Estimating and Projecting Forest Biomass Energy Potential in China: A Panel and Random Forest Analysis

Abstract

Understanding the spatiotemporal evolution of forest biomass energy potential is essential for supporting low-carbon land-use planning and regional energy transitions. China, characterized by pronounced spatial heterogeneity in forest resources and ecological conditions, provides an ideal case for examining how biophysical endowments and management factors shape biomass energy potential. This study constructs a province-level panel dataset for China covering the period from 1998 to 2018 and investigates long-term spatial patterns, regional disparities, and driving mechanisms using spatial visualization, Dagum Gini decomposition, and fixed-effects estimation. The results reveal a gradual spatial reorganization of forest biomass energy potential, with the national center of gravity shifting westward and northwestward, alongside a moderate dispersion of high-potential clusters from coastal areas toward the interior. Interregional transvariation is identified as the dominant source of regional inequality, indicating persistent structural differences among major regions. To explore future dynamics, a random forest model is employed to project provincial forest biomass energy potential from 2018 to 2028. The projections suggest moderate overall growth, smoother distributional structures, and a partial reduction in extreme provincial disparities. Central, southwestern, and northwestern provinces are expected to emerge as important contributors to future growth, reflecting ecological restoration efforts, expanding plantation forests, and improved forest management. The findings highlight a continued upward trend in national forest biomass energy potential, accompanied by a spatial shift toward inland regions and evolving regional disparities. This study provides empirical evidence to support region-specific development strategies, optimized spatial allocation of forest biomass resources, and integrated policies linking ecological sustainability with renewable energy development.

1. Introduction

Global energy transitions have accelerated in recent decades as nations confront the dual challenges of resource depletion and climate change. Among various renewable sources, biomass energy has attracted increasing attention for its renewability, carbon neutrality, and abundant reserves [1,2]. Within this sector, forest biomass energy—derived from forestry residues, logging by-products, and energy plantations—plays a particularly important role due to its stable reserves, high conversion potential, and ecological benefits [3,4]. Recent studies emphasize that developing forest biomass energy can effectively mitigate greenhouse gas emissions while enhancing energy security and supporting rural economies [5].

In China, the transition to a low-carbon economy has made renewable energy development a strategic priority. The Guidelines for Green and Low-Carbon Industrial Transformation (2024) explicitly highlight forest biomass as a key sector for green growth and carbon neutrality. China’s forest resources have expanded steadily over the past two decades, providing an important ecological foundation for the development of forest biomass energy. China’s forest resources have expanded steadily over the past two decades, with national forest area increasing continuously from 1998 to 2018. This growing resource base highlights China’s substantial potential for forest biomass utilization. However, despite China’s vast forest resources, the forest biomass industry remains underdeveloped compared with other renewable energy sectors. High feedstock costs and weak technological innovation continue to limit industrial competitiveness [6,7]. Meanwhile, the lack of consistent policy incentives and insufficient financial support have further constrained large-scale application [8,9].

Existing research on forest biomass energy has provided valuable insights into its importance and utilization potential. For instance, regional assessments show that forest biomass can significantly contribute to emission reduction targets and renewable substitution [10,11]. Other studies focus on technological progress, discussing advances in gasification, combustion, and biochemical conversion processes that improve conversion efficiency [12,13,14]. Comparative analyses also demonstrate that the physical and chemical characteristics of different wood species strongly affect their energy properties [13]. In addition, research on biomass supply chains has emphasized the importance of optimizing logistics and processing systems to improve cost-effectiveness [15,16]. Although these studies have enriched theoretical understanding, most are qualitative in nature or employ conventional statistical models that cannot fully capture nonlinear relationships among ecological, economic, and spatial variables. As a result, their predictive capacity for long-term biomass energy potential under complex regional conditions remains limited.

Another limitation in current research lies in the insufficient attention to regional heterogeneity. China’s forest biomass resources vary significantly across provinces due to differences in forest structure, management intensity, and economic development [17,18]. Few studies have systematically quantified these spatial differences or examined their impact on biomass energy potential. Furthermore, advanced analytical approaches such as machine learning remain underused in biomass potential forecasting, despite their demonstrated advantages in capturing complex multivariate interactions, nonlinear effects, and threshold behaviors in resource–energy systems. This methodological gap is particularly evident in large and heterogeneous countries such as China, where forest biomass energy potential is jointly shaped by ecological endowments, management practices, and socioeconomic conditions.

