Application of Internet of Things (IoT) Technologies in Green Stormwater Infrastructure (GSI): A Bibliometric Review

Abstract

This bibliometric review elucidates the emerging intersection of Internet of Things (IoT) technologies and Green Stormwater Infrastructure (GSI), demonstrating the potential to reshape urban stormwater management. The study analyzes a steadily increasing corpus of literature since 2013, pointing out considerable international collaboration. Prominent contributions originate from the United States, Canada, Italy, China, and Australia, underscoring the global acknowledgement of the potential of IoT-enhanced GSI. Diverse GSI applications such as green roofs, smart rain barrels, bioretention systems, and stormwater detention ponds have demonstrated enhanced efficiency and real-time control with IoT integration. However, existing literature reveals several challenges, notably the requirement of advanced monitoring, the development of predictive optimization strategies, and extensive scalability. Comprehensive cost–benefit analyses are also critical for the widespread acceptance of IoT-integrated GSI. Current research addresses these challenges by exploring innovative strategies such as microbial-fuel-cell-powered soil moisture sensors and large-scale RTC bioretention systems. Emphasis is also on the need for security measures against potential digital threats. Future research needs to focus on real-time data-based monitoring plans, model validation, continuous optimization, and supportive policy frameworks. As the world confronts urban development, climate change, and aging infrastructure, IoT and GSI synergism presents a promising solution for effective stormwater management and enhancement of cultural ecosystem services. Continued exploration in this promising domain is crucial to pave the way for smarter, greener urban environments.

1. Introduction

In accordance with United Nations projections, the world’s urban populace will experience an increment of 2.5 billion by 2050 [1]. The burgeoning urbanization, however, imposes significant environmental, economic, and social burdens on cities [2,3]. These challenges range from amplified poverty, unemployment, crime rates, and political unrest to intensifying environmental concerns such as biodiversity loss, pollution, and an escalating frequency of natural disasters.

Of the numerous adverse impacts of urbanization, stormwater hazards hold particular prominence due to their substantial contribution to global-scale natural disasters, exacerbated by ongoing climate change. Acknowledging these hazards, diverse global initiatives have sought to implement ecosystem services techniques and nature-based solutions (NBS), including green stormwater infrastructure (GSI) [4,5,6]. GSI carries significant implications for attenuating the impacts of heat island effect, boosting stormwater management capacity, mitigating climate change effects, and reducing environmental pollution [7,8,9]. The preponderance of literature supporting GSI attests to the substantial groundwork in this sphere, substantiating the well-established nature of knowledge in this field.

However, the advent of the Information Age has introduced a set of unique challenges to stormwater management. A significant evolution in data collection methods has been spurred by the proliferation of the Internet of Things (IoT) [10,11]. Conceptualized by the MIT Auto-ID Laboratory in 1999, the IoT has fueled investigations into the transformation of urban systems into “smart cities”, enabled by technological advancements in computation, information and communication technology, IoT, robotics, and autonomous systems [12,13,14]. The emergence of “smart cities” has led to the exploration of technologies that can be harnessed to more efficiently monitor and manage GSI and NBS. These advancements are critical for helping cities adapt to climate change and deliver ecosystem services more effectively [15,16,17]. Such interventions pave the way for sustainability, transforming cities into intricate social–ecological–technological systems, wherein technology and environmental functionality dictate the necessary changes [17,18,19].

Incorporating smart tools and technologies into GSI has been identified as a natural next step for stormwater management. A plethora of smart stormwater management systems have been proposed, mostly by integrating sensing, control, communication, and computing capabilities. These advancements can augment the operation, performance, and capacity of existing or new infrastructure through enhanced data capture, analysis, and control of infrastructure systems. These methods have been tested in the realm of water management for monitoring, controlling, and managing water volume and quality, and for supporting decision-making practices in water distribution networks [20,21,22,23,24,25]. Despite these strides, the integration of emergent smart technologies in GSI faces multiple challenges. One significant barrier is the lack of standardization in the terminology used in IoT literature, which impedes efficient information retrieval and summarization [26]. For a concept to attain widespread dissemination, innovation, and further adaptation, it necessitates maturation, particularly when delineating requisite standards to orchestrate workflows across diverse applications. However, discerning the degree of relevance remains an intricate endeavor [27]. In pragmatic applications, synchronization of proprietary formats across devices facilitates manufacturers in amalgamating data from heterogeneous formats, thereby engendering a standardized system [28,29]. A judicious approach might encompass a comprehensive aggregation and classification of salient terminologies. This could subsequently be bolstered by promoting ubiquitous recognition and utilization of these terms, culminating in the standardization of protocols in the operational phase. The second obstacle lies in defining the relevance of the literature to the IoT, as the term “smart” might be restricted to a buzzword merely illustrating a new development trend. Consequently, this vague representation risks oversimplifying a plethora of related applications [26]. For the purposes of this paper, “smart” stormwater systems are defined as those that utilize the Internet of Things (IoT) and associated technologies to achieve more responsive, efficient, and sustainable management of stormwater runoff. This involves the integration of sensors, controllers, actuators, and wireless communication modules. By fitting stormwater management components such as rainwater harvesting and water quality monitoring equipment with these devices, cities can function as real-time, distributed treatment plants. These systems can be linked to existing stormwater infrastructure through reliable and cost-effective actuators, such as valves and pumps, allowing for precise water flow control. Moreover, real-time data on water quantity and quality are collected, stored in cloud databases, and subsequently analyzed for further network-wide distribution as required. This dynamic nature of “smart” systems, which can adapt their operations in response to individual storm events, sets them apart from traditional static stormwater infrastructure that does not offer such real-time monitoring and control capabilities [30].

