Establishment and Application of Biomass Model for Vegetation Condition Assessment After Ecological Restoration—Yixing Quarry Case Study

Abstract

Biomass is a vital index used to evaluate the vegetation rebuilding effect of mining slopes after restoration. It is essential to establish models for estimating the biomass and carbon storage of the vegetation community on mining slopes. Therefore, this paper establishes models for the biomass and carbon storage of such vegetation, taking an abandoned quarry after ecological restoration in Yixing City, Jiangsu Province, as the research object. Firstly, the variables of the biomass estimation models were determined based on the correlation analysis results; the vegetation biomass model was comprehensively selected, and the accuracy of the optimal models was verified. Meanwhile, the carbon storage calculation model was established in combination with the carbon content and the growth pattern of vegetation. The results showed that (1) the optimal models were the cubic and linear functions, respectively, for the shrubs and herbs, while the relevant variables of the shrub and the herb plants were the average height multiplied by the diameter of each shrub plant (DH) and the average height multiplied by the coverage rate (CH), respectively, with the verification results of R² > 0.814, R_S > 2.8%, and R_MA > 6%; and (2) in the restored mining slopes, the vegetation biomass was 120.264 t, including 10.586 t of herbs and 109.678 t of shrubs, and the vegetation carbon storage was 50.585 t, including 3.705 t of herbs and 46.880 t of shrubs. The proposed models have good prediction accuracy and reliability after quantitative evaluation and can be applied to the biomass estimation and carbon storage calculation of restored mining slopes, providing a reference for the environmental sustainability of post-mining areas and other ecologically restored slopes.

1. Introduction

About 80% of mineral resources are developed in the form of open-pit mining in China, leading to exposed rock slopes with infertile soil, sparse vegetation, etc. [1,2]. Although vegetation rebuilding on mining slopes can be affected by the slope conditions, such as the height and slope angle [3,4], more attention is paid to the stability and sustainability of mining ecosystems while open-pit mines in China are under restoration [5,6]. Considering that vegetation rebuilding is one of the key factors in the ecological restoration of mines [7,8], the model establishment of the biomass estimation and carbon storage calculation of vegetation rebuilt on mining slopes is of great practical significance to evaluate the vegetation rebuilding effect.

Ecological restoration techniques for mining slopes have gradually been combined with engineering measures, considering the impact of the slope rock mass quality [4,9] and the post-mining ecosystem situation [10,11]. Nowadays, vegetation techniques for mining slopes, such as vegetation concrete and planting bags [12,13], are commonly used, and a polymer stabilizer is applied to improve the stability of the external soil on the rock slopes, providing conditions for better vegetation growth [14,15,16]. Also, scholars have evaluated the effect of the vegetation community rebuilt on mining slopes from different perspectives. Gastauer and Zhang et al. studied the ecological succession patterns, vegetation characteristics, and community stability to enhance slope revegetation and rehabilitation [17,18,19]. Shang, Shen, and Lopez-Marcos et al. discussed the impact of slope conditions on vegetation restoration, such as the slope dip and slope aspect [20,21,22]. Zou, Mounsey, and Zhou et al. evaluated the ecological restoration effect of vegetation on mine slopes using remote sensing, environmental driver analysis, and the AHP-FCE method [23,24,25].

Among the vegetation types, shrubs and herbs are widely adopted due to their characteristics of high biomass and wide ecological adaptability [26,27]. As vegetation biomass is important in assessing vegetation resilience in restored mines, some scholars have studied and estimated it. Hu and Kumar studied the biomass characteristics of herbs and shrubs at different stand years in a restored mining area [28,29], while mechanistic models, satellite data from Landsat images, etc., were combined to estimate vegetation biomass [30,31,32]. The interrelationship of vegetation biomass and soil properties in mining areas is discussed through qualitative analysis [33], soil micromorphological techniques [34], and vegetation ecological indexes [35,36]. Further, research related to the carbon sequestration capability and carbon storage provides different methods to evaluate vegetation’s carbon sequestration potential, using quadrat surveys and field studies [25,37,38], LiDAR and hyperspectral data [39], particle swarm optimization algorithms [40], etc.

According to the literature review, studies related to vegetation rebuilding on mining slopes mainly concentrate on engineering measures, vegetation community stability, etc. As for the carbon sequestration capability, the direct measurement method is mainly used for long-term stand vegetation, while the remote sensing method used for large mining areas lacks accuracy verification. Due to the unevenly distributed vegetation and complex slopes in small-scale mining areas, the direct measurement method can provide accurate vegetation structural data to elucidate the situation of the vegetation community via field investigation.

The Yuguotang Quarry was taken as the research object, considering the fact that it underwent open-pit mining, leading to serious surface landscape destruction and soil erosion, and it was restored with related measures with a relatively good vegetation restoration effect. Therefore, based on the field investigation and laboratory tests, the main aims of this paper are as follows: firstly, we aim to establish models for biomass estimation through vegetation structure data and correlation analysis results, with the verification of the model accuracy; in addition, we seek to determine the carbon storage calculation models of the vegetation according to the carbon content and the growth pattern of the vegetation; moreover, we aim to calculate the biomass and carbon storage of the vegetation on the restored mining slopes. This study can provide ideas and guidelines for establishing biomass and carbon storage models for the vegetation communities on restored mine slopes and other similar forms of engineering in slope restoration, with practical significance for the environmental sustainability of post-mining areas.

