A Bibliometric Analysis of the Three-North Shelter Forest Program

Abstract

The Three-North Shelter Forest Program (TNSFP) is a large-scale ecological restoration project that has attracted worldwide attention. It covers 4.069 million km² across 13 provinces in northern China, including northwestern, north-central, and northeastern regions. Bibliometric analysis provides a structural overview of the research in this field and offers insights into key research fronts. We conducted a literature review of the Web of Science Core Collection (WoSCC) from 1990 to 2024 using HistCite for a comprehensive literature analysis and CiteSpace for visualizing research trends and co-citation networks. Based on the literature data from the WoSCC, we performed a bibliometric visualization review of the TNSFP. We observe a rising trend in research on the TNSFP, with the number of publications steadily increasing, especially after 2011. Remote Sensing emerged as the leading journal during the study period, accounting for 8.84% of the total publications. China is the leading contributor to research in this field, comprising 99.32% of the publications, with the Chinese Academy of Sciences (CAS) being the primary research institution, accounting for 36.05%. Research on the TNSFP is interdisciplinary, with Environmental Sciences serving as its primary focus. Ecological restoration and climate change are likely to be the main trends in future research. This study provides a comprehensive overview of the TNSFP’s research landscape, offering insights that can inform policy decisions, guide future research directions, and support on-the-ground conservation and afforestation strategies.

1. Introduction

Environmental degradation has emerged as a critical global issue, contributing to a host of environmental and socio-economic problems such as natural disasters, food insecurity, biodiversity loss, and climate instability [1]. Recent assessments show that nearly 30% of the Earth’s land surface is degraded, impacting the livelihoods of more than 3.2 billion people worldwide and exacerbating these interconnected challenges [2,3]. In recent decades, rapid urbanization, deforestation, and unsustainable land use practices have significantly accelerated the degradation of ecosystems worldwide [4]. Land degradation lies at the heart of agricultural and forest land loss, especially in arid and semiarid regions where vegetation loss and soil erosion are particularly severe. The continued expansion of desertified areas not only diminishes land productivity but also poses serious threats to both human and animal health, contributing to dust storms, water scarcity, and the displacement of vulnerable populations [5,6,7]. These escalating challenges underscore the need for long-term, science-based interventions capable of restoring degraded land and ensuring ecological sustainability.

Ecological restoration is a key strategy for rebuilding ecological balance and reversing environmental deterioration. It seeks to re-establish the structure and function of damaged ecosystems, enhance biodiversity, and strengthen the resilience of natural systems against future disturbances. As the international community advances toward sustainable development goals, ecological restoration has gained prominence in global environmental governance, receiving growing attention through policy initiatives such as the UN Decade on Ecosystem Restoration (2021–2030) [5]. In this global context, China has implemented one of the most ambitious ecological restoration initiatives—the Three-North Shelter Forest Program (TNSFP). Launched in 1978 and organized by the central government, the TNSFP is a large-scale, state-led afforestation and windbreak project designed to combat desertification and control soil erosion. It strategically establishes windbreaking forest strips across the “Three Norths”—Northeast, North, and Northwest China—to halt the southward expansion of the Gobi Desert and stabilize fragile ecosystems. The program is implemented in multiple phases and coordinated at national and provincial levels, involving forest farms, local authorities, and rural communities. As one of the world’s largest shelterbelts, it spans over 35 million hectares and stretches across northern China, forming a “Green Great Wall” [8,9]. It stands as the largest afforestation initiative globally [10]. For comparison, the Great Green Wall of the Sahel aims to restore 100 million hectares across Africa, while India’s Green India Mission targets 5 million hectares. The TNSFP has led to the planting of approximately 66 billion trees across 13 provinces, significantly enhancing forest coverage and ecosystem services. This extensive afforestation has not only mitigated desertification but also improved carbon sequestration and biodiversity. Commonly planted species include Populus (poplar), Salix (willow), Elaeagnus angustifolia (oleaster), Caragana korshinskii (Korshinsk peashrub), and Hippophae rhamnoides (sea buckthorn). The selection of these species is tailored to the specific ecological conditions of each region within the program’s vast area [11,12,13]. The TNSFP plays a pivotal role in a range of ecological protection strategies, including dust storm mitigation, carbon sequestration, soil erosion control, and biodiversity enhancement. These efforts have been central to interdisciplinary research on ecological restoration and sustainable land management [8,14,15,16].

Extensive research has shown that the TNSFP has achieved remarkable progress in land restoration, climate regulation, and environmental protection [17,18,19]. A growing body of literature recognizes the importance of the TNSFP in addressing the ecological challenges of northern China. Studies of the TNSFP show the importance of long-term planning, afforestation techniques, and interregional cooperation [20]. However, most studies in the field of the TNSFP have only focused on specific projects or localized effects, and few have systematically mapped the research landscape [21,22,23,24]. Previous studies of the TNSFP have not dealt with its knowledge structure or long-term development trends in much detail. Such approaches, however, have failed to address the need for a holistic, visual understanding of the evolution and hotspots of this field.

The specific objective of this study is to conduct a bibliometric analysis of the research on the TNSFP from 1990 to 2024. This study sets out to investigate the development trends, collaborative networks, and emerging research topics in this field. We initially used HistCite to gain a preliminary understanding of the citation history and to generate historiographic maps that highlight the chronological development of influential publications. However, due to HistCite’s discontinued support and limited analytical capabilities, we subsequently adopted CiteSpace as the primary tool for bibliometric visualization and knowledge domain mapping. CiteSpace, which remains under active development, is widely recognized for its robust performance in uncovering research fronts, co-citation clusters, and emerging intellectual trends. This study provides new insights into the research structure and thematic evolution of the TNSFP. It offers an important contribution by identifying key journals, countries, institutions, and research fronts, potential directions for future research. Understanding these patterns enables a more comprehensive assessment of the achievements and persistent challenges of the TNSFP, while highlighting the integrative nature of this study across ecology, forestry, and bibliometric science. By mapping the intellectual structure and evolution of TNSFP-related research, this work provides strategic insights for both scholars and policymakers. Moreover, it contributes to global climate adaptation efforts and supports the implementation of sustainability-oriented policies through evidence-based approaches to land restoration and ecological resilience.

