Dynamic impacts of urbanization development on carbon storage and NPP: spatiotemporal responses in the Wanjiang urban belt (2000–2020)

Introduction: Urban agglomerations, as pivotal spatial units in global urbanization processes, are central to urban science, their land use/cover changes directly regulate regional carbon cycles and profoundly affect ecosystem carbon storage and net primary productivity (NPP). To clarify the dynamic changes, driving factors, and future trends of these two key ecological indicators in Wanjiang Urban Agglomeration, multiple models and analytical methods were integrated.

Methods: The Carbon Storage module of the InVEST model was used to assess regional carbon stocks, while the improved CASA model quantified multi-period NPP using remote sensing and climate data. A random forest algorithm combined with the SHAP method analyzed the independent effects of topographic, anthropogenic, and natural factors, and the PLUS model combined with Markov Chain forecast 2040 carbon storage.

Result: Results show urban area expanded by 2,230.87 km² (3.57% annual growth), primarily from cultivated land (4.52 × 10 km² to 4.26 × 10 km²). Annual carbon storage loss reached 0.25 × 10 tC and NPP loss 17.49 × 10⁴ tC, with both first rising then falling. Carbon storage is predominantly driven by topographic factors (with thresholds) and secondarily by natural factors, while NPP is jointly regulated by topographic and climatic drivers. Anthropogenic factors exert the weakest influence on both. The 2040 forecast indicates continued carbon storage decline.

Discussion: This work clarifies key dynamic patterns and driving mechanisms, providing scientific support for regional ecological protection and advancement of “carbon peaking and carbon neutrality” goals.

1 Introduction

Land use change ranks as the second-largest driver of the dramatic global rise in atmospheric CO₂ concentrations, surpassed only by fossil fuel combustion (Quay et al., 1992). Beyond its role in greenhouse gas emissions, it also acts as a key force shaping variations in Earth system carbon storage (Zhang et al., 2015). Variations in carbon sequestration capacity across land use types drive changes in overall carbon storage (Cantarello et al., 2011; Ni, 2013). These carbon dynamics, in turn, amplify pressing environmental crises such as global warming, desertification, and sea level rise (Hauer et al., 2020). Urbanization, as a dominant form of land use change in contemporary societies, further intensifies these carbon-related challenges. Cities account for ∼78% of global anthropogenic greenhouse gas emissions (Li et al., 2019; Huang et al., 2019), requiring balanced urban development and carbon reduction—particularly amid China’s urbanization. Projections show urban areas will cover 67% of Earth’s land by 2050 (11% rise from 2020, Shi et al., 2023). Urbanization converts high-carbon landscapes (croplands, forests, grasslands) to low-carbon built-up areas (Grimm et al., 2008; De Carvalho and Szlafsztein, 2019; Wang et al., 2022), reducing terrestrial carbon storage and shifting NPP (net primary productivity, a key carbon sequestration indicator). Quantifying these spatiotemporal impacts is critical for low-carbon urban planning (Hutyra et al., 2011).

Urbanization-driven carbon dynamics leave terrestrial ecosystems’ global carbon cycle role uncertain (Zhao and Running, 2010; Potter et al., 2012; Ahlström et al., 2012; Zhuang et al., 2023), largely from urbanization, climate change, and deforestation. As a key carbon sequestration indicator (Roxburgh et al., 2005), NPP (organic carbon accumulation post-plant respiration) is hard to estimate via time-consuming traditional methods (Sonti et al., 2022; Huo et al., 2023) or sparse eddy covariance flux tower data (Baldocchi, 2003). The widely used CASA (Carnegie–Ames–Stanford Approach) model accurately captures spatiotemporal NPP variations, suiting urbanization-related NPP response analysis (Tang et al., 2024a; Liang et al., 2015; Potter et al., 2012). Empirical evidence from China underscores the link between urbanization and NPP losses: 13.77 Tg C (1982–2015, He et al., 2017), 11.60 × 10⁻³ Pg C (2000–2010, Wen et al., 2019), and 0.137 Tg C in the Pearl River Delta (2000–2010, Jiang et al., 2015). Urban vegetation and soil drive carbon sequestration, but urban expansion disturbs these processes, necessitating impact quantification across urbanizing contexts.

Carbon storage reflects terrestrial ecosystems to remove carbon dioxide from the atmosphere (Chen et al., 2021). The InVEST (Integrated Valuation of Ecosystem Services and Trade offs) model’s carbon module estimates regional carbon storage via four carbon pools (Wu and Wang, 2023; Tang et al., 2024b). China’s urban expansion has accelerated carbon storage loss since 1980 (108.99 Tg C by 2010, 87.20 Tg C in 2000–2010), primarily due to the conversion of cropland and forest to urban land (Liu et al., 2019). Three major urban agglomerations—the Beijing–Tianjin–Hebei region (Kang et al., 2025; He et al., 2016; Xie et al., 2018), the Yangtze River Delta (Wang et al., 2025a; Wu et al., 2024; Wang et al., 2023), and the Pearl River Delta (Cai et al., 2024; Yan et al., 2018; Li et al., 2022) — have been identified as carbon loss hotspots. However, research on the Wanjiang City Belt (WJUA) — a key Yangtze River Economic Belt component (Gui et al., 2022) — is scarce. Accelerated urban sprawl and industrial expansion in the WJUA have disrupts its carbon cycle, yet existing studies lack independent driver analysis (Wang et al., 2025b), hindering understanding of its urbanization-driven carbon dynamics.

To address these limitations, this study systematically couples three robust modeling approaches: the InVEST model (for detailed carbon storage assessment across multiple ecosystem pools), the CASA model (for high-accuracy, large-scale estimation of NPP and carbon sequestration, based on photosynthetically active radiation and light-use efficiency (Zhang et al., 2024)), and the PLUS–Markov Chain model (for projecting future land use changes (Wu et al., 2024)). Using these tools, we analyze the spatiotemporal variations in ecosystem carbon storage and NPP in the WJUA from 2000 to 2020 — directly addressing the need to quantify spatiotemporal responses to urbanization development. We further apply the Random Forest algorithm combined with the SHAP (SHapley Additive exPlanations) method to quantitatively evaluate the relative influence and contribution priorities of socioeconomic, natural, and topographic factors on the evolution of carbon storage and NPP. Additionally, we simulate and predict ecosystem carbon storage in the WJUA under a natural development scenario for 2040. The findings of this study are expected to provide a scientific basis for optimizing ecological planning and supporting the implementation of China’s “dual-carbon” strategy in the middle and lower reaches of the Yangtze River Basin.

2 Materials and methods

2.1 Study area

WJUA located in Anhui Province within the middle and lower reaches of the Yangtze River Basin, serves as the core demonstration zone for industrial transfer within the Yangtze River Delta (YRD) region. The WJUA comprises eight prefecture-level cities—Hefei, Wuhu, Ma’anshan, Tongling, Anqing, Chizhou, Chuzhou, and Xuancheng—along with Jin’an District and Shucheng County of Lu’an City (Figure 1). The total area covers approximately 75,877.49 km². Geographically adjacent to the YRD, China’s most economically dynamic urban agglomeration, the WJUA benefits from strong economic and industrial linkages with the region. The area is characterized by a subtropical monsoon climate, with an annual mean temperature of 15 °C–17 °C and average precipitation ranging from 800 mm to 1,600 mm. Riverine plains and low mountains dominate its landscape, providing favorable natural and locational conditions for both agricultural productivity and industrial development. Economically, the WJUA functions as a key growth pole of Anhui Province, contributing over 60% of the provincial GDP in 2020 and attracting more than 6.2 trillion CNY in large-scale investments. With its rapid industrial clustering, urban expansion, and land use transformation, the Wanjiang Urban Agglomeration represents a typical and representative region for examining the interactive mechanisms between urbanization and carbon dynamics (Wang et al., 2025b).

