Analysis of the Driving Force of Spatial and Temporal Differentiation of Carbon Storage in Taihang Mountains Based on InVEST Model

Abstract

The Taihang Mountains are an important ecological barrier in China, and their ecosystems have good carbon sink capacity. Studying the spatial-temporal variation characteristics and driving factors of carbon storage in the Taihang Mountains ecosystem provides decision-making for the construction of “dual carbon” projects and the improvement of ecological environment quality in this region. This paper takes the area in the Taihang Mountains as the research area, based on the land use and carbon density data of 2005, 2010, 2015, and 2019 of the Taihang Mountains, calculates the carbon storage in the region with the Integrated Valuation of Ecosystem Services and Trade-offs (InVEST) model, explores the main factors affecting the spatial differentiation of carbon storage in this region, and analyzes their driving mechanisms by Geodetector. The results show that: (1) From 2005 to 2019, the land use of the Taihang Mountains changed somewhat. The area of forest and construction land increased slightly, while the area of farmland and grassland decreased. (2) The current carbon storage in the Taihang Mountains ranges from 1472.91 × 10⁶ t to 1478.17 × 10⁶ t (t is the abbreviation of ton), and shows a decreasing trend, which is due to the decrease in forest and the increase in construction land. (3) Slope and Normalized Difference Vegetation Index (NDVI) are the main driving factors affecting the spatial variation of carbon storage in the Taihang Mountains ecosystem. Temperature, precipitation, and population density are the secondary factors affecting the spatial variation of carbon storage. (4) The synergy between the driving factors is more potent than the individual factor, which is the most evident between NDVI and slope. This means some areas may have more abundant carbon storage under the combined effect of slope and NDVI.

1. Introduction

The current research showed that “carbon peak and carbon neutralisation” (double carbon) is an effective way to deal with climate change and reduce carbon emissions [1,2,3,4]. In the general debate of the 70th session of the United Nations General Assembly, China proposed “dual carbon” construction goals and committed to reducing the adverse effects of carbon emissions. Enhancing the carbon storage potential of ecosystems is an important way to reduce carbon emissions [5] and achieve the goal of “double carbon”. The Taihang Mountains are an essential ecological barrier in Eastern China, and their complex landforms and variable climatic conditions are of ecological importance [6,7]. The 14th Five-Year Plan proposed that the Yanshan-Taihang Mountains Ecological Containment Area is an important national functional area in China; its ecological significance includes water and ecological protection. However, in recent years, carbon emissions in the Taihang Mountains have increased sharply [8], and the ecological environment is also relatively fragile [9]. How to reduce emissions and rebuild the ecosystem scientifically is the primary task of ecological construction in Taihang Mountain. Tasks include quantitative assessment of the impact of driving factors on the spatial distribution of carbon stocks in the Taihang Mountains to explore scientific ways to improve the carbon stocks of regional ecosystems, which is of great significance for improving the Taihang Mountain ecosystem and reducing carbon emissions.

At present, the methods of studying carbon storage in ecosystems are mainly classified as field survey [10], remote sensing inversion [11], and model simulation [12]. Field survey can be used to estimate carbon storage data accurately, but it has a smaller scope and can be disruptive to the research environment when collecting data [13,14]. Remote sensing inversion can be used for large-scale carbon storage research, mainly based on specific ecosystems and partial carbon data for research [15,16,17]. Model simulation methods can effectively estimate, predict, and evaluate carbon storage at various scales by estimating regional carbon storage with the help of FORECAST, HASM, and other models. Currently, Integrated Valuation of Ecosystem Services and Trade-offs (InVEST) model is the most widely used model, which has less data, fast calculation speed, and accurate estimation results. Scholars estimate and evaluate carbon storage from different perspectives [18,19,20] and scales [21,22] through the InVEST model. Scholars have used the InVEST model to study carbon storage assessment in different regions from different perspectives and scales [23,24]. Li analyzed the variation rules of carbon density and storage in the Taihang Mountains under different gradients of driving factors through the InVSET model [25]. However, there are relatively few studies on the driving mechanism of the spatial distribution of carbon stocks in ecosystems. Studying the effect of driving factors on the spatial distribution of carbon storage in ecosystems is a new method to study its driving mechanism, which is more scientific and accurate.

