Study on the influence mechanism of land use on carbon storage under multiple scenarios: A case study of Wenzhou

2.3.1 Data sources

The basic data required for the SD model were obtained from the statistical yearbooks of Zhejiang Province and Wenzhou City. The data required for the PLUS model, the LUCC data, were the 30 m accuracy remote sensing data published by Wuhan University. The 30 m accuracy data provide detailed LUCC information with high accuracy, which can ensure the detailed capture of the complexity of LUCC changes and better reflect the spatial heterogeneity of land use, and also solve the problem of excessive data computation, which is suitable for the prediction of land use and CS [55,56].The prediction drivers included 8 environmental factors and 13 economic and social factors, and the limiting factors were the open water and ecological function protection areas (Table 1), and the above factors are the main reasons influencing the LUCC changes. Meteorological data were obtained from the National Meteorological Science Data Center (http://data.cma.cn/), and the rest of the data were obtained from the Resource and Environmental Science and Data Center of the Chinese Academy of Sciences (www.resdc.cn) and Geospatial Data Cloud (www.gscloud.cn) (Table 1).

2.3.2 Data processing

The data processing flow is as follows:

1. The raw vector data of LUCC were extracted and cropped by mask by ArcGIS 10.8 software, and then classified into six land use types, namely, cultivated land, woodland, grassland, water, construction land, and unused land by applying the reclassification tool.

2. The digital elevation model data were processed by ArcGIS 10.8 software to obtain slope and direction data.

3. Distances to waters, public service points, and roads were calculated using Euclidean distances in ArcGIS 10.8 software.

4. Combined with the meteorological station data and statistics of Wenzhou’s average annual temperature and average annual precipitation from 2005 to 2020, the raster data of Wenzhou at 30 km resolution were generated by spatial interpolation with ArcGIS10.8 software.

5. The raster data of NDVI, GDP, population, night lights, and limiting factors were all processed by ArcGIS10.8 software, and after the unified coordinate system (CGCS2000_3_Degree_GK_CM_108E) was carried out, the raster data were processed by resampling and cropping. At the same time, it was ensured that the number of rows and columns of the 2005 and 2020 LUCC data was the same, and the projected coordinates of all raster data were identical.

6. The SD model uses the underlying statistical yearbook data, which was standardized mainly through SPSS.26 Descriptive Statistics.

2.4.1 Scenarios of future

Based on China’s “Dual-Carbon” goal and the goal of territorial spatial ecological governance, Wenzhou City serves as a pioneer area for low-carbon ecology, high-quality development, and ecological security governance of territorial space. We assume that Wenzhou will follow the path of low-carbon, ecological and high-quality sustainable development in the future. At the same time, the impacts of relative and absolute conflicts between future economic and social development and ecological low-carbon on LUCC and CS were considered. Four scenarios are set up:

1. CS under the NDS: According to the development trend of Wenzhou City from 2005 to 2020, it is sufficient to set up a natural development scenario and make basic restrictions on the transfer of land, without considering the restrictions on the change in LUCC by external policies.

2. HUS: Under the high speed of urban development, the high economic growth has led to extreme encroachment of urban construction land and extremely high occupancy of other land [57]. LUCC changes in the urbanization process are mainly reflected in a 40% increase in the probability of transferring cultivated land, woodland, grassland, and water to construction land, and a 100% increase in the probability of transferring unused land to construction land. This shift not only increases the area of construction land in cities, but also reflects the high intensity demand for land resources in cities [58]. Therefore, the probability of shifting cultivated land, woodland, grassland, water, and unused land to construction land is higher in this scenario.

3. LHDS: Construction land is the main factor leading to high carbon emissions [59]. In order to improve the low-carbon and ecological nature of economic development, alleviate the absolute conflict between economic development and ecological low-carbon, Wenzhou needs to take the sustainable development of the green low-carbon circular economy model to improve the CS comprehensively. Therefore, the transfer of woodland and grassland to construction land should be restricted. At the same time, the unused land should be transferred to land categories with higher CS such as woodland, grassland, and cultivated land [60].

4. ESGS: In order to ensure urban ecological quality and ecological security, human interference in ecologically sensitive areas, ecologically fragile areas, and prohibited development zones should be minimized to create conditions for ecological governance and protection of national land space [61]. In this scenario, only land categories with low CS are allowed to be converted to land categories with high CS, and the occupation of construction land and cultivated land to forest land is restricted, and the protection of land categories with high CS, such as woodland, water, and grassland, is strengthened [62]. The LUCC prediction transfer parameters for the above four scenarios are given in Table 2.

