Assessing the Spatiotemporal Patterns and Afforestation Impacts on Land-Use Carbon Storage in the Yellow River Basin Using Multi-Source Remote Sensing Products

Abstract

Afforestation plays a vital role in reshaping land systems and enhancing carbon sequestration, particularly in ecologically fragile regions. However, the carbon implications and spatial dynamics of large-scale planted-forest (PF) expansion in the Yellow River Basin (YRB) remain insufficiently understood. Focusing on the YRB, this study integrates multi-source land-use, forest type, and carbon datasets to evaluate land-use transitions (2000–2020) and quantify changes in total ecosystem carbon (TEC), aboveground carbon (AGC), and PF-derived AGC (PF-AGC) from 2005 to 2020 under the IPCC-based accounting framework. The results show cumulative land-use conversion of 118,481 km², with forest land expanded to 11.89% of the basin, mainly due to afforestation efforts in the middle reaches. TEC followed a rise–decline–rebound trajectory, yielding a net gain of 1.96 × 10⁸ t, while AGC increased by 4.37 × 10⁸ t. With the expansion of PF, PF-AGC contributed 1.60 × 10⁸ t (36.61% of AGC gains), primarily sourced from grassland (40.51%), natural forests (35.15%), and cropland (23.56%). PFs were dominated by young stands (≤40 years), spatially clustered in the middle–lower reaches, and exhibited higher carbon sink potential than natural forests. Spatially, AGC and PF distributions underwent staged reconfiguration. Standard deviational ellipse and centroid analyses revealed eastward shifts and axis changes in AGC, and southwestward migration of PFs, indicating PF expansion as a major driver of carbon redistribution. These findings clarify the forest age–land-use–carbon nexus and highlight the spatial impact of afforestation, offering critical insights for region-specific low-carbon strategies and sustainable land governance in the YRB.

1. Introduction

Global climate change is one of the most formidable challenges of the twenty-first century. Persistent warming and frequent extreme events threaten ecological security and socioeconomic development. The Paris Agreement calls for restraining the rise in global mean temperature to well below 2 °C and pursuing efforts to limit it to 1.5 °C, underscoring the urgency of reducing greenhouse gas emissions while enhancing carbon sinks [1]. Therefore, achieving carbon neutrality has become a global consensus and national priority in China. Forest carbon sinks, which stores half of terrestrial ecosystem carbon, regulate atmospheric CO₂, and stabilize the carbon cycle, play a central role [2,3,4,5]. Since 1999, large-scale ecological programs, such as the Grain-for-Green Program (GGP), have enhanced carbon sequestration, reshaped ecosystem structure and function, and improved habitat quality. Owing to their rapid growth and high sequestration rates, plantations have become pivotal for achieving carbon peak and neutrality goals [6,7,8,9,10].

The Yellow River Basin (YRB) is a key ecological barrier and a heavily human-influenced economic corridor in China. It encompasses alpine ecosystems, grassland, forest, cropland, wetland, and urban ecosystems, with several subregions exhibiting among the highest ecosystem carbon densities in China. This makes the YRB a strategic national carbon sink that plays a critical role in maintaining ecological security and promoting sustainable development [11]. Although research on basin-scale carbon storage has expanded in recent years, most studies have emphasized macro-level drivers, including socioeconomic development, climate variability, and land-use restructuring [12,13,14,15,16]. However, the specific contributions of planted forests (PFs) and the mechanisms underlying their ability to shape the spatiotemporal evolution of regional carbon storage remain poorly understood. As a core element of large-scale ecological restoration in China, afforestation can enhance regional carbon sinks, regulate ecosystem functions, and support carbon peak and neutrality goals. However, the limited systematic assessment of PF carbon sinks limits our understanding of their role in the carbon cycle and hinders evidence-based decisions regarding silvicultural management and spatial planning. Therefore, a spatiotemporally explicit analysis is required to evaluate the role of PFs in the carbon stock dynamics of the YRB, elucidate the contribution mechanisms, and explore optimization pathways [17,18,19].