2. Study Area

2.1. General Introduction of Study Area

This paper takes the region within the mining influencing area of the abandoned open-pit limestone quarry, the Yuguotang Quarry, as the research object; this area has suffered damage from mining activities, and has the conditions for conducting the ecological restoration measurements. The Yuguotang Quarry is located in the Xinjie District, in the northeastern part of Zhangzhu Town, Yixing City, Jiangsu Province, China (119°41′32.49″ E and 31°18′11.12″ N), where the terrain is low-mountain and hilly. The original landform has been destroyed (Figure 1), leading to an obvious height difference and the formation of residual hills, owing to mining activities such as local mining and later filling. The maximum elevation of the quarry is 141.0 m and the lowest elevation is 41 m, with a height difference of 74 m. The quarry has an irregular shape, with the mining influencing area being 750 m in length from north to south and 800 m in width from east to west, with a total area of 27.29 hm². According to the topographic analysis, the quarry can be divided into slope area and wasteland, with 22.56 hm² of wasteland and 4.74 hm² of slope area.

The area is located on the east side of the north wing of the Zhangzhu syncline, where the monocline and the strata generally tilt to the south. The tectonic movement is not strong and fault structures are not developed in the region. The formation lithology of the quarry is relatively uniform, mainly including the Lower Triassic strata Qinglong Group stratum (T₁q) and the Quaternary residual soil (Q₄^el), consisting of medium-thin layered limestone and dolomitic limestone, interbedded with mudstone limestone, and residual soil with the general thickness less than 2 m. Due to the lithology, the mining object in the quarry is limestone, for building materials, and the open-pit mining method is used. The groundwater in the area can be divided into (1) pore water in the quaternary soil, (2) karst water, and (3) bedrock fissure water in the limestone strata. The main source of groundwater is the infiltration of atmospheric precipitation.

Yixing City belongs to a subtropical southern monsoon climate with abundant rainfall. The rainfall in the area is mostly concentrated from June to September, accounting for about 50% of the annual rainfall. In Yixing City, the annual average precipitation is 1199.5 mm, with a maximum daily rainfall of 1738 mm and an average rainfall during the flood period of 584.1 mm.

According to the field investigation data before restoration, the land use of the post-mining quarry can be concluded; it is mainly composed of mining land and a pond (Figure 2). The rock slope after excavation is steep, with a slope angle of 60–80° and a relatively large cutting range, while the slope top exceeds the watershed of the original landform. Also, there is a pond with water accumulation after mining.

The vegetation community in the area mainly consists of evergreen broad-leaved forests. The undamaged area surrounding the mining area has good tree development, with species such as camphor, poplar, and elm, shown in the forest land in Figure 2. Due to mining activities, the vegetation in other forest land has been damaged to a certain extent, where the vegetation community is composed of primitive vegetation with lower coverage and growth density compared to the forest land. The wasteland around the mining area developed with naturally growing herbs during a period after the mining activities were completed, as shown by the grassland in Figure 2. However, the original vegetation community within the quarry was almost destroyed, causing exposure of the rock slope surface with scattered plant development.

2.2. Ecological Assessment of Abundant Quarry

The slopes (type A) of the quarry are divided into several detailed areas, considering the slope conditions and vegetation status (Figure 3 and Figure 4). According to the results of the field investigation before the ecological restoration, the information and description of each slope area are shown in

Table 1. Description and information of the detailed slope areas before restoration.

Slope Area	Slope Description	Slope Stability	Vegetation Status	Projected Area (m2)	Surface Area (m2)
A1	Circular arc shape on plane and steep or ladder shape in vertical direction, slope angle of 65–75°, slope direction of 87°, slope length of 250 m, slope width of 15–20 m, maximum slope height of 40 m	Good stability, developed joints, and local collapse	Completely exposed slope surface	5714	6313
A2	Nearly linear shape on the plane and steep in the vertical direction, slope angle of 60–70°, slope direction of 312°, slope length of 120 m, slope width of 7–16 m, maximum slope height of 40 m	Stable, developed joints and local collapse	Completely exposed slope surface	4623	5083
A3	Flat terrain with semicircle shape on plane, slope angle of 60–70°, elevation of 128–130 m	/	Completely exposed surface	947	1738
A4	Circle shape on the plane and steep in the vertical direction, slope angle of 60–65°, slope direction of 30°, slope length of 750 m, slope width of 20–37 m, maximum slope height of 50 m	Good stability, developed joints, loose rock mass, and local collapse	Completely exposed slope surface	26,155	34,746
A5	Nearly linear shape on the plane and steep in the vertical direction, slope angle greater than 65°, slope direction of 230°, slope length of 105 m, slope width of 50 m, maximum slope height of 35 m	Poor stability, developed joints, and loose rock mass	Completely exposed slope surface	5880	6111
A6	Nearly linear shape on the plane and gentle in the vertical direction, slope angle of 40°, slope direction of 329°, slope length of 115 m, slope width of 10 m, maximum slope height of 10 m	Good stability, little deposit on the slope surface	Affected by original vegetation, vegetation developed on the slope surface	1055	1377
A7	Nearly linear shape on plane and gentle in vertical direction, slope angle of 30–50°, slope direction of 135°, slope length of 150 m, slope width of 10–20 m, slope height of 3–9 m	Good stability, little deposit on the slope surface	Affected by original vegetation, vegetation developed on the slope surface with severe soil erosion	1857	2145

.