2. Materials and Methods

2.1. Data Sources

We retrieved bibliometric data directly from the Web of Science Core Collection (WoSCC) using our institution’s standard subscription access [25]. WoSCC is one of the most well-known and comprehensive scientific literature databases, covering the most important and influential research outputs worldwide [26]. The specific combination of sub-databases included in this subscription (e.g., SCIE, SSCI, or BKCI) is determined by Clarivate’s licensing arrangements, and we were unable to independently verify whether certain collections—such as the Book Citation Index—were included. As a result, some literature types, particularly books or book chapters, may not be represented in our dataset. Furthermore, the WoSCC primarily indexes English-language, peer-reviewed journals, which may limit coverage of Chinese-language research, especially important in the context of the TNSFP. We acknowledge this limitation and suggest that future studies integrate additional databases such as CNKI, Google Scholar, or OpenAlex to ensure broader linguistic and regional representation. Our study spanned the period from 1 January 1990, to 1 January 2025. The search formula was as follows: TS = (“Three North Shelter”) OR TS = (“Three North Shelterbelt”) OR TS = (“Three North Afforestation”) OR TS = (“Three North Region”) OR TS = (“Three North Protection Forest”). A total of 147 papers were initially retrieved. We cleaned the data by including only English-language publications. The literature type was limited to “article” and “review.” After filtering, a total of 139 papers were included, consisting of 132 articles and 7 reviews. These publications were exported from the WoSCC in plain text format. All available information, including the number of documents, citation counts, authors, affiliations, countries, titles, keywords, publication years, journals, and references, was downloaded and integrated for bibliometric analysis and visualization [27].

2.2. Bibliometric Analysis Methods

This paper presents a comprehensive overview of publication output, journal distribution, leading countries and institutions, disciplinary interactions, key research themes, collaboration networks, and research trends related to the TNSFP.

2.2.1. Bibliometric Methods

The methods and tools used in this study are as follows: CiteSpace and HistCite were employed for bibliometric analysis and data visualization. The CiteSpace software, a freely available literature analysis tool programmed by Dr. Chaomei Chen at the College of Information Science and Technology at Drexel University, is widely used for recognizing and visualizing research data [28]. This tool reveals development trends and rules of the data from critical points of view. HistCite is a citation mapping and analysis tool developed by Dr. Eugene Garfield. It graphically illustrates the relationships between various scientific works within a specific field [29]. This allows users to quickly trace the development history of a field, identify key literature in that domain, and access the most recent significant contributions. It should be noted that HistCite is no longer under active development, which limits its compatibility with contemporary Clarivate systems.

In this study, the retrieved literature data were first imported into CiteSpace (6.4.R1 version) and HistCite. A new database was then established to store both the original data and the analytical results. Information was extracted through methods such as social network analysis and cluster analysis, employing a series of algorithmic operations to construct data networks and cluster maps [30,31]. Furthermore, the evolution of keywords and research trends was examined using timeline visualizations and cluster analysis, aiming to explore the dynamic changes in research hotspots [32].

2.2.2. Analysis Methods

To analyze collaborative relationships, we conducted a Social Network Analysis (SNA). SNA is a powerful tool for revealing structural patterns and interdependencies within collaborative networks [33,34]. Network density, which indicates the level of connectivity and reflects the overall integrity and complexity of the network, is calculated as defined in Equation (1).

$$Density={\displaystyle \frac{2m}{n(n-1)}}$$

(1)

In Equation (1), m represents the actual number of relationships, and n represents the nodes in the network. When the density of a network reaches 1, every individual is connected to all others; conversely, a density of 0 indicates that no relationships exist between individuals. A density above 0.1 is generally seen as a sign of sufficient network connectivity [35]. Lower-density networks can still offer valuable insights when modularity and weighted mean silhouette values are high [36].

In the WoSCC, discipline classification is primarily based on the Web of Science Subject Categories (WCs). This schema encompasses approximately 250 subject areas across science, social sciences, and arts and humanities. Each journal indexed in the WoSCC is assigned to one or more of these categories, and articles inherit the categories of their respective journals. This classification system facilitates detailed bibliometric analyses by allowing for comparisons within specific subject areas. However, the assignment of multiple categories to a single journal can lead to overlapping coverage, potentially complicating analyses. To address this, some studies have implemented article-level reclassification methods to assign unique subject categories to individual articles, thereby enhancing the precision of discipline-specific analyses [37].

Modularity is a key metric used to assess the degree of clustering among nodes in a network, with $Q$ representing the modularity value, as defined in Equation (2).

$$Q={\displaystyle \frac{1}{2m}}{\sum }_{ij}[A_{ij}-{\displaystyle \frac{k_i{k}_j}{2m}}]\delta (c_i,c_j)$$

(2)

In Equation (2), $m$ denotes the total number of edges in the network, $k\mathrm{i}$ represents the sum of the weights of edges connected to node $i$, and $\delta (c\mathrm{i}, c\mathrm{j})$ is an indicator function that equals 1 if nodes $i$ and $j$ belong to the same cluster, and 0 otherwise. A $Q$ value greater than 0.3 is generally considered indicative of a meaningful cluster structure [38,39].

The Weighted Mean Silhouette $S\left(i\right)$ for each node $i$ is used in CiteSpace to evaluate the consistency of clustering results and is defined in Equation (3):

$$S\left(i\right)={\displaystyle \frac{b\left(i\right)-a\left(i\right)}{max\{a\left(i\right), b\left(i\right)\}}}$$

(3)

In Equation (3), $a\left(i\right)$ denotes the average distance between node $i$ and all other nodes within the same cluster, while $b\left(i\right)$ represents the minimum average distance from node $i$ to all nodes in any other cluster. In general, a weighted mean silhouette value above 0.5 indicates a reasonable clustering structure, while values exceeding 0.7 are regarded as highly convincing and reflect a well-defined knowledge domain [40,41,42].