Approaches for carbon storage estimation fall into two broad categories: field-based inventories, which combine plot surveys, remote-sensing interpretation, and biomass-to-carbon conversion factors achieve high accuracy but are costly and thus limited in spatial extent [20]; and process-based dynamic global vegetation models, which simulate vegetation–climate interactions [21,22] but are excluded from regional-level applications by parameter complexity and data requirements. In contrast, the Integrated Valuation of Ecosystem Services and Trade-offs (InVEST) carbon model supports large-scale assessments because of its operational simplicity and modest data requirements. However, its assumption of fixed carbon densities limits the representation of environmental and anthropogenic effects [23]. The IPCC-based accounting framework improves the characterization of spatial heterogeneity in ecosystem carbon storage [24], but it does not explicitly account for the effects of forest age. Incorporating age structure can therefore yield more realistic estimates of current carbon stocks and future sequestration potential.

In this study, we analyzed the spatiotemporal evolution of land use in the YRB from 2000 to 2020. Within the IPCC-based accounting framework, our specific objectives were to: (i) quantify total ecosystem carbon (TEC), aboveground carbon (AGC), and carbon storage in natural forests (NFs) and PFs; (ii) identify and trace the land-use origins and transformation pathways of PFs, and quantify their contributions to regional carbon stock increases; (iii) assess the carbon sink potential associated with forest age structure by distinguishing the effects of different stand development stages; and (iv) characterize the spatial patterns and directional shifts in AGC and PF expansion using standard deviational ellipse (SDE) and centroid shift analyses. By integrating multi-source remote sensing products with an IPCC-based accounting framework, this study elucidates how afforestation drives changes in carbon storage and provides decision-making support for ecosystem restoration, carbon neutrality, and sustainable land management in the YRB.

2. Materials and Methods

2.1. Study Area and Data Sources

2.1.1. Study Area

The YRB is divided into three sections: (i) the upper reaches—from the headwaters to Hekou Town, including Qinghai, Sichuan (Zoige/Tangke), Gansu, Ningxia, and western Inner Mongolia; (ii) the middle reaches—from Hekou Town to Taohuayu (Zhengzhou, Henan), covering southern Inner Mongolia, northern Shaanxi, western Shanxi, and western Henan; and (iii) the lower reaches—from Taohuayu to the Bohai Sea, mainly spanning eastern Henan and Shandong. The main stem is 5464 km long and drains 752,400 km². The basin has a pronounced west–east stepwise topography, with three terraces and strong climatic gradients (Figure 1). The first terrace on the Qinghai–Tibet Plateau generally lies above 3000 m and is characterized by low temperatures, limited precipitation, and arid conditions. The second terrace is the Loess Plateau (1000–2000 m), which has a relatively gentle relief and marked ecological fragility. Intense summer rainfall drives severe soil erosion and delivers large sediment loads to the Yellow River, making the YRB a priority area for afforestation and ecological restoration. The third terrace corresponds to the North China Plain, which lies mostly below 100 m and has a mean annual temperature of approximately 13–15 °C under a monsoonal climate, with hot summers and cold winters [25].