To circumvent these challenges, a thorough understanding of the current state of the field is crucial. By leveraging the bibliometric tools “Citespace” and “Bibliometrix R package”, we were able to trace the progression of the field and zero in on its salient aspects. For instance, Liu et al.’s [31] study employed Citespace and the R package to delve into the nuances of green roof ecosystem services research. Aria’s work [32] underscores the adaptability of the Bibliometrix R package, with numerous review articles benefiting from its capabilities. This underscores bibliometrics as a potent approach in identifying pivotal research themes and offering robust research evaluation. This study emphasizes “smart” GSI, exclusively delving into IoT-related technologies. By meticulously scrutinizing the quintessential lexicon and acumen in the domain, it elucidates a theoretical rationale for the convergence of these disciplines, distinguishing it from extant literature. Preceding scholarship has yet to furnish a temporal narrative detailing the evolution of this emergent interdisciplinary nexus. Moreover, employing a bibliometric approach to discern the foundational literature in this avant-garde intersection not only buttresses the understanding of its current developmental trajectory but also underscores imminent challenges and aspirations.

This paper sets out to delve into the existing literature on the integration of IoT technologies in GSI for stormwater management, with a focus on three main objectives: (1) offering a dynamic review of the research in this domain and synthesizing it comprehensively; (2) encapsulating how the IoT correlates with GSI, pinpointing the directions of state-of-the-art research and delineating its constraints; and (3) forecasting future trajectories and limitations of this sector.

2. Terminology Clarification

In this exploration of green infrastructure’s role in stormwater management, the subtle yet distinct applications of IoT are addressed. Recognizing the potential for confusion given the broad spectrum of terms employed across the literature [26,33], it is essential to define the specific language used in this review upfront. This linguistic clarity not only facilitates accurate discourse on the topic but also paves the way for distinct discussions related to ancillary topics such as urban planning, urban ecology, and urban landscapes.

• Internet of Things/IoT:

The term “IoT” denotes the information network resulting from interconnecting various sensing devices to the internet. The overarching objective of IoT is to facilitate connections and interactions between people and objects through an information network [34].

• Smart/Smarter Technology/Intelligent Control Technology:

“Smart technology” is characterized by the ability to sense, monitor (collect data), communicate, manage, analyze, integrate, regulate, or optimize equipment in a methodical manner. This broad category encompasses standalone devices with inherent “smart” functionalities as well as technologies that can be retrofitted to improve the operational efficiency of other tools or networks. Owing to the diverse range of technologies encompassed within “smart technologies”, it becomes challenging to collate all applications. Furthermore, varying degrees of “smartness” may exist, necessitating the discernment of what can be technically classified as “smart” [26].

• Real-Time Control/RTC:

The term “smart” finds its roots in the established research field of real-time control (RTC) for stormwater management [26,35,36,37,38]. RTC is predicated on the continuous monitoring of process data (such as water levels and flow) and dynamic adjustments of flow conditions using flow control devices (e.g., pumps, sluice gates, and movable weirs) for the real-time management of existing urban design systems [39,40]. RTC systems can be classified as local control systems or system-wide control systems, based on their complexity and control scope [39,41]. RTC potentially offers a cost-effective solution, contingent upon the system type and scale [42].

• Information and Communication Technology/ICT:

Information and Communication Technology (ICT) refers to the technical and application systems leveraging contemporary information technologies such as computers, communication, and network technology for acquiring, processing, transmitting, and sharing information. ICT’s utility extends across multiple sectors, including data collection, monitoring, intelligent control, and optimization in stormwater management systems.

• Active Control:

An active control system involves the use of a device, such as an actuator, to influence an asset in some way. Either local or global control systems may drive the actuator. Notably, assets utilizing mechanical principles, such as vortex flow controls to regulate flow velocities, are not included in active control [26].

3. Methodology

3.1. Literature Search Strategy

The ensuing discourse provides a systematic conceptual framework to facilitate a logical narrative (Figure 1).

For conducting a robust bibliometric analysis, the Web of Science Core Collection database is frequently utilized [43]. The analysis encapsulates the tracking of numerical data and keyword shifts over time, facilitating an understanding of the evolving themes of interest [44,45]. Search phrases such as “IoT*”, “green infrastructure*”, and “stormwater” were used to filter the titles, abstracts, and keywords in the Web of Science database covering the years 1990–2023. The “*” denotes words that are related or similar to the base word. This search strategy yielded 145 documents, complete records of which were downloaded to extract relevant information including title, abstract, authorship, document type, publication date, journal, and references. Following a thorough full-text reading, 53 documents were retained for further analysis (Figure 2).