Figure 1

Figure 1. Study area. (a) China (b) DEM of WJUA.

2.2 Data and methods

This study utilized five phases of land use data for the years 2000, 2005, 2010, 2015, and 2020, which were obtained from the National Geomatics Center of China (http://www.globallandcover.com). Three categories of driving factors influencing carbon storage and NPP were analyzed, namely, topographic (Digital Elevation Model [DEM], slope), anthropogenic (Gross Domestic Product [GDP], population density, main/secondary/tertiary roads, highways, railways), and natural (rivers, annual mean temperature, annual mean precipitation). GDP and population density data were sourced from the Data Center for Resources and Environmental Sciences, Chinese Academy of Sciences (https://www.resdc.cn). To ensure consistency with road and railway datasets, all driving factors were constrained to the years 2015 and 2020. Information on roads (motorways, primary roads, secondary roads, and tertiary roads), railways, and rivers was derived from OpenStreetMap (OSM; https://www.openstreetmap.org). DEM data (acquired in 2000) was obtained from the Geo-spatial Data Cloud (https://www.gscloud.cn), with slope data derived from the DEM. Data used in the InVEST and CASA models are presented in Table 1. All datasets were standardized to the Albers Conic Equal Area coordinate system and resampled to a 30 m spatial resolution to ensure consistency in column numbers and pixel size.

Table 1

Table 1. Data used in the InVEST and CASA models. (Meteorological data include monthly mean temperature, monthly total solar radiation and monthly total precipitation).

The NPP dataset utilized in this study was generated using an improved CASA model. Conceptually, the CASA model calculates NPP as the product of vegetation’s Absorbed Photosynthetically Active Radiation (APAR) and light use efficiency (ε)—a foundational framework established in early seminal work (Potter et al., 1993; Field et al., 1995). The core formula governing this calculation is expressed as follows:

$NPP\left(x,t\right)=APAR\left(x,t\right)\times \mathrm{\epsilon }\left(\mathrm{x},\mathrm{t}\right)$

Where: NPP (x,t) represents the net primary productivity at spatial location x and time t. APAR (x,t) (i.e., photosynthetically active radiation absorbed by vegetation at x and t) is jointly determined by total solar radiation, the Normalized Difference Vegetation Index (NDVI), and vegetation absorption coefficients—key variables that capture the fraction of solar radiation available for photosynthesis and its uptake by vegetation. ε(x,t) (i.e., light use efficiency at x and t) integrates the effects of three critical modulating factors: temperature stress coefficient, water stress coefficient, and maximum vegetation light use efficiency (Wen-Quan et al., 2007). Notably, the specific improvements made to the CASA model and the detailed validation results of the modified framework have been thoroughly documented in our previous study (Tang et al., 2024a).

The carbon storage dataset was constructed using the Carbon Storage and Sequestration module of the InVEST model (Version 3.9.0), which assesses regional carbon storage capacity by quantifying the storage of ecosystem carbon pools (Hamel et al., 2015). Within the carbon storage module, total carbon storage is calculated as the sum of four carbon pools (Table 2). The specific calculation methods for each pool are as follows: Vegetation carbon storage is the product of the area of each land use type and its corresponding vegetation carbon density; Soil carbon storage is computed by multiplying the organic carbon density of the 0–100 cm soil layer by the area of the respective land use type; Litter carbon and deadwood carbon are estimated based on proportional coefficients relative to vegetation biomass (adopting 12% and 8%, respectively) (Tang et al., 2024b). Carbon density values were sourced from previous research (Sun et al., 2023). Drawing on relevant studies and references from analogous regions, a carbon density table tailored to land use types in this study was compiled and refined. Future land use patterns were projected using the Markov Chain method within the PLUS model. Operating at a 30 m × 30 m grid resolution, the PLUS model employs the Land Expansion Analysis Strategy (LEAS) to identify expansion rules for different land use types (e.g., the correlation between forestland expansion and slope gradient or distance to water sources). It also utilizes a multi-type random patch seeding mechanism to simulate the spatial distribution of newly added land, effectively addressing the limitation of “homogeneous expansion” inherent in traditional models (Liang et al., 2021). Model validation indicated that the Kappa coefficient between the simulated 2020 land use pattern and actual data was 0.82, meeting the required prediction accuracy standards.

Table 2

Table 2. Carbon density of different land use types (Unit of carbon storage: C/hm²).

To address inter-dependencies among predictor variables, the Random Forest algorithm was implemented on the MATLAB 2022 platform to evaluate the relative importance of 12 feature variables for both carbon storage and NPP—with higher importance values indicating a more pronounced influence on the target variables. Partial dependence plots derived from the Random Forest model were used to visualize response functions (Schwalm et al., 2017; Sun et al., 2016), a widely adopted approach in environmental modeling for interpreting complex relationships. Partial dependence analysis focuses on quantifying the marginal effect of a single covariate (or independent variable) on the response variable, while holding all other covariates constant at their respective values. This analytical framework is particularly valuable for extracting insights from large datasets, especially when Random Forest models are dominated by low-order interactions (Lu et al., 2025). Additionally, it serves as a robust tool for identifying key features, detecting non-linear relationships between predictors and responses, and gaining a mechanistic understanding of the overall behavior of the predictive model—all of which enhance the interpretability of how driving factors shape carbon storage and NPP dynamics in the study area.

3 Results

3.1 Land use changes in the WJUA from 2000 to 2020

The spatial distribution of land use types across the WJUA from 2000 to 2020 is illustrated in Figure 2. Cropland and forest dominated the regional landscape, with cropland consistently accounting for over 56% of the total area, primarily concentrated north of the Yangtze River and in the northern portion of Xuancheng. Forest covers approximately 30% of the study area, with primary distributions south of the Yangtze River and in western Anqing. In contrast, impervious surfaces, water bodies, grassland, and barren land occupy relatively small areas, together accounting for less than 14% of the total area.

Figure 2

Figure 2. Land-cover of WJUA from 2000 to 2020. (a) 2000 (b) 2010 (c) 2020.

Notable temporal shifts in land use were observed between 2000 and 2020, with the most pronounced changes occurring in cropland and built-up areas. Cropland area decreased by 5.79% over the 20-year period, with 5.19% of this reduction occurring in the first decade (2000–2010). Conversely, urban built-up areas expanded by a factor of 1.72, with the most substantial growth centered in Hefei, followed by Wuhu, Ma’anshan, and Chuzhou. Overall, built-up land increased by 71.50% across the entire study period, with a 31.18% expansion in the first decade, indicating a marked acceleration in urbanization dynamics during the latter decade (2010–2020).

Tables 3,4 present the land use transition matrices for the WJUA during the periods 2000–2010 and 2010–2020, respectively. While the Sankey diagram in Figure 3 visually illustrates the conversion relationships among different land use types across the full 20-year study period.

