Bibliometric analysis is a valuable research methodology for studying scientific literature, allowing researchers to identify global research trends based on academic publication outputs (Alsharif and Baharun, 2020). Scopus, recognized as one of the leading multidisciplinary bibliographic databases, is widely used for conducting bibliometric analysis (Md Khudzari et al., 2018). With its extensive journal coverage and robust citation tracking system, Scopus enables detailed performance evaluations and publication trend analyses across multiple scientific fields (Wang and Waltman, 2016). The bibliometric analysis of this study employed research articles collected from the Scopus database. The methodology framework is presented in Fig. 1, with each stage of data collection, processing, and analysis outlined. The search query “TITLE-ABS-KEY (“deep tunnel*” OR “tunnel” OR “deep storage*” AND “stormwater” OR “storm water” OR “CSO” OR “sewer overflow*” OR “wastewater*” OR “waste water”)” was used to identify relevant articles published between 1968 and 2024. In this study, the term “deep tunnel” was used as a general term to include a range of related structures and systems, such as “deep tunnel sewerage systems”, “deep tunnel sewage systems”, “deep tunnel drainage systems”, and “deep tunnel”. This broader term captures the different terminologies used in the literature to describe underground infrastructures designed for stormwater and sewage management. The search query was further filtered only to include articles published in English resulting in a total of 585 papers. The papers were saved as a Research Information System (RIS) file and processed into a corpus using Cortext Manager, an online bibliometric analysis tool that generates clusters of related keywords to identify the principal themes within the datasets (El Akkari et al., 2018).

The research contributions of various constituents within the potential application of deep tunnel systems were assessed. Metrics such as publication numbers, citations, and authorship patterns were utilized to gain insights into the productivity of researchers, institutions, and countries. This analysis helped identify leading contributors and observe trends over time. Moreover, performance analysis provided descriptive data on the productivity and impact of different entities, forming a solid foundation for bibliometric studies (Moral-Muñoz et al., 2020). In addition, VOSviewer and CorText Manager were used to perform science mapping techniques. These tools include citation and co-citation analysis to explore the intellectual structure and relationships between research constituents, such as authors, institutions, and topics. Specifically, VOSviewer was used to create bibliometric maps based on network data, enabling the visualization of relationships between scientific articles, journals, researchers, and keywords (Van Eck and Waltman, 2010). On the other hand, CorText facilitated the mapping of co-occurrences among keywords, the analysis of collaboration networks between countries, and the tracking of publication patterns across years and journals (Maniquiz-Redillas et al., 2022).

The original search query “TITLE-ABS-KEY (“deep tunnel*” OR “tunnel” OR “deep storage*” AND “stormwater” OR “storm water” OR “CSO” OR “sewer overflow*” OR “wastewater*” OR “waste water”)” used for the bibliometric analysis yielded limited data on water quality focusing primarily on other aspects of deep tunnel systems research such as urban planning considerations, role in flood control and drainage, and computation of hydraulic properties and flow trajectories (He et al., 2024; Sun et al., 2024). Thus, the comprehensive review utilized a separate search query focusing on urban influent water quality rather than deep tunnels. This reflects the current lack of research that links deep tunnels to water quality improvement, underscoring a gap in the literature. The alternative query was designed to capture insights into stormwater runoff management and water quality treatment in urban settings.

The search query “TITLE-ABS-KEY (“urban” AND “influent*” OR “inlet*” AND “CSO*” OR “combined sewer overflow*” OR “stormwater*” OR “storm water*” OR “sewer overflow*” OR “drainage*” AND “water quality” OR “runoff quality” OR “concentration*”)” limited to open-access and English-language articles was used resulting in 108 documents. This approach enabled a comprehensive evaluation of various water treatment practices, particularly for urban stormwater runoff by examining factors influencing treatment efficiency. The data included information on stormwater sources from different land uses, treatment facility types, influent and effluent pollutant concentrations, and pollutant removal efficiencies. An understanding of stormwater influent variations is deemed essential for the development of effective stormwater quality guidelines and the implementation of deep tunnel systems (Ho et al., 2013).