In order to explore the spatio-temporal variation of carbon storage and its driving factors in the Taihang Mountains, based on the land use data of this region in 2005, 2010, 2015, and 2019, this paper analyzed the distribution of carbon storage in this region through the InVEST model, and identified the main driving factors causing the variation through Geodetector. This study can provide a basis for the policy making of ecological restoration and land use adjustment in the Taihang Mountains.

2. Materials and Methods

2.1. Overview of the Study Area

The Taihang Mountains are located 110°39’–116°55’ east longitude and 34°91’–40°65’ north latitude (Figure 1), which spans the four administrative regions of Hebei, Beijing, Henan, and Shanxi, with a total area of about 13.58 × 10⁴ km². The average annual temperature in the study area is 10.3 °C, and the average annual precipitation is 501.8 mm, which belongs to the warm temperate continental monsoon climate. The climate is characterized by shorter spring and autumn seasons, more precipitation in summer, and cold and dry winters [26]. There are many types of land use in the Taihang Mountains. The eastern and southern parts are mainly cultivated land, the northern and western areas are mainly grassland and forest, and the construction land is mainly concentrated in the provincial capital and surrounding cities. Different land use types and vegetation structures have a great influence on ecosystem carbon storage and carbon sink capacity [27]. The area in the Taihang Mountains is rich in vegetation resources such as forest and grassland, with good ecological functions, and is the primary source of carbon storage in North China.

2.2. Data Sources

At the annual scale, the spatial and temporal changes of ecosystem carbon storage are relatively weak. Therefore, this paper takes five years as the time interval to select the data of the study area and selects the land use image data of the four time points of 2005, 2010, 2015 and 2019. The land use data of the areas in the Taihang Mountains in 2005, 2010, 2015, and 2019 were obtained from the Resource and Environment Science and Data Center (https://www.resdc.cn/ (accessed on 28 January 2022)). According to the research needs, this paper took satellite remote sensing images as the data source. It classified the land use types into six types: arable land, grassland, forest, water, construction land, and unused land through visual interpretation. The carbon density data of this study came from relevant data of existing references [22,25], and the data were revised according to the research needs to obtain the carbon density of various land use in the Taihang Mountains (

Table 1. Carbon density of various land use types in the Taihang Mountains (t·km⁻²).

Type	Cabove	Cbelow	Csoil	Cdead
farmland	2.19	0.42	90.16	0.00
grassland	38.99	7.80	103.98	1.90
forest	0.65	3.38	83.69	0.10
water	6.38	0.00	170.13	0.00
construction land	5.63	0.00	68.99	0.00
unused	0.00	0.00	0.00	0.00

Note: The data were obtained with corrections based on literature 22 and 25. C_above: all living organisms on the surface; C_below: all living organisms below the surface; C_soil: in mineral and organic soils; C_dead: all dead organic matter.

).

Considering the natural conditions of the Taihang Mountains and referring to relevant literature [28,29,30,31,32,33,34], this paper selects elevation, slope, Normalized Difference Vegetation Index (NDVI), temperature, precipitation, soil texture, population density, and GDP as driving factors for spatial differentiation of regional carbon storage. Elevation data were obtained from the geospatial data cloud platform (https://www.gscloud.cn/ (accessed on 18 March 2022)) with a resolution of 90 m. Slope data were calculated and extracted by ArcGIS software. The NDVI of each period was obtained from the Resource and Environment Science and Data Center (https://www.resdc.cn/ (accessed on 20 March 2022)), which was transformed by Mosaic, Projection, with a resolution of 1 km. The annual average temperature and precipitation data of the Taihang Mountains were collected from 390 meteorological stations in various periods, all of which were obtained from the National Meteorological Science Data Center (http://data.cma.cn/ (accessed on 18 March 2022)). These data were interpolated separately using ANUSPLIN software to obtain the spatially interpolated data of annual mean temperature and annual precipitation for each period. The soil data were compiled according to the 1:1 million soil types map and the second soil census data. The data included three types: sand, silt, and clay, and the value was the percentage. Population density data were downloaded from WorldPop (https://www.worldpop.org/ (accessed on 31 March 2022)) and released by the world map of population density. The population density datasets for each period were extracted from these downloaded population data by ArcGIS software with a resolution of 30 m. The GDP data came from the “China GDP Spatial Distribution Kilometer Grid Dataset” product under the Resources and Environment Data Center (https://www.resdc.cn/ (accessed on 20 March 2022)) of the Chinese Academy of Sciences, with a resolution of 30 m.