3.2. Data Processing and Analysis

The data recorded in this review encompass aspects such as source countries, themes, intercountry cooperation networks, scientific fields, and research institutions. Visualization tools “Citespace” and “Bibliometrix R package” were used to represent these data graphically [32,46,47]. The “Bibliometrix” R package offers a comprehensive and robust analysis, well suited for visualizing data that are directly statistical in nature. In parallel, “CiteSpace” was employed to uncover implicit patterns across various knowledge domains. The frequency of keywords within the dataset is represented by text and point size, while the width of lines connecting points indicates the co-occurrence frequency of two keywords. This hotspot view effectively presents a quick understanding of the relevant direction in a field. Through the use of these quantitative analysis tools, essential features of the knowledge base in this field are elucidated, including global publication status, literature growth rate, influential authors, most cited articles, and emerging research hotspots and trends. Graphical knowledge maps were also created to further understand the intersection of GSI and IoT.

The primary steps involved in using the bibliometric tools include: (1) Importing the screened 53 papers into Bibliometrix and CiteSpace in text format; (2) Extracting visual information such as publication outputs, authors, and countries/regions in Bibliometrix, and subsequently reviewing and summarizing practices in the relevant fields based on this information; (3) Employing CiteSpace to categorize research in the field by identifying hotspots and temporal changes; (4) Leveraging the knowledge network diagram created by the previous steps to gain insights into the field, analyze its limitations, and provide recommendations.

4. Results and Discussion

4.1. Overview and Practices

4.1.1. Overview

Graphical analyses illustrate that scholarly discourse surrounding the integration of IoT and GSI began only as recently as 2013. Despite periodic fluctuations, the yearly count of relevant publications shows an upward trajectory as of June 2023 (Figure 3). The per annum literature growth rate stands at 21.48%, although the factors contributing to this fluctuation remain unknown. Of the 53 publications identified, each garnered an average of 11.98 citations, while the co-authorship index was measured at 4.28. International collaboration among authors was determined to be 28.3%. Predominantly, authors from the United States, Canada, Italy, China, and Australia contributed to these 53 papers, with 38, 14, 9, 8, and 7 authors, respectively. These metrics indicate that the field is in its nascent stage but gaining recognition globally, with different nations conducting diverse experiments and practices to investigate its potential. The authors making the most significant contributions to this domain are Oberascher, M., Rauch, W., and Sitzenfrei, R., with substantial work in the study of smart rain barrels. The ten most cited papers predominantly concentrate on RTC and research on GSI practices (

Table 1. Ten most cited works in the field of IoT and GSI.

Number	Publications	Authors	Citations
			Average per Year	Total
1 [48]	Improving the performance of stormwater detention basins by real-time control using rainfall forecasts	Gaborit, E.; Muschalla, D.; Vallet, B.; Vanrolleghem, P.A.; Anctil, F.	5.18	57
2 [49]	Emerging investigators series: building a theory for smart stormwater systems	Mullapudi, A.; Wong, B.P.; Kerkez, B.	6.43	45
3 [50]	The role of monitoring sustainable drainage systems for promoting transition towards regenerative urban built environments: a case study in the Valencian region, Spain	Perales-Momparler, S.; Andres-Domenech, I.; Hernandez-Crespo, C.; Valles-Moran, F.; Martin, M.; Escuder-Bueno, I.; Andreu, J.	6.14	43
4 [51]	Ecohydraulic-driven real-time control of stormwater basins	Muschalla, D.; Vallet, B.; Anctil, F.; Lessard, P.; Pelletier, G.; Vanrolleghem, P.A.	4.2	42
5 [52]	lmproved reliability of stormwater detention basin performance through water quality data-informed real-time control	Sharior, S.; McDonald, W.; Parolari, A.J.	6.2	31
6 [53]	Integrated stormwater inflow control for sewers and green structures in urban landscapes	Lund, N.S.V.; Borup, M.; Madsen, H.; Mark, O.; Arnbjerg-Nielsen, K.; Mikkelsen, P.S.	6	30
7 [54]	Potential advantages in combining smart and green infrastructure over silo approaches for future cities	Kaluarachchi, Y.	7.25	29
8 [55]	An integrated optimization and rule-based approach for predictive real-time control of urban stormwater management systems	Shishegar, S.; Duchesne, S.; Pelletier, G.	5.2	26
9 [56]	Research Development on Sustainable Urban Infrastructure From 1991 to 2017: A Bibliometric Analysis to Inform Future Innovations	Du, H.B.; Liu, D.Y.; Lu, Z.M.; Crittenden, J.; Mao, G.Z.; Wang, S.; Zou, H.Y.	5.2	26
10 [20]	Stated preferences for smart green infrastructure in stormwater management	Meng, I.; Hsu, D.	4.8	24

). As can be seen in Table 1, the most researched and cited are those with applications on detention basins. A lot of the stormwater management research related to the IoT has yet to be aggregated and defined in greater detail, as the systematic framework of the topic has not yet been fully formed. The fact that there are numerous papers on review kinds in this field further supports this.

The advent of GSI in the historical timeline is relatively recent. During this era, RTC has been demonstrated to allow cost-effective applications that increase network capacity while preserving future adaptive options through retrofitting or control-rule iterations and upgrades [57,58,59]. Although the majority of research focuses on modeling studies over practical implementation, the Louisville and Jefferson County Metropolitan Sewer District (Louisville MSD) in Kentucky and the MAGES system in Paris stand as remarkable examples of network-scale smart stormwater systems in practice [60].