Table 3

Table 3. Land use transfer matrix from 2000 to 2010. (The row data are from 2000, and the columns are for 2010. Unit: km²).

Table 4

Table 4. Land use transfer matrix from 2010 to 2020. (The row data are from 2010, and the columns are for 2020. Unit: km²).

Figure 3

Figure 3. Sankey Diagram of land use transfer.

During the first decade (2000–2010), cropland underwent the most substantial changes, decreasing by 2,344.83 km², with a total outflow of 3,484.14 km², accounting for 72.53% of the total outflow area. As shown in Figure 3, the majority of this converted cropland was redistributed to three land use types: impervious surfaces, forest, and water. Benefiting from the Grain for Green project, forest area increased by 1,141.34 km², primarily converted from cropland. Grassland experienced a slight decline. Water area increased by 237.82 km², mainly due to recent farmland-to-wetland initiatives and the South-to-North Water Diversion projects, with 92.88% of the increase originating from cropland. Barren land slightly increased, showing minimal change. Impervious surfaces expanded by 972.86 km², 89.25% of which originated from cropland.

During 2010–2020, the largest land outflows occurred from cropland and forest, whereas cropland and impervious surfaces experienced the greatest inflows. Forest decreased by 833.36 km², with 96.75% converted to cropland. Cropland decreased by 271.41 km², primarily converting to impervious surfaces and forest (Figure 3). Water decreased by 138.51 km², mainly flowing into cropland. Grassland and barren land experienced minor declines. Impervious surfaces increased by 1,258.02 km², 91.52% of which originated from cropland.

Overall, impervious surfaces expanded from 3,120.10 km² in 2000 to 5,350.97 km² in 2020, an increase of 2,230.87 km² (71.50%). Over the 2000–2020 period, cropland had the largest outflow, primarily converting to impervious surfaces and forest, followed by forest, of which 91.82% flowed into cropland. The largest land in-flows were observed in impervious surfaces, cropland, and forest, in that order.

3.2 Changes in carbon storage and NPP from 2000 to 2020

Figures 4a–e, Figures 5a–e illustrate the spatial distributions of carbon storage and NPP across five time points from 2000 to 2020, with lighter colors indicating higher carbon storage or NPP per unit area. High values of both carbon storage and NPP are primarily concentrated south of the Yangtze River and in the Dabie Mountain region of Anqing city, where the land surface is predominantly forested. Low values are mainly observed in water bodies, approaching zero. Urban areas exhibit slightly higher values than water, while cropland shows higher values than impervious surfaces but lower than forests. A notable pattern is observed in the northwest of the study area, particularly in Hefei—the provincial capital of Anhui—which has undergone continuous urban expansion during the study period, accompanied by a progressive decline in both carbon storage and NPP.

Figure 4

Figure 4. The carbon storage results from the InVEST model (Unit: t C/hm², CS: Abbreviation for Carbon Storage). (a) 2000 (b) 2005 (c) 2010 (d) 2015 (e) 2020 (f) Change of CS from 2000 to 2020.

Figure 5

Figure 5. The NPP results from an improved CASA model (Unit: g C/m²). (a) 2000 (b) 2005 (c) 2010 (d) 2015 (e) 2020 (f) Change of NPP from 2000 to 2020.

The spatial change distributions of carbon storage and NPP are shown in Figures 4f, 5f, where red shades indicate increases, blue shades indicate decreases, and blank areas represent no change. Figure 4f shows that areas with increased carbon storage over 2000–2020 are mainly located at the edges of forests and grasslands, accounting for 3.48% of the total WJUA area. Decreases in carbon storage are primarily observed around urban areas, particularly in Hefei and cities along the Yangtze River, accounting for 21.07% of the total area. Cropland decreased substantially, with most of the outflow converting to impervious surfaces. Given that cropland has a much higher carbon density than impervious surfaces, the conversion of high-carbon-density land to low-carbon-density land led to a reduction in carbon storage across the WJUA. Benefiting from ecological restoration projects in Anhui Province, forest area increased slightly. As forests have the highest carbon density among all land use types, the region south of the Yangtze River experienced the smallest decline in carbon storage. Overall, carbon storage in the WJUA exhibited a gradual decreasing trend from 2000 to 2020, with a total loss of 5.04 × 10⁴ t C, corresponding to a 0.93% reduction.

Figure 5f illustrates the NPP changes over the 20-year period. Areas with increased NPP are mainly located in cropland north of the Yangtze River, where 0.82% of the area experienced an NPP increase exceeding 200 g C/m², and 27.33% of the area had increases in the range of 0–100 g C/m². Decreases in NPP primarily occurred around urban areas and forested regions, with 29.03% of the area experiencing reductions greater than 100 g C/m². Over the 20 years, NPP in the WJUA decreased by a total of 3.498 × 10⁶ t C.

Based on the values of carbon storage and NPP (Figure 6), both indicators exhibited generally consistent trends from 2000 to 2020. In the early period, both increased over time, whereas in the later period, they showed varying degrees of decline. Specifically, from 2000 to 2005, carbon storage increased from 10.79 to 10.86 t C/ha, and NPP rose from 487.24 to 531.25 g C/m². After 2005, carbon storage remained relatively stable from 2005 to 2010 before declining rapidly to 10.78 t C/ha by 2020. NPP, in contrast, declined more sharply, reaching 441.14 g C/m² by 2020. The maximum value of carbon storage occurred in 2010, whereas NPP peaked in 2005. Overall, the rate of decline in NPP was greater than that of carbon storage.

Figure 6

Figure 6. Temporal trends of NPP and carbon storage in the WJUA from 2000 to 2020.

3.3 Comparative analysis of the driving factors of carbon storage and NPP

The effects of 12 drivers (topographic, anthropogenic and natural) on carbon storage and NPP are illustrated in Figures 7–10, respectively. The response functions of the top six drivers regulating carbon storage are presented in Figure 7: slope, DEM, temperature, precipitation, distance to highways, and distance to railways. Notably, topographic and natural factors exhibited distinct thresh-old-dependent effects, whereas anthropogenic factors exerted more uniform, lower-dynamism influences. Specifically, slope (Figure 7a) triggered a rapid increase in carbon storage at low values, stabilizing beyond a critical threshold—indicating negligible impact above a certain magnitude. DEM (Figure 7b) displayed an analogous trend: carbon storage rose with elevation before plateauing, as other factors assume greater regulatory importance at higher altitudes. Temperature (Figure 7c) supported carbon storage within a narrow optimal range but induced marked declines when exceeded. In contrast, precipitation (Figure 7d) sustained stability across a broad range with no obvious threshold. For anthropogenic factors, neither distance to highways (Figure 7e) nor railways (Figure 7f) caused significant carbon storage fluctuations, indicating low variability in infrastructure impacts. The remaining six drivers are omitted due to substantially weaker influences.

Figure 7

Figure 7. Response functions of carbon storage: Single-variable response of carbon storage to the top six driving factors with other variables held constant. (a) Slope (b) DEM (c) Temperature (d) Precipitation (e) Highway (f) Railroad.