The bibliometric analysis revealed notable growth in research on deep tunnel systems for stormwater management and flood mitigation. Fig. 2 exhibits the number of published articles and cumulative citations from 1968 to 2023. After 2010, a notable acceleration in the publication trend was observed, culminating in 27 articles and a cumulative citation count of 6,285 in 2023. This upward trend in publications aligns with a global focus on infrastructure solutions to mitigate the impacts of increasingly severe weather patterns. The sharp rise in studies during the 21^st century may be linked to the implementation of deep tunnel systems in major cities such as Chicago, Kuala Lumpur, Sydney, and Tokyo (Nagare et al., 2020), where these systems have been deployed to manage urban flood risks more effectively. Such infrastructure has gained attention as a viable response to climate-induced flooding challenges. Generally, the rising interest in deep tunnel system research aligns with a global movement toward sustainable infrastructure solutions aimed at enhancing urban resilience.

The top contributing journals about deep tunnel systems are presented in Table 1. Among the 585 obtained articles, Tunnels and Tunneling International led with 63 publications. This journal has established a primary source of research and case studies on underground infrastructure, focusing on the technical aspects of tunnel design and practical applications within urban environments. The main themes in Tunnels and Tunneling International centered on challenges and innovations in large-scale tunneling projects. The journal emphasized construction techniques, advanced tunnel-boring machine (TBM) technologies, and strategies for overcoming complex geotechnical conditions. Case studies on urban stormwater management and flood control, where deep tunnel systems are used to manage CSO were frequently featured in the said journal. Additionally, Tunnels and Tunneling International includes interdisciplinary topics such as corrosion protection, sustainable construction materials, and integrating tunnels into existing urban infrastructure (Marshall et al., 2007; Shiomi, 2006).

Tunneling and Underground Space Technology contributed 19 publications, often focusing on innovative underground technologies. Emphasis was placed on seismic resilience and structural analysis of tunnels, particularly in earthquake-prone regions, as well as on the interactions between tunnels and geological conditions (Avunduk et al., 2021; Avunduk and Copur, 2018; Sousa et al., 2018). Meanwhile, the Journal of Hydraulic Engineering and Civil Engineering contributed 15 publications each, whereas both World Tunneling and Water Science and Technology had 12 publications each.

Table 2 presents the top 10 most-cited journals in deep tunnel systems research, with the Journal of Hydraulic Engineering leading with 517 citations, followed by the Chemical Engineering Journal with 377 citations. This high citation count reflected the interdisciplinary nature of deep tunnel systems research, encompassing civil engineering, environmental science, and material science. A major theme in Journal of Hydraulic Engineering articles involved managing complex flow behaviors in stormwater systems. Both experimental and numerical modeling approaches were commonly used to simulate transient flows and optimize tunnel designs for resilience under extreme weather conditions (Muller et al., 2017; Vasconcelos and Leite, 2012; Vasconcelos and Wright, 2011).

A comprehensive keyword co-occurrence network analysis was performed to map the research landscape regarding deep tunnel systems. From 5,543 unique keywords obtained from 585 documents, only 336 were mentioned at least five times, illustrating intense emphasis on several topics. Fig. 3 categorizes these terms into six distinct clusters, each denoted by a color that indicates thematic study inclinations. The keyword analysis revealed consistent themes within the research domain, centered on urban water infrastructure and management, with prominent terms like wastewater treatment, stormwater, sewer network, and tunneling.

The red cluster is structured around several interconnected topics within wastewater treatment and pollution mitigation, with each keyword representing a distinct yet related area. Core keywords such as water pollution, phosphorus, organic pollutants, and pollutant removal are emphasized, reflecting the cluster’s breadth across chemical and biological treatment methods, including adsorption, coagulation, and membrane filtration. Research within this cluster addresses both organic and inorganic pollutants, including polycyclic aromatic hydrocarbons, toxic metals, nitrogen, and phosphorus, which are commonly found in tunnel wash waters and construction wastewater. High pollutant removal efficiencies are highlighted through combined treatment methods, such as organic sorbent materials and multi-stage filtration systems, aligning with stringent regulatory standards across diverse wastewater contexts (Kim et al., 2015; Paruch and Roseth, 2008).