2.4.2 SD models

System dynamics treat complex systems as systems with multiple information causal feedback mechanisms, which is able to deal with nonlinear and time-varying problems and perform quantitative simulations [63]. Meanwhile, the SD model can reveal the complex relationship among LUCC, economic development, and ecological protection. And by simulating different development modes, it can effectively assess which development mode is most conducive to the long-term sustainability of the city [64,65]. Therefore, the SD model has been widely used in the simulation of urban composite systems and LUCC changes [66].

This study constructs a base model based on the linear regression relationship between the economic and social subsystem, the natural resource subsystem and the LUCC subsystem (Figure 3). The economic and social subsystem has the greatest impact on LUCC, and economic and social development leads to an increase in investment in all types of social fixed assets and materials for survival, accelerating the expansion of cultivated land and construction land, and increasing the cost of ecological and environmental management. The population subsystem reflects the changes in urban and rural population, and the increase in population and GDP growth inevitably leads to a further increase in construction land, thus indirectly affecting the spatial pattern and occupancy of LUCC. The natural resources subsystem mainly affects the water use and temperature of cultivated land, woodland, grassland, and water. The LUCC subsystem reflects the spatial pattern and transfer of LUCC under different development scenarios.

Figure 3

System dynamics model.

Based on the LUCC data, economic and social data, and natural resource data of Wenzhou City from 2005 to 2020, the interactions between the subsystems and variables were simulated. After several calibrations, the parameters between the variables were determined and the model was constructed (Figure 3). The simulation period of the SD model is 2005–2035 with a time step of 1 year. The historical simulation phase is 2005–2020, and the real data in 2020 is used to evaluate the accuracy of the simulation model. The scenarios are simulated with a base year of 2020, and the parameters and policy constraints of different scenarios are input to simulate the four scenarios of LUCC demand in Wenzhou City in 2035.

2.4.3 PLUS models and model validation

The PLUS model is a patch-generated LUCC change simulation model, which contains two modules, the Land Expansion Analysis Strategy (LEAS) model and the multi-type stochastic seed-based (CARS) [15,26]. In this study, we extracted the land expansion data of Wenzhou City from 2005 to 2020 using the LEAS module, and analyzed the impact of drivers on land expansion through the Random Forest algorithm to identify and quantify the key drivers affecting the change in LUCC. The CARS module utilizes the analysis results of the LEAS module to further simulate the LUCC. Parameters such as domain weights are to be set in this module, and domain weights are set according to the proportion of the expansion area of different land types of the total area. The parameters are as follows: (1) LEAS: the value of the decision tree is 20, the sampling rate is 0.01, the mTny is 14, and the number of parallel threads is set to 1. (2) CARS: the neighborhood range is set to 3, the Thread is 6, the diminishing threshold coefficient is 0.5, the diffusion coefficient is 0.1, and the seed probability of the random patches is 0.0001.

Based on the 2005 and 2020 LUCC data (Figure 4), 23 impact factors were selected (Figure 5), including elevation, slope, mean annual temperature, mean annual precipitation, land type, mean annual humidity, NDVI, distance to off-river rivers, distance to important transportation routes, GDP, nighttime lighting, population density, distance to settlements, distance to points of public services, distance to governmental sites, and Ecological Reserve data. Among them, surface waters and ecological protection zones are the limiting factors. Surface waters are mainly composed of rivers, lakes, and coastal waters, which block the expansion and encroachment of urban land use with their natural barrier characteristics, while ecological protection zones are composed of river origins, water conservation areas and animal and plant protection zones, which limit the expansion of the city as an important ecological protection site. Using LUCC data in 2005 as a simulation reference map, the PLUS model was run to obtain the LUCC simulation results in 2020. The accuracy of the PLUS model was evaluated by comparing the simulation results with the real LUCC data in 2020. If the accuracy of the results is high, the LUCC demand data predicted by the SD model for the four scenarios in Wenzhou City in 2035 are input into the PLUS model. Based on the LUCC data in 2020, the changes in the spatial and temporal pattern of LUCC in Wenzhou City in 2035 were simulated.

Figure 4

LUCC in (a) 2005 and (b) 2020.

Figure 5

Driver and limiting factors.