In response to this methodological gap, machine learning methods have been increasingly applied in biomass assessment, forest resource estimation, and energy forecasting due to their flexibility and strong predictive performance [19]. Compared with traditional econometric or parametric approaches, machine learning models are better suited to handling nonlinear relationships, high-dimensional data, and complex interactions among ecological and socioeconomic drivers [20]. Studies have demonstrated that machine learning techniques can significantly improve estimation accuracy and forecasting robustness in biomass and renewable energy research [21].

Building on these methodological considerations, this study develops a comprehensive framework for evaluating and forecasting China’s forest biomass energy potential. The research first quantifies provincial-level potential and analyzes its spatiotemporal evolution. It then identifies key influencing factors—such as forest stock volume, economic scale, and technological investment—and constructs a random forest (RF) model to project future potential trends. The RF model, known for its robustness and interpretability, enables a more accurate assessment of the nonlinear relationships among multi-dimensional drivers of biomass energy development [22].

This paper makes three main contributions. First, it provides a systematic measurement of China’s forest biomass energy potential using multi-period forestry data. Second, it identifies the primary socio-economic and ecological determinants influencing potential differences across regions. Third, it applies a machine learning–based prediction model to enhance forecasting accuracy, offering a methodological innovation beyond traditional econometric approaches. The analytical framework developed in this study also offers potential reference for developing countries that face comparable challenges in biomass resource assessment, technological capacity, and forest-based industrial development, thereby providing broader theoretical insights and policy guidance for optimizing forest biomass resource allocation and supporting sustainable low-carbon energy planning.

2. Materials and Methods

2.1. Data Sources and Indicator Selection

To clearly define the geographical scope and temporal coverage of the analysis, this study focuses on mainland China and employs provincial-level panel data covering the period from 1998 to 2018, which constitutes the historical observation window of the research. The study area and its regional classification are presented in Appendix A.

Forest resource data used in this study are primarily obtained from China’s National Forest Resource Inventory, a nationwide and periodically conducted official forest survey organized by the national forestry authority. The inventory provides standardized and comparable information on forest area, forest structure, and forest management activities at the provincial level. Data from five inventory periods spanning 1998–2018 are employed to ensure national coverage, temporal consistency, and interprovincial comparability. These inventories constitute the authoritative national framework for forest resource statistics in China.

Data on forestry production and management indicators are obtained from two official national statistical sources. The China Forestry Statistical Yearbook is an annually published statistical compilation released by China’s national forestry administration, reporting province-level information on forestry production, output, and management activities. The Summary Table of Annual Cutting Quotas is an official administrative dataset issued by the same authority, documenting provincially approved harvesting limits and related management indicators across years.

This study constructs a comprehensive and systematic estimation model for forest biomass energy potential, based on three primary sources of forest biomass: forest residues, forestry residues, and harvested materials from energy forests. Forest residues refer to remaining forest resources that grow in non-protected areas and have not been harvested for industrial use. These mainly include residues from shrubland areas, pruning residues from economic forests, and maintenance residues from trees growing on non-forest land. Forestry residues are by-products generated during routine forest management and maintenance activities. They primarily consist of residues from seedling pruning, thinning and trimming of poles, and tending residues from young and middle-aged stands. Harvested materials from energy forests mainly refer to forest resources composed of fuelwood forests and oil-producing energy tree species.

To estimate the potential of forest biomass resources and energy, this study employs nine specific indicators, including shrubland area, economic forest area, number of trees on non-forest land, seedling yield, tending area of young and middle-aged forests, and the annual allowable cutting quota, among others.

Table 1. Definitions, Data Sources, and Units of Indicators Used for Estimating Provincial Forest Biomass Energy Potential in China (1998–2018).

Category	Indicator	Definition/Meaning	Unit	Data Source
Forest Growth Residues	Shrubland area	Area of shrubland that generates residual biomass through harvesting and tending	104 ha	National Forest Resource Inventory Report; China Forestry Statistical Yearbook
	Economic forest area	Area of economic forests generating pruning and management residues	104 ha	China Forestry Statistical Yearbook
	Number of trees on non-forest land	Number of scattered trees on farmland, villages, roadsides, etc., generating trimming residues	104 trees	China Forestry Statistical Yearbook
Forestry Production Residues	Seedling yield	Quantity of seedlings producing trimming and pruning residues during cultivation	104 units	China Forestry Statistical Yearbook
	Tending area of young and middle-aged forests	Area undergoing tending and pruning, producing silvicultural residues	104 ha	National Forest Resource Inventory Report
	Annual allowable cutting quota	Officially approved harvesting quota, producing harvesting residues	104 m3	Summary Table of Annual Cutting Quotas
	Commercial timber output	Output of processed timber, producing processing residues	104 m3	China Forestry Statistical Yearbook
	Bamboo output	Bamboo used for energy-related processing, generating bamboo residues	104 culms	China Forestry Statistical Yearbook
Energy Forests	Fuelwood forest area	Area of dedicated fuelwood forests producing biomass for energy	104 ha	China Forestry Statistical Yearbook