To quantify driver importance, attribution analysis was performed via the SHAP method. Results aligned with Figure 7: slope and DEM emerged as dominant drivers of carbon storage changes, with absolute mean SHAP values exceeding 0.30 (Figure 8a), followed by natural factors, while anthropogenic factors exerted relatively weaker impacts. Meanwhile, SHAP beeswarm plots (Figure 8b) visualized the magnitude and patterns of all 12 drivers. Topographic and natural factors exhibited concentrated SHAP value distributions, serving as core drivers with non-unidirectional effects. Among anthropogenic factors, SHAP values for highways, railways, and roads were scattered—highways, in particular, showed negative outliers, suggesting potential local inhibitory effects. GDP exhibited a wide SHAP values, including high positive outliers, indicating heterogeneous economic development impacts. Population density and distance to rivers had narrow distributions, reflecting low variability in their influences.

Figure 8

Figure 8. Identifying the drivers of patterns of carbon storage. (a) The SHAP importance (averaged absolute SHAP values) for recovery time. (b)The summary plot of SHAP values in random forest machine learning.

For NPP, the top six response functions drivers are DEM, temperature, slope, precipitation, distance to rivers, and GDP (Figure 9). Topographic and natural factors exerted strong driving effects, with elevation and temperature showing distinct threshold, while anthropogenic factors had weak direct impacts. Specifically, NPP increased rapidly with rising DEM (Figure 9a) and stabilized at approximately 500 m elevation—beyond which elevation’s promotional effect weakened. NPP remained high at 13 °C–16 °C (Figure 9b) but declined significantly above 16 °C, reflecting an optimal temperature range with high temperatures inhibition. At low slope (Figure 9c), NPP fluctuated slightly before stabilizing, indicating weak, non-trending impacts. Precipitation (Figure 9d) influenced NPP evenly without significant thresholds, while NPP remained stable with variations in distance to rivers (Figure 9e), suggesting minimal direct river impacts. GDP’s (Figure 9f) inhibitory effect was confined to low-GDP ranges, stabilizing above a certain level. The remaining six drivers are omitted due to weaker NPP impacts.

Figure 9

Figure 9. Response functions of NPP: Single-variable response of NPP to the top six driving factors with other variables held constant. (a) DEM (b) Temperature (c) Slope (d) Precipitation (e) River (f) GDP.

For feature importance (Figure 10a), consistent with carbon storage, topographic and natural factors were core NPP drivers: DEM, temperature, and slope had significantly higher SHAP values than other factors, followed by precipitation. Anthropogenic infrastructure factors had the lowest importance and weak direct impacts. SHAP beeswarm plots (Figure 10b) revealed concentrated, bipolar SHAP value distributions for DEM, slope, temperature, and precipitation—reflecting their bidirectional regulatory effects on NPP. Distance to rivers showed low impact variability. While GDP and population density had moderate SHAP value ranges but limited overall contributions. Anthropogenic infrastructure factors displayed low SHAP value dispersion, further confirming NPP impacts.

Figure 10

Figure 10. Identifying the drivers of patterns of NPP. (a) The SHAP importance (averaged absolute SHAP values) for recovery time. (b) The summary plot of SHAP values in random forest machine learning.

3.4 Projected land use and carbon storage in 2040

All Figure 11a illustrates the projected land use in 2040 based on the Markov model. Overall, land use dynamics in the WJUA are not pronounced, with impervious surfaces continuing to expand, primarily through outward growth from existing urban areas. The quantitative changes can be observed in the land use transition matrix for 2020–2040 (Table 5). By 2040, the largest land inflow is expected for built-up land, totaling 1,679.44 km², followed by cropland with an inflow of 1,255.33 km². Cropland remains the largest contributor to out-flows, with 58.75% converting to impervious surfaces, 18.97% to forest, and the remainder to water.

Figure 11

Figure 11. Distribution map of land use, carbon storage, and carbon storage change in 2040. (a) Landuse of 2040 (b) CS of 2040 (c) Change of CS from 2020 to 2040.

Table 5

Table 5. Land use transfer matrix from 2020 to 2040 (Unit: km²).

Overall, impervious surfaces are projected to increase by 1,028.90 km², water by 436.43 km², and forest area will increase slightly. In contrast, cropland is expected to decrease by 1,540.32 km², while barren land and grassland experience minor reductions.

In this study, the carbon storage of WJUA in 2040 was estimated using the PLUS-Markov Chain and InVEST models (Figure 11b). The average carbon storage stood at 10.76 t C/hm², maintaining a consistent downward trend. Regarding the spatial pattern of changes (Figure 11c), the reduction in carbon storage was predominantly concentrated in the vicinity of water bodies, which is likely linked to further urban expansion within the study area—given that most cities here are situated along rivers or lakes. In contrast, increases in carbon storage were primarily detected in the Dabie Mountains in the western part of the study area, as well as around state-owned forest farms in the northern and southeastern regions. This trend is presumably closely associated with the implementation of ecological protection initiatives. Compared with the preceding 2 decades, carbon storage in the study area exhibited a continued decreasing trend from 2020 to 2040; however, the magnitude of this reduction was less pronounced than that observed in the earlier 2 decades.

4 Discussion

4.1 Model validation

Reliable quantification of carbon storage and NPP is fundamental to understanding regional ecological carbon cycling. To ensure the credibility of our findings, we conducted rigorous validation of the core model outputs and cross-referenced our results with existing literature on Anhui Province. A comparison of our carbon storage estimates with those reported in prior studies (Mei et al., 2024; Wang et al., 2025b) revealed a high degree of consistency. Specifically, carbon storage hotspots were consistently concentrated in the western Dabie Mountains and southern Anhui Mountains, while low-value areas were distributed in the Jianghuai Hills. Notably, carbon storage in cropland-dominated regions was significantly lower than that in southern mountainous areas, and all studies documented a gradual decreasing trend in carbon storage over 2000–2020 period. Furthermore, NPP estimates from the improved CASA model were validated against MODIS NPP products (MOD17A3H) retrieved from the National Aeronautics and Space Administration (NASA) Land Processes Distributed Active Archive Center (https://ladsweb.modaps.eosdis.nasa.gov/search/), which have a spatial resolution of 500 m. As illustrated in Supplementary Figure S1, the validation yielded a high coefficient of determination (R² = 0.85) and a root mean square error (RMSE) of 25.93 g C/m², demonstrating the robust reliability of the CASA-modeled NPP. Furthermore, key input parameters (e.g., vegetation indices, meteorological variables) have been cross-validated against multiple independent datasets (Tang et al., 2024a; Tang et al., 2024b) to mitigate parameter-induced uncertainties. Collectively, this cross-validation confirms the reliability of both our carbon storage and NPP estimates.

4.2 Mechanistic analysis of the dynamic differences between carbon storage and NPP

A key finding of this study is that NPP declined at a faster rate than carbon storage across the WJUA. Over 90% of urban expansion in the region stemmed from cropland, which has a relatively high baseline NPP. Consequently, the conversion of cropland to impervious surfaces drove substantial NPP losses, consistent with previous observations (Liu et al., 2019). Furthermore, NPP represents an annual ecological increment, rendering it highly sensitive to short-term disturbances such as built-up area expansion and climate variability. In contrast, carbon storage is a long-term accumulative stock. The soil carbon pool exhibits a distinct lag effect, with soil organic carbon decomposition cycles spanning 50–100 years, leading to slower short-term declines. This aligns with findings in Shenzhen (Wang et al., 2024b). Notably, regions experiencing NPP reductions largely coincide with areas of carbon storage decline, however, regions with increasing NPP did not exhibit corresponding gains in carbon storage. This discrepancy likely arises from the sequential process through which NPP contributes to stable soil carbon: litter generated by enhanced NPP must undergo microbial decomposition and humification before being transformed into stable soil organic carbon. In the short term, even with elevated NPP, changes in the soil carbon pool remain imperceptible, precluding synchronous increases in total carbon storage. For instance, during 2000–2005, NPP increased while carbon storage continued to decline gradually (Figure 6).