The blue cluster centered on stormwater and urban runoff management, branching into interconnected topics that address the challenges posed by urbanization and intense rainfall events. Key areas included the effects of infrastructure construction and flood events on pollution and sediment levels in water systems, with studies examining both metal and organic contamination in rivers. Within this context, deep tunnel systems have been identified as an emerging infrastructure solution for managing stormwater overflow and mitigating flooding (Martínez-Santos et al., 2015; Wu et al., 2016). Meanwhile, the green cluster focused on sewer networks and tunneling infrastructure, with major keywords highlighting TBM technologies for large-diameter drainage tunnels and the role of geological factors in guiding project decisions (Newman et al., 2016; Shen et al., 2012). The yellow and cyan clusters explored topics of wastewater treatment plant efficiency, waste recycling, and the complexities of constructing durable water infrastructure that adheres to environmental standards. Collectively, each cluster addressed various aspects of water management and infrastructure that respond to the demands of urban environments.

Distinct patterns in research activity across various years and regions are shown in the contingency matrix presented in Fig. 4(a). The keyword “United States” was strongly correlated with research from 1994 and 2008, indicating early leadership in deep tunnel systems research. Particularly, Chicago’s tunnels and reservoirs to manage CSOs and urban flooding were explored (Pluth et al., 2021). Germany and the United Kingdom led sustainable urban systems and flood risk mitigation research, launching initiatives to improve flood resilience and environmental sustainability in 2018 (Blätter et al., 2021; Grafe and Hilbrandt, 2019). China’s research activity intensified between 2019 and 2023, aligning with the rapid urbanization of cities like Guangzhou, where managing heavy rainfall has become a priority, exemplified by deep tunnel systems application (Wu et al., 2016).

A keyword co-occurrence matrix is presented in Fig. 4(b). Terms like “tunneling” and “sewer network” were revealed prominent in 2007. Over time, terms such as “wastewater treatment,” “water quality,” and “wastewater” gained prominence, driven by urbanization and factors like residential and commercial development, wastewater treatment capacity, CSO frequency, and stormwater drainage efficiency (Pluth et al., 2021). Research on deep tunnel systems has since diversified, focusing not only on construction but also on water quality monitoring. This shift marks an evolving role for deep tunnel systems, which now addresses water quality concerns alongside stormwater storage (Grafe and Hilbrandt, 2019). Consequently, deep tunnel systems are regarded as essential infrastructure for both flood mitigation and pollution control, bolstering climate resilience in urban areas.

Characterizing urban runoff quality is essential for the effective design of deep tunnel systems and the optimization of urban water quality management. A detailed assessment of pollutant types and concentrations in runoff, such as sediments, nutrients, and heavy metals, enables deep tunnel systems to be more effectively tailored to target specific contaminants, control pollutant discharges, manage primary pollution sources, and improve wastewater treatment options to mitigate environmental impact (Sousa et al., 2018). This data supports informed decisions on storage and treatment capacities necessary for managing urban stormwater runoff (Ho et al., 2013). Ultimately, characterizing runoff quality informs strategies that safeguard public health, maintain water quality standards, and enhance climate resilience in urban environments.

Influent pollutant concentrations in urban stormwater runoff across different land use types are presented in Fig. 5. Specifically, TSS, TN, and TP were three of the most commonly analyzed water quality parameters in the reviewed papers. Mixed urban areas showed the highest range of TSS concentrations, with a mean of 238.3 mg/L. In a study of a mixed urban area by Nejad and Zecchin (2020), a high influent TSS of 550.9 mg/L was recorded, largely due to the high imperviousness (77%) of the area, with asphalt and mixed surfaces promoting runoff carrying fine solids and particulates. This urban setting captures intense runoff with minimal natural filtration. Conversely, in a study by Kuster et al. (2022), a much lower influent TSS concentration of 12 mg/L was reported, as stormwater from a predominantly residential and commercial watershed with reduced imperviousness led to lower sediment loads.