provides a detailed description of all indicators used in the estimation model.

The indicators listed in Table 1 were selected based on three main considerations: conceptual relevance, data availability, and consistency with existing literature on forest biomass energy assessment.

First, the selected indicators correspond directly to the main physical sources of forest biomass energy, including forest growth residues (e.g., shrubland area, economic forest area, and scattered trees on non-forest land), forestry production residues (e.g., seedling cultivation, forest tending, harvesting, and processing activities), and dedicated energy forests (fuelwood forest area) [23]. These categories jointly represent the dominant pathways through which forest biomass resources are generated and collected for energy use.

Second, all indicators are officially reported in China’s National Forest Resource Inventory and the China Forestry Statistical Yearbook, ensuring long-term continuity, interprovincial comparability, and data reliability over the study period (1998–2018). This criterion is particularly important for constructing a consistent provincial-level panel dataset suitable for spatiotemporal analysis.

Third, the indicator system is consistent with widely adopted measurement frameworks in previous studies on forest biomass energy potential, which typically estimate biomass resources by linking forestry activity indicators (area, volume, or output) with corresponding conversion coefficients [24]. By following this established indicator–coefficient approach, the present study ensures methodological transparency and facilitates comparison with existing empirical research.

2.2. Calculation Method for Total Forest Biomass Energy Potential

The forest biomass energy potential is conceptually divided into three sequential components: the theoretical potential, the available (technical) potential, and the utilizable (economic) potential.

The theoretical potential refers to the maximum amount of renewable energy embedded in forest biomass resources under ideal conditions, without considering technical, economic, or ecological constraints. It represents the upper physical limit of energy that could theoretically be obtained from forest biomass. In this study, the theoretical potential is calculated by multiplying the quantities of different types of forest biomass resources by their corresponding conversion coefficients.

Considering that the conversion coefficients reported in the existing literature exhibit relatively limited variation across regions and studies, the mode of the reported values was adopted as the representative conversion coefficient. This choice is intended to enhance robustness and cross-regional comparability in a national-scale assessment, rather than to capture fine-grained ecological heterogeneity at the local level.

The available (technical) potential refers to the portion of the theoretical potential that can be feasibly collected and supplied for energy use under current forest management practices, technological conditions, and accessibility constraints. It excludes biomass components that are technically inaccessible, ecologically restricted, or reserved for non-energy purposes.

The utilizable (economic) potential represents the fraction of the available biomass resources that can be effectively converted into usable energy after accounting for economic feasibility, processing efficiency, and sustainability considerations. This stage reflects realistic development conditions under existing technological and institutional frameworks and serves as the basis for estimating the final forest biomass energy potential.

The calculation of forest biomass energy potential is obtained by multiplying the amount of utilizable forest biomass resources by the standard coal conversion coefficient. According to the China Energy Statistical Yearbook, this study adopts a conversion coefficient of 0.571. The specific calculation formula for the total forest biomass energy potential is shown in Equations (1)–(4):

$$L_a=\sum _{m=1}^{n}F{R}_m\ast u_m+\sum _{n=1}^{n}T{C}_n\ast v_n+HY\ast \omega$$

(1)

$L_a$ represents the total physical reserves of forest biomass resources, expressed as the aggregated physical quantity of biomass prior to the application of availability and utilization constraints. In Equation (1), $L_a$ is calculated by converting forestry statistical indicators into biomass quantities using corresponding conversion coefficients and then summing them across resource categories. $F{R}_m$ denotes the biomass generated during forest growth and is estimated based on growth-related indicators reported in units of 10⁴ ha or 10⁴ trees: shrubland biomass and economic forest biomass are calculated as “shrubland area or economic forest area (10⁴ ha) × conversion coefficient (t·ha⁻¹)”, while biomass from scattered trees on non-forest land is calculated as “number of trees (10⁴ trees) × conversion coefficient (kg·tree⁻¹)”, with the results subsequently converted into tons for aggregation.