2.3. Research Methods

2.3.1. InVEST Model

The InVEST model, jointly developed by Stanford University, the Nature Conservancy, and the World Wide Fund for Nature, is an open source ecosystem service function assessment model, which can be well used to assess regional carbon storage and its value [35]. This paper uses the carbon storage module in the InVEST model to analyze the carbon storage changes in terrestrial ecosystems in the Taihang Mountains between 2005 and 2019, which classified the carbon density of each land use type as four primary carbon pools: (1) Aboveground living biomass(C_above): All living organisms above the soil. (2) Belowground living biomass(C_below): All living organisms below the soil. (3) Soil carbon(C_soil): the organic carbon of mineral soils. (4) Dead organic matter: dead organic matter on the soil, includes litterfall, fallen or standing dead wood. According to the distribution of land use types in the Taihang Mountains, the InVEST model analyzes and calculates the average carbon density (C_total) of various land use types. The specific formulas are as follows:

$${\mathrm{C}}_{\mathrm{i}}={\mathrm{C}}_{\mathrm{i},\mathrm{above}}+{\mathrm{C}}_{\mathrm{i},\mathrm{below}}+{\mathrm{C}}_{\mathrm{i},\mathrm{soil}}+{\mathrm{C}}_{\mathrm{i},\mathrm{dead}}$$

(1)

$${\mathrm{C}}_{\mathrm{total}}={\displaystyle \sum _{\mathrm{i}=1}^{\mathrm{n}}{\mathrm{C}}_{\mathrm{i}}\times {\mathrm{S}}_{\mathrm{i}}}$$

(2)

i is the type of land use in type i; C_i is the total carbon density of i land use type(t·hm⁻²); C_i,above ground carbon density (t·hm⁻²) of i land use type; C_i,below is the subsurface carbon density of i land use type (t·hm⁻²); C_i,dead is the carbon density of dead organic matter of i land use type (t·hm⁻²); S_i is the area of i land use type(m²); C_total is the total carbon storage of terrestrial ecosystems (t).

2.3.2. Random Forest

The principle of random forest method is to use the importance evaluation method of random forest to analyze the importance of each driving factor and screen out the driving factors with the strong influence on the spatial differentiation of carbon storage. The random forest method [36] was used to further screen environmental factors (slope, DEM, NDVI), climate factors (precipitation, temperature), soil factors (clay, sand, and silty soil content), and human disturbance factors (population density, GDP). After random forest calculation, the importance ranking results of driving factors are shown in Figure 2. Further, the driving factors with less than 30% importance (clay content, silt soil content, GDP) were removed, and the driving factors with higher importance (slope, NDVI, population density, air temperature, precipitation, DEM, sand soil content) were screened.

2.3.3. Geodetector

Geodetector is a new statistical method proposed by Prof. Jinfeng Wang, which is used to detect the spatial differentiation among geographical objects and analyze the driving factors of their spatial [37]. Geodetector can be used not only to detect the spatial differentiation of driving factors, but also to detect the degree of interaction between two factors on dependent variables. In this paper, factor detection and interaction detection of Geodetector are used to analyze the driving factors and their interactions affecting the spatial differentiation of carbon storage in the Taihang Mountains.

(1) Factor detector: Factor detector is used to detect the spatial differentiation of Y; and to detect how much a specific factor X explains the spatial differentiation of attribute Y. Factor detection results are measured by q value, whose expression is:

$$\mathrm{q}=1-\frac{{\displaystyle \sum _{\mathrm{i}=1}^{\mathrm{I}}{\mathrm{n}}_{\mathrm{i}}{\mathsf{\delta }}_{\mathrm{i}}^2}}{\mathrm{n}{\mathsf{\delta }}^2}$$

(3)

q is the explanation intensity of driving factor X to Y, and the range is (0–1). i is the amount of carbon storage in the study area; n_i and n are the number of samples in type i research area and the total number of samples in the research area, respectively. δ_i² and δ² are the variances of carbon storage in type I study area and the variance of overall carbon storage in the study area, respectively.