2.3. Ecological Restoration Measures

According to the description in Table 1 and the diagram of each slope area in Figure 4, it can be concluded that the terrain, landscape, land resources, and vegetation community were severely damaged, while potential geological hazards developed. Therefore, the treatment measures were taken for the quarry slope ecological restoration, considering the sustainable development of the environment and ecosystem of the post-mining area. The ecological restoration of the quarry began in June 2022, with a construction period of 9 months.

The quarry slope ecological restoration can be divided into slope reinforcement and vegetation–soil system rebuilding.

(1)

Slope reinforcement

The treatment measures for slope reinforcement include slope cutting, protective net rainfall interception, and a drainage system. After slope cutting, the slopes were 45°, with 2–3 levels of slope surface, and the steps had a width of 4 m. To avoid the threat of falling gravel on the slopes, protective nets were set. To reduce the damage of the rainwater on the vegetation and soil on the slope, the drainage system was set according to the terrain conditions, the maximum average rainfall intensity of one hour, the size of the catchment area, etc.

(2)

Vegetation-soil system rebuilding

To realize the rebuilding of the vegetation–soil system on the slope surface, normal spraying, net-suspended spray seeding, and external-soil spray seeding techniques were applied, using mixed soil with processed seeds, polymer soil stabilizer, and soil with fertile and organic content. The thickness of the covering soil layer was about 0.3 m, while grass seeds were selected to be similar to the surrounding natural plants to ensure vegetation coordination between the restoration area and the natural environment.

3. Methodology

3.1. Assessment of Vegetation Community Condition After Remediation

To clarify the ecological restoration effect of vegetation cover establishment on the mining slopes, field investigation and laboratory tests were conducted, concentrating on the assessment of vegetation community condition after remediation.

3.1.1. Field Investigation

According to the field investigation results, it is known that the vegetation types on the slope surfaces are shrubs and herbs as the dominant vegetation, with high coverage (Figure 5). To reduce the impact of human activities on the vegetation community, the quadrats were set at the edge of the vegetation community.

Since vegetation sampling and data collection were carried out in the quadrats, the following should be set as the principles: (1) the vegetation within the quadrats can represent the current growth status; (2) the location of quadrats should be recorded by the installation of PVC pipes and GPS; (3) the quadrat size should consider the growth features of vegetation. The requirements of the quadrat setting are shown in

Table 2. Requirement of the quadrat setting.

Vegetation Type	Quadrat Size	Sampling Method	Sampling Requirement
Herb	1 m × 1 m	Cutting method	Cut off the aboveground herbs in the quadrat.
Shrub	1 m × 1 m	Partly cutting method	Cut off a quarter of aboveground shrubs within the quadrat corner.

.

Due to the difference in growth features between the shrubs and herbs, the vegetation structural data can be divided into vegetation community data and (per) plant data with different measurements. The structural data for vegetation community can represent the rebuilding effect of the vegetation community, while that per plant is recorded for the growth status. The vegetation structural data for the shrubs and herbs are shown in

Table 3. Vegetation structural data for the herbs and shrubs.

Vegetation Type	Data Type	Recorded Content
Herb	Vegetation community	Connectivity of each herb on the slope (α), coverage rate of the quadrat (C).
	Plant	Average height (H), maximum height (Hmax).
Shrub	Vegetation community	Connectivity of each shrub on the slope (α), coverage rate of the quadrat (C), total shrub counts in the quadrat (n).
	Per plant	Height of per-shrub plant (H), diameter of per-shrub plant (D).

. Among the recorded content, the connectivity of vegetation on the slope is the ratio of the area of a single plant on the slope to the total area of the slope, while the coverage rate of the quadrat is the ratio of the area of the plants in the quadrat to the total area of the quadrat. As for the height of the plants and the diameter of per-shrub plant, it is directly measured in the field investigation.

3.1.2. Laboratory Testing

To obtain the vegetation biomass and calculate vegetation carbon storage, laboratory treatment and tests were carried out after sample preparation.

The dry weight of vegetation is usually taken as the vegetation biomass due to the influence of the moisture content on the vegetation. Therefore, the drying method was used to obtain the dry weight of the vegetation samples to eliminate the influence of moisture on biomass measurement with a uniform drying temperature and time. As for the dry weight of the aboveground herbs and shrubs, as the biomass, the vegetation should be naturally air-dried indoors for 10 days, cut into suitable lengths, dried in an oven at 70–80 °C for 8 h, and lastly measured for the dry weight with an accuracy to 0.1 g [41].