$T{C}_n$ refers to biomass residues produced during forest management and production activities and is estimated in a similar indicator–coefficient framework. Specifically, residues from seedling cultivation are calculated as “seedling yield (10⁴ units) × conversion coefficient (kg·unit⁻¹)”; residues from tending of young and middle-aged forests are calculated as “tending area (10⁴ ha) × conversion coefficient (t·ha⁻¹)”; and harvesting- and processing-related residues are calculated as “annual allowable cutting quota or commercial timber output (10⁴ m³) × conversion coefficient (t·m⁻³)”. In addition, residues from bamboo production are calculated as “bamboo output (10⁴ culms) × conversion coefficient (kg·culm⁻¹)”.

$HY$ indicates biomass from fuelwood forests and is calculated as “fuelwood forest area (10⁴ ha) × conversion coefficient (t·ha⁻¹)”. The parameters $u_m$, $v_n$, and $\omega$ denote the corresponding conversion coefficients that transform the above area-, volume-, and quantity-based forestry indicators into comparable biomass quantities (t). The specific numerical values of all conversion coefficients used in these calculations are provided in

Table 2. Conversion, availability, and utilization coefficients used for estimating provincial forest biomass energy potential in China (1998–2018).

Indicator	Shrubland	Economic Forest	Economic Forest	Seedlings	Tending of Young and Middle-Aged Forests	Forest Harvesting	Commercial Timber	Bamboo	Fuelwood Forest
Conversion Coefficient	10	7.2	2	0.125	7.2	1.17	0.9	5	16
Availability Coefficient	0.32	1	1	1	0.11	0.57	0.20	0.21	0.24
Utilization Coefficient	0.56	0.23	0.34	0.66	2.21	0.06	0.24	0.27	1

Note: In the conversion coefficients, the units for shrubland, economic forests, tending of young and middle-aged forests, and fuelwood forests are t/hm²; for scattered trees, seedlings, and bamboo, the units are kg/tree; and for forest harvesting and commercial timber, the units are t/m³. The conversion coefficients reported in Table 2 correspond to the modal values reported most frequently across the relevant literature, adopted to ensure consistency and cross-regional comparability at the national scale.

, ensuring consistency with official statistical units and the reproducibility of the estimation process.

$$L_b=\sum _{i=1}^n{L}_{ai}\cdot {\lambda }_i$$

(2)

$L_b$ represents the amount of forest biomass resources that can be obtained after excluding non-energy-use components. It is derived by multiplying each type of physical reserve $L_{ai}$ by its corresponding availability coefficient ${\lambda }_i$, which reflects the proportion of biomass that can actually be collected under practical forest management conditions [23].

$$L_c=\sum _{i=1}^n{L}_{bi}\cdot f_i$$

(3)

$L_c$ denotes the portion of available forest biomass resources that can be further utilized for energy production. It is obtained by multiplying the available biomass $L_{bi}$ by the utilization coefficient $f_i$, which specifies the fraction of biomass suitable for conversion into bioenergy [24].

$$L=0.571\times \sum _{i=1}^n{L}_{ci}$$

(4)

$L$ represents the total forest biomass energy potential, calculated by multiplying the utilizable biomass resources by the standard coal conversion coefficient of 0.571, as specified in the China Energy Statistical Yearbook. This coefficient converts forest biomass into an equivalent amount of standard coal to enable uniform energy accounting. The specific values of the conversion coefficients were determined through a review of relevant literature (as shown in Table 2) [25,26,27].

2.3. Projection of Forest Biomass Energy Potential Using Random Forest

To ensure the accuracy of future forecasts for China’s forest biomass energy potential, it is first necessary to determine the influencing factors of the total forest biomass energy potential. Therefore, this study constructs the following empirical model, as shown in Equation (5):

$$\ln {PFBE}_{cg}={\alpha }_{cg}+{\beta }_1{VAL}_{cg}+{\beta }_2\ln {PGDP}_{cg}+{\beta }_3{EDU}_{cg}+{\beta }_4{ORG}_{cg}+{\epsilon }_{cg}$$

(5)

In Equation (5), $c$ and $g$ represent the i-th province and g-th year, respectively; ${\alpha }_{ct}$ is a constant term; ${\epsilon }_{cg}$ denotes the random error term; ${\beta }_c$ represents the elasticity coefficient of each indicator, which indicates that when other factors remain unchanged, a 1% change in the factor will lead to a ${\beta }_c$% change in the total forest biomass energy potential. Definitions of the model variables are as follows:

${PFBE}_{cg}$: Total forest biomass energy potential—the dependent variable.