Kerkez et al. [30] introduced a consortium titled “Open Storm”, dedicated to leveraging the potential of low-cost sensors and actuators for transitioning static stormwater systems towards dynamic control systems (Open Storm) [61]. This initiative seeks to optimize adaptive control by integrating centralized grey and distributed green infrastructure, transforming the RTC paradigm into a community-based stormwater management perspective [62]. The study identifies the limitations of earlier RTC technologies in urban drainage, which were primarily applied to large-scale infrastructure projects. It underscores the prospect of cost-effectiveness and adaptive potential when implementing systems at a catchment scale [63]. The realization that the application of decentralized, coordinated smart technology for stormwater management at a catchment level could offer substantial benefits for environmentally friendly and resilient systems has spurred a series of experiments on GSI practices in recent years [26,64,65,66]. These include green roofs, rain barrels, bioretention ponds, and stormwater detention ponds, which are currently the primary focus of the literature.

4.1.2. Practices

The application scales of GSI are diverse and can be categorized into four distinct approaches. At the property level, the focus lies on green roofs and smart rain barrels, while bioretention systems are utilized for drainage at the street level. Stormwater detention ponds have a broader scope of application, ranging from neighborhood to watershed scales. An overview of these strategies’ design, operation, and experimental results, in addition to how IoT technology intervenes in each to enhance existing beneficial effects or to broaden the range of potential applications for the device, is presented [67].

(1)

Green roofs

Urban expansion often results in an increase in impervious surface area, disrupting natural water cycles and contributing to the prevalent urban challenges of flooding and urban heat islands. In contemporary cities, rooftops constitute approximately 40–50% of the total impervious surface area. GSI, including green roofs, is broadly acknowledged as a best practice for stormwater management. Green roofs offer additional benefits such as mitigation of urban heat island effects [68,69], and building energy savings through evapotranspiration and insulation [70,71,72]. However, plant health is pivotal to the success of green roofs, and negligence during prolonged drought can lead to adverse impacts. Traditional rainwater harvesting systems can also encounter limitations [37,73,74]. Hence, IoT technology can be leveraged for tasks such as monitoring soil conditions or irrigating plants.

The use of IoT technology in the context of green roofs is primarily observed in the monitoring and optimization of these systems for enhanced performance. Research indicates that green roofs have the capacity to hold stormwater, thereby mitigating peak flows and runoff [75,76,77]. Principato et al. [78] examined the use of movable gates with RTC in an urban basin to reduce combined sewer overflows, comparing scenarios using IoT and RTC with passive green roofs. They constructed a remote monitoring system to measure evapotranspiration from a large green roof test patch using a modified weighing lysimeter [78]. In another study [72], a large green roof test plot was monitored remotely to measure evapotranspiration using a customized weighing lysimeter. The system included a network of digital load cells communicating with a microcomputer via the i2c protocol, uploading data to cloud storage regularly. The real-time data capture of test module ET rates endorsed the significance of active sensor systems in advancing green roof technology and integrating ET performance into the design process. These findings imply that advancements in green roof performance monitoring technology could allow for more efficient deployment and maintenance. Extended green roofs, also known as passive green roofs, are designed to require less irrigation and maintenance. A previous study on RTC at the same basin reported that the arrangement coupling RTC with a passive green roof was most effective in reducing overflows [79].

Green roofs operated by RTC can retain water for an extended duration by closing a control valve, which enhances the heat reduction impact and generates anaerobic zones for water treatment [67,80,81]. Similar outcomes can be achieved by merging various configurations of RTC sluice gates with scenarios where the system includes green infrastructure (e.g., green roofs, permeable paving) [59].

(2)

Smart rain barrels

Rain barrels serve as compact rainwater collection devices, typically positioned under a house’s gutter to accumulate rainwater from the roof. This collected rainwater finds utility in nonpotable applications such as watering gardens, irrigating lawns, or washing vehicles. Rain barrels principally function to reduce rainwater flow into drains, consequently alleviating pressure on city drainage systems. An advancement on this conventional technology is the Smart Rain Barrel (SRB), an IoT-empowered microstorage system specifically designed to enable sophisticated rainwater harvesting methods. It typically comprises a standard rain barrel paired with a remotely controllable release valve, forming a control system that enables novel modes of operation. Communication modules and other auxiliary components may also be included.

SRB has been extensively explored in a series of studies conducted by Oberascher, M. et al. [80,81,82,83]. In one of these investigations, the SRB was incorporated into a smart city pilot project that monitored every water inflow and outflow on the campus of the University of Innsbruck in Austria. The study employed weather forecasts and time-controlled filling levels of different GSI structures, along with connected sewer systems, for RTC. It was demonstrated that installation location and the storage capacity of the rain barrels could result in flood reduction ranging from 18 to 40%, even though only a rudimentary automated control system was employed [82]. Another study [84] aimed to address the limitations of small water storage capacity and to leverage high-resolution weather forecasts to increase the retention volume while ensuring adequate rainwater collection for irrigation. The research built upon existing SRB models to scrutinize the impact of hypothetical SRB retrofits on the urban water infrastructure (both drainage and water supply systems) of an Alpine city with 2900 residents. The open-source program “Smartin” was developed to merge Python programs SWMM5 and EPANET2 into a coupled model. The study discovered that compared to an unmanaged rain barrel, simple control strategies could significantly augment the performance of an integrated system. For example, such strategies were found to decrease combined sewer overflows and water demand [85].