Before the test of vegetation carbon content, the vegetation samples should be ground and screened to obtain test samples with a size less than 0.075 mm, and should then be dried at 105 °C for 2 h for later tests. The test of vegetation carbon content should follow the requirement of combustion–infrared absorption spectrometry [42] with a carbon element analyzer (EA4000-FS125, Jena GmbH, Jena, Germany).

3.2. Establishment of the Vegetation Biomass Calculation Model

3.2.1. Data Analysis

Vegetation biomass is a vital index of the vegetation carbon storage calculation. The establishment of vegetation biomass estimation models can be conducted considering the vegetation growth characteristics, while taking the dry weight of vegetation as the biomass and the collected data as the independent variables. The selected independent variables are shown in

Table 4. Independent variables of the herbs and shrubs.

Vegetation	Independent Variables	Code of Variable	Vegetation	Independent Variables	Code of Variable
Herb	Average height (m)	H	Shrub	Average height (m)	H
	Coverage rate of the quadrat (%)	C		Diameter of per-shrub plant	D
	Product of average height and coverage rate (m)	CH		Plant projected area (m2)	A (A = πD2/4)
	Product of average height and square of coverage rate (m)	C2H		Plant volume (m3)	V (V = AH, m3)
	Product of square of average height and coverage rate (m2)	CH2		Product of average height and diameter of per-shrub plant (m2)	DH
				Product of square of diameter of per-shrub plant and average height (m3)	D2H

.

The correlation between independent variables and vegetation biomass is evaluated based on the Pearson correlation coefficient method [43,44]. The independent variable with significant correlation should be selected for the fitting variable according to the Pearson correlation coefficient (r), which is calculated as Equation (1):

$$r={\displaystyle \frac{{\sum }_{\mathrm{i}=1}^{\mathrm{n}} (x_i-\overline{X})(y_i-\overline{Y})}{\sqrt{{\sum }_{\mathrm{i}=1}^{\mathrm{n}} (x_i-\overline{X}{)}^2}\sqrt{{\sum }_{\mathrm{i}=1}^{\mathrm{n}} (x_i-\overline{Y}{)}^2}}}$$

(1)

where r is the Pearson correlation coefficient, $x_i$ is the value of each independent variable, $y_i$ is the value of vegetation biomass, $\overline{X}$ is the average value of each independent variable, and $\overline{Y}$ is the average value of vegetation biomass.

When |r| is closer to 1, it indicates a higher degree of correlation between the two variables, 0.8 ≤ |r| < 1 indicates a high correlation, 0.5 ≤ |r| < 0.8 a moderate correlation, 0.3 ≤ |r| < 0.5 a low correlation between variables, and |r| < 0.3 extremely weakly correlated or unrelated variables.

3.2.2. Model Establishment and Verification

Regression analysis can reflect the relationship between predictor variables and the independent variable, while the regression coefficients in different regression equations can reflect the sensitivity of the predictor variable to the independent variable [45,46]. To establish the vegetation biomass estimation model, the six most commonly utilized regression function models were adopted for regression analysis (

Table 5. Mathematical expression of the function models for regression analysis.

Model Type	Mathematical Expression
Linear function model	W = a + bx
Quadratic function model	W = a + bx + cx2
Cubic function model	W = a + bx + cx2 + dx3
Logarithmic function model	W = a + b·lnx
Power function model	W = a·xb
Exponential function model	W = a·ebx

Here, W is vegetation biomass as a predictor variable, a is the regression constant, and b, c, d, etc., are the regression coefficients.

), including linear function, quadratic function, cubic function, logarithmic function, power function, and exponential function model [47,48].

Among the regression function models, the optimal biomass estimation model should be adopted, with the highest determination coefficient (R²) and a significance of the regression test (P) less than 0.01. The coefficient of determination (R²) can be calculated as in Equation (2):

$$R^2=1-{\displaystyle \frac{{\sum }_{\mathrm{i}=1}^{\mathrm{m}} ({\widehat{y}}_i-y_i{)}^2}{{\sum }_{\mathrm{i}=1}^{\mathrm{m}} (y_i-\overline{y}{)}^2}}$$

(2)

where R² is the determination coefficient, $y_i$ is the measured vegetation biomass, ${\widehat{y}}_i$ is the estimation value of vegetation biomass, and $\overline{y}$ is the average measured vegetation biomass.

As for the accuracy verification of the optimal model, the Pearson correlation coefficient (r), average error (R_S), and average relative error (R_MA) are utilized as the indexes for the comparative analysis between the estimation values and measured values [49,50]. When the index values are smaller, the estimation model has higher accuracy with less significant differences and a stronger correlation. The average error (R_S) and average relative error (R_MA) are calculated as in Equations (3)–(5):

$$R_S={\displaystyle \frac{\sum (y_i-{\widehat{y}}_i)}{\sum y_i^{\prime }}}\times 100\mathrm{\%}$$

(3)

$$R_{MA}={\displaystyle \frac{1}{\mathrm{n}}}\sum _{\mathrm{i}=1}^{\mathrm{n}} \left|{\displaystyle \frac{y_i-{\widehat{y}}_i}{y_i}}\right|$$

(4)

where R_S is the average error, R_MA is the average relative error, $y_i$ is the measured vegetation biomass, and ${\widehat{y}}_i$ is the estimation value of vegetation biomass.