${VAL}_{cg}$: VAL measures the share of forestry output value in the regional economy and reflects the relative economic importance of forestry activities, including tree planting, forest tending, timber processing, and related market transactions. A higher proportion indicates stronger regional reliance on forestry production and a more developed forestry industrial system. Such regions typically exhibit better coordination between forest resource cultivation, processing, and commercialization, which facilitates the stable supply and efficient utilization of forest biomass resources [28]. Consequently, a higher forestry output share is expected to be positively associated with forest biomass energy potential.

${PGDP}_{cg}$: PGDP represents the overall level of regional economic development and serves as a proxy for infrastructure conditions, investment capacity, and technological diffusion. Previous studies suggest that economic development can influence forestry production through improved transportation networks, access to capital, and policy implementation capacity [29]. However, its effect on forest biomass energy potential may be indirect and mediated by institutional and management factors rather than acting as a direct driver of biomass resource availability.

${EDU}_{cg}$: EDU captures the human capital structure of the forestry sector by measuring the proportion of forestry practitioners with higher education. A higher education level generally implies stronger technical skills, better adoption of modern forest management practices, and greater capacity for efficient biomass collection and utilization [30]. Well-educated forestry personnel are also more capable of implementing advanced silvicultural techniques and resource-saving technologies, which can enhance both forest productivity and biomass energy potential.

${ORG}_{cg}$: ORG reflects the effectiveness of forest health management and ecological protection efforts. Pest and rodent outbreaks can significantly reduce forest growth, biomass accumulation, and the availability of harvestable residues. Effective pest-control systems help maintain stable forest productivity and reduce biomass losses caused by biological disturbances [31]. Therefore, higher pest and rodent control rates are expected to contribute positively to forest biomass energy potential by safeguarding forest resource quality and long-term sustainability.

Based on the identified influencing factors of total forest biomass energy potential, this study constructs a random forest-based projection model to generate forward-looking estimates beyond the latest observed inventory year. It should be clarified that the random forest model is not employed as a conventional time-series forecasting approach. Instead, it is trained on multi-period provincial panel data, where temporal information is implicitly reflected through inventory periods and cross-sectional variation. The model is used to project future forest biomass energy potential by learning nonlinear relationships between the dependent variable and its explanatory factors. The overall structure of the random forest-based projection framework is illustrated in Figure 1.

Decision trees are based on repeated multivariate regression or classification of data to achieve a comprehensive evaluation and future prediction of the assessment objects. A random forest consists of multiple decision trees and integrates randomly sampled data and variables to generate aggregated results from multiple trees. The final prediction is determined by selecting the outcome that appears most frequently. The specific classification formula is as follows:

$$f\left(x\right)=q{\left\{h_y\left(x_y^w\right)\right\}}_{y=1}^{w\_tree}$$

(6)

In this equation: ${ x}_y$ represents the y-th test sample that contains w attribute features; $h_y\left(x_y^w\right)$ denotes the prediction result of the y-th decision tree; q refers to the classification result with the highest prediction frequency; and w_tree represents the total number of decision trees.

To provide a holistic view of the methodological design and analytical logic of this study, an integrated research framework is presented in Figure 2.

3. Results and Discussion

3.1. Time-Series Analysis of China’s Forest Biomass Energy Potential

The results indicate that China’s forest biomass energy potential exhibits clear temporal fluctuations from 1998 to 2018, with a declining tendency in the later period. As shown in Figure 3, the kernel density distributions remain relatively slender across inventory years, with no evidence of bimodal or multimodal patterns.

The distribution peak increases from 1998 to 2003, followed by a noticeable downward shift in 2008. Although a partial recovery is observed in 2013, the distribution in 2018 shifts downward again, corresponding to a lower overall level of forest biomass energy potential. Changes in the width and tail of the distributions suggest moderate variation in dispersion over time, while no persistent polarization is detected. Overall, the kernel density results reveal a pattern of stage-specific fluctuation combined with a gradual decline in the national-level distribution of forest biomass energy potential.

3.2. Spatial Analysis of China’s Forest Biomass Energy Potential

The results reveal pronounced spatial heterogeneity in forest biomass energy potential across China. As shown in Figure 4, higher levels of forest biomass energy potential are consistently concentrated in the Southwest and Central South regions, whereas lower levels are predominantly observed in the Northwest and parts of North and Northeast China.