(3)

Bioretention cells

Bioretention cells serve as water treatment systems that are either naturally occurring or artificially engineered. They are usually characterized by a biodiverse environment such as a wetland, pond, or flooded area [86]. Employing the mechanisms intrinsic to biodiversity and natural ecosystems, bioretention cells purify and store water. They achieve this by transforming pollutants and nutrients within the water into biomass and sediment, aided by the actions of plants, algae, microorganisms, and other organisms [87].

These water treatment systems not only curtail water pollutants but also regulate water flow and quality, thereby enhancing the water’s overall condition. Their applications range from urban water management and ecological restoration to flood control, making them a sustainable solution for water management [88]. Bioretention cells are particularly advantageous and show immense potential in addressing stormwater quality. However, the efficacy of these systems in treating water quality varies across studies. Such variations in outcomes are often attributed to suboptimal design, inconsistent pollutant loading, and the challenge of removing lower pollutant concentrations compared to higher ones [86,87,88].

One study [89] analyzed multiple RTC schemes to improve water quality in bioretention towers, indicating that water quality improvement is contingent on nutrient presence. Another key aspect of bioretention cell design relates to effluent storage for reuse, requiring effective treatment of contaminants and pathogens to adhere to region-specific reuse legislation [90,91,92]. Another research endeavor [92] investigated two cost-effective strategies to treat stormwater for collection and reuse through bioretention basins. The authors discerned that RTC enabled bioretention to lessen the adverse impacts of both short- and long-term drought periods and to mitigate the effects of mass influx on fecal microbial processing. Yet, the study did not ascertain how inaccurate predictions influence the bioretention performance of RTC. As a result, there is a need for additional research to evaluate the feasibility of utilizing rainfall forecasting models for RTC-based bioretention [67].

(4)

Stormwater detention ponds

Traditional stormwater detention ponds are designed to mitigate flooding. In one study [93], a two-stage detention basin was constructed. The upper stage remains dry except during storm events, while the lower stage includes a small outlet to facilitate pollutant settling. Dry detention basins, which remain dry even during rainstorms, have also been considered to prevent the propagation of mosquitoes in stagnant water [94]. Such factors must be taken into account in planning for both conventional and sustainable drainage systems [95,96]. Leveraging the power of IoT in stormwater detention ponds opens up new possibilities for effective stormwater management. By integrating sensors, actuators, and wireless communication devices, real-time monitoring of weather conditions and rainfall parameters can be achieved, allowing dynamic flow control at multiple sites [30,97]. This smart system integration can elevate the capabilities of stormwater ponds, making their management strategies more adaptable and responsive to changing climatic and environmental conditions [53,98,99,100].

Smart stormwater ponds primarily control flooding by retaining a portion of incoming water and regulating outflows [101]. The volume of water they discharge is largely determined by the type and size of the outlet structure. Typically, stormwater detention ponds incorporate both a secondary overflow outlet to manage pond storage during the rainy season, and at least one primary outlet to control regulated flow from the pond, such as a weir, orifice, or riser-type outflow [102]. Stormwater detention ponds are further classified as in-line and off-line. Off-line basins do not interfere with sediment movement and fish migration, unlike in-line basins that can cause upstream flooding due to water accumulation [101]. The integration of IoT in off-line basins enables the optimization of the drainage system’s performance based on its state, primarily controlling inflow and initiating diversion after the river flow reaches a critical threshold. For example, one study [48] compared flood reduction between active flow diversion control and passive structural control applied to an off-line detention basin. The results showed a 2–3-fold higher flood reduction with active control.

A variety of experimental studies have been conducted to explore the effect of IoT integration on stormwater detention ponds. One study [97] presents a simple yet effective RTC setup for flood control. This approach represents a smart, integrated method for global communication through agents in the drainage system. It is primarily composed of parallel detention basins with modular valve openings to control downstream flows [67]. Recent research has accomplished multi-objective scenarios, such as stormwater detention and pollutant removal, by using RTC technology to control environmental flows and shape streamflow [49,51,103,104,105,106,107,108].

Further studies, in addition to experimental research, include models of the behavior of both the drainage system and the smart stormwater detention pond. For instance, the literature [109] compared the effectiveness of reactive and predictive RTC systems in terms of overflow reduction and detention time extension. The results showed that despite its imperfections, the predictive system was superior. This suggests that IoT applications can provide a window of opportunity to predict the system and optimize its operation. Researchers who have implemented new operation rules for detention basin control have found that it meets a wide range of hydrologic, hydraulic, and water quality objectives [49]. Moreover, smart stormwater detention ponds can respond to weather forecasts by discharging stored water prior to rainfall events [49,110,111]. Advances in smart technology have been made to forecast and minimize the risk of urban flooding as well as detect and reduce combined sewer overflows [53,112,113].