3.3. Establishment of the Vegetation Carbon Storage Prediction Model

As the main approaches for the carbon sequestration of the restored mines, vegetation biomass and carbon content are key to improving vegetation carbon storage via photosynthesis [25,51,52]. Therefore, the prediction model of vegetation carbon storage is established according to the vegetation biomass and carbon content.

The shrubs and herbs rebuilt on the slopes have a high growing speed for rapid restoration and dense and complex root systems to reinforce the overlying soil layer. Therefore, the total vegetation carbon storage consists of the carbon storage of shrubs and herbs, as in Equation (5):

$${\mathrm{C}}_P={\mathrm{C}}_{Sh}+{\mathrm{C}}_{He}$$

(5)

where ${\mathrm{C}}_P$ is the total vegetation carbon storage, ${\mathrm{C}}_{Sh}$ is the shrub carbon storage, and ${\mathrm{C}}_{He}$ is the herb carbon storage.

Herbs often grow in the form of vertically grown plants and are irregularly distributed, with a large number of pre-plants and dense roots per unit area. Shrubs grow vertically with a single diameter and multiple branches, with roots buried at a greater depth. The carbon storage calculation models were determined based on the differences in growth characteristics and shapes of herbs and shrubs. The carbon storage of shrubs and herbs can be calculated as in Equations (6) and (7):

$${\mathrm{C}}_{He}=\sum _{\mathrm{i}=1,\mathrm{j}=1}^{\mathrm{n}}{S}_{\mathrm{i}}\times {\alpha }_{\mathrm{ij}}\times W_{\mathrm{ij}}\times {\delta }_{\mathrm{ij}}$$

(6)

$${\mathrm{C}}_{Sh}=\sum _{\mathrm{i}=1,\mathrm{j}=1}^{\mathrm{n}}{S}_{\mathrm{i}}\times {\alpha }_{\mathrm{ij}}\times W_{\mathrm{ij}} \times {\delta }_{\mathrm{ij}}\times n$$

(7)

where ${\mathrm{C}}_{He}$ is the herb carbon storage, ${\mathrm{C}}_{Sh}$ is the shrub carbon storage, $S_{\mathrm{i}}$ is the surface area of different slope areas i, ${\alpha }_{\mathrm{ij}}$ is the connectivity of herb/shrub type j on the slope areas i, $W_{\mathrm{ij}}$ is the vegetation biomass of herb/shrub type j on the slope areas i, ${\delta }_{\mathrm{ij}}$ is the vegetation carbon content of herb/shrub type j on the slope areas i, and n is the total shrub count in the quadrat.

4. Results

4.1. Correlation Analysis Between the Biomass and Variables

Based on the field investigation results, the dominant shrub vegetation rebuilt on the slopes is Amorpha fruticosa L. and Indigofera tinctoria L., while the dominant herb vegetation is Solidago canadensis L. and Erigeron canadensis L. Therefore, the estimation models for the common herb and shrub biomass were established according to correlation analysis, model optimal selection, and accuracy verification.

The aboveground biomass of Amorpha fruticosa L. is mainly concentrated between 50 and 200 g/per plant, with that of Indigofera tinctoria L. between 100 and 300 g/per plant or concentrated between 50 and 200 g/per plant, Solidago canadensis L. between 100 and 800 g/m², and Erigeron canadensis L. between 200 and 600 g/m² (Figure 6). The correlation analysis between the biomass and independent variables was conducted via SPSS 27.0.1 and Excel 2021. The results of the correlation analysis between the biomass and independent variables are shown in Figure 7 and

Table 6. The correlation analysis results between the biomass and variables for shrub plants (D: diameter of per-shrub plant, A: plant projected area, V: plant volume, D²H: square of diameter of per-shrub plant multiplied by average height, DH: average height multiplied by diameter of per-shrub plant, H: average height, W: vegetation biomass).

Vegetation Type	Correlation Coefficients	D	A	V	D2H	DH	H	W
Amorpha fruticosa L.	D	1.000	0.981	0.856	0.856	0.766	0.357	0.675
	A	0.981	1.000	0.876	0.876	0.754	0.323	0.676
	V	0.856	0.876	1.000	1.000	0.962	0.689	0.868
	D²H	0.856	0.876	1.000	1.000	0.962	0.689	0.868
	DH	0.766	0.754	0.962	0.962	1.000	0.851	0.900
	H	0.357	0.323	0.689	0.689	0.851	1.000	0.781
	W	0.675	0.676	0.868	0.868	0.900	0.781	1.000
Indigofera tinctoria L.	D	1.000	0.981	0.856	0.856	0.766	0.357	0.675
	A	0.981	1.000	0.876	0.876	0.754	0.323	0.676
	V	0.856	0.876	1.000	1.000	0.962	0.689	0.868
	D2H	0.856	0.876	1.000	1.000	0.962	0.689	0.868
	DH	0.766	0.754	0.962	0.962	1.000	0.851	0.900
	H	0.357	0.323	0.689	0.689	0.851	1.000	0.781
	W	0.675	0.676	0.868	0.868	0.900	0.781	1.000