In 1998 and 2003, relatively high-potential areas are mainly clustered in the Central South and eastern Southwest regions, while most provinces in the Northwest and western Southwest remain at lower levels. Although the overall spatial configuration remains broadly stable in 2008, the distribution becomes more dispersed, indicating increasing differentiation among provinces without clear polarization.

By 2013, provinces in Central South, East China, and parts of North China emerge as relatively higher-potential areas, whereas the Northwest, western Southwest, and Northeast regions continue to exhibit comparatively lower levels. The spatial pattern in 2018 shows a similar core–periphery structure, with persistent high-potential clusters in the Southwest and Central South regions and consistently lower levels in the Northwest.

From a regional aggregation perspective, forest biomass energy potential exhibits distinct regional trajectories over time. The Central South and Southwest regions display notable fluctuations, while East China maintains relatively stable spatial positions. In contrast, the Northwest, North China, and Northeast regions generally remain in lower-potential positions throughout the study period, reinforcing the persistence of regional spatial disparities.

Figure 5 summarizes the spatial trajectory, average distribution, and regional inequality of forest biomass energy potential in China from 1998 to 2018.

As shown in Figure 5a, the center of gravity of forest biomass energy potential shifts gradually eastward and slightly southward over time. The standard deviational ellipses become more elongated, indicating an expansion in spatial dispersion across provinces.

Figure 5b reveals a clear and persistent spatial pattern. Higher average forest biomass energy potential is concentrated in the Southwest and Central South regions, forming contiguous high-potential clusters, whereas provinces in the Northwest, North China, and parts of the Northeast consistently exhibit lower potential levels. This core–periphery structure remains stable throughout the study period.

According to Figure 5c, regional inequality in forest biomass energy potential remains pronounced. The Dagum decomposition shows that between-region differences account for the largest share of overall inequality, followed by transvariation intensity, while within-region differences contribute relatively little. This indicates that disparities are primarily driven by structural differences across major regions.

At the regional aggregation level, total forest biomass energy potential exhibits divergent regional trajectories over time. Most regions experience a declining tendency after 2008, with the Northeast showing the steepest decrease and East China the smallest reduction, whereas the Northwest displays a modest upward trend. Persistent differences among provinces within the same regions further highlight substantial spatial heterogeneity.

3.3. Influencing Factors Analysis of Forest Biomass Energy Potential

Table 3. Descriptive statistics of influencing factors for provincial forest biomass energy potential in China (1998–2018).

Variable	Mean	Standard Deviation	Maximum	Minimum
Per capita GDP (CNY)	32,377	27,376	164,158	2489
Proportion of forestry output (%)	4.54	3.65	41.00	0.41
Share of highly educated staff in grassroots forestry stations (%)	42.15	22.10	97.00	1.23
Forest pest and rodent control rate (%)	72.00	21.83	100.00	5.00

reports the descriptive statistics of the key variables affecting forest biomass energy potential. Substantial variation is observed across provinces in terms of economic development, forestry structure, human capital, and forest management conditions, indicating pronounced heterogeneity in the underlying determinants.

The panel regression results are presented in

Table 4. Fixed-effects panel regression results for provincial forest biomass energy potential in China (1998–2018).

Variables	Coef.	Std.Err.	t
VAL	0.028 ***	0.006	4.654
lnPGDP	0.004	0.045	0.086
EDU	–0.007 ***	0.002	–3.744
ORG	–0.002 **	0.001	–2.157
_Cons	6.553 ***	0.385	17.015
F-statistic	0.000 ***	R2-statistic	0.135

Note: ***, **, and * indicate significance levels at 1%, 5%, and 10%, respectively.

. The proportion of forestry output value (VAL) exhibits a significantly positive association with forest biomass energy potential, suggesting that regions with a stronger forestry economic base tend to exhibit higher potential levels. In contrast, the share of highly educated staff in grassroots forestry stations (EDU) shows a significant negative relationship with forest biomass energy potential. Similarly, the forest pest and rodent control rate (ORG) is negatively associated with forest biomass energy potential at the 5% significance level.

Per capita GDP (lnPGDP) does not display a statistically significant effect, indicating that overall economic development alone does not directly translate into higher forest biomass energy potential. The regression results highlight that structural and management-related factors play a more prominent role than general economic conditions in shaping interprovincial differences in forest biomass energy potential.