Numerous prior investigations have employed the Geographical Detector method to identify and quantify the driving factors of carbon storage and NPP (Cai et al., 2024; Zhang et al., 2024; Mei et al., 2024; Wang et al., 2024a). However, this approach cannot fully eliminate interdependencies among factors, which may lead to inflated q-values for topographic variables such as slope (Wang et al., 2024b). To address this limitation, the Random Forest algorithm combined with the SHAP method was utilized herein to assess the independent impacts of each factor on carbon storage and NPP, thereby avoiding confounding effects stemming from factor inter-dependencies. This contrasts with findings from investigations conducted in Wuhan (Wang et al., 2022), Dalian (Wu et al., 2024), and Nanchang (Fu et al., 2025), where roads were identified as primary drivers of carbon storage—these efforts did not account for the independent effects of individual drivers or exclude potential confounding from other variables. In contrast to these conclusions, carbon storage in the present work was predominantly regulated by topographic factors (slope, DEM) with distinct threshold effects, while natural factors (temperature, precipitation) played secondary stabilizing roles and anthropogenic infrastructure exerted weak, uniform impacts. Consistent with recent findings, the spatial distribution of carbon storage is strongly shaped by topographic conditions (Gai et al., 2024; Gc et al., 2025; Mi et al., 2024). Under the framework of nature-based solutions (NbS), topography also provides critical implications for carbon storage protection: safeguarding the integrity of topographic conditions and the continuity of ecological lands is essential to sustaining their carbon sequestration capacity (Li et al., 2023). The same analytical framework was applied to explore NPP drivers, revealing that NPP was co-governed by topographic (DEM, slope) and climatic (temperature) factors—with temperature exhibiting a clear optimal range (13 °C–16 °C). This implies that agricultural and forestry practices in Anhui Province could be optimized within this temperature window. For instance, selecting climate-adapted crop and forest species would help to maximize NPP and associated carbon sequestration. Such optimization supports national “double carbon” goals while safeguarding food security, addressing a key trade-off in sustainable land use. These findings align with those from Anhui Province, where partial correlation analysis explored NPP-driver relationships and demonstrated that NPP across all land cover types was more strongly affected by temperature than precipitation (Tang et al., 2024a). Parallel observations have been reported in China’s karst areas, where NPP was linked to natural factors, with slope being the primary influencing factor at the annual scale (Zhang et al., 2023), and in the Yellow River Basin, where natural factors exerted significantly stronger impacts on NPP than anthropogenic drivers (Wang et al., 2024c). Collectively, these consistencies underscore the credibility of using the Random Forest algorithm combined with the SHAP method to disentangle the independent effects of drivers on carbon storage and NPP. More importantly, the identified driving mechanisms forge a robust nexus between local-scale ecological processes and macro-level sustainability aspirations. By underscoring the predominant role of natural factors and their intrinsic threshold effects, the findings yield actionable guidance for integrating carbon storage augmentation and NPP optimization into territorial spatial planning and low-carbon urban development—core pillars of China’s national ecological civilization initiative, which align with global Sustainable Development Goals (SDGs 13 and 15).

4.3 Limitations

In the present study, acquiring long-term time series and large-scale field measurements remains a significant challenge. This limitation not only constrains the iterative improvement of carbon storage estimation models but also reduces the precision of their outputs (Yan et al., 2016). More critically, the outputs of existing models generally lack validation against systematic field data, limiting the reliability of the results. Moreover, the core input parameters of the InVEST model consist of carbon density values for four carbon pools corresponding to each land use type. However, carbon density itself exhibits significant spatial heterogeneity. For instance, differences in carbon density across forest stands of varying ages within the same vegetation type are not captured (Wang et al., 2025c), further increasing parameter uncertainty. Although this study employed the coupled InVEST-PLUS model to simulate the future spatial patterns of carbon storage in the study area—providing some insights into potential future trends—the model does not account for potential influences from future climate change or socioeconomic developments. Consequently, the simulation results inevitably carry a degree of uncertainty, and the model’s explanatory power regarding the long-term dynamics of carbon storage remains limited.

5 Conclusion

To comprehensively characterize the spatiotemporal dynamics of ecosystem carbon storage and NPP in the WJUA during 2000–2020, we employed the InVEST model and an improved CASA model for comparative analysis. Building on this foundation, we employed a Random Forest approach with the SHAP method to systematically quantify the independent effects of anthropogenic, natural, and topographic factors on the dynamics of these two indicators. Furthermore, we applied the coupled InVEST-PLUS model simulate and project ecosystem carbon storage in the WJUA under a natural development scenario for 2040. The main findings are as follows.

1. Declining Carbon Sequestration Capacity: Multiple disturbances, including urbanization and human activities, have reduced the carbon sequestration capacity of the WJUA. Between 2000 and 2020, both carbon storage and NPP exhibited gradual declines, with total reductions of 0.93% and 9.46%, respectively.

2. Relationship with Land use Change: Carbon storage in the WJUA is closely linked to land use dynamics. Cropland accounts for 56% and forest for 30% of the study area, with changes in cropland predominantly driving variations in both carbon storage and NPP. Forests, possessing substantially higher carbon density and NPP per unit area than other land use types, serve as the main carbon sinks in the region. The continuous conversion of cropland to built-up land has led to the transformation of high-carbon-density land into low-carbon-density land, representing the primary reason for the observed declines in carbon storage and NPP. Carbon storage was predominantly regulated by topographic factors (slope, DEM) with threshold effects, supported secondarily by natural factors, while anthropogenic infrastructure had the weakest impacts. NPP was co-governed by topographic and climatic (temperature, 13 °C–16 °C optimal range) drivers, with GDP as a anthropogenic regulator. These differences underscore the need for targeted management to enhance carbon sequestration and vegetation productivity.

3. Future Projections under Natural Development: Simulation results indicate that, under a natural development scenario, carbon storage in the WJUA is projected to continue declining through 2040. These projections underscore the need for targeted ecological management strategies, including strengthened protection of cropland, cultivation of forest resources, and optimization of urban ecological spatial patterns, to mitigate potential ecological risks associated with declining carbon storage and to support the stability and sustainability of regional ecosystems.

Data availability statement

The datasets presented in this study can be found in online repositories. The names of the repository/repositories and accession number(s) can be found below: The minimum dataset required to replicate our research results (mainly includes land use data, GDP data, population data, annual mean temperature, and annual mean precipitation, roads, railways, and rivers, DEM data and Vector boundary data) is located in the manuscript. In addition, the land use data, GDP data, population data, annual mean temperature, and annual mean precipitation, roads, railways, and rivers, DEM data and Vector boundary data are third-party data, and the authors had no special access privileges to the data and that other researchers will be able to access the data in the same manner as the authors. They can be obtained in the following ways: 1. land use data product is available from http://www.globallandcover.com. 2. GDP data, population data, annual mean temperature, and annual mean precipitation are available from https://www.resdc.cn. 3. roads, railways, and rivers are available from https://www.openstreetmap.org. 4. DEM product is available from https://www.gscloud.cn. 5. Vector boundary data is available from https://www.webmap.cn/.