(2) Interaction detector: interaction detector can analyze whether there is an interaction among the driving factors and the intensity of the interaction. It can analyze the joint effect between different driving factors on the dependent variable Y. The interaction between driving factors is shown in

Table 2. Types of interactions.

Description	Interaction
q(x1∩x2) < Min[q(x1), q(x2)]	Weaken, nonlinear
Min[q(x1), q(x2)] < q(x1∩x2) < Max[q(x1), q(x2)]	Weaken, uni-
q(x1∩x2) > Max[q(x1), q(x2)]	Enhance, bi-
q(x1∩x2) = q(x1) + q(x2)	Independent
q(x1∩x2) > q(x1) + q(x2)	Enhance, nonlinear

Note: The table was compiled based on Reference [37].

.

3. Results

3.1. Land Use Change Results

By analyzing the spatial distribution and changes of land use in the Taihang Mountains from 2005 to 2019 (Figure 3,

Table 3. Land use change in the Taihang Mountains from 2005 to 2020 (unit: area-km², proportion-%).

Type	2005		2010		2015		2019		Area Change Rate from2005 to 2019
	Area	Pct	Area	Pct	Area	Pct	Area	Pct
farmland	50,460.70	37.26%	50,224.40	37.08%	49,980.6	36.90%	47,133.4	34.81%	−2.45%
grassland	39,485.90	29.15%	39,527.70	29.18%	39,505.3	29.17%	40,027.8	29.56%	0.41%
forest	37,143.70	27.42%	36,922.40	27.26%	36,877.5	27.23%	35,522.1	26.23%	−1.19%
water	1981.70	1.46%	1956.27	1.44%	1960.76	1.45%	1853.44	1.37%	−0.09%
construction land	6169.43	4.56%	6607.65	4.88%	6914.25	5.11%	10,704.8	7.91%	3.35%
unused land	198.92	0.15%	201.909	0.15%	201.909	0.15%	163.055	0.12%	−0.03%

Area: Land use type area. Pct: Percentage of land use type area.

), it can be seen that the types of land use in the region have evident spatial heterogeneity on the whole, and the variation range of land area of each land use type is small during the monitoring time.

In terms of spatial distribution, the land use types were classified as farmland, forest, grassland, water, construction land, and unused land in the Taihang Mountains. Farmland was the largest land use type, between 34.81% and 37.26% in the Taihang Mountains, mainly distributed in the low altitude areas of the Taihang Mountains, distributed in small amounts in the mid-altitude areas such as Jincheng and Changzhi. It was followed by forest and grassland which occupy the second and third place, respectively, in the region, mainly distributed in the higher altitude mountain and hilly areas of the Taihang Mountains. Construction land was located in the fourth place with a relatively low percentage between 4.56% and 7.91%, mainly distributed in the central cities of the provinces and larger cities. Water and unused land were the two types with the lowest proportion of all land use types. Water was distributed in the mid and low elevation areas of the Taihang Mountains, such as larger lakes and rivers. Unused land was mainly bare land and saline land with a small amount distributed in the Taihang Mountains.

From the perspective of the time change, there was a small range change in land use types in different periods in the Taihang Mountains. Between 2005 and 2019, the area of forest and construction land increased slightly. Among them, construction land showed the largest increase, mainly in cities and surrounding areas. The water area change showed a characteristic of decreasing first and then increasing. At the same time, the land use area of arable land, grassland, and unused land showed a feature of decreasing year by year, among which the most decreased was arable land.

3.2. Spatial and Temporal Distribution of Carbon Storage

According to the spatial distribution of carbon storage (Figure 4), there was apparent spatial heterogeneity of carbon storage in the Taihang Mountains. Combined with the analysis in Figure 2, the overall distribution of carbon storage was closely related to land use types, which showed that the high-value areas were mainly located in the areas with high vegetation covers, such as forest and grassland, while the low-value areas were mainly in the areas with weak carbon sink capacities such as arable land and construction land. The specific distribution position of carbon storage was mainly high in the northeast and southwest and scattered in other regions.