2.1.2. Data Sources and Pre-Processing

Land-use information was obtained from the China Land Use Dataset (1985–2021), which was derived from 335,709 Landsat scenes and provides annual national maps. To ensure temporal consistency with the national classification, the maps were reclassified into six primary classes: cropland (CL), forest land (FL), consisting of both PF and NF, grassland (GL), water land (WL), built-up land (BL), and unused land (UL). Biomass inputs were obtained from the Vegetation Biomass Distribution Atlas of China (2001–2020) and accessed via the PIE-Engine remote-sensing and geoinformation cloud platform. PF and NF distributions were obtained from the 1990–2020 national dataset released by the National Ecoscience Data Center. Forest age was obtained from the 2020 national forest-age spatial dataset on the same platform. Soil organic carbon (SOC) density was derived from the National Soil Information Grid of China (NSIGC; 2010–2018), provided by the National Earth System Science Data Center. During pre-processing, point samples from the “2010s China Terrestrial Ecosystem Carbon Density” dataset was used as the reference. Using ArcGIS 10.8, the sample locations were paired with collocated NSIGC pixels, and 85 outliers were removed. Ordinary least-squares regressions were then fitted separately by 5-year stratum (R² > 0.70 for all strata). The resulting regression functions were applied using Raster Calculator to produce SOC rasters for 2001–2005, 2006–2010, and 2011–2015. Because of insufficient reference data for 2016–2020, the unadjusted NSIGC SOC product was used for this period. Vegetation regionalization data were from the Vegetation Regionalization Vector Dataset of China; YRB boundary vectors from “China’s Nine Major River Basins.” Both datasets are available at the Resource and Environment Science Data Center, Chinese Academy of Sciences. All datasets were clipped to the YRB boundary, re-projected to the Albers Equal Area Conic projection, and co-registered prior to analysis. To accurately analyze land-use transitions and planted forest origins, 30 m × 30 m data were used, while all datasets in the IPCC-based accounting framework were resampled to 1 km × 1 km. The data sources and specifications are listed in

Table 1. Data sources used in this study.

Data Type	Dataset Name	Format	Resolution	Source
Multi-Source Remote Sensing Products	Land-use data	Raster	30 m	http://doi.org/10.5281/zenodo.4417809 (accessed on 17 March 2025)
	Biomass data	Raster	500 m	https://engine.piesat.cn/ (accessed on 17 March 2025)
	Plantation and natural forest data	Raster	1000 m	http://www.nesdc.org.cn (accessed on 19 March 2025)
	Forest age data	Raster	30 m
Soil carbon density	Soil carbon density data	Raster	1000 m	https://www.geodata.cn (accessed on 20 March 2025)
Regional datasets	Vegetation regionalization data	Vector	-	https://www.resdc.cn (accessed on 20 March 2025)
	Yellow River Basin boundary data	Vector	-

.

2.2. Research Methods

2.2.1. Land-Use Change Analysis

Single land-use dynamic degree: The annual rate of change for land-use class i over period T (years) was computed as follows:

$$K_i ={\displaystyle \frac{A_{i,t_2}-A_{i,t_1}}{A_{i,t_1}T}} \times 100\%$$

(1)

where K_i is the dynamic degree of land-use type i, $A_{i,t_1}$ and $A_{i,t_2}$ are the areas of class i at the beginning and end of the period, respectively, T is the time interval between t₁ and t₂.

Land-use transition matrix: A transition matrix characterizes the direction and magnitude of conversion among land-use types:

$$S_{\mathit{ij}} = \left[\begin{array}{cccc}{S}_{11}& S_{12}& \dots & S_{1n}\\ S_{21}& S_{22}& \dots & S_{2n}\\ \dots & \dots & \dots & \dots \\ S_{n1}& S_{n2}& S& S_{\mathit{nn}}\end{array}\right]$$

(2)

where S_ij represents the area transferred from land-use type i at the beginning of the study period to type j at the end of the study period and n denotes the number of land-use categories.

2.2.2. Carbon Storage Accounting Model

TEC was estimated within the IPCC-based accounting framework as follows:

$$C_{\mathit{total}}=C_{\mathit{above}}+C_{\mathit{below}}+C_{\mathit{soil}}+C_{\mathit{dead}}$$

(3)

where $C_{\mathit{total}}$ is the TEC and $C_{\mathit{above}}$, $C_{\mathit{below}}$, $C_{\mathit{soil}}$, and $C_{\mathit{dead}}$ are the AGC, belowground carbon, SOC, and dead organic carbon (DOC) storage, respectively.

Aboveground biomass carbon is estimated as follows:

$$C_{\mathit{above}}=\sum _i{B}_{(\mathit{above},i)}\times {\mathit{CF}}_i\times S_i$$

(4)

where $B_{(\mathit{above},i)}$ is the density of aboveground biomass for class i; ${\mathit{CF}}_i$ is the carbon fraction coefficient for class i; $S_i$ is the area for class i. Coefficients followed the 2006 IPCC Guidelines and national parameters [24]: 0.50 (FL), 0.393 (GL), 0.45 (CL), and 0.40 (BL). Where applicable, a belowground fraction of 0.366 is commonly used [26,27].