Author contributions

JF: Data curation, Methodology, Investigation, Writing – original draft, Software, Visualization. HT: Writing – review and editing, Writing – original draft, Visualization, Validation, Funding acquisition, Software. JY: Writing – review and editing, Supervision, Formal Analysis.

Funding

The author(s) declared that financial support was received for this work and/or its publication. This research was funded by the Anhui Province University Research Project (No.2024AH051857), and Anhui Young Talent Educators Development Initiative (No.YQYB2025038).

Acknowledgements

The paper was partially supported by National Natural Science Foundation of China (42271301), Anhui University Excellent Research and Innovation Project (No. 2022AH010094). Authors appreciate the reviewers for their invaluable comments which have led to significant improvement in the paper.

Conflict of interest

The author(s) declared that this work was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Generative AI statement

The author(s) declared that generative AI was not used in the creation of this manuscript.

Any alternative text (alt text) provided alongside figures in this article has been generated by Frontiers with the support of artificial intelligence and reasonable efforts have been made to ensure accuracy, including review by the authors wherever possible. If you identify any issues, please contact us.

Publisher’s note

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.

Supplementary material

The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fenvs.2026.1747268/full#supplementary-material

References

AhlströM, A., Miller, P. A., and Smith, B. (2012). Too early to infer a global NPP decline since 2000. Geophys. Res. Lett. 39, L154031–L154036. doi:10.1029/2012gl052336

CrossRef Full Text | Google Scholar

Baldocchi, D. D. (2003). Assessing the eddy covariance technique for evaluating carbon dioxide exchange rates of ecosystems: past, present and future. Glob. Change Biol. 9, 479–492. doi:10.1046/j.1365-2486.2003.00629.x

CrossRef Full Text | Google Scholar

Cai, J. H., Chi, H., Lu, N., Bian, J., Chen, H. Q., Yu, J. K., et al. (2024). Analysis of spatiotemporal predictions and drivers of carbon storage in the Pearl river Delta urban agglomeration via the PLUS-InVEST-GeoDetector model. Energies 17, 5093–5115. doi:10.3390/en17205093

CrossRef Full Text | Google Scholar

Cantarello, E., Newton, A. C., and Hill, R. A. (2011). Potential effects of future land-use change on regional carbon stocks in the UK. Environ. Sci. & Policy 14, 40–52. doi:10.1016/j.envsci.2010.10.001

CrossRef Full Text | Google Scholar

Chen, Y., Yue, W. Z., Liu, X., Zhang, L. L., and Chen, Y. A. (2021). Multi-scenario simulation for the consequence of urban expansion on carbon storage: a comparative study in Central Asian republics. Land 10, 608–624. doi:10.3390/land10060608

CrossRef Full Text | Google Scholar

DE Carvalho, R. M., and Szlafsztein, C. F. (2019). Urban vegetation loss and ecosystem services: the influence on climate regulation and noise and air pollution. Environ. Pollut. 245, 844–852. doi:10.1016/j.envpol.2018.10.114

PubMed Abstract | CrossRef Full Text | Google Scholar

Field, C. B., Randerson, J. T., and MalmströM, C. M. (1995). Global net primary production: combining ecology and remote sensing. Remote Sens. Environ. 51, 74–88. doi:10.1016/0034-4257(94)00066-V

CrossRef Full Text | Google Scholar

Fu, S., Peng, Y. X., Xu, B. X., Feng, Y. X., Yu, J., Zeng, X. J., et al. (2025). Investigating the effects of urban expansion on carbon stocks and developing optimization strategies: a case study of nanchang. Ecol. Front. 45, 1486–1497. doi:10.1016/j.ecofro.2025.06.017

CrossRef Full Text | Google Scholar

Gai, Z., Zheng, W., Faye, B., Wang, H., and Du, G. (2024). Temporal and spatial characteristics and influencing factors of carbon storage in black soil area under topographic gradient. Land 14 (16), 14010016–14010034. doi:10.3390/land14010016

CrossRef Full Text | Google Scholar

Gc, S., Kafle, G., Ayer, S., Khamcha, R., Poudel, S., Prabhakar, A., et al. (2025). Landscape-level assessment of topographic influences on organic carbon storage in forests of far Western Nepal. Plant-Environment Interact. 6 (2), 70039–70050. doi:10.1002/pei3.70039

PubMed Abstract | CrossRef Full Text | Google Scholar

Grimm, N. B., Faeth, S. H., Golubiewski, N. E., Redman, C. L., Wu, J. G., Bai, X. M., et al. (2008). Global change and the ecology of cities. Science 319, 756–760. doi:10.1126/science.1150195

PubMed Abstract | CrossRef Full Text | Google Scholar

Gui, L. Z., Hu, X. W., Li, X. R., and Zheng, M. (2022). Study on the influence of undertaking industrial transfer on the sustainability development of Wanjiang city belt. Sustainability 14, 14993–15009. doi:10.3390/su142214993

CrossRef Full Text | Google Scholar

Hamel, P., Chaplin-Kramer, R., Sim, S., and Mueller, C. (2015). A new approach to modeling the sediment retention service (InVEST 3.0): case study of the Cape fear catchment, North Carolina, USA. Sci. Total Environ. 524, 166–177. doi:10.1016/j.scitotenv.2015.04.027

PubMed Abstract | CrossRef Full Text | Google Scholar

Hauer, M. E., Fussell, E., Mueller, V., Burkett, M., Call, M., Abel, K., et al. (2020). Sea-level rise and human migration. Nat. Rev. Earth & Environ. 1, 28–39. doi:10.1038/s43017-019-0002-9

CrossRef Full Text | Google Scholar

He, C. Y., Zhang, D., Huang, Q. X., and Zhao, Y. Y. (2016). Assessing the potential impacts of urban expansion on regional carbon storage by linking the LUSD-urban and InVEST models. Environ. Model. & Softw. 75, 44–58. doi:10.1016/j.envsoft.2015.09.015

CrossRef Full Text | Google Scholar

He, C. Y., Liu, Z. F., Xu, M., Ma, Q., and Dou, Y. Y. (2017). Urban expansion brought stress to food security in China: evidence from decreased cropland net primary productivity. Sci. Total Environ. 576, 660–670. doi:10.1016/j.scitotenv.2016.10.107

PubMed Abstract | CrossRef Full Text | Google Scholar

Huang, K. N., Li, X., Liu, X. P., and Seto, K. C. (2019). Projecting global urban land expansion and heat island intensification through 2050. Environ. Res. Lett. 14, 114037–114049. doi:10.1088/1748-9326/ab4b71

CrossRef Full Text | Google Scholar

Huo, Z. M., Liu, G. Y., Yan, N. Y., Li, H., Shah, A. M., Chen, Y., et al. (2023). Estimating the contribution of community landscape construction to urban carbon neutrality: methodology and database construction. Environ. Res. Lett. 18, 104035–104055. doi:10.1088/1748-9326/acfb24

CrossRef Full Text | Google Scholar

Hutyra, L. R., Yoon, B., Hepinstall-Cymerman, J., and Alberti, M. (2011). Carbon consequences of land cover change and expansion of urban lands: a case study in the Seattle metropolitan region. Landsc. Urban Plan. 103, 83–93. doi:10.1016/j.landurbplan.2011.06.004