4.2. Hotspots and Trends

4.2.1. Keywords and Clusters

A deeper exploration into the hotspots and trends was conducted using Citespace, resulting in a more visually expressive representation. Information on research topics was extrapolated from keywords and clusters to discern further the research hotspots and trends. Six clusters were identified: #0 optimization, #1 modeling, #2 bibliometric analysis, #3 multi-agent systems, #4 resilience, and #5 water quantity control (Figure 4).

Clusters #1 and #2 pertain to modeling and optimization of smart rainwater management systems, which require consideration of rainwater’s physical characteristics, data collection, intelligent control, optimization strategy, and intelligent monitoring. This ensures optimal rainwater collection and utilization. In software engineering, multi-agent systems (Cluster #3) are a recurrent research theme, particularly in distributed and open network environments. Such systems consist of numerous forms of intelligence capable of independent perception, decision-making, and action, along with communication and collaboration for achieving common goals. In the realm of IoT, smart devices and sensors could act as intelligent agents in multi-agent systems, facilitating advanced intelligent functions through mutual communication and collaboration. This cluster suggests ongoing exploration for upgrades on various physical devices, sensors, and smart terminals. Future requirements for this field, represented by clusters #4 and #5, need to account for the adaptability and resilience of GSI in the face of uncertainties and unexpected events. Moreover, there is a necessity to optimize and control rainwater harvesting and utilization based on quantitative hydrological control techniques to maximize the utilization benefits.

The timeline of keywords was curated and is depicted in Figure 4. The node sizes and timescales within the six clusters offer a dynamic perspective. In such a timeline analysis, the node size reflects the importance of the keywords, while the degree of connectedness indicates the association’s progression. The figure signifies that concepts have undergone significant evolution throughout the decade, with a surge in relevant terms. Diverse inquiry methods have been utilized, including practical, experimental, and historical approaches to probe the field’s frontier and advancements. The upcoming phases might involve further review and standardization of terminology in this area, along with the development of comprehensive implementation matrix frameworks for integrating IoT with GSI through various literature reviews. A notable gap in this field is the deficiency of novel methodologies for improved quantitative control of hydrology.

Figure 5 showcases the top 10 emerging words, illuminating the novelty and progressive nature of the keywords. The citation text is rather concise, leading to a short appearance duration of each word. RTC emerged as the earliest keyword, suggesting that the integration of IoT and GSI originated from the RTC research field, primarily applied to hydrological runoff control. Current research mainly targets optimizing existing centralized base model networks, as urban environments become smarter and greener, and the concept of NBS further allows integration of distributed stormwater management facilities and IoT. The primary constraints to this field’s development can be attributed to the optimization of methods, technology, and associated costs.

4.2.2. Applications

A plethora of studies have focused on the centralized use of IoT in urban drainage systems [114,115]. Even though a few have considered centralized and decentralized urban green infrastructure to address urban flooding impacts [116], there is little literature on smarter distributional studies. Research on globally intelligent systems is just getting started, and in 2017, Garofalo et al. [113] advanced the field by developing a decentralized RTC premised on the multi-agent paradigm. This system, integrated with the SWMM hydrodynamic simulation model, facilitated complex emergent behaviors grounded on interactions between agents possessing simple behaviors. This control system saw implementation in the urban drainage system in Cosenza, Italy, illustrating the potential of decentralized techniques to decrease complexity while enhancing hydraulic performance when juxtaposed against centrally managed systems.

In the context of GSI, the choice of control method is contingent on the scale of the application. For instance, larger-scale GSI, such as retention pools, often resort to traditional centralized systems necessitating continuous prediction and optimization. Green roofs, which require more intricate monitoring and control equipment due to their target spatial distribution, pose more challenges than centralized systems. To comprehend the constraints impeding the development of GSI, two broad categories have been identified and reviewed: monitoring and control, and forecasting and optimization.

(1)

Monitoring and control

GSI is gaining prominence in environmental management, as referenced by [87], who present a four-step adoption matrix for effective stormwater management practices. The initial stage emphasizes the significance of monitoring and control for these systems. This phase necessitates a robust modeling framework, including the selection of appropriate methods and algorithms and the identification of model parameters and variables. Continual adjustments and updates to the control parameters of current numerical models of stormwater systems are crucial for their validation and calibration. As these models are used, they must be adapted and simulated to meet the desired limit values of these parameters. Factors such as climate change impacts and urban developmental conditions across diverse cities should be considered to enhance model performance. Concurrently, a wealth of real-time data such as water quality measurements, water levels, flow rates, and precipitation data should be collected, forming the basis for subsequent control steps [87]. The enhancement of control systems may incorporate time-based criteria that take into account particle size distribution and associated hydraulic residence times. By adding water quality sensors and defining thresholds as control parameters, further system improvements can be achieved. When these data are linked to the system’s digital twin, RTC programs can be designed to optimize system performance by adjusting thresholds.