2.4.4 InVEST models

In this study, the CS module of the InVEST model was utilized to assess the CS of terrestrial ecosystems in Wenzhou City in 2035, based on LUCC data and carbon density data. Among them, the carbon module divides the ecosystem CS into four basic carbon pools: above-ground biogenic carbon, below-ground biogenic carbon, soil carbon, and dead organic carbon [67]. The calculation formula is as follows:

where i is the number of LUCC types; C sum is the sum of ecosystem CS; C i − above , C i − below , C i − soil , C i − dead are the carbon densities in above-ground, below-ground, soil, and dead organic matter in the ith type of LUCCs, respectively, and A i is the area corresponding to the ith type of LUCC.

Carbon density is an important input parameter for the accurate assessment of CS by the InVEST model, which varies in different climatic zones [68]. In this study, the carbon density was corrected with reference to the carbon density of Jiangsu and Zhejiang regions and the data from previous carbon density studies [69,70]. The carbon density of dead organic matter for each LUCC type was estimated by referring to the IPCC 2006 national greenhouse gas emission inventory [71]. In this study, the carbon density correction method proposed by Alam et al. [72] and Shui et al. [73] was used for the correction, and the correction formula was as follows:

where C SP and C BP denote soil carbon density and biomass carbon density calculated by mean annual rainfall, respectively; C BT represents biomass carbon density calculated by mean annual temperature; and MAP and MAT represent mean annual rainfall and mean annual temperature, respectively. The corrected carbon density of each LUCC type in the study area is shown in Table 3.

3.1.1 Characterization and spatial mechanisms of LUCC from 2005 to 2020

From 2005 to 2020, the areas of woodland, grassland, and water in Wenzhou City decreased by 407.55, 2.02, and 55.66 km², respectively (Table 4). And the percentage also showed a decreasing trend, with woodland decreasing from 75.61 to 72.06%, grassland decreasing from 0.03 to 0.01%, and water decreasing from 2.28 to 1.79%. Among all LUCC types, Woodland had the largest decrease in area and constructed land had the largest increase in area. Cultivated land, unused land, and constructed land increased by 169.78, 0.065, and 295.38 km², respectively, all of which showed an upward trend, with cultivated land increasing from 17.39 to 18.87% and constructed land increasing from 4.69 to 7.27%. The land that has not been transformed accounts for 91.6% of the total land, the amount of shifting of cultivated land accounts for 4.6%, the amount of shifting of construction land accounts for 2.7%, the amount of shifting of woodland accounts for 1%, and the amount of shifting of the rest of the types of land is small accounting for about 0.01% (Table 4 and Figure 6). The results of the analysis are as follows: (1) Compared with the west, the terrain of the east coast is flat, with good conditions for cultivation and construction, and the expansion of cultivated land and construction land is more obvious. (2) The high speed of economic development and the rapid growth of urban and rural populations have increased the demand for food, leading to the occupation of large areas of ecological land (woodland, grassland, and water) in the west and north by cultivated land and construction land. (3) The rise of a large number of private enterprises in townships and counties has led to the rapid economic and social development of townships and increased demand for construction land, accelerating the expansion of construction land in the ecological area in the western and northern regions in the form of a point distribution (Figure 6).

Figure 6

Distribution of LUCC space transfers from 2005 to 2020. (a) Year 2005. (b) Year 2020. (c) 2005–2020 land transfer.

3.1.2 Characterization and spatial mechanisms of CS from 2005 to 2020

Using the InVEST model to assess the CS in Wenzhou City in 2005 and 2020, the CS was 2.28 × 10⁵ kt and 2.22 × 10⁵ kt, respectively (Table 5), with a total decrease of 6.12 × 10³ kt, and the CS of woodland decreased by 8.76 × 10³ kt, and there was a decreasing trend in the CS from 2005 to 2020, with the total amount decreasing by 2.4%. In terms of LUCC types, the CS in descending order, is woodland, cultivated land, construction land, water, grassland, and unused land. From 2005 to 2020, the transfer and dynamics of CS of each land type (Table 6 and Figure 6), CS conversions of cultivated land, construction land, and woodland were 2.032 × 10⁴ kt, 3.39 × 10³ kt, and 1.20 × 10³ kt, respectively, which are the main factors affecting the change in CS in Wenzhou from 2005 to 2020, and the CS transfer of the remaining land use is relatively small and less influential.