Belowground biomass carbon is estimated as follows:

$$C_{\mathit{below}}=\sum _i{B}_{(\mathit{above},i)}\times R_i\times {\mathit{CF}}_i\times S_i$$

(5)

where $R_i$ denotes the root-to-shoot ratio for class i. For CL, R = 0.19 based on national averages; For FL, values were derived from IPCC default tables by climatic zone. For GL, R varied by GL type.

Soil carbon storage is calculated as follows:

$$C_{\mathit{soil}}=\sum _{j=1}^{n_i}\left(C_{\mathit{ij}}\times S_j\right)$$

(6)

where $C_{\mathit{ij}}$ is the soil carbon density in grid cell j of type i, $S_j$ is the grid cell area, and is the number of cells of type i.

DOC storage: Under the IPCC Tier 1 approach, non-forest stable land uses (CL, GL, BL, WL, and UL) were assumed to have zero net changes in DOC [24]. For FL, IPCC default factors for forest conversion were applied. Consistent with Tier 1, DOC was not separately estimated for GL (litter and woody debris were accounted for within the aboveground parameterization), and the DOC values for BL, WL, and UL were set to zero.

2.2.3. Spatiotemporal Analysis

SDE and centroid migration were used to characterize the spatial structure and displacement of the AGC. SDE summarizes the dispersion and orientation of AGC, whereas centroid analysis quantifies shifts in the weighted mean center across periods [28,29]. The migration distance between the two periods is calculated as follows:

$$D=\sqrt{{\left({\overline{X}}_2-{\overline{X}}_1\right)}^2+{\left({\overline{Y}}_2-{\overline{Y}}_1\right)}^2}$$

(7)

Centroid coordinates were calculated as follows:

$$\overline{X}={\displaystyle \frac{{\sum }_{i=1}^n(X_i·W_i)}{{\sum }_{i=1}^n{X}_i}}, \overline{Y}={\displaystyle \frac{{\sum }_{i=1}^n(X_i·W_i)}{{\sum }_{i=1}^n{X}_i}}$$

(8)

where ($\overline{X}$, $\overline{Y}$) are the centroid coordinates; ($X_i,Y_i$) are the coordinates of element i; W is the weight; and n is the total number of elements.

3. Results

3.1. Spatiotemporal Evolution of Land Use in the YRB

Between 2000 and 2020, cumulative land-use transitions in the YRB amounted to 118,481 km². A Sankey diagram reveals intensive, reciprocal conversions among classes, which are most pronounced in the middle–lower reaches, indicating high spatiotemporal dynamics and structural complexity. In terms of land-use transitions (Figure 2a), FL, WL, and BL were the principal recipient classes, whereas CL and UL were the main donor classes. Forest gains originated chiefly from GL and CL. Despite localized reversals, FL exhibited a net increase consistent with large-scale ecological restoration, such as the GGP. BL expanded mainly at the expense of CL and GL, reflecting rapid urbanization; WL originated largely from CL and GL; and UL was primarily transformed into GL and CL and declined continuously.

In terms of spatial distribution, FL, CL, and BL were concentrated in the middle–lower reaches, GL dominated the upper–middle reaches, and WL and UL were concentrated upstream. Spatial changes (Figure 2b) showed FL expansion in the middle reaches, notably in Shaanxi and Shanxi; CL expansion on the Hetao and Ningxia plains; GL increases across the upper–middle reaches, notably around Yulin, Ordos, and Lanzhou; sharp BL growth in the middle–lower reaches, especially near Xi’an and Zhengzhou; upstream WL expansion consistent with glacier-melt signals; and a sustained upstream decline in UL.

Regarding class composition, FL, CL, and GL were dominant. FL covered approximately 11.89% of the basin and increased continuously with a positive and rising dynamic degree (Figure 2c). CL accounted for approximately 23.64% and exhibited an overall decline, with a decreasing and then increasing trajectory. GL accounted for approximately 57.10% and increased overall but with marked fluctuations (Figure 2c). WL, BL, and UL together accounted for approximately 7.37%, with WL varying substantially, BL increasing rapidly, and UL declining steadily (Figure 2c).