Centrality metrics offer a quantified method to identify crucial points in a network with different specialties or tipping points [43,44]. Centrality can be mathematically defined as Equation (4).

$$Centrality(nod{e}_i)={\sum }_{i\ne j\ne k}{\displaystyle \frac{p_{jk}(i)}{p_{jk}}}$$

(4)

In Equation (4), $p_{jk}$ represents the number of shortest paths between node j and node k, and $p_{jk}\left(i\right)$ is the number of those paths that go across node i.

Burst detection can identify variables that vary significantly over a short period of time [45]. CiteSpace employs Kleinberg’s burst detection algorithm to identify sudden increases in citations [46]. In a burst time $t_1$ to $t_2$, the burst strength is defined as Equation (5).

$$Burst Strength=\sum _{t_1}^{t_2}(\sigma \left(0,r_t,d_t\right)-\sigma \left(1,r_t,d_t\right))$$

(5)

In Equation (5), $r_t$ stands for the relevant documents out of a total of $d_t$ in the $t^{th}$ batch, and $\sigma \left(0,r_t,d_t\right)$ represents the cost of the automaton.

3. Results

3.1. Research Outputs and Publication Trends

3.1.1. Yearly Outputs

As shown in Figure 1, the overall trend in publications related to the TNSFP has been steadily rising. Notably, an increase began after 2010. Prior to 2015, the annual number of publications remained below 10. However, post 2015, a rise occurred, with annual outputs doubling the average of previous years. This pattern suggests that the TNSFP has gradually emerged as a growing research focus. The publication trend can be broadly divided into three phases. From 2004 to 2010, research activity was minimal, with only two publications recorded in 2004 and none in the subsequent 6 years. This publication gap may be attributed to practical and ecological challenges during that period. Notably, afforestation efforts from the 1980s to the early 2000s primarily involved monoculture plantations, particularly of poplar trees, which became increasingly vulnerable to pest infestations and environmental stress in arid and semi-arid regions. In response, forest management strategies shifted toward mixed-species plantations after 2000 to improve ecosystem stability and resilience. However, these strategic adjustments required time for implementation and ecological observation before yielding publishable scientific results. Additionally, much of the effort during 2005–2010 was focused on large-scale ecological engineering and restoration in degraded areas such as the Loess Plateau. Scientific publications may have lagged behind on-the-ground progress. Moreover, natural disturbances, such as the severe snowstorm in 2008 that damaged newly planted forests, likely redirected institutional efforts toward recovery rather than academic output during this period [47]. Between 2011 and 2014, the field began to gain traction, with 3–5 papers published annually, marking its early developmental stage. A notable growth phase began in 2015, as the number of publications rose to 10 and continued to increase steadily, peaking at 15 in both 2020 and 2021. In 2022 and 2024, the number of publications peaked at 17 articles.

In addition to publication volume, Figure 1 also presents the trend in citations, which reflects the academic influence of the published works. The Total Global Citation Score (TGCS) measures the number of times a publication has been cited across the WoSCC database, while the Total Local Citation Score (TLCS) indicates citations within the analyzed dataset [48]. The TGCS peaked in 2015 at 665 citations, while the TLCS reached its highest point in the same year at 46 citations. However, both metrics have shown a noticeable decline in recent years, with the TLCS dropping to zero in 2022 and 2024 and the TGCS decreasing to just 14 citations in 2024.

3.1.2. Distribution of Journals

Table 1. Top 20 journals publishing papers on research of the TNSFP.

	Journal	Recs	TLCS	TGCS
1	Remote Sensing	13	17	417
2	Sustainability	10	0	84
3	Agricultural and Forest Meteorology	7	27	354
4	Forests	7	0	103
5	Land Degradation & Development	6	8	134
6	Frontiers in Plant Science	4	0	25
7	Journal of Cleaner Production	4	11	352
8	Ecological Indicators	3	23	339
9	Journal of Hydrology	3	0	88
10	Science of the Total Environment	3	0	181
11	Catena	2	0	62
12	Energies	2	0	67
13	Environmental Earth Sciences	2	28	145
14	Environmental Monitoring and Assessment	2	8	57
15	Environmental Research Letters	2	0	28
16	Forest Ecology and Management	2	0	67
17	Frontiers in Environmental Science	2	0	12
18	International Journal of Remote Sensing	2	4	46
19	Journal of Arid Land	2	2	28
20	Land Reclamation in Ecological Fragile Areas	2	1	3

Recs stands for the publication amount. TGCS stands for Total Global Citation Score, and TLCS stands for Total Local Citation Score.

presents the top 20 journals publishing literature studies related to the TNSFP, including their publication counts (Recs), Total Local Citation Score (TLCS), and Total Global Citation Score (TGCS). The journal Remote Sensing leads with the highest number of publications, contributing 13 articles and accounting for 8.84% of the total. It also demonstrates strong global influence, with a TGCS of 417. However, its TLCS is 17, which is relatively moderate. This suggests that while the journal enjoys broad interdisciplinary reach, its impact within the core research community may be more limited. Sustainability ranks second in publication count with 10 articles, making up 6.80% of the total. Despite a TGCS of 84, it has a TLCS of 0, indicating broader but less targeted academic engagement. Agricultural and Forest Meteorology shows strong performance in both global and local impact. With seven publications, it records a TLCS of 27 and a TGCS of 354, even surpassing Remote Sensing in local citations despite having fewer papers. Environmental Earth Sciences stands out for its citation efficiency. Although it has published only two papers, it achieves a TLCS of 28 and a TGCS of 145, highlighting its niche but influential role in the literature. Other journals, such as the Journal of Cleaner Production, with four publications and a TGCS of 352, and Ecological Indicators, with three publications and a TGCS of 339, underscore the interdisciplinary nature of research in this field. In contrast, Forests, with seven publications, and Science of the Total Environment, with three publications, reflect a stronger emphasis on global impact, with TGCS values of 103 and 181, respectively.