From the changes in carbon storage of land use types (

Table 4. Changes in carbon storage of various land use types in the Taihang Mountains (×10⁶ t).

Type	2005–2009		2010–2014		2015–2019		Total
	Storage	Pct	Storage	Pct	Storage	Pct	Storage
farmland	−2.19	−0.13	−2.26	−0.14	−26.41	−1.71	−30.87
grassland	0.64	0.06	−0.34	−0.01	7.98	0.65	8.27
forest	−1.94	−0.12	−0.39	−0.02	−11.90	−0.75	−14.24
water	−0.45	−0.03	0.08	0.01	−1.89	−0.12	−2.26
construction land	3.27	−0.13	2.29	−0.14	28.29	−1.71	33.84
unused land	0.00	0.06	0.00	−0.01	0.00	0.65	0.00
Total	−0.68	−0.29	−0.63	−0.31	−3.95	−2.99	−5.26

Storage Area: Carbon storage by land use type. Pct: Percentage of carbon storage by land use type.

), the carbon storage in the Taihang Mountains showed a decreasing trend year by period. From 2005 to 2019, the carbon storage in the Taihang Mountains decreased by −5.26 × 10⁶ t overall, with the most significant decrease of −3.95 × 10⁶ t in 2015–2019. In terms of land use types, farmland and grassland decreased by −30.87 × 10⁶ t and −14.24 × 10⁶ t during the whole period, respectively. Arable land was the main land use type with reduced carbon storage. Forest and construction land increased by 8.27 × 10⁶ t and 33.84 × 10⁶ t, respectively. Construction land was the land use type with the most significant increase in carbon storage.

From the spatial distribution of carbon storage in different periods (Figure 5), the changes from 2005 to 2009 were mainly in most areas of Hebei, Henan, and Beijing, with weak changes and random distributions. Carbon storage in Shanxi remained unchanged. From 2010 to 2014, carbon storage mainly showed a slight decrease. From 2015 to 2019, the changes in carbon stocks were relatively noticeable, but those changes were randomly distributed and had no apparent regularity.

3.3. Spatial and Temporal Variation Drivers of Carbon Storage

The factor detection results of the Geodetector (Figure 6) showed that altitude, slope, NDVI, temperature, and other factors have different degrees of influence on the spatial distribution of carbon storage in ecosystems of the Taihang Mountains. Figure 6 showed that slope and NDVI were the main driving factors affecting the spatial differentiation of carbon storage in ecosystems. Their explanatory power was significantly greater than that of other factors. This indicated that during this period, the change in the geographical environment and the dense degree of vegetation dominated the change in carbon storage in the region to a certain extent. The influence of other driving factors on the spatial variation of carbon storage in the Taihang Mountains was relatively stable. Except for the driving factor of sand content, the q values of all factors were greater than 0.05, and the explanatory power was more than 5%, indicating that these factors also played a certain role in the spatial differentiation of carbon storage. The factor of sand content had little explanatory power for the spatial variation of carbon storage in the region. The q value was less than 0.05, and the explanatory power was less than 5%, indicating that it had little influence on the spatial distribution of carbon storage in the Taihang Mountains.

Table 5. Carbon density of various land use types in the Taihang Mountains (t·km⁻²).