The analysis of four typical areas in Wenzhou City from 2005 to 2020 shows (Figure 6) that the areas with significant CS reduction are mainly areas A and D, where woodland and grassland are reduced in a large area, mostly distributed in point form in ecological service areas, with a wide range and a high density. The fringe areas of the main urban areas of B and C, where the CS is reduced, are distributed in piecemeal form, with a large area, and are mostly areas of woodland and cultivated land. The part of CS increasing is mainly the new cultivated land and construction land along the east coast and river, and the space of CS increasing is distributed in a narrow band with a small area. The spatial change in CS is consistent with LUCC in Wenzhou from 2005 to 2020, which is influenced by economic and social development (Figure 7). In the past 10 years, ecological CS technology has mainly focused on the exploration of loamy carbon sequestration, carbon capture, utilization, and storage (CCUS) technology, and ecological spatial restoration is considered to be the best natural solution with low cost and high CS. Therefore, in the future, Wenzhou should increase the investment in ecological and environmental management and restoration of land space to solve the problem of declining CS in Wenzhou at the level of ecological CCUS [74].

Figure 7

Dynamics of CS from 2005 to 2020. (a) 2005 CS, (b) 2020 CS, (c) 2020 CS intensity, and (d) 2005–2020 carbon transfer region.

3.3.1 Change analysis of LUCC in multi-scenario from 2020 to 2035

Based on the PLUS model for the 2005–2020 LUCC simulation, it can be seen that from the accuracy calibration process, the PLUS model has higher sensitivity to the accuracy and quality of the driving vector data, and the use of 500 and 1,000 m low-precision driving factors will directly reduce the accuracy of the LEAS module and increase the LUCC simulation error. In this study, the data used are mostly 30 m, and the overall accuracy of PLUS model is 91.86%, and the kappa coefficient is 0.91, which is high, indicating that the model is effective for LUCC simulation results. Based on the LUCC data in 2020, the LUCC demand simulated by SD under the four scenarios is substituted in the PLUS model, and the simulation results are as follows (Figures 9 and 10). The results show that cultivated land and construction land account for about 30% of the total land under NDS and HUS, and the transfer ratio of the two is 43 and 53%, respectively. Woodland accounted for 67–69% of the total land, and the changes in woodland, grassland, and water were all less than 3%, and most of the land shifted to construction land and cultivated land. Because of the absolute priority of economic development, economic and social development and low-carbon ecology are in absolute conflict, which is not conducive to the realization of the goal of “Dual-Carbon” and territorial spatial ecological governance. Under the LHDS and ESGS, woodland accounts for 70–73% of the total land, and from the perspective of land transfer, the transfer ratio is 46%, while cultivated land and construction land account for about 27% of the total land, and the transfer ratio is 24 and 28%, respectively. As a result of advocating low-carbon and high-quality development, ecological protection and ecological governance of land space, and focusing on the protection of ecological land, the area of this type of land transferred to cultivated land and construction land is relatively small, and the trend of returning and expanding the area of woodland and grassland is obvious, and the conflict of interests between economic development and ecological and low-carbon is relatively small, mostly relative, which is conducive to low-carbon and high-quality development.

Figure 9

Multi-scenario land modeling, land transfer space, and LUCC ratio in 2035. (A) NDS 2035 Land prediction; (a) NDS 2020–2035 transfer region; (B) HUS 2035 Land prediction; (b) HUS 2020–2035 transfer region; (C) LHDS 2035 Land prediction; (c) LHDS 2020–2035 transfer region; (D) ESGS 2035 Land prediction; (d) ESGS 2020–2035 transfer region.

Figure 10

Transfer of different land types. (a) 2020–2035 (NDS). (b) 2020–2035 (HUS). (c) 2020–2035 (LHDS). (d) 2020–2035 (ESGS).

3.3.2 Analysis of spatial mechanism of LUCC change 2020–2035

From the analysis of the four typical change regions in (Figures 9 and 11), it can be seen that the expansion of LUCC space under different scenarios is highly differentiated. Under NDS and HUS, the rapid expansion of construction land leads to a significant reduction in cultivated land in regions B and C. The expansion of construction land in regions A and D occupies a large amount of ecological space, such as woodland and grassland, with a significant reduction in woodland, and the construction land is widely distributed in a large point-like manner within the ecological zone. Under HUS, the expansion of construction land within the ecological zone seriously breaks the ecological protection space. The expansion of construction land and cultivated land under the LHDS is similar to that of the ESGS, and the space of woodland and grassland in the ecological space of regions A and D is better preserved and not extensively destroyed. The land transfer in regions B and D is dominated by woodland, construction land, and cultivated land, which is mainly transferred to the space of construction land, with the overall expansion of a smaller magnitude.

Figure 11

Transfer intensity of LUCC in multi-scenario. (a) NDS 2020–2035 transfer intensity. (b) HUS 2020–2035 transfer intensity. (c) LHDS 2020–2035 transfer intensity. (d) ESGS 2020–2035 transfer intensity.