3.1.3. Annual Publication by Country

Figure 2 illustrates the publication trends related to the TNSFP from 1990 to 2024. As shown, research articles in this field began to emerge in 2004. Since 2011, the number of publications has exhibited a rapid upward trend, followed by a steady output in recent years. Notably, China has consistently contributed the majority of publications, highlighting its significant domestic research advantage in this area.

3.2. Contributor Performance and Collaboration Patterns

3.2.1. Productive Countries

Note that the number of publications attributed to each country is based on whether a paper includes at least one author from that country, as per the WoSCC’s data reporting. This means that a single paper with authors from multiple countries will be counted once for each country, but the actual number of authors per country is not reflected in these counts. We summarized the top 10 contributing countries to the research on the TNSFP in

Table 2. Top 10 productive countries studying the TNSFP.

	Country	Recs	TLCS	TGCS
1	China	146	240	4839
2	USA	25	86	1401
3	Australia	4	7	106
4	Canada	3	32	408
5	Belgium	1	0	8
6	France	1	0	12
7	Germany	1	0	1
8	Italy	1	3	22
9	Japan	1	28	132
10	Mongolia	1	0	10

Recs stands for the publication amount. TGCS stands for Total Global Citation Score, and TLCS stands for Total Local Citation Score. Publication and citation counts were generated using HistCite, which applies a full counting method. Each country represented by at least one author of a given publication receives full credit for the publication and all its citations.

. China has generated the majority of publications and citations in this field, comprising about 99.32% of the total publications, with 240 TLCS and 4839 TGCS. The USA follows with 17.01%, 86, and 1401, respectively.

3.2.2. Productive Institutions

A total of 198 institutions have published articles related to the TNSFP. In

Table 3. Top Contributing Institutions to the TNSFP.

	Institution	Recs	TLCS	TGCS
1	Chinese Academy of Science	53	166	2437
2	Beijing Normal University	20	51	1229
3	Beijing Forestry University	19	19	302
4	University of Chinese Academy of Science	16	21	669
5	Chinese Academy Forestry	15	15	516
6	Peking University	9	3	416
7	Nanjing Forestry University	6	5	155
8	Tsinghua University	6	0	151
9	University of Maryland	6	20	242
10	Northeast Forestry University	5	3	112

Recs stands for the publication amount. TGCS stands for Total Global Citation Score, and TLCS stands for Total Local Citation Score.

, we highlight the top 10 institutions with the highest publication volume. Chinese institutions hold a dominant position in this field, accounting for over 90% of the publications among the top 10 contributing institutions. Among them, the Chinese Academy of Sciences (CAS) stands out as the leading research institution, contributing 36.05% of the total publications and publishing nearly three times as many papers as the second-ranked institution. Meanwhile, the CAS has the highest citation counts, with a TLCS of 166 and a TGCS of 2437. Other major contributors include Beijing Normal University and Beijing Forestry University, both of which contribute a significant portion of the total publications.

3.2.3. Scientific Collaboration of Countries

In this study, we analyzed the country-level collaboration network, as shown in Figure 3. The country network consisted of 13 collaborative pairs among 13 countries, with a network density of 0.1667, indicating relatively sparse international cooperation in this field. Within the network, China ranked first in both publication output and centrality, with a centrality score of 1.59. China ranked first in both the number of publications and international collaborations related to the TNSFP, indicating its dominant role in this research field [49,50,51]. The United States held the second position, with a centrality of 0.02, indicating far less engagement compared to China.

3.2.4. Scientific Collaboration of Institutions

As shown in Figure 4, in the institutional collaboration network, the Chinese Academy of Sciences (CAS) ranked first in both publication count and centrality, with values of 53 and 0.80, respectively. This indicates that CAS was highly active not only in publishing but also in establishing connections with other institutions in this field. Although the China Academy of Forestry (CAF) and the Institute of Remote Sensing & Digital Earth (IRSD) did not have particularly high publication counts, they ranked second and third in centrality. This suggests that both institutions placed strong emphasis on collaboration. In contrast, the Shenyang Institute of Applied Ecology (SIAE), despite producing a substantial number of publications, had a centrality value of 0. This implies that SIAE conducted its research largely independently, with minimal collaboration.

3.3. Thematic and Disciplinary Structure

3.3.1. Discipline Interaction

In Figure 5a, we can see that there were 87 interdisciplinary and mutual application relationships among 35 disciplines. The overall map density was 0.1462, indicating relatively close interdisciplinary connections. In Figure 5b, Environmental Sciences ranked first, with a count of 71, accounting for 51.08% of the total, and had a centrality of 0.48. Following closely, Geosciences, Multidisciplinary ranked second with a count of 28 (20.14%) and a centrality of 0.17.

The discipline interaction networks are illustrated in Figure 5, highlighting both the extent and strength of interdisciplinary cooperation in research on the TNSFP. We can infer that Environmental Sciences and Geosciences are the core disciplines in this field. Research on the TNSFP includes Forestry, Remote Sensing, Imaging Science, Meteorology, Soil Science, Environmental Studies, and Water Resources, all of which are closely related to environmental research [52]. Among these disciplines, Environmental Sciences recorded both the highest centrality and the largest number of publications, indicating its dominant position within the network. Remote Sensing had a lower publication count but ranked high in centrality, suggesting a strong interdisciplinary linkage. Forestry, by contrast, exhibited a high number of publications but low centrality, pointing to limited its connectivity with other fields.

3.3.2. Keywords Analysis

The relationships between keywords in academic literature can be reflected through their co-occurrence frequency, with higher frequencies indicating stronger thematic connections [53,54,55,56]. Typically, a higher co-occurrence frequency indicates a stronger connection between two research topics. The co-occurrence map helps visualize research domains and the evolution of research hotspots. We conducted a keyword co-occurrence analysis and obtained a total of 293 keywords. We selected the top 10 keywords based on their significance. Figure 6 presents the most influential themes and reveals the interconnectedness among major research areas related to the TNSFP.