The analysis of TN concentrations revealed that urban roads had the highest mean concentration at 3.6 mg/L, with concentrations ranging from 1.5 mg/L to 5.2 mg/L. Guerrero et al. (2020) noted that vehicular activity consistently contributes nitrogen through atmospheric deposition and runoff, leading to relatively steady TN levels on urban roads compared to residential areas. In a study by Lucke et al. (2014), mixed-use urban landscapes exhibited inconsistent TN levels due to varied pollutant sources and land uses. The highest mean TP concentration was observed in urban roads at 0.49 mg/L, followed by urban residential at 0.24 mg/L, and mixed urban at 0.21 mg/L. Mixed urban areas displayed the TP variability likely due to the mix of residential, commercial, and industrial activities that contribute phosphorus from diverse sources such as fertilizers, detergents, and organic waste (Kuster et al., 2022).

Table 3 presents the influent concentrations of key water quality parameters in urban stormwater runoff from mixed land uses, highlighting the wide range of pollutants that impact urban water quality. The pH of runoff averaged near-neutral at 7.39 ± 0.79. In a study by Ho et al. (2024), pH was found to fluctuate due to influences like vehicle emissions, industrial pollutants, or concrete dust from construction. Dissolved oxygen (DO) levels ranged from 2.06 mg/L to 10.43 mg/L, with lower values suggesting high organic pollution environments may deplete oxygen, impacting aquatic life. The average chemical oxygen demand (COD) was 221.17 ± 90.98 mg/L, while biochemical oxygen demand (BOD) averaged 17.59 ± 22.46 mg/L, these values highlight the potential for organic pollution across locations, suggesting sources like organic waste, leaf litter, or roadside debris. Moreover, heavy metals, such as lead (Pb) and cadmium (Cd), were also detected, with Pb at 6.89 ± 3.92 µg/L and Cd at 0.72 ± 1.35 µg/L, likely originating from urban dust and road runoff.

Low impact development (LID) systems help reduce the volume of runoff entering stormwater infrastructure by promoting infiltration, evapotranspiration, and on-site water retention (Kim et al., 2024). This reduces the burden on deep tunnel systems, which are often designed to handle overflow during heavy rain events. With LID systems managing smaller rainfall events, deep tunnel systems can focus on larger, less frequent events, improving overall system efficiency. Studies included in the comprehensive review often utilized treatment facilities such as constructed wetlands (CW), bioretention filters (BF), bioretention systems (BR), and detention basins (DB) in order to effectively address pollutant removal.

Fig. 6 compares the removal efficiencies of various LID systems for three key pollutants: TSS, TN, and TP. CW consistently demonstrated the highest removal efficiencies, achieving 90% for TSS, 68% for TN, and 79% for TP, reflecting their strong capacity for sediment capture and nutrient reduction. In contrast, DB showed the lowest removal rates for TN and TP, with efficiencies of 10% and 13%, respectively, indicating a limited ability to address nutrient pollution. This low removal efficiency underscores the limitations of DB in treating nutrient-rich stormwater, as they primarily focus on peak flow reduction rather than pollutant removal (Humphrey et al., 2022). This limited effectiveness contrasts with CW, which utilizes permanent pools and vegetation to create anaerobic conditions that promote nitrogen reduction through microbial processes, underscoring the advantages of CW in pollutant treatment (Al-Rubaei et al., 2017).

Despite exhibiting low nutrient removal efficiencies, DB achieved high TSS removal at 68% which could be attributed to sedimentation. These systems store stormwater to prevent flooding and allow suspended solids to settle out before being released into the environment or storm sewers (Lange et al., 2023). BF focuses on capturing and filtering pollutants through media layers, such as soil or engineered media, before releasing treated water into the drainage system, typically without infiltration (Drapper and Hornbuckle, 2015). BF exhibited varied removal efficiencies ranging from 38% to 54% for TSS, TN, and TP. Meanwhile, BR, a stormwater management practice that uses a vegetated basin with soil layers to filter and infiltrate runoff, promoting pollutant removal and groundwater recharge (Lange et al., 2020), showed moderate removal efficiencies, achieving 74.48% for TSS, 26.28% for TN, and 39.33% for TP, suggesting they may be less effective than CW, particularly for TSS and TP. Intense storms with high flow rates can bypass the filtration capacity, particularly for dissolved nutrients, which require specific soil conditions and adequate retention times for effective treatment through biochemical and physiochemical processes (Shrestha et al., 2018).