3.2. Spatiotemporal Patterns of Carbon Storage in the YRB

Based on the IPCC-based accounting framework [24], TEC values in the YRB were 76.56, 85.43, 72.28, and 78.52 × 10⁸ t in 2005, 2010, 2015, and 2020, respectively, exhibiting an increasing–decreasing–rebounding trajectory with an overall upward tendency. From 2005 to 2020, TEC registered a net gain of 1.96 × 10⁸ t. Spatially, TEC was higher in the west than in the east and higher in the south than in the north (Figure 3a). The Qinghai–Tibet Plateau formed a high-value zone where low temperatures suppressed soil organic matter decomposition, whereas inland sandy and ecologically fragile areas, notably the Ordos Basin, showed lower stocks.

At the basin scale, the response of SOC was complex. Prior studies indicated that afforestation affects vegetation and soil pools differently [30,31,32]. For example, increases in vegetation carbon do not necessarily translate into synchronous SOC accumulation, with ratios modulated by species and site conditions and exhibiting pronounced spatial heterogeneity [33]. Accordingly, AGC better reflects the immediate effects of afforestation. AGC in the YRB totaled 19.69, 20.43, 20.07, and 24.06 × 10⁸ t in 2005, 2010, 2015, and 2020, mirroring the TEC pattern but with a steeper rise during 2015–2020. In addition, the net increase from 2005 to 2020 was 4.37 × 10⁸ t. Spatially, AGC exhibited a clear south–north gradient (Figure 3b), with high values concentrated in the middle and lower reaches.

For PF aboveground carbon (PF-AGC), the values in the YRB were 0.74, 1.24, 1.60, and 2.34 × 10⁸ t in 2005, 2010, 2015, and 2020, respectively. Over 2005–2020, PF-AGC increased by 1.60 × 10⁸ t, accounting for approximately 36.61% of the basin-wide AGC net increase. Spatially, high PF-AGC and expansion belts were concentrated in the middle reaches (notably in Shaanxi and Shanxi) (Figure 3c). The PF-AGC trajectory aligned with the 2000–2020 forest expansion pattern inferred from land-use conversions, indicating strong spatiotemporal coherence.

3.3. Assessment of Afforestation Impacts on Carbon Storage in the YRB

3.3.1. Effects of Afforestation on Land-Use and Carbon Storage

During 2000–2020, the cumulative conversion to PF totaled 56,191 km², indicating a pronounced expansion. Source attribution (Figure 4a) identified GL, NF, and CL as the primary contributors, accounting for approximately 40.51%, 35.15%, and 23.56%, respectively. Conversions from GL to PF aligned with grazing exclusion and degraded GL restoration, whereas conversions from CL to PF were temporally and spatially consistent with the GGP. Inputs from WL, BL, and UL were minor and contributed little to the overall expansion, accounting for approximately 0.15%, 0.63%, and 0.0027%, respectively. Commas are only used for numbers with five or more digits. We have removed them in four-digit numbers. Please confirm.

Time-series analysis (Figure 4c) showed that transfers into PF totaled 5566 km² during 2000–2005, sourced primarily from CL and GL with limited input from NF. PF-AGC accounted for approximately 0.74 × 10⁸ t, which is consistent with a modest basin-level sink and early implementation of the GGP. During 2005–2010, transfers increased to 13,612 km², with conversions from GL to PF becoming dominant, NF-to-PF increasing, CL-to-PF persisting, and PF-AGC reaching 1.24 × 10⁸ t. Over 2010–2015, transfers reached 14,826 km² and the expansion rate moderated, with GL-to-PF and NF-to-PF contributions roughly balanced, CL-to-PF remaining steady, and PF-AGC increasing to 1.60 × 10⁸ t. During 2015–2020, the expansion peaked at 21,387 km², with GL remaining the largest source, followed by NF and CL, and PF-AGC rising to 2.34 × 10⁸ t, further strengthening its relative contribution to the basin-level sink.