3.4. Prediction of Forest Biomass Energy Potential Using Random Forest Model

Based on the random forest-based projection model, China’s forest biomass energy potential was projected beyond the latest officially released inventory year (2018). Model performance was evaluated using the coefficient of determination (R²) and the mean squared error (MSE). The results indicate strong predictive capability, with an average R² of 0.9233 and relatively low prediction errors across provinces, suggesting that the projection outcomes provide a reliable representation of future potential under existing structural conditions.

As illustrated in Figure 6, the projected forest biomass energy potential across China’s six major regions exhibits differentiated regional trajectories over the forecast horizon. Overall, the projections indicate a moderate increase relative to the 2018 baseline, reflecting nonlinear relationships captured by the explanatory variables rather than a direct extrapolation of historical trends.

From a regional perspective, the Central South, Northwest, and North China regions display relatively stable projection paths, characterized by mild fluctuations over time. The Southwest and East China regions exhibit comparatively flat trajectories, indicating limited expansion in projected potential, with East China recording the lowest growth rate among all regions. In contrast, the Northeast region demonstrates more pronounced variability, with a noticeable increase in projected potential during the early projection period followed by a slight moderation toward the end of the forecast horizon. Despite this fluctuation, the projected level in the Northeast remains above the 2018 baseline.

Overall, the random forest-based projections highlight persistent regional heterogeneity in future forest biomass energy potential, with differences in projected dynamics across regions remaining evident throughout the forecast period. Detailed province-level goodness-of-fit statistics, including both R² and MSE, are reported in Appendix B. Additional analyses of the projected results, including distributional characteristics, regional inequality structure, and spatial evolution patterns, are presented in Appendix C.

3.5. Discussion

The results of this study indicate that China’s forest biomass energy potential experienced a steady but moderate evolution between 1998 and 2018, accompanied by a gradual narrowing of regional disparities. This overall pattern is consistent with existing assessments suggesting that only a limited share of China’s theoretical biomass resources can be sustainably utilized under ecological and technical constraints. For example, Xu et al. estimated that approximately 27% of theoretical biomass resources are realistically exploitable, although improved spatial allocation could generate substantial climate mitigation benefits [32]. Similarly, Nie et al. projected that even under long-term development scenarios, biomass residues from agricultural and forestry systems would contribute only a modest share to primary energy demand [33]. Together, these findings support the view that forest biomass energy functions primarily as a complementary component within China’s renewable energy system rather than a dominant substitute for fossil energy.

At the same time, it should be recognized that the present analysis is conducted at the national scale to ensure internal consistency and cross-regional comparability. While this framework is appropriate for identifying broad temporal trends and aggregate patterns, future research could further refine the assessment by incorporating more ecologically differentiated parameters as higher-resolution and harmonized forestry data become available. Such extensions would enable a more nuanced understanding of local-level biomass dynamics within China’s highly diverse forest ecosystems.

From a spatial perspective, the observed convergence in forest biomass energy potential across regions aligns with prior studies on forest carbon and productivity dynamics. Yin et al. reported significant spatial spillover effects in forest carbon-sink efficiency, with high-efficiency clusters concentrated in eastern and southwestern China [34]. The kernel density and fixed-effect results in this study similarly indicate that regions characterized by stronger forestry sectors and more effective pest-control systems tend to exhibit higher biomass energy potential. This suggests that spatial coordination of forestry production and ecological management plays an important role in shaping regional biomass outcomes.

The factor analysis further identifies three variables—forestry output share, education level of forestry workers, and forest pest-control effectiveness—as key drivers of the spatiotemporal variation in forest biomass energy potential. This finding is consistent with the conclusions of Ibitoye et al., who emphasized the importance of technical human capital and biological pest control for sustaining long-term forestry productivity and ecological resilience [31]. In addition, Lin et al. demonstrated that higher forestry output efficiency is closely associated with improved resource–environment coordination, lending support to the positive relationship between forestry output share and biomass potential observed in this study [28].