All ten keywords significantly contributed to the evolution of research in this domain, with almost all showing centrality values exceeding 0.1. Among the keywords, “Climate Change” stands out with the highest centrality and publication count, reaching values of 0.33 and 35, respectively. “Desertification” shows a centrality value of 0.23 and a publication count of 20. Similarly, “Loess Plateau” demonstrates notable influence, with a centrality of 0.18 and a count of 19. In addition, the keywords “Climate” and “Dynamics” also exhibit relatively high centrality values, while “Ecological Restoration” stands out for its higher publication count. “Climate Change,” “Desertification,” and “Loess Plateau” were identified as high-frequency keywords, highlighting the core themes, primary research directions, and key study areas of the TNSFP [57].

To explore the intellectual structure and research trends in the field of the TNSFP, we employed CiteSpace to conduct a keyword co-occurrence cluster analysis. The clustering was based on co-occurring keywords within the same publications, with cluster labels generated using the log-likelihood ratio (LLR) algorithm [58]. As shown in Figure 7, the timeline visualization captures the temporal progression and shifting focus of major research themes in the field.

The network comprises 296 nodes and 1300 links, where each node represents a keyword and each link indicates co-occurrence relationships. Keywords that first appeared in a particular year are marked accordingly, while the accumulation of frequency over time is visualized through the thickness of the colored rings surrounding each node. Each color represents a specific year, allowing for an intuitive understanding of keyword trends across time. Meanwhile, the clustering analysis identified 10 major keyword clusters, including “#0 precipitation”, “#1 conversion model”, “#2 ecological restoration”, “#3 tree rings”, “#4 spatiotemporal variation”, “#5 carbon density”, “#6 vegetation index”, “#7 spatio-temporal variation”, “#8 wind speed”, and “#9 air temperature”. In the timeline view, each cluster is represented as a horizontal line, with the associated keywords arranged from left to right according to their earliest occurrence. The vertical ordering reflects the relative size of the clusters, which correlates with the number of keywords contained. The quality of the clustering results is evaluated using the modularity (Q) and weighted mean silhouette value (S). The modularity Q value is 0.5211, exceeding the commonly accepted threshold of 0.3, indicating a significant and well-partitioned network structure. In addition, the weighted mean silhouette value (S) of 0.831, which surpasses the threshold of 0.5, suggests that the clustering is highly convincing. Values above 0.7 are generally considered indicative of a reliable and coherent cluster structure, supporting the structural validity of the keyword clusters presented in Figure 7. This approach enables a clear visualization of the shifts in research focus within the TNSFP over time in response to changing research priorities and environmental challenges.

Figure 8 displays the relatively high-frequency keywords in time order. We can identify relatively high-frequency keywords for recent years: climate change, desertification, ecological restoration, loess plateau, dynamics, climate, cover, afforestation, land use, etc. By tracking keyword evolution in publications from a time zone perspective, we find that research derived from the TNSFP has been increasingly expanding. The evolution of keywords can be divided into three distinct phases. Phase I: 2004–2010, Phase II: 2011–2019, Phase III: 2020–2024.

During Phase I (2004–2010), high-frequency keywords were relatively scarce, with “Remote Sensing” being the most prominent among them. During this period, research on the TNSFP was relatively limited and primarily focused on the field of remote sensing [59]. The relatively small node size of this keyword reflects, to some extent, that studies on the project were still in their early stages during Phase I. In Phase II (2011–2019), research keywords related to the TNSFP experienced an explosive increase, with the emergence of terms such as “climate change”, “desertification”, “ecological restoration” and “loess plateau”. During this period, the scope of research expanded significantly, focusing primarily on climate and ecological aspects [60,61,62]. This phase marked a trend toward diversification in the study of the TNSFP. Phase III, spanning from 2020 to 2024, features keywords such as “biodiversity” and “conservation”, indicating that research on the TNSFP has continued to expand into broader biological and ecological restoration domains [63].

3.4. Knowledge Base and Emerging Research Fronts

3.4.1. Co-Citation Cluster

Figure 9 illustrates the co-citation network related to the TNSFP, comprising 491 nodes and 1630 links. High modularity (Q = 0.8296) and weighted mean silhouette value (S = 0.9402) indicate a well-structured and reliable clustering result. To highlight the core research directions, only the six most representative co-citation clusters are displayed in the figure: #0 north shelterbelt region, #1 northern China, #2 spatial variation, #3 Gansu-Ningxia region, #4 shelter forest, and #5 intrinsic water-use efficiency. To maintain clarity, clusters with fewer connections or peripheral relevance were excluded, which may result in non-sequential cluster numbering in the visualization. Cluster #0 north shelterbelt region, Cluster #1 northern China, and Cluster #3 Gansu-Ningxia region represent the core research areas, primarily clustered in northern and northwestern China. Cluster #2 spatial variation reflects a growing interest in the heterogeneous responses across both space and time. In this field, the main focus lies in the in-depth analysis of spatial changes in the land use, vegetation cover, and productivity of the TNSFP. Cluster #4 shelter forest continues to represent foundational studies on the structure, function, and effectiveness of shelterbelts, with ongoing discussions on their role in wind erosion control and ecosystem services. Lastly, Cluster #5 intrinsic water-use efficiency reflects more recent investigations into plant physiological responses under afforestation, particularly how water-use strategies adapt to climatic stressors in arid zones. Meanwhile, these clusters point toward emerging research frontiers, indicating a shift toward finer-scale ecological assessment and regional adaptability within the TNSFP.

3.4.2. Cited Publication

A clear research emphasis has emerged around ecological restoration and environmental change in the context of the TNSFP. Most of the publications were concentrated in ecological and environmental journals, reflecting strong disciplinary alignment.

Table 4. The top 5 cited publications in the Three-North Shelter Forest field.