Spatially, PF expansion was concentrated in the middle–lower reaches, forming belts and patches along the main valleys, tributary corridors, and agricultural zones, whereas expansion on the Qinghai–Tibet Plateau was limited by low temperatures and soil constraints (Figure 4b). Overall, PF priority areas overlapped with severe soil erosion zones and densely populated regions, underscoring the strategic role of afforestation at the ecological–socioeconomic interface.

3.3.2. Comparative Analysis of PFs and NFs

From 2005 to 2020, PF-AGC increased from 0.74 × 10⁸ t to 2.34 × 10⁸ t, an increase of approximately 216.22%. Over the same period, NF aboveground carbon rose from 2.90 × 10⁸ t to 3.28 × 10⁸ t, an increment of 0.38 × 10⁸ t (approximately 13.10%). The share of PFs in FL aboveground carbon increased from approximately 19.73% to 46.43% (Figure 5a), indicating a markedly strengthened basin-scale contribution of PF.

Given that major afforestation species in the basin, such as Pinus tabulaeformis, Platycladus orientalis, and Populus spp., typically exhibit a growth inflection at approximately 40 years, and stands ≥71 years [34], especially in upper mountainous areas, often show signs of senescence, forest age was partitioned into five classes: young forest (YF, ≤10 years), middle-aged forest (MiF, 11–25 years), near-mature forest (NmF, 26–40 years), mature forest (MaF, 41–70 years), and overmature forest (OF, ≥71 years). For PFs, stands ≤40 years accounted for 86.66% of the area (40,608 km²), with an age-class composition of 2.43% YF, 44.25% MiF, 39.98% NmF, 12.31% MaF, and 1.03% OF. For NFs, stands ≤40 years accounted for 40.31% (25,398 km²), with 0.69% YF, 9.11% MiF, 30.51% NmF, 48.88% MaF, and 10.81% OF (Figure 5c). Collectively, these results indicate a young age structure across the basin, with PF exhibiting a higher carbon-sink potential than NF.

Spatially, FL aboveground carbon was concentrated in the middle reaches and eastern plains, where MiF stands predominated, and the upper mountainous areas, where NmF and MaF stands dominated (Figure 5d). NFs were mainly distributed in the upper–middle mountainous regions and dominated by NmF and MaF stands, with scattered over-mature patches (Figure 5e). PFs were widely distributed in the middle–lower reaches and dominated MiF stands (Figure 5f). Overall, recent gains in forestland aboveground carbon and basin-wide carbon sink potential were driven primarily by PFs. The basin’s forest structure remains in an early successional stage, indicating that substantial sink potential has yet to be fully realized and continued support remains for national dual-carbon goals.

3.3.3. Effects of Afforestation on Carbon Storage in the YRB

Statistical analyses revealed clear stage-specific shifts in the spatial patterns of AGC and PF carbon from 2005 to 2020. In terms of SDE metrics (Figure 6d), the ellipse area and oblateness reached minima in 2010, increased thereafter, but declined overall across the full period. This trajectory indicated spatial concentrations up to 2010, followed by progressive diffusion.

Spatially, during 2005–2010 (Figure 6e), the AGC centroid shifted southwest, whereas the SDE semi-major axis shortened, indicating aggregation toward the middle reaches. From 2010 to 2015 (Figure 6f), the GL AGC declined sharply, whereas the forest AGC, especially in PFs, increased. Because GL was concentrated in the upper–middle reaches and forest in the middle reaches, these contrasting changes resulted in a pronounced centroid shift. The migration distance reached 37.22 km (Figure 6a), which was the largest during the study period. From 2015 to 2020 (Figure 6g), a marked increase in GL AGC tempered the tendency of the centroid to migrate northeast.