Economic factors, however, exhibited limited explanatory power in this study, a finding that may appear counter-intuitive but is consistent with existing evidence. Mi et al. showed that bioenergy’s contribution to sustainable development in China is primarily driven by policy frameworks and technological progress rather than income growth, and their comparative analysis between China and the United States further demonstrated that bioenergy influences sustainability mainly through structural transformation and innovation effects rather than per capita GDP expansion [35]. In the context of forest biomass energy potential, this result suggests that economic development alone does not directly translate into higher resource potential, which is fundamentally constrained by forest endowments, management intensity, and institutional arrangements. As such, the non-significant effect of per capita GDP may indicate a form of structural decoupling between macroeconomic growth and biomass energy potential at the supply side, rather than a weakening of economic relevance per se. This implies that strengthening the knowledge base, policy coordination, and regulatory consistency of China’s forestry sector could be more effective in enhancing biomass energy potential than relying solely on continued macroeconomic expansion.

Methodologically, the strong predictive performance of the random forest model highlights its suitability for capturing the nonlinear interactions among ecological, economic, and management variables. Previous studies by Ji et al. and Song et al. have demonstrated the robustness of random forest approaches in estimating forest biomass and volume using multisource data [17,18]. Similarly, Yan et al. emphasized that complex nonlinear relationships among vegetation, land use, and socio-economic factors necessitate flexible machine-learning methods for biomass energy assessment [8]. The results of this study further confirm the value of such approaches for analyzing forest biomass energy potential under heterogeneous regional conditions. From a policy perspective, the findings are broadly consistent with international experiences emphasizing coordinated forest governance and sustainability safeguards in forest biomass utilization [36,37].

Finally, although the random forest projections suggest a non-linear future trajectory, the overall upward trend remains subject to ecological and structural constraints. As cautioned by Xu et al. and Nie et al., the sustainable supply ceiling of biomass resources represents a persistent limitation [32,33]. These findings underscore the importance of continued methodological refinement and integrated analytical frameworks in future research to better capture the systemic role of forest biomass energy within broader bioenergy and carbon-sink systems [38].

4. Conclusions and Suggestions

4.1. Conclusions

This study develops an integrated empirical framework to assess and project forest biomass energy potential in China at the provincial level by combining inventory-based estimation, panel data analysis, and machine-learning-based projection. By linking historical assessment with forward-looking projection within a unified analytical structure, the study extends existing biomass energy research beyond static estimation or single-method analysis [39].

The empirical results indicate that China’s forest biomass energy potential experienced stage-specific fluctuations over the period 1998–2018, without the emergence of strong regional polarization. Although regional trajectories differ, overall disparities remained relatively stable, suggesting a balanced national pattern rather than divergence.

The influencing-factor analysis provides new evidence on the determinants of forest biomass energy potential. Sector-specific production structure, human capital in forestry management, and forest protection capacity are identified as statistically significant drivers, whereas general economic development, represented by per capita GDP, does not exert a direct effect. This finding refines existing understandings by distinguishing forestry-specific mechanisms from broader macroeconomic influences.

Building on these determinants, the random forest–based projection reveals that forest biomass energy potential is likely to exhibit a moderate increase relative to the 2018 baseline, followed by gradual moderation over the projection horizon. Importantly, this projection reflects nonlinear structural relationships among influencing factors rather than a mechanical extrapolation of past trends.

This study contributes to the literature in three main aspects. First, it provides a provincial-scale, inventory-consistent assessment of forest biomass energy potential over a long time span. Second, it integrates econometric analysis and machine-learning projection to capture both linear and nonlinear relationships within a single framework. Third, it offers new empirical insights into the long-term evolution and regional differentiation of forest biomass energy potential in China, thereby enriching the existing knowledge base on national bioenergy development.

4.2. Suggestions

Based on the empirical findings, the following recommendations are proposed to support the development of forest biomass energy.

First, the efficiency of the forest biomass value chain should be strengthened. Priority should be given to improving residue collection, processing, and utilization systems in order to transform existing forest resources into a stable and economically viable biomass energy supply [40].

Second, human capital development and forest management capacity should be explicitly prioritized. Strengthening professional training, technical education, and pest and rodent control systems is essential for improving biomass productivity and ensuring the sustainable use of forest resources.

Third, differentiated development strategies should be adopted to account for spatial heterogeneity in forest resources and industrial foundations. Regions with stronger economic and technological bases should focus on efficiency enhancement and innovation, while forest resource–abundant regions should prioritize utilization capacity and ecological protection [41].

These recommendations not only apply to China but also provide reference implications for other developing countries with heterogeneous forest resources and emerging bioenergy sectors [42]. In such contexts, coordinated value-chain development, capacity building in forest management, and region-specific policy design represent practical and transferable approaches for promoting forest-based bioenergy while safeguarding ecological sustainability [43].