and

Table 7. The correlation analysis results between the biomass and variables for herb plants (H: average height, C: coverage rate of the quadrat, CH: average height multiplied by coverage rate, C²H: square of coverage rate multiplied by average height, CH²: coverage rate multiplied by the square of average height, W: vegetation biomass).

Vegetation Type	Correlation Coefficients	H	C	CH	C2H	CH2	W
Solidago canadensis L.	H	1.000	−0.145	0.915	0.777	0.940	0.951
	C	−0.145	1.000	0.242	0.472	0.114	0.134
	CH	0.915	0.242	1.000	0.964	0.963	0.982
	C2H	0.777	0.472	0.964	1.000	0.890	0.915
	CH2	0.940	0.114	0.963	0.890	1.000	0.969
	W	0.951	0.134	0.982	0.915	0.969	1.000
Vegetation Type	Correlation Coefficients	H	C	CH	C2H	CH2	W
Erigeron canadensis L.	H	1.000	−0.191	0.771	0.603	0.863	0.803
	C	−0.191	1.000	0.446	0.631	0.21	0.414
	CH	0.771	0.446	1.000	0.971	0.934	0.988
	C2H	0.603	0.631	0.971	1.000	0.851	0.946
	CH2	0.863	0.21	0.934	0.851	1.000	0.936
	W	0.803	0.414	0.988	0.946	0.936	1.000

.

According to the correlation analysis results, it can be seen that the correlation coefficients are 0.900, 0.868, and 0.868, respectively, between the biomass of Amorpha fruticosa L. and the factors DH, V, and D²H. Due to the maximum correlation coefficient of 0.900, DH was selected as the fitting variable. Similarly, based on the results of the correlation coefficients, DH, CH, and CH were selected as the fitting variables for Indigofera tinctoria L., Solidago canadensis L., and Erigeron canadensis L., with the maximum correlation coefficients of 0.895, 0.982, and 0.988.

4.2. Establishment, Optimal Selection, and Accuracy Verification of Biomass Models

Regression analysis was utilized for the establishment of a vegetation biomass model based on the selected fitting variable. Further, the estimation models of the vegetation biomass were established by the regression analysis among the different function models and comprehensively selected considering the coefficient of determination and the F-test value. The accuracy of the optimal biomass estimation model was evaluated by combining the average error and the average relative error. To verify the accuracy of the optimal biomass estimation model, the estimated biomass was calculated based on the optimal biomass model and compared with the measured values.

The biomass estimation models for Amorpha fruticosa L. are shown in

Table 8. Mathematical expression of the function models for regression analysis (DH: average height multiplied by diameter of per-shrub plant, **: represents extremely significant correlation).

Function Type	Fitting Variable	Regression Parameters				Coefficient of Determination (R2)	Regression Test Significance (P)	Joint Hypotheses Test (F)
		Constant	Coefficients
		a	b	c	d
Linear function	DH	26.680	104.426	/	/	0.809 **	<0.001	97.691
Quadratic function		153.529	109.487	/	/	0.717 **	<0.001	58.242
Cubic function		21.306	113.814	−3.089	/	0.810 **	<0.001	46.860
Logarithmic function		46.239	35.405	59.285	−13.657	0.814 **	<0.001	30.668
Power function		4.084	0.667	/	/	0.777 **	<0.001	80.126
Exponential function		59.406	0.667	/	/	0.777 **	<0.001	80.126

, obtained from the regression analysis using linear function, quadratic function, cubic function, logarithmic function, power function, and exponential function. The regression analysis results of the different function models show that the best-fitting effect is relatively good when DH is the fitting variable. The cubic function has the highest fitting accuracy, with a maximum coefficient of determination of 0.814 and the highest F-test value, indicating statistical significance. Therefore, the cubic function fitting model was selected as the optimal model for the biomass estimation model of Amorpha fruticosa L., as is shown in Equation (8):

W = 46.239 + 35.405x + 59.285x² − 13.657x³

(8)

where W is the estimated biomass of Amorpha fruticosa L. and x is the fitting variable (DH).

The verification results show that the average error (R_S) between the estimated and measured values is 2.51% and the average relative error (R_MA) is 8%, which meets the standards of R_S < 30% and R_MA < 20%. Meanwhile, external validation of the biomass model for Amorpha fruticosa L. showed that the linear correlation coefficient r between the estimated and measured values is 0.991, greater than 0.9, indicating a high correlation level (p < 0.001).