CrossRef Full Text | Google Scholar

Jiang, C., Wu, Z. F., Cheng, J., Yu, Q., and Rao, X. Q. (2015). Impacts of urbanization on net primary productivity in the Pearl river Delta, China. Int. J. Plant Prod. 9, 581–598. doi:10.22069/IJPP.2015.2464

CrossRef Full Text | Google Scholar

Kang, J. Q., Zhang, L. L., Meng, Q. Y., Wu, H. T., Hou, J. Y., Pan, J., et al. (2025). Land use and carbon storage evolution under multiple scenarios: a spatiotemporal analysis of Beijing using the PLUS-InVEST model. Sustainability 17, 1589–1609. doi:10.3390/su17041589

CrossRef Full Text | Google Scholar

Li, Y. L., Han, M. Y., Liu, S. Y., and Chen, G. Q. (2019). Energy consumption and greenhouse gas emissions by buildings: a multi-scale perspective. Build. Environ. 151, 240–250. doi:10.1016/j.buildenv.2018.11.003

CrossRef Full Text | Google Scholar

Li, X. C., Luo, Y. H., and Wu, J. S. (2022). Decoupling relationship between urbanization and carbon sequestration in the Pearl river Delta from 2000 to 2020. Remote Sens. 14, 526–546. doi:10.3390/rs14030526

CrossRef Full Text | Google Scholar

Li, G., Cheng, G., Liu, G., Chen, C., and He, Y. (2023). Simulating the land use and carbon storage for nature-based solutions (NbS) under multi-scenarios in the three gorges reservoir area: integration of remote sensing data and the RF–Markov–CA–InVEST model. Remote Sens. 15 (1), 5100–5123. doi:10.3390/rs15215100

CrossRef Full Text | Google Scholar

Liang, W., Yang, Y. T., Fan, D. M., Guan, H. D., Zhang, T., Long, D., et al. (2015). Analysis of spatial and temporal patterns of net primary production and their climate controls in China from 1982 to 2010. Agric. For. Meteorology 204, 22–36. doi:10.1016/j.agrformet.2015.01.015

CrossRef Full Text | Google Scholar

Liang, X., Guan, Q., Clarke, K. C., Liu, S., Wang, B., and Yao, Y. (2021). Understanding the drivers of sustainable land expansion using a patch-generating land use simulation (PLUS) model: a case study in Wuhan, China. Comput. Environ. Urban Syst. 85, 101569–101582. doi:10.1016/j.compenvurbsys.2020.101569

CrossRef Full Text | Google Scholar

Liu, X. P., Pei, F. S., Wen, Y. Y., Li, X., Wang, S. J., Wu, C. J., et al. (2019). Global urban expansion offsets climate-driven increases in terrestrial net primary productivity. Nat. Commun. 10, 13462–13469. doi:10.1038/s41467-019-13462-1

PubMed Abstract | CrossRef Full Text | Google Scholar

Lu, M. G., Sun, H. W., Yang, Y., Xue, J., Ling, H. B., Zhang, H., et al. (2025). Assessing recovery time of ecosystems in China: insights into flash drought impacts on gross primary productivity. Hydrology Earth Syst. Sci. 29, 613–625. doi:10.5194/hess-29-613-2025

CrossRef Full Text | Google Scholar

Mei, Z. J., Li, C., Zhao, J., Li, Z. X., Chen, K. Y., Huang, X., et al. (2024). The temporal and spatial evolution characteristics and driving factors of ecosystem service bundles in Anhui province, China. Land 13, 736–752. doi:10.3390/land13060736

CrossRef Full Text | Google Scholar

Mi, Y., Li, S., and Wu, B. (2024). Study on the variation of carbon storage in the chang-zhu-tan urban agglomeration in China based on topographic relief. Front. Environ. Sci. 12 (1), 1481540–1481554. doi:10.3389/fenvs.2024.1481540

CrossRef Full Text | Google Scholar

Ni, J. (2013). Carbon storage in Chinese terrestrial ecosystems: approaching a more accurate estimate. Clim. Change 119, 905–917. doi:10.1007/s10584-013-0767-7

CrossRef Full Text | Google Scholar

Potter, C. S., Randerson, J. T., Field, C. B., Matson, P. A., Vitousek, P. M., Mooney, H. A., et al. (1993). Terrestrial ecosystem production - a process model-based on global satellite and surface data. Glob. Biogeochem. Cycles 7, 811–841. doi:10.1029/93gb02725

CrossRef Full Text | Google Scholar

Potter, C., Klooster, S., and Genovese, V. (2012). Net primary production of terrestrial ecosystems from 2000 to 2009. Clim. Change 115, 365–378. doi:10.1007/s10584-012-0460-2

CrossRef Full Text | Google Scholar

Quay, P. D., Tilbrook, B., and Wong, C. S. (1992). Oceanic uptake of fossil fuel CO2: carbon-13 evidence. Sci. (New York, N.Y.) 256, 74–79. doi:10.1126/science.256.5053.74

PubMed Abstract | CrossRef Full Text | Google Scholar

Roxburgh, S. H., Berry, S. L., Buckley, T. N., Barnes, B., and Roderick, M. L. (2005). What is NPP? Inconsistent accounting of respiratory fluxes in the definition of net primary production. Funct. Ecol. 19, 378–382. doi:10.1111/j.1365-2435.2005.00983.x

CrossRef Full Text | Google Scholar

Schwalm, C. R., Anderegg, W. R. L., Michalak, A. M., Fisher, J. B., Biondi, F., Koch, G., et al. (2017). Global patterns of drought recovery. Nature 548, 202–205. doi:10.1038/nature23021

PubMed Abstract | CrossRef Full Text | Google Scholar

Shi, K. F., Yu, B. L., Ma, J. J., Cao, W. D., and Cui, Y. Z. (2023). Impacts of slope climbing of urban expansion on global sustainable development. Innovation 4, 100529. doi:10.1016/j.xinn.2023.100529

PubMed Abstract | CrossRef Full Text | Google Scholar

Sonti, N. F., Groffman, P. M., Nowak, D. J., Henning, J. G., Avolio, M. L., and Rosi, E. J. (2022). Urban net primary production: concepts, field methods, and Baltimore, Maryland, USA case study. Ecol. Appl. 32, e2562–e2578. doi:10.1002/eap.2562

PubMed Abstract | CrossRef Full Text | Google Scholar

Sun, H. W., Gui, D. W., Yan, B. W., Liu, Y., Liao, W. H., Zhu, Y., et al. (2016). Assessing the potential of random forest method for estimating solar radiation using air pollution index. Energy Convers. Manag. 119, 121–129. doi:10.1016/j.enconman.2016.04.051

CrossRef Full Text | Google Scholar

Sun, F. F. F., Hong, W., Luo, H., Yu, J., Fang, L., and Miao, Y. (2023). Evolution analysis and prediction of carbon storage in Anhui province based on PLUS and InVEST model. J. Soil Water Conserv. 37, 151–158. doi:10.13870/j.cnki.stbcxb.2023.01.021

CrossRef Full Text | Google Scholar

Tang, H., Fang, J. W., and Yuan, J. (2024a). Climate change and land Use/Land cover change (LUCC) leading to spatial shifts in net primary productivity in Anhui province, China. Plos One 19, e0307516–e0307541. doi:10.1371/journal.pone.0307516