From 2005 to 2020, the orientation of the PF SDE evolved toward the southwest, and its centroid shifted southwest (Figure 6c). These shifts are consistent with the evolution of AGC, characterized by progressive inward contraction at both ends of the semi-major axis and a slight southward lengthening of the semi-minor axis, underscoring the influence of PF on AGC. Overall, from 2005 to 2020 (Figure 6b), sustained increases in PF carbon across the middle–lower reaches were associated with an eastward displacement of the AGC centroid.

4. Discussion

4.1. Land-Use Reconfiguration and Policy Context

From 2000 to 2020, the YRB underwent substantial land use reconfiguration, with the most intense and structurally complex changes occurring in the middle and lower reaches. FL expanded continuously, primarily at the expense of CL and GL, with particularly strong gains in Shaanxi and Shanxi. This expansion was spatiotemporally consistent with large-scale ecological programs, most notably the GGP, with grazing exclusion, GL restoration, and NF protection program localized reversals from FL back to CL or GL indicating coexisting land-use competition and ecological fragility [2,35,36]. Overall, forest expansion not only reshaped the land system quantitatively but also consolidated the role of forests as the basin’s core carbon sink [2,37,38].

4.2. Carbon Storage Dynamics and Afforestation Effects

From 2005 to 2020, TEC exhibited a rising–declining–rebounding trajectory, primarily driven by climate warming, enhanced precipitation variability, and large-scale land-use changes, including afforestation and CL conversion, in the YRB [39,40,41]. Given the complexity and lag in SOC formation and turnover, AGC more directly captures afforestation effects [35]. Over the same period, AGC increased by approximately 22.19% and the middle–lower reaches progressively emerged as the basin’s carbon sink core [42]. PF-AGC registered a net increase of 1.60 × 10⁸ t, with high-growth belts closely matching the Shaanxi–Shanxi expansion corridor, underscoring the contribution of large-scale ecological projects to national dual-carbon goals [43,44].

4.3. Sources and Impacts of PF Expansion

From 2005 to 2020, the cumulative area converted to PFs totaled 56,191 km², whereas the net PF area increase was 29,500 km², indicating that a considerable proportion of conversions did not persist as net gains, possibly due to land-use adjustments, stand degradation, or conversion to NFs. PF expansion mainly originated from GL, NF, and CL. While this generally strengthened the basin-scale carbon sink, trade-offs may arise, such as increased water consumption and reduced biodiversity when PF replaces NF or when species composition becomes overly uniform [45,46,47,48,49]. Spatially, although PF clusters in erosion-prone, densely populated middle–lower reaches aid ecological restoration and soil–water conservation, indiscriminate large-scale expansion may elevate ecological risks and trigger secondary degradation [50,51,52,53].

4.4. PF-Driven Carbon Gains and Sink Potential

During 2005–2020, forest AGC gains were primarily PF-driven, with PF-AGC increasing by approximately 216.22% and the PF share of forest AGC increasing by approximately 26.70 percentage points [53]. PF was dominated by stands ≤40 years (86.66% of PF area), implying high carbon-sequestration potential. Spatially, middle-aged PFs were prevalent in the middle reaches and eastern plains, whereas near-mature to mature NFs dominated the upper mountainous areas [54]. Management should shift from area expansion to quality-oriented stable sequestration by promoting mixed-species and uneven-aged stands, optimizing stand density and structure, avoiding NF replacement, and implementing zonal and age-class management under water–carbon co-constraints [55].

4.5. Spatial Evolution of Carbon Storage and Planted Forest Role

SDE diagnostics revealed a stage-specific evolution in AGC and PFs from 2005 to 2020. The SDE area and oblateness decreased to a minimum in 2010, increased thereafter, and declined overall across the period, indicating a transition from spatial convergence to the subsequent diffusion of carbon storage [19]. From 2010 to 2015, GL AGC declined markedly, whereas forest AGC, particularly PF-AGC, increased rapidly, driving the largest centroid shift during the study period (37.22 km). This shift reflects the combined effects of GL–forest structural reconfiguration and PF expansion. From 2015 to 2020, a rebound in GL AGC attenuated the ongoing northeastward migration of the centroid [53]. Over the same period, the PF SDE orientation and centroid shifted southwest. Concurrently, the AGC ellipse centroid moved eastward, the semi-major axis contracted, and the semi-minor axis extended slightly southward, indicating that PF expansion was a key driver of AGC spatial reorganization [56].