Title	Year	Author	Journal	Freq
China and India lead in greening of the world through land use management [64]	2019	Chen C	Nature Sustainability	22
Characteristics, drivers and feedbacks of global greening [65]	2020	Piao SL	Nature Sustainability	12
Payoff from afforestation under the Three-North Shelter Forest Program [21]	2020	Cao SX	Journal of Cleaner Production	10
Assessing the Three-North Shelter Forest Program in China by a novel framework for characterizing vegetation changes [66]	2017	Qiu BW	ISPRS, Journal of Photogrammetry and Remote Sensing	9
Multiple afforestation programs accelerate the greenness in the ‘Three North’ region of China from 1982 to 2013 [67]	2016	Zhang Y	Ecological Indicators	9

displays the five most frequently cited publications related to research on the TNSFP.

As shown in Table 4, one of the most frequently cited works is by Chen et al. (2019) [64], titled “China and India lead in greening of the world through land-use management”, published in Nature Sustainability (22 citations). In a related study, Piao et al. (2020) [65], in “Characteristics, drivers, and feedbacks of global greening” (Nature Sustainability, 12 citations), explored greening as an emerging climate feedback mechanism. Cao et al. (2020) [21] assessed the direct impact of the TNSFP in their study titled “Payoff from afforestation under the Three-North Shelter Forest Program”, published in Journal of Cleaner Production (10 citations). Their research provided empirical evidence showing that afforestation projects in semi-arid to arid northern China yielded significant ecological payoffs, including reductions in desertification and improvements in local microclimates. Focusing on methodological innovation, Qiu et al. (2017) [66] introduced a novel analytical tool in the study “Assessing the Three-North Shelter Forest Program in China by a novel framework for characterizing vegetation changes” (ISPRS, Journal of Photogrammetry and Remote Sensing, nine citations). This work presented the COTTS framework, which integrates MODIS data to monitor temporal and spatial patterns of vegetation change from 2001 to 2015, offering a more precise way to evaluate the program’s effectiveness. Meanwhile, Zhang et al. (2016) [67], in the article “Multiple afforestation programs accelerate the greenness in the ‘Three North’ region of China from 1982 to 2013” (Ecological Indicators, nine citations), analyzed long-term NDVI data from multiple satellite sources. The study revealed a significant three-decade increase in vegetation greenness, primarily driven by human-led afforestation and sustained ecological interventions against climate change.

3.4.3. Strongest Citation Burst

To detect emerging research trends and sudden shifts in scholarly focus, citation burst detection was employed. A citation burst indicates a sharp increase in citations over a specific time span, reflecting a publication’s rapid academic influence. This method, commonly applied in bibliometric studies, enables the identification of key research turning points. The timeline visualization presents the temporal patterns of these bursts, where the blue bar marks the publication year, the red segment denotes the burst period, and the light blue portion shows the pre-publication years. Figure 10 highlights a selection of highly influential papers published between 2004 and 2024 that experienced significant citation bursts.

As shown in Figure 10, the strongest citation bursts reveal a shifting research focus on the TNSFP, China’s largest afforestation project, from early vegetation assessments to broader ecological and global environmental implications. The earliest burst appeared with studies such as Duan et al., which utilized long-term AVHRR NDVI data from 1982 to 2006 to assess vegetation dynamics across the TNSFP region. Their study revealed spatial and temporal patterns of vegetation change, highlighting areas of greening and degradation, and identified precipitation as the primary climatic factor influencing NDVI trends, particularly in temperate desert regions of northwestern China [68]. Following this, Wang et al. conducted an assessment of the TNSFP’s effectiveness in addressing desertification and dust storms. While acknowledging the program’s large scale and long-term efforts, the study suggested that more comprehensive and systematic evaluations are needed to fully understand its environmental impacts [69]. Zhang et al. observed a significant increase in vegetation greenness in China’s “Three North” region from 1982 to 2013. Their analysis of satellite-derived NDVI data indicated that this trend was largely associated with multiple afforestation programs, highlighting the positive impact of sustained ecological restoration efforts [67]. Chen et al. highlighted China’s significant role in global greening trends, attributing substantial increases in vegetation cover to land-use management practices, including large-scale afforestation efforts such as the TNSFP. Their study emphasized that these initiatives have contributed notably to reversing land degradation and enhancing ecological restoration [64]. Qiu et al. introduced the COTTS framework, using MODIS 500 m 8-day composite data (2001–2015) to assess vegetation dynamics in the TNSFP. This method detected vegetation change patterns with 90.08% accuracy and a kappa coefficient of 0.8688. The study refined the evaluation of the program’s ecological impact, revealing spatial variability in vegetation trends [66]. Feng et al. suggested that large-scale revegetation efforts could have implications for regional water resources. Their findings offer valuable insights for assessing water sustainability in projects like the TNSFP [70]. Cao et al. (2020) conducted a cost–benefit analysis of the TNSFP, revealing that traditional afforestation methods yielded negative net benefits due to high costs, and underscored the need for tailored approaches to maximize ecological and economic outcomes in large-scale afforestation projects [21]. Piao et al. (2020) identified CO₂ fertilization as a key driver of global greening, with regions like China, including the TNSFP area, seeing significant greening due to afforestation. Their findings highlight the need to balance ecological benefits with potential hydrological impacts in large-scale projects [65]. Their work underscored the program’s relevance in shaping Earth system processes and influencing climate-related outcomes.