PubMed Abstract | CrossRef Full Text | Google Scholar

Tang, H., Liu, X., Xie, R. J., Lin, Y. Q., Fang, J. W., and Yuan, J. (2024b). Response of carbon energy storage to land use/Cover changes in Shanxi province, China. Energies 17, 3284. doi:10.3390/en17133284

CrossRef Full Text | Google Scholar

Wang, Z., Zeng, J., and Chen, W. X. (2022). Impact of urban expansion on carbon storage under multi-scenario simulations in Wuhan, China. Environ. Sci. Pollut. Res. 29, 45507–45526. doi:10.1007/s11356-022-19146-6

PubMed Abstract | CrossRef Full Text | Google Scholar

Wang, S. C., Wei, G. E., Gao, M. M., and Fan, Y. M. (2023). Investigating the mechanism of urbanization on the net primary productivity of vegetation in the Yangtze River economic belt: a comprehensive analysis from global and local effects. Front. Ecol. Evol. 11, 1231487–1231500. doi:10.3389/fevo.2023.1231487

CrossRef Full Text | Google Scholar

Wang, C., Wang, L. S., Zhao, W. Z., Zhang, Y. Y., and Liu, Y. Y. (2024a). Analysis of spatiotemporal change and driving factors of NPP in Qilian Mountains from 20 0 0 to 2020. Rangel. Ecol. & Manag. 96, 56–66. doi:10.1016/j.rama.2024.05.005

CrossRef Full Text | Google Scholar

Wang, R. Y., Mo, X. Y., Ji, H., Zhu, Z., Wang, Y. S., Bao, Z. L., et al. (2024b). Comparison of the CASA and InVEST models' effects for estimating spatiotemporal differences in carbon storage of green spaces in megacities. Sci. Rep. 14, 5456–5468. doi:10.1038/s41598-024-55858-0

PubMed Abstract | CrossRef Full Text | Google Scholar

Wang, X. L., He, W. X., Huang, Y. L., Wu, X., Zhang, X., and Zhang, B. W. (2024c). Exploring spatial non-stationarity and scale effects of natural and anthropogenic factors on net primary productivity of vegetation in the yellow River basin. Remote Sens. 16, 3156. doi:10.3390/rs16173156

CrossRef Full Text | Google Scholar

Wang, J. Y., Shao, Z. F., Fu, P., Zhuang, Q. W., Chang, J. H., Jing, P. Q., et al. (2025a). Unraveling the impact of urban expansion on vegetation carbon sequestration capacity: a case study of the Yangtze river economic belt. Sustain. Cities Soc. 120, 106157–106171. doi:10.1016/j.scs.2025.106157

CrossRef Full Text | Google Scholar

Wang, Y. C., Ang, Y. F., Zhang, Y. Y., Ruan, Y. F., and Wang, B. Y. (2025b). Identification of ecological functional areas and scenario simulation analysis of the Wanjiang urban belt from a trade-Off/Synergy perspective. Land 14, 444. doi:10.3390/land14030444

CrossRef Full Text | Google Scholar

Wang, Y. D., Zhang, D., He, C. Y., Qi, T., and Li, P. H. (2025c). Simulating the effects on carbon sinks caused by urban expansion using a coupled urban-ecological model. J. Clean. Prod. 520, 146184–146196. doi:10.1016/j.jclepro.2025.146184

CrossRef Full Text | Google Scholar

Wen, Y. Y., Liu, X. P., Bai, Y., Sun, Y., Yang, J., Lin, K., et al. (2019). Determining the impacts of climate change and urban expansion on terrestrial net primary production in China. J. Environ. Manag. 240, 75–83. doi:10.1016/j.jenvman.2019.03.071

PubMed Abstract | CrossRef Full Text | Google Scholar

Wen-Quan, Z., Yao-Zhang, P., and Jin-shui, Z. (2007). Remote sensing estimation of terrestrial vegetation net primary produc tivity in China. Chin. J. Plant Ecol. 31, 423–434. doi:10.17521/cjpe.2007.0050

CrossRef Full Text | Google Scholar

Wu, F., and Wang, Z. Y. (2023). Assessing the impact of urban land expansion on ecosystem carbon storage: a case study of the Changzhutan metropolitan area, China. Ecol. Indic. 154, 110688–110702. doi:10.1016/j.ecolind.2023.110688

CrossRef Full Text | Google Scholar

Wu, Q., Wang, L., Wang, T. Y., Ruan, Z. Y., and Du, P. (2024). Spatial-temporal evolution analysis of multi-scenario land use and carbon storage based on PLUS-InVEST model: a case study in Dalian, China. Ecol. Indic. 166, 112448–112462. doi:10.1016/j.ecolind.2024.112448

CrossRef Full Text | Google Scholar

Xie, W. X., Huang, Q. X., He, C. Y., and Zhao, X. (2018). Projecting the impacts of urban expansion on simultaneous losses of ecosystem services: a case study in Beijing, China. Ecol. Indic. 84, 183–193. doi:10.1016/j.ecolind.2017.08.055

CrossRef Full Text | Google Scholar

Yan, Y., Zhang, C., Hu, Y. F., and Kuang, W. H. (2016). Urban land-cover change and its impact on the ecosystem carbon storage in a dryland city. Remote Sens. 8, 6–23. doi:10.3390/rs8010006

CrossRef Full Text | Google Scholar

Yan, Y. C., Liu, X. P., Wang, F. Y., Li, X., Ou, J. P., Wen, Y. Y., et al. (2018). Assessing the impacts of urban sprawl on net primary productivity using fusion of Landsat and MODIS data. Sci. Total Environ. 613, 1417–1429. doi:10.1016/j.scitotenv.2017.09.139

PubMed Abstract | CrossRef Full Text | Google Scholar

Zhang, M., Huang, X. J., Chuai, X. W., Yang, H., Lai, L., and Tan, J. Z. (2015). Impact of land use type conversion on carbon storage in terrestrial ecosystems of China: a spatial-temporal perspective. Sci. Rep. 5, 10233–10245. doi:10.1038/srep10233

PubMed Abstract | CrossRef Full Text | Google Scholar

Zhang, W., XI, M. Z., Liu, H. D., and Zheng, H. (2023). Low sensitivity of net primary productivity to climatic factors in three karst provinces in southwest China from 1981 to 2019. Ecol. Indic. 153, 110465–110478. doi:10.1016/j.ecolind.2023.110465

CrossRef Full Text | Google Scholar

Zhang, L. K., Guan, Q. C., Li, H., Chen, J. W., Meng, T. Y., and Zhou, X. (2024). Assessment of coastal carbon storage and analysis of its driving factors: a case study of Jiaozhou Bay. China. Land 13, 1208. doi:10.3390/land13081208

CrossRef Full Text | Google Scholar

Zhao, M. S., and Running, S. W. (2010). Drought-induced reduction in global terrestrial Net Primary production from 2000 through 2009. Science 329, 940–943. doi:10.1126/science.1192666

PubMed Abstract | CrossRef Full Text | Google Scholar

Zhuang, Q. W., Shao, Z. F., Li, D. R., Huang, X., Li, Y. Z., Altan, O., et al. (2023). Impact of global urban expansion on the terrestrial vegetation carbon sequestration capacity. Sci. Total Environ. 879, 163074–163091. doi:10.1016/j.scitotenv.2023.163074

PubMed Abstract | CrossRef Full Text | Google Scholar