4.6. Uncertainties and Recommendations Under the Dual-Carbon Strategy

Although the PF dataset used in this study exhibits high classification accuracy (ranging from 77.33 ± 0.67% to 81.78 ± 0.59%) and aligns well with the National Forest Resources Inventory at the provincial level, it still bears notable uncertainties. In addition, rising atmospheric CO₂ concentrations in the YRB may also have contributed to the observed increases in forest carbon stocks [57,58,59]. Future research should integrate UAV-based observations, satellite-derived products, atmospheric CO₂ concentrations, and field-based surveys to obtain more detailed and spatially explicit information on species composition, stand age, and species-specific biomass, thereby improving the accuracy of carbon storage assessment, especially for mixed forests. Such integration would enable a more accurate and comprehensive assessment of plantation conditions, particularly in heterogeneous landscapes. While PFs have demonstrated significant short- to medium-term gains in AGC, their effects on SOC remain insufficiently understood. Existing evidence suggests that mixed-species plantations can enhance SOC by approximately 12% compared to monocultures [60,61]. In northern China, ecosystem carbon sinks are dominated by biomass carbon, accounting for roughly 74% of the total, with SOC contributing about 26%. Thus, failing to account for timber harvesting and forest age-class transitions may lead to systematic overestimation of long-term carbon sequestration potential [62,63]. To support the region’s transition toward carbon peaking and neutrality, afforestation strategies in the middle–lower basin corridor should shift from extensive area expansion to quality-oriented interventions. These include prioritizing species diversity, optimizing stand structure to balance water and carbon benefits, avoiding the replacement of NFs to preserve their long-term sink stability, and strengthening age-class management to accelerate the maturation of carbon-rich stands. Such refinements can help ensure that current carbon gains are sustained over the long term, providing both ecological resilience and strategic value under China’s dual-carbon objectives [60,64,65].

5. Conclusions

5.1. Land-Use Change (2000–2020)

Land-use change in the YRB was substantial, with a cumulative transfer area of 118,481 km². FL accounted for approximately 11.89% of the basin and was concentrated in the middle reaches. The forest area increased continuously, the dynamic degree remained positive, and most conversions into FL occurred in the middle reaches.

5.2. Carbon Dynamics and PF Gains (2005–2020)

TEC followed an increasing–decreasing–rebounding trajectory and posted a net gain of 1.96 × 10⁸ t, with a higher spatial pattern in the west than the east and higher pattern in the south than the north. AGC, which better captures near-term afforestation effects than TEC, increased by 4.37 × 10⁸ t. PF-AGC increased by 1.60 × 10⁸ t, contributing 36.61% of the AGC net gains, with high values and expansion belts concentrated in the middle reaches.

5.3. Sources and Spatial Patterns of PF Expansion (2005–2020)

PFs expanded markedly, with 56,191 km² cumulatively transformed into PFs, sourced mainly from GL (40.51%), NF (35.15%), and CL (23.56%). The expansion was concentrated in the middle–lower reaches, forming belt- and patch-like patterns.

5.4. PF and NF Carbon (2005–2020)

PF carbon increased from 0.74 × 10⁸ t to 2.34 × 10⁸ t, an increase of 216.22%, whereas NF increased by 13.10%. The share of PF in regional forest carbon increased from 19.73% to 46.43%. PFs were dominated by stands ≤40 years and mainly distributed in the middle–lower reaches, indicating higher sink potential than NFs.

5.5. Spatial Evolution of Carbon Storage (2005–2020)

The SDE of AGC and PFs exhibited stage-specific changes. The SDE area and oblateness decreased to a minimum in 2010, increased thereafter, and declined overall across the period. The AGC centroid shifted southwest, and the SDE semi-major axis contracted, indicating aggregation toward the middle reaches. Over the same period of PF expansion, the PF SDE orientation and PF centroid also shifted southwest, corroborating the AGC evolution.