4. Discussion

The analysis above reveals that the research field of the TNSFP has experienced significant development over the past few decades. Prior to 2011, the field was in its infancy, with relatively few publications related to the program. However, from 2011 onwards, the number of publications increased dramatically, signaling a phase of rapid development. The annual distribution of publications serves as an indicator of research development and scholarly interest over time [71,72]. Despite the relatively modest number of publications in the field, earlier studies received a high average citation count, reflecting superior quality and notable academic influence. However, citation metrics have shown a declining trend in recent years, suggesting that while research output continues to grow, its academic impact may be waning. This highlights a need for more impactful and widely recognized research in the future. A significant portion of the research on the TNSFP has been published in journals dedicated to Environmental Sciences, as well as in related fields such as Remote Sensing and Sustainability. Remote sensing, in particular, has seen the highest volume of publications, underscoring its importance in the monitoring and assessment of the region’s ecological improvements. This is reflective of the growing reliance on advanced technologies to collect and analyze environmental data. China has been the dominant contributor to TNSFP research, not only in terms of publication volume but also in fostering international cooperation. This is largely attributed to the fact that the TNSFP is based in China and was launched by the Chinese government, which places strong emphasis on its outcomes and long-term impact [73]. This suggests that China is not only the most productive country in the field of the TNSFP but also the most active in terms of global collaboration. The United States ranks second in terms of publications, but with significantly fewer studies. Tests show that while international collaboration is prevalent, the USA’s involvement primarily relies on cooperative research rather than independent studies. Similarly, although Australia and Canada ranked third in publication output, their zero centrality values indicate that their research activities were largely independent, with minimal integration into international collaboration networks. The collaboration between institutions, however, has been relatively weak, indicating a crucial area for future improvement. Enhanced cooperation, both among countries and between institutions, is vital for advancing the research and implementation of the TNSFP.

Within China, the Chinese Academy of Sciences (CAS) stands out as the leading institution, contributing nearly three times the number of publications of the second-ranked institution, Beijing Normal University, with Beijing Forestry University following closely behind. This dominance of Chinese institutions reflects their central role in driving research in this field and maintaining a leadership position in TNSFP studies. An important underlying factor contributing to this institutional dominance is financial support. Since the TNSFP is a major national initiative organized and funded by the Central Government, funding allocation plays a critical role in determining research capacity. Institutions with greater financial resources are better positioned to acquire study materials, conduct fieldwork, access advanced technologies, and support publication efforts. As a result, disparities in funding may directly influence both the quantity and quality of research output. In addition to funding, geographical location may also influence institutional involvement and academic centrality. For example, the Shenyang Institute of Applied Ecology (SIAE), located in Shenyang at the eastern end of the shelterbelt and far from the Gobi Desert—the primary focus area of many afforestation efforts—may have received comparatively less research funding. This geographic positioning might also partly explain its lower level of participation in national collaboration networks, as reflected by its zero centrality value in the institutional collaboration analysis. Simultaneously, the research direction has also been shaped by global agendas, particularly the United Nations’ ongoing promotion of climate change mitigation, ecological restoration, and socio-economic sustainability. These international priorities have increasingly influenced national research agendas, encouraging interdisciplinary approaches and aligning TNSFP studies with broader global environmental and developmental goals. Environmental Sciences has been the primary discipline in the TNSFP research, with significant contributions from related fields such as Geosciences, Forestry, Agriculture, Remote Sensing, and Imaging Science. A major focus of the research has been the monitoring of environmental improvements in the TNSFP, particularly in terms of vegetation coverage and soil biomass, which are critical indicators of ecological progress. In addition, the program’s impact on other environmental factors, such as changes in atmospheric composition, the local water cycle, and the carbon cycle, has also been a key research focus. Remote sensing and imaging technologies have been widely employed in data collection and information extraction, highlighting the crucial role of technology in advancing research in this field [24]. Given the China-specific nature of the TNSFP, and to address the limitations of relying solely on the Web of Science Core Collection (WoSCC), we further incorporated a comprehensive literature search using the China National Knowledge Infrastructure (CNKI) database. Applying the same retrieval criteria and period (1 January 1990, to 1 January 2025), we identified 1122 relevant publications. As shown in Figure 11, our bibliometric analysis of the CNKI dataset, focusing on institutions, keywords, and countries, revealed a more granular institutional distribution, dominated by Chinese institutions. Notably, the State Forestry Administration Three-North Protected Forest Construction Bureau emerged as the most prolific institution in CNKI but was underrepresented in WoSCC due to the predominance of Chinese-language publications. Meanwhile, institutions like the Chinese Academy of Sciences (CAS) and the Shenyang Institute of Applied Ecology (SIAE) showed strong presence in both databases. These findings underscore the complementary nature of CNKI and WoSCC: CNKI captures more localized, policy-driven institutional activity, whereas WoSCC reflects broader international collaboration networks and visibility. Together, they provide a more comprehensive and representative overview of institutional involvement in TNSFP research.

A closer look at the shift in research keywords reveals several key trends in the evolution of the field. Early studies primarily concentrated on environmental issues such as desertification, climate, and vegetation cover, reflecting the initial goals of the TNSFP to combat land degradation. Over time, however, the research focus broadened to include topics such as climate change and ecological restoration, reflecting a growing interest in ecosystem functioning and its global environmental implications [74]. Later studies have shifted their attention to vegetation dynamics, afforestation, and the effectiveness of related policies. The emergence of region-specific keywords, such as the Loess Plateau, indicates a growing emphasis on localized studies within the broader theTNSFP framework. More recently, research has integrated ecological processes with human-environment interactions and sustainable land management practices, signaling a transition from problem-driven assessments to more holistic, system-based evaluations of long-term ecological and socio-environmental impacts [75]. A comparison of keyword analysis between CNKI and WoSCC reveals significant thematic differences. In the CNKI dataset, the top keywords are generally broad and practice-oriented, such as Countermeasures, The Three-North Project, Construction, Shelter Forest, Suggestion, Problem, Engineering Construction, Effectiveness, Construction Achievements, and The Three North Regions. Among these, Countermeasures had the highest frequency. These keywords reflect the emphasis of Chinese-language literature on policy implementation, practical challenges, and localized project outcomes, aligning with the administrative and engineering-based nature of TNSFP in China. The WoSCC keyword network is characterized by more specialized and academic terms. All top ten keywords in WoSCC contribute significantly to the evolution of research, with nearly all exhibiting centrality values above 0.1. Climate Change stands out with the highest centrality (0.33) and publication count (35), followed by Desertification (centrality 0.23, count 20) and Loess Plateau (centrality 0.18, count 19). Other keywords such as Climate, Dynamics, and Ecological Restoration also show either high centrality or frequency. These terms reflect global environmental concerns and suggest that WoSCC-indexed literature approaches TNSFP from broader scientific and international perspectives. This contrast highlights the complementary nature of the two databases: CNKI captures grounded, policy-driven, and regionally tailored insights, while WoSCC reflects academic, theoretical, and globally relevant research themes. The integration of both provides a more comprehensive and balanced understanding of TNSFP research, as shown in Figure 12.