For each GSI method implemented in urban drainage, various specialist guides are available for consultation [117,118]. Such guides suggest monitoring parameters related to water quality and system water balance, including inflow and outflow rates, precipitation, soil moisture, and soil infiltration rate, in addition to water storage levels where applicable. Infiltration devices designed for specific water treatments, such as green roofs and rainwater ponds, require particular attention to these parameters. Water quality monitoring typically focuses on evaluating total solids (TSS), metals, phosphorus, and nitrogen levels [67]. Nonetheless, these guidelines may not be applicable in all situations and may not adequately address specific monitoring needs. With the increasing influence of IoT, comprehensive real-time monitoring of stormwater systems has become imperative. As the number of devices grows, decentralized devices covering larger spatial scales become more beneficial due to the impracticality of monitoring all parameters associated with a GSI [67]. Therefore, monitoring multiple distributed GSI can be costly and time-consuming. Conversely, for GSI covering smaller spatial scales (such as green roofs and rain barrels), it might be essential to simplify and limit the number of monitored parameters.

In relation to green roofs and rain barrels, which lack intake structures due to their spatial distribution characteristics, monitoring reservoir inflow becomes crucial for assessing inputs and optimizing system control to achieve goals such as water quality improvement or runoff volume and peak reduction. For both technologies, IoT applications deem outflow monitoring essential, especially if a facility includes a reuse reservoir, wherein the outflow from the facility equals the input to the reservoir. Monitoring inflow, storage level, and outflow for detention ponds is imperative to prevent overflows, which could result in damages and floods in the vicinity of the facility [67]. In some cases, evapotranspiration factors may also need to be considered. Although less relevant for infiltration monitoring, monitoring soil moisture significantly influences green roofs and stormwater detention ponds. IoT devices can manage the internal water storage of the device, thereby controlling the dry/wet cycle, which affects pollutant removal performance and the mineralization and accumulation of organic matter [89,119,120]. Previous moisture conditions can also influence plant water balance performance and metal uptake by highly accumulating plants [121,122].

Monitoring water quality-related parameters is relatively complex and heavily relies on recent advancements in sensor technology and IoT. Total solids, a key indicator of water quality for detention ponds, enable immediate identification and remediation of water quality contamination issues. The efficacy of detention ponds in reducing pollutants depends on various factors. However, these can be clarified and validated by tracking retention duration, eliminating suspended particles and their associated pollutants, and allowing UV disinfection to occur during the day [51,67,115,123]. Current IoT technologies facilitate real-time, multidirectional communication between sensors, actuators, and centralized control systems, enabling real-time data collection and subsequent remote control. However, due to a variety of cost and technical reasons, centralized monitoring of large facilities is prevalent [124,125]. Despite many cities and utilities beginning to establish distributed sensor networks to assist with this [126], and some practical research already conducted [60,127], these networks primarily focus on monitoring, and active control at the network scale has not yet been achieved [128].

(2)

Forecasting and optimization

GSI extends beyond monitoring and RTC, incorporating predictive optimization to better align with practical applications. To simulate hydrological processes at varying catchment scales, development and application of optimization algorithms founded on objective functions and stormwater management scenarios are imperative.

One strategy for optimization involves data collection to adjust the rule set for off-line optimization. Basic data gathering allows targeted maintenance of stormwater management facilities exhibiting unusual water levels. It also enables manual iterative investigation of the implications of applying different rule sets to simulations [129], or conducting network-wide fundamental response operations [57,130]. For smart stormwater system optimization, it becomes evident that the analysis, optimization, and learning processes must be structured to allow proactive real-time optimization of the system’s operation [131]. This necessity arises due to hydraulic modeling, which often requires processing for several hours. Such time constraints can be a setback in situations requiring decision-making within minutes [132]. Present strategies remain confined to developing optimization at the network scale and to the real-time perspective essential for coordinating decentralized infrastructure [37,133]. Consequently, innovative and sophisticated real-time stormwater system optimization methods are beginning to find limited application [127].

Optimal decisions can also be made based on both internal and external network data through contemporary applications that integrate with rainfall radar and forecasting [48,134,135]. This enhances the ability to optimize decision-making, considering the current and projected states of the network, as demonstrated by the Paris MAGES system in practice [127]. By integrating data and extending optimization performance from a hydraulic focus to include other external factors and interactions between subsystems, optimization efforts can be enhanced. The management of novel green infrastructure can realize significant stormwater benefits along with environmental and social functions [136,137,138]. Due to the evolution of smart management across various subsystems, stormwater management is now an integral component of the smart city agenda [139].

4.3. Limitations and Recommendations

The future trajectory of integrating IoT in GSI is contingent upon various determinants. Key constraints encompass aspects such as the pace of technological advancements, associated costs, the need for standardization, evolving regulations, and market demand. While there is significant promise, widespread adoption may still be a decade or two away, especially considering that this research domain is in its nascent stages. Existing research on the application of IoT in GSI has focused on the study of RTC. Although small-scale applications, such as green roofs and rain barrels, exhibit potential for decentralized forms of implementation, the complexity of these systems necessitates significant scaling to significantly impact urban runoff at the watershed drainage scale. We envision a possible future transition and intelligent control of small- and large-scale through medium-scale drainage systems such as bioswales [140]. These methods also entail greater costs due to the requirement for a higher number of sensors and more expensive data transmission structures than other technologies. Consequently, the majority of experiments remain in the monitoring phase.