For the same reason, the biomass estimation models for Indigofera tinctoria L., Solidago canadensis L., and Erigeron canadensis L. were established, selected, and verified. The biomass estimation models are shown in Figure 8 and

Table 9. Selected biomass estimation models and the accuracy verification results (**: represents extremely significant correlation).

Vegetation Type	Fitting Variable	Correlation Coefficient	Model Expression	Coefficient of Determination (R2)	Regression Test Significance (P)	Joint Hypotheses Test (F)	Average Error (RS)	Average Relative Error (RMA)
Indigofera tinctoria L.	DH	0.895	$W=-50.819+627.961x$	0.965 **	<0.001	1224.113	0.76%	5%
Solidago canadensis L.	CH	0.982	$W=0.441+590.047x$	0.975 **	<0.001	989.299	0.91%	3%
Erigeron canadensis L.	CH	0.988	$W=525.005-985.204x+752.612x^2-160.45x^3$	0.856 **	<0.001	51.997	2.8%	6%

. The above results indicate that the selected models have excellent accuracy and can be applied to biomass estimation.

4.3. Vegetation Carbon Storage Calculation

Combined with the current situation of the vegetation community assessed via the field investigation, the vegetation community structural data are shown in

Table 10. Data for the carbon storage calculation of the vegetation on the mining slopes.

No.	Surface Area (m2)	Dominant Vegetation Type	Vegetation Connectivity on the Slope (%)	Vegetation Carbon Content (%)	Vegetation Biomass of per Unit Area (g)	Vegetation Carbon Storage (t)
A1	6313	Erigeron canadensis L.	43.7	30.627	1013.6	0.856
		Solidago canadensis L.	10.4	33.180	990.9	0.216
		Amorpha fruticosa L.	24.3	44.266	4622.7	3.139
A2	5083	Erigeron canadensis L.	8.0	36.280	460.6	0.068
		Amorpha fruticosa L.	33.8	44.922	3719.7	2.871
		Indigofera tinctoria L.	21.9	45.046	3317.6	1.664
A3	1738	Amorpha fruticosa L.	90.0	41.732	3509.7	2.291
A4	34,746	Solidago canadensis L.	15.6	35.421	489.7	0.940
		Erigeron canadensis L.	20.4	36.733	394.5	1.027
		Amorpha fruticosa L.	31.2	43.018	4208.3	19.625
		Indigofera tinctoria L.	29.7	41.771	4011.0	17.290
A5	6111	Erigeron canadensis L.	43.0	39.830	571.2	0.598

, presenting the test results of the vegetation carbon content. Since the vegetation community has been developing on the surface of slope A₆ and slope A₇, vegetation rebuilding was carried out on slope areas A_1–5 as a key ecological restoration measure for the slope, while vegetation rebuilding measures were not conducted on slopes A₆ and A₇. Therefore, the slope areas A_1–5 are taken into consideration for the vegetation carbon storage calculation and biomass calculation.

Combined with the field investigation results, the shrub planting area is 2.71 hm², including 1.57 hm² for Amorpha fruticosa L. and 1.14 hm² for Indigofera tinctoria L., while the herb planting area is 1.90 hm², including 1.29 hm² for Erigeron canadensis L. and 0.61 hm² for Solidago canadensis L. Based on the optimal biomass estimation model and the calculation results, it can be found that the total vegetation biomass reaches 120.264 t. The shrub biomass is 109.678 t, with 64.593 t for Amorpha fruticosa L. and 45.085 t for Indigofera tinctoria L., while the herb biomass is 10.586 t, with 7.281 t for Erigeron canadensis L. and 3.305 t for Solidago canadensis L. According to the carbon content of vegetation via laboratory tests, the carbon storage of the rebuilt vegetation is 50.585 t, consisting of 3.705 t of herb carbon storage and 46.880 t of shrub carbon storage.

5. Discussion

5.1. Analysis of Vegetation Rebuilding Effect on the Mining Slopes

According to the field investigation and the calculation results of the biomass and the carbon storage of the rebuilt vegetation on the mining slopes, it can be seen that the vegetation community rebuilt on the slopes grew rapidly after restoration, with a relatively good amount of biomass and carbon storage. Therefore, more analysis is needed to evaluate the vegetation rebuilt effect combined with the related qualitative assessment and the quantitative calculation results.

5.1.1. Analysis of the Biomass with the Rebuilt Vegetation Effect

The biomass of herbs and shrubs is greatly affected by growth characteristics, and this has an impact on the establishment of the biomass estimation model, and especially on the selection of the independent variables.

For instance, according to the correlation analysis results, the biomass of herbs is greatly affected by the average height, while that of shrubs is influenced by the diameter and height. Also, the analysis results indicate that the shrub diameter multiplied by height (DH) is the fitting variable for the biomass model in the form of the cubic function, while that for herbs is the coverage multiplied by height (CH) in the linear function. Also, the aboveground biomass of a shrub plant consists of the stems and leaves, with higher height, which is much larger than the aboveground biomass of the herbs. Therefore, the vegetation biomass per unit area of the shrubs is higher than that of the herbs.