(a) Interaction detection results of various driving factors in 2005
	Slope	NDVI	Population density	Temperature	Precipitation	DEM	Sand content
Slope	0.145263
NDVI	0.261068	0.107138
Population density	0.175947	0.18995	0.058719
Temperature	0.19432	0.213544	0.100619	0.066508
Precipitation	0.212274	0.182488	0.166709	0.181148	0.079104
DEM	0.197217	0.237445	0.113396	0.149461	0.195397	0.079148
Sand content	0.175611	0.138308	0.081373	0.106838	0.118021	0.114162	0.013254
(b) Interaction detection results of various driving factors in 2010
	Slope	NDVI	Population density	Temperature	Precipitation	DEM	Sand content
Slope	0.146408
NDVI	0.295082	0.173866
Population density	0.174676	0.239022	0.070465
Temperature	0.195646	0.257381	0.098310	0.064256
Precipitation	0.201914	0.223937	0.140304	0.174254	0.074503
DEM	0.202309	0.266731	0.106707	0.157744	0.182107	0.080421
Sand content	0.175562	0.197178	0.087353	0.107572	0.110939	0.116426	0.014702
(c) Interaction detection results of various driving factors in 2015
	Slope	NDVI	Population density	Temperature	Precipitation	DEM	Sand content
Slope	0.147723
NDVI	0.271326	0.162433
Population density	0.168483	0.192730	0.046709
Temperature	0.206462	0.212121	0.096644	0.076255
Precipitation	0.221082	0.221373	0.130925	0.224103	0.047442
DEM	0.2037	0.218641	0.104872	0.143506	0.229366	0.081873
Sand content	0.176818	0.185499	0.062399	0.114538	0.091057	0.117868	0.014741
(d) Interaction detection results of various driving factors in 2019
	Slope	NDVI	Population density	Temperature	Precipitation	DEM	Sand content
Slope	0.145575
NDVI	0.277027	0.207217
Population density	0.173123	0.232681	0.055401
Temperature	0.208208	0.260988	0.103817		0.081079
Precipitation	0.196302	0.249764	0.116512	0.171902	0.171902	0.050695
DEM	0.197016	0.263565	0.113819	0.159376	0.159376	0.186644	0.091127
Sand content	0.171153	0.224737	0.071306	0.114256	0.114256	0.082632	0.119960

Enhance, nonlinear Enhance, bi-.

showed the detection results of the pairwise interactions of seven factors for the spatial differentiation of ecosystem carbon storage. The pairwise interaction relationship of these factors was nonlinear enhanced or bivariate enhanced, indicating that any factor combined with other factors can enhance the influence on the spatial differentiation of carbon storage. In each period, the explanatory power of slope combined with NDVI type ranged from 0.2610 to 0.2950. This type had the most explanatory power among the synergistic types of the spatial differentiation of carbon storage in each ecosystem, indicating that the regions with good slope and NDVI values may have more carbon storage. Most of the synergistic types between NDVI and other factors were nonlinear enhancement types, which further proves that NDVI was the main driving factor affecting the spatial differentiation of carbon storage. There was a nonlinear enhancement type between sand content factor and other factors in all years, indicating that the effect of sand content factor combined with other factors was far better than that of the effect alone. In short, the synergistic effect among these driving factors was not a simple superposition relationship, which significantly impacted the spatial differentiation of carbon storage.

4. Discussion

4.1. Changes of Carbon Storage

The total carbon storage in the overall ecosystem of the Taihang Mountains decreased year by year during 2005–2019. Specifically, the carbon storage of all land use types showed slight changes in different periods, with the rate of change ranging from 0.06% to 3%. In terms of land types, the area of grassland and arable land dramatically reduced, while the area of construction land significantly increased. This indicates that the policy of “returning farmland to the forest” [38] implemented in the Taihang Mountains area in this period achieved a good effect, and also indicates that the urbanization development in this period has generated a greater demand for construction land [39]. Therefore, the increase of construction land area and arable land area, and the decrease of grassland area and forest area will weaken the carbon sequestration capacity of the ecosystem in the region, which is consistent with the research results of spatio-temporal changes in carbon storage in other regions [40,41,42].