Distributed IoT applications in stormwater detention ponds and bioretention can achieve commendable results in runoff control and water quality improvement due to their larger operational scale. Depending on the design strategies of monitoring and equipment, the implementation costs can be relatively low. The continuous optimization of models for prediction indicates a larger potential for development in this domain. Future research on IoT-controlled bioretention systems and stormwater detention ponds should incorporate additional hydrologic quantification through lab and field studies. Active control techniques using depth sensor measurements and soil moisture readings at various medium depths should be developed further to balance water quality and hydrologic objectives. System program design should incorporate adaptive controls and utilize weather forecasts [89].

Moreover, upgrading existing bioretention systems with IoT is not always feasible, depending largely on the construction specificities of bioretention and stormwater detention ponds, as well as the local implementation context. However, by actively controlling an outflow valve, existing systems can prevent complete drainage of detention ponds or cisterns to maintain optimal soil moisture content. This strategy helps avoid drastic soil moisture changes, thereby creating an efficient soil microenvironment for vegetation-assisted rhizoremediation. Therefore, field trials and research work are necessary to adapt IoT to large-scale detention systems under various circumstances [141].

The adoption of RTC technologies is expected to increase with the availability of low-cost, low-power sensors for IoT applications in the future [37,142], exemplified by the fusion of sensor innovation and sustainable bioenergy production based on eco-friendly concepts such as plant-based microbial fuel cells. Studies on microbial-fuel-cell-powered soil moisture sensors have shown promise in this domain [143]. The subsequent step could involve integrating RTC bioretention systems with other stormwater control strategies at the district or city scale, leveraging intelligent process controls of various connected stormwater management units through optimization algorithms [144]. In the realm of IoT, where the integration of sensors, actuators, and intelligent devices is predominant, formal verification is paramount. Techniques such as model checking, process algebra, and theorem proving address many of the security challenges posed by the IoT, enhancing the dependability of the underlying infrastructure [145,146]. Embracing such formal verification techniques in IoT can offer greater extensibility and heterogeneity, potentially fast-tracking the real-world deployment of GSI.

IoT’s capabilities enable real-time oversight of GSI’s functionality and water usage, offering avenues for optimized system management. This not only aids in averting resource wastage and potential breakdowns but also paves the way for cost-effective maintenance strategies. However, the initial investment required to incorporate IoT into GSI cannot be overlooked. While there is an undeniable ecological advantage, the integration may face resistance from urban dwellers. The alterations brought about in the urban milieu could be overwhelming and disrupt familiar routines, leading to hesitancy in embracing this change, despite its evident environmental benefits [17]. Some critiques also point towards a metric-focused GSI potentially sidelining democratic planning and silencing emerging voices in local governance [147]. Moreover, the economic opportunities spawned by IoT investments could inadvertently exacerbate societal inequalities, further marginalizing vulnerable groups [148]. Such multifaceted challenges necessitate further exploration as we advance with IoT-driven innovations. While smart city technologies undeniably warrant contemplation in forthcoming urban strategies, ensuring paramountcy in execution, it remains imperative to safeguard the exigencies of marginalized demographics and the broader public welfare [149]. This entails transcending a mere accentuation on eco-efficiency.

The assessment of existing literature reveals knowledge gaps that could direct future research demands and innovation challenges. These include creating maintenance and monitoring plans based on real-time data analysis, validating modeling results in the physical world, and continuously optimizing the model to predict outcomes. A supporting framework that considers policy, cost–benefit analysis, management, among others, is essential for the widespread use of improved stormwater treatment options [150]. As infrastructure transitions to intelligent stormwater systems that can be networked and controlled, cybersecurity becomes a critical consideration to protect systems from potential digital threats [30]. While smart technologies are revolutionizing city operations, it is important to ensure that enhancing stormwater regulation services also ensures desirable daily experiences, crucial cultural ecosystem services, and resilience against climate change and aging infrastructure [100]. We also expect to see further optimization in the future by adding IoT facilities to the combination of grey and green infrastructure [151].

5. Conclusions

The integration of IoT with GSI is an emerging field with remarkable potential to revolutionize urban stormwater management. This bibliometric review offers an insightful analysis of the current state of research, exhibiting a steady increase in publications and expanding global collaborations since the inception of this field in 2013. While the field is still in its early stages, significant contributions from countries including the United States, Canada, Italy, China, and Australia indicate a worldwide recognition of the potential of IoT-enhanced GSI. The versatility of GSI applications, including green roofs, smart rain barrels, bioretention systems, and stormwater detention ponds, allows for diverse implementation scales. IoT technology enhances these applications by broadening their range, improving their efficiency, and offering RTC and optimization. Furthermore, the integration of IoT and GSI is expected to provide resilience in the face of environmental uncertainties and optimize rainwater harvesting and utilization. However, despite the substantial potential of the IoT–GSI intersection, the existing body of research reveals several limitations and challenges. These include the necessity for additional monitoring and control mechanisms, the development of predictive optimization strategies, and the requirement of extensive scalability for smaller applications such as green roofs and rain barrels to make a significant impact. Additionally, a comprehensive cost–benefit analysis must be considered for the successful and widespread adoption of these technologies.

Innovative approaches to address these challenges are currently under investigation, such as the application of microbial-fuel-cell-powered soil moisture sensors and the integration of RTC bioretention systems with other stormwater control strategies at larger scales. While IoT’s ecological benefits for urban landscapes are undeniable, their socio-economic