Therefore, the calculation results of the biomass demonstrate that the total biomass of shrubs is almost 10 times that of the biomass of herbs—the biomass of shrubs is 40.47 t/hm² and that of the herbs is 5.57 t/hm², which demonstrates that the vegetation on the mining slopes can not only be the key index for the field investigation and the observation object, but also influence the biomass.

Moreover, combined with the field investigation, it can be concluded that the vegetation growth characteristics and slope conditions can influence the distribution of vegetation biomass, subsequently influencing the vegetation biomass. Although both the shrubs and herbs grow in the form of an aggregated vegetation community, the root burial depth and vegetation weight decide the distribution scope of vegetation on the slopes. Due to the light weight and shallow root depth of the herbs, they can have a larger growth space on the relatively steeper slopes, while the shrubs need thicker soil cover for the roots to adapt to root growth and to provide nutrients for growth. For example, the shrubs were mainly concentrated on the lower part of the slope areas A1, A2, and A5 while the herbs could be distributed on the slope surface because the thickness of the soil cover layer in the lower part of the slopes is relatively larger than the slope surface.

5.1.2. Analysis of the Carbon Storage with the Rebuilt Vegetation Effect

Based on the laboratory test results of the carbon content of Amorpha fruticosa L., Indigofera tinctoria L., Solidago canadensis L., and Erigeron canadensis L. shown in Table 9, it can be concluded that the shrubs have a higher carbon content than the herbs no matter which slope area they are in. The carbon content of the shrubs ranges from 41.732 to 44.922%, while that of the herbs ranges from 30.627 to 39.830%.

Considering the larger area and the higher biomass of the shrubs in the study area, the shrubs have the greater carbon sequestration potential. Therefore, it can be concluded that the shrubs have advantages of high growth speed in the short-term living period and higher carbon content, leading to fast changes in the landscape, land use, and carbon sequestration.

Therefore, to realize vegetation rebuilding on mining slopes with a higher recovery speed and to make the vegetation part of the carbon sequestration pathway, with great potential for the restoration of mining slopes, the selection of vegetation types for the preferred vegetation should consider the growth characteristics—herbs can be adopted for steep slopes, and shrubs can be applied in relatively gentle slopes and the lower parts of steep slopes.

5.2. Limitations of This Study and Future Work

In this paper, the study provides an approach to evaluate the effect of the vegetation rebuilt on restored mining slopes, with the application of models for biomass and carbon storage. However, there are still some limitations to the method used in the study.

Firstly, the sampling and data collection were carried out on the edge of the vegetation community to reduce the impact of human activities on the vegetation community. The amount of samples and data may not be enough for a more accurate analysis.

Also, the vegetation biomass models mainly concentrate on the vegetation structural data and the growth features, while the slope (slope angle, slope direction, slope height, etc.), environment (temperature, properties of the mixed soil, restoration technologies, etc.), and time conditions were not taken into consideration. This could potentially affect the uncertainty of the results.

Moreover, the model was established according to the data collected in one field investigation, and lacked the subsequent data to eliminate the potential uncertainty of the results caused by the time of data acquisition.

In addition, although the effect of the vegetation on the mining slopes was analyzed with the models of the biomass and carbon consequence, this heavily relied on field investigation and laboratory tests. Therefore, more approaches should be taken for mutual verification, such as analyzing the remote sensing data and supplied investigation data.

Therefore, to fulfill and deepen the work on the biomass and carbon storage of the vegetation on the restoration slopes, more quarries after ecological restoration and the subsequent investigation should be examined for data collection to enrich the establishment of the biomass model with long-term effects, while other techniques can be introduced to further verify the accuracy of the biomass estimation.

6. Conclusions

This paper established models for the biomass estimation and carbon storage calculation of the vegetation community rebuilt on post-mining rock slopes, combined with field investigation, laboratory tests, data analysis, etc. After the comprehensive selection of the optimal model and the accuracy verification, the biomass and carbon storage of the vegetation were calculated to evaluate the vegetation rebuilding effect.

The results show that the structure characteristics of vegetation are the key index of the biomass, while different indexes should be applied for the biomass estimation models of each vegetation type due to the vegetation community features, such as the fitting variables of the diameter multiplied by height (DH) for shrub biomass and the coverage multiplied by height (CH) for herb biomass. The established vegetation biomass models had the greatest accuracy with the tests of the determination coefficient (R²), average error (R_S), average relative error (R_MA), significance of regression test (P), Joint Hypotheses Test (F-test), etc., indicating the effectiveness and statistical significance of the results. Also, the vegetation carbon storage calculation results show that there is a difference in carbon storage between the shrubs and herbs, owing to the slope conditions and the growth features, which should be taken into consideration for the subsequent mining slope management.

Therefore, it can be concluded that the vegetation biomass estimation model has applicability and effectiveness, offering useful applications for biomass estimation and carbon storage calculation of the vegetation communities. This study provides an approach to calculate the biomass and carbon storage for the vegetation community on restored post-mining slopes with model establishment, using field investigation, laboratory tests, correlation analysis, regression analysis, accuracy testing, and other methods as the reference for other similar engineering works on slopes for ecological restoration.