4.2. Driver Factors of Spatial Differentiation of Carbon Storage

The analysis results of Geodetector showed that slope and NDVI are the main influencing factors in the spatial differentiation of carbon storage in the Taihang Mountains. This result is consistent with Miaoyu Liet al.’s [43] conclusion that NDVI is the dominant factor affecting the spatial distribution of ecosystem carbon storage in the Loess Plateau region through path analysis. However, compared with Geodetector, the path analysis method is simple in calculation and straightforward in results, but it requires that the degree of collinearity between independent variables not be too high. Olorunfemi [44] studied carbon storage in different land use types in Ekiti State, Southwestern Nigeria, and found that about 50% of the carbon storage is contained in natural forests with high vegetation coverage; Eid [45] studied the effect of different vegetation cover on soil organic carbon stocks, which showed that areas with vegetation cover > 75–100% have higher carbon storage. Compared with other land types (such as arable land and construction land), high-coverage vegetation (such as forest and grassland) has better carbon sequestration capacity. Therefore, priority can be given to increasing regional vegetation coverage for an effective carbon sink. In addition, Buraka [46] studied that the land use type along the slope has an obvious effect on carbon density; Cheng [47] established the soil database of Anhui Province through GIS technology and found that the carbon density increased with the increase of slope; Hamere [48] studied the change in carbon storage along the gradient of the slope in the Geduo forest and concluded that the slope has a significant impact on litter carbon storage. These studies illustrate that slope is an important factor which greatly affects the spatial distribution of vegetation and soil carbon density. Based on the interaction detection results of Geodetector, it can be found that the combination of different driving factors can enhance the influence on the spatial differentiation of carbon storage, and the effect of slope and NDVI type was the strongest. Therefore, it is necessary to consider the impact of different geographical conditions and environmental factors on regional carbon sequestration and strengthen the vegetation coverage in higher altitude areas when implementing the “dual carbon” construction in the future. For example, in the measures to improve the carbon storage of the regional ecosystem in the Taihang Mountains, priority should be given to the role of slope and vegetation coverage, focusing on strengthening the surface vegetation in the high elevation area, increasing the forest area on the whole, and controlling the expansion of construction land in the area with high vegetation coverage.

4.3. Deficiencies and Prospects

At present, scholars’ research methods for influencing factors of carbon storage spatial differentiation [49,50,51,52,53] mainly adopt qualitative analysis, while this paper attempts to quantitatively study the driving factors of carbon storage using Geodetector, which provides a new idea for the study of carbon sinks in ecosystems. It is certain that this paper also has the following shortcomings: (1) The number of some driving factors is insufficient. For example, only sand, clay, and silt content were selected as soil texture factors. However, the driving mechanism of soil factors for carbon sequestration in the ecosystem is very complex, so it may be necessary to select more comprehensive soil factors. (2) The results obtained from calculating driving factors using Geodetector do not pay attention to the direction of action of driving factors. Therefore, it is necessary to strengthen further the influence of soil factors on ecosystem carbon sink and study driving factors using spatial statistical analysis methods in future studies.

5. Conclusions

The carbon storage in the Taihang Mountains has changed from 2000 to 2019, mainly due to the changes in land use types; environmental factors such as slope and NDVI dominated the spatial distribution of carbon storage in the region. The concrete conclusions are as follows:

(1) From 2005 to 2019, the land use types in the Taihang Mountains changed to a relatively small extent. During the whole study period, the land area of forest and construction land increased slightly, and the land use area of cultivated land and grassland decreased by period.

(2) From 2005 to 2019, the total value of carbon storage in the Taihang Mountains area ranged from 1472.91 × 10⁶ t to 1478.17 × 10⁶ t, showing a decreasing trend. From the perspective of spatial distribution, the carbon storage mainly decreased from the high altitude and lush vegetation areas to the surrounding areas. From the perspective of land use type, forest accounted for the largest proportion of carbon storage in the region, followed by farmland, grassland, construction land, water area, and unused land. Combined with land use change, the decrease of forest land and the increase of construction land are the main reasons for the decrease in carbon storage in the Taihang Mountains.

(3) According to the factor detection results, the spatial variation of ecosystem carbon storage in the Taihang Mountains is mainly affected by topographic and environmental factors. Slope and NDVI have a significantly greater impact on the spatial variation of carbon storage in the region than other factors, which were the main driving factors affecting the spatial variation of carbon storage in the Taihang Mountains ecosystems. Other factors (e.g., temperature, sand content, etc.) also played a role in the region’s spatial variation of carbon storage. According to the interaction detection results, the interactions among the drivers are more robust than those of a single factor on the spatial variation of carbon storage. Among them, the NDVI synergistic slope type had the most potent synergistic effect, indicating that the area under the synergistic effect of slope and NDVI have more abundant carbon storage.