substantial impact on the form of the corridors. Some studies employed a multiplex resistance surface, integrating data such as DEM, land use type, and NDVI (Yang et al., 2024). For the Fuzhou GSSP, after consulting with experts from relevant planning and management departments, we used land use type as the basis for the resistance surface. By considering the actual conditions, we ensured that the corridor simulations accurately reflected the situation on the ground.

Using the gravity model to quantify the G a b between GPAs, and combining the GPA’s d P C , Fuzhou GSSP constructed the first-level corridors. However, in the construction process, we found that although the gravity model’s calculation process was relatively cumbersome, it was consistent with the d P C of GPA, and it could be helpful for the rationality of the first-level corridors selected. However, particular circumstances require special treatment. For example, the water-based GPA 10, Min River, and its d P C value was high, but the gravitational value with other GPAs was deficient because its resistance value was high. Moreover, due to its relatively large span, in the process of generating corridors by GIS, the location of the source point after generalization could not be determined accurately. Therefore, although it was a crucial GPA, it did not account for an essential status in ecological corridors construction.

As in the research of Lu et al. (2025) and Yang et al. (2017b), the d P C of GPA17 was low, and the gravitational values between it and other GPAs were also minimal. However, it was a coastal wetland, a critical ecological protection area, and an essential ecological park. The connection between this GAP and urban areas was deemed as highly necessary, so it was connected to the first-level corridors.

In terms of network structure, we introduced the loop index α , the average connection index β , the network connection degree λ , and the cost ratio index C R to evaluate the simulated first-level network. Related studies have confirmed that this approach has been effective in optimizing network structure (Yang et al., 2024).

Basing on the previous research (Lu et al., 2025; Yang et al., 2024), we identified green strategic nodes of minimum-maximum resistance at the intersections of the minimum and maximum resistance corridors. Unlike the method of identifying the ridgeline of the resistance surface (Fu et al., 2022), the approach used in the Fuzhou GSSP identified fewer nodes. However, this method was more straightforward and better suited for large-scale ecological planning, such as city-level planning. During the identification process, we found that certain GPAs, such as GPA 6, 7, 8, 9, and 11, played a crucial role in the connectivity of GPAs (Figure 13). These GPAs acted as stepping stones and key building blocks for the green network, effectively linking urban green spaces with suburban green spaces, and can therefore be considered strategic GPAs.

This study focused solely on establishing the ecological structure of the GSS at the city level. However, to fully realize the potential of the green space system, more attention must be paid to its ecological and service functions (Cao et al., 2024). GSSP needs to integrate multiple functions with the local urban development context to ensure the rationality of the ecological structure and provide further guidance for urban GSSP.

The methodology adopted in the Fuzhou GSSP quantified relevant planning indicators, making the planning process more scientific and rational. The preliminary results of the green network in Fuzhou GSSP were largely consistent with the actual local conditions, providing functional guidance for future planning. However, in real-world planning projects, specific circumstances must still be considered, and adjustments should be made according to local conditions. As an ecological planning approach, this method can be applied to other planning areas such as Territorial Space Planning (Ministry of Natural Resources, 2025) and Nature Reserve System Planning, particularly in addressing landscape fragmentation. Secondly, this paper has not considered the needs of biodiversity conservation enough, and the response of different species to different environments varies. Therefore, subsequent studies can combine species distribution modeling and habitat suitability analysis to provide more precise protection for specific species in the region.