3.4.1 Dynamic analysis of CS from 2020 to 2035

The dynamic simulation of CS was unfolded based on the SD-PLUS-InVEST model, which has a longer development time, greater data demand, higher sensitivity to the accuracy of the LUCC data, and more complex requirements than the static model, but the dynamic model has high accuracy and many prediction scenarios, which makes it more suitable for predicting spatial changes in CS over long time periods [75]. The results of the four scenarios simulated by SD-PLUS were substituted in the InVEST model to assess the changes in CS in Wenzhou City under different scenarios from 2020 to 2035 (Table 9 and Figure 12). The variability of CS under different scenarios is significant (Figure 10). In 2035, the CS under NDS, HUS, LHDS, and ESGS is 2.191 × 10⁵ kt, 2.142 × 10⁵ kt, 2.231 × 10⁵ kt, and 2.226 × 10⁵ kt, with an increment of −1.676, −3.692, 0.189, and −0.149%, respectively. Under the four scenarios, the CS in Wenzhou City from 2020 to 2035 shows different degrees of changes, and the CS of NDS, HUS, and ESGS decreases by 3.730 × 10³ kt, 8.214 × 10³ kt, and 33.09 kt, respectively, and the CS under LHDS increases by 42.05 kt. The largest changes in CS of land use types under the four scenarios were in woodland and cultivated land, with CS of woodland decreasing by 6.3 × 10³ kt and 1.338 × 10⁴ kt under NDS and HUS, respectively, and CS of cultivated land increasing by 1.78 × 10³ kt and 3.91 × 10³ kt, respectively; CS of woodland under NDS and HUS was 325 kt and 20 kt, respectively, and CS of cultivated land under NDS increased by 8.10 × 10² kt, cultivated land CS decreased by 1.32 × 10³ kt under HUS. CS of construction land under NDS, HUS, and ESGS showed an upward trend with an increase of 6.80 × 10³ kt, 1.58 × 10³ kt, and 9.90 × 10³ kt, respectively, and a decrease of 7.2 × 10² kt under LHDS, while the remaining land use types showed smaller changes in the total CS withdrawal. CS in Wenzhou City tends to have an overall decreasing trend from 2020 to 2035, with the largest decreases under NDS and HUS, the small decrease of CS under ESGS and the small increase under LHDS all benefit from the constraints of the policies of ecological protection, low-carbon development and territorial spatial ecological governance, which emphasize the strict protection of woodland, cultivated land, grassland, and water in development, and the effective coordination of the absolute and relative conflicts between the economic and social development and the low-carbon ecological interests.

Figure 12

Intensity of change in CS and heat map of transfers. (A) NDS 2035 carbon transfer intensity; (a) NDS 2035 carbon transfer. (B) HUS 2035 carbon transfer intensity; (b) HUS 2035 carbon transfer. (C) LHDS 2035 carbon transfer intensity; (c) LHDS 2035 carbon transfer. (D) ESGS 2035 carbon transfer intensity; (d) ESGS 2035 carbon transfer.

3.4.2 Spatial analysis of CS from 2020 to 2035

The spatial distribution of CS in 2035 is shown in Figure 13. Under NDS and HUS, the areas with decreasing woodland and decreasing CS are widely distributed in a sporadic point-like manner in regions A and D. The reason for this is that the urbanization rate in the western and northern woodland areas has increased, and the construction land for economic development and township settlements has increased, leading to the encroachment of a large amount of woodland and grassland space, and the encroachment of the economic and social development space on the space for low-carbon ecological services is in an absolute conflict, which results in the space with higher CS becoming smaller.

Figure 13

Spatial distribution and dynamic transfer of CS. (A) NDS 2035 CS; (B) HUS 2035 CS; (C) LHDS 2035 CS; (D) ESGS 2035 CS. (a) NDS 2020–2035 carbon transfer region; (b) HUS 2020–2035 carbon transfer region; (c) LHDS 2020–2035 carbon transfer region; (d) ESGS 2020–2035 carbon transfer region.

The changes in CS under LHDS and ESGS are similar, with woodland significantly higher than cultivated land. Under LHDS, the area of CS increase is significantly larger than the area of CS decrease, and the areas of CS decrease are mainly located in the areas of coastal cultivated land and construction land expansion in B and D. The construction land and cultivated land in the ecological reserves in A and D under LHDS and ESGS are significantly reduced compared with those in NDS and HUS. At the same time, the occupation of space for economic and social development and space for low-carbon ecological services is in a balanced state, with mostly relatively conflicting and non-conflicting interests, and there is a significant increase in the space with higher CS in the region.