Future research on the TNSFP is anticipated to place greater emphasis on various aspects of ecological recovery, especially related to soil, water, and atmospheric conditions. As forest coverage reaches certain thresholds and the shelterbelts begin to fulfill their protective functions, subsequent studies are likely to emphasize remote sensing monitoring, spatial changes in land use and vegetation cover, as well as ecological restoration processes. To assess the long-term effects of the TNSFP, scholars are likely to develop more comprehensive evaluation systems by employing methodologies such as integrated modeling approaches, participatory research frameworks, and multi-scale ecological assessments. Given the different afforestation timelines and ecological contexts across regions, research priorities are expected to vary by location and stage of development. Additionally, future research may expand to include perspectives from economics and the humanities, in addition to ecology, thereby providing a more interdisciplinary and integrated understanding of the program’s long-term environmental, social, and economic impacts [76,77,78].

Despite the strengths of this analysis, certain limitations must be acknowledged. The analysis tools and data preprocessing methods used in this study were not without flaws. Although a precision retrieval strategy and manual data cleaning were employed to filter out irrelevant articles, some omissions were still detected, which could impact the accuracy of the research front analysis to a degree [79]. Additionally, the focus on highly cited and influential articles may have resulted in the exclusion of more recent studies with innovative ideas that have yet to gain significant attention [80]. Furthermore, the notable discrepancy between the global citation score (TGCS) and the local citation score (TLCS) may be partially attributed to the dominance of Chinese-language publications that are not fully indexed in international databases such as Web of Science. Since many important studies concerning the TNSFP are published in Chinese for domestic policy and implementation purposes, their citations may not be reflected in TLCS, potentially underestimating local academic influence. Due to limited access to platforms such as InCites, normalized citation indicators like the Category Normalized Citation Impact (CNCI) could not be included, which may affect the comparability of citation performance across fields. While this study primarily relied on TGCS and TLCS for citation analysis, the limitations of unnormalized metrics are acknowledged. As a supplementary exploration, FWCI values were retrieved for selected journals, institutions, and countries. Journals such as Remote Sensing (FWCI = 13.0), Sustainability (10.0), and Agricultural and Forest Meteorology (7.0) demonstrated strong field-normalized citation impact. Similarly, institutions like the Chinese Academy of Sciences and Beijing Normal University, which ranked highly in TGCS and TLCS, also exhibited high FWCI scores of 3.31 and 3.23, respectively. At the national level, China presented a notably high FWCI of 2.36, compared to 1.00 for Canada and the UK. These observations suggest a degree of alignment between raw citation metrics and normalized impact indicators in this research field. Although FWCI was not incorporated into the main analytical framework to maintain methodological consistency, these supplementary results support the overall reliability of the findings. Future research may benefit from the integration of broader data sources such as Google Scholar or OpenAlex, along with standardized metrics like FWCI, to achieve a more balanced and comprehensive evaluation. Ongoing improvements in open bibliometric platforms are expected to further enhance the accuracy and completeness of such analyses. We hope that their limitations can be addressed in future updates, further improving the accuracy and comprehensiveness of research analyses in the field.

5. Conclusions

In this study, to investigate the research trends of the TNSFP, we conducted a bibliometric analysis of relevant publications retrieved from the WoSCC. This analysis aimed to explore publication trends, major research directions, and potential future research trajectories. The main findings are as follows:

(1)

Research on the TNSFP remained limited before 2011 but grew rapidly thereafter. Despite the modest volume, publications have high citation rates, indicating strong academic influence. Major studies have appeared in environment-oriented journals like Remote Sensing and Sustainability. China dominates the field, with the Chinese Academy of Sciences (CAS) leading in publication volume.

(2)

Interdisciplinary connections were relatively strong, as indicated by a density of 0.1462 based on 87 links among 35 disciplines. Research on the TNSFP is centered around Environmental Sciences and Geosciences, with strong interdisciplinary integration involving fields such as Forestry, Remote Sensing, Soil Science, and Ecology.

(3)

Research on the TNSFP initially focused on remote sensing, with later emphasis on keywords such as climate change, desertification, and ecological restoration.

(4)

Citation studies of the TNSFP highlight key themes like vegetation dynamics, afforestation impact, and global greening. Authors such as Chen C and Cao SX have made significant contributions, while future research will focus on broader environmental impacts.

In conclusion, research on the TNSFP has experienced rapid growth since 2011, with increasing academic influence particularly in the fields of Environmental Sciences and Geosciences. The evolution of TNSFP research reflects a shift from technical remote sensing studies to holistic environmental assessments, underscoring its growing role in global ecological restoration discourse. This transformation highlights the program’s significance not only for national policy and ecological practice but also for informing large-scale afforestation and desertification control initiatives worldwide.

Looking ahead, future research is expected to move beyond general environmental impacts to address several critical and specific areas. These include the long-term ecological sustainability of shelterbelts, especially in arid and semi-arid zones where vegetation resilience remains uncertain; the socio-economic impacts on local communities, including changes in livelihoods and land use patterns; and the integration of climate adaptation strategies to enhance the program’s resilience to shifting climatic conditions. Furthermore, advances in technology—such as remote sensing, artificial intelligence, and big data analytics—offer new opportunities to monitor, evaluate, and optimize afforestation outcomes at multiple spatial and temporal scales. Addressing these emerging research areas will not only strengthen the scientific foundation of the TNSFP but also enhance its relevance to global efforts in ecosystem restoration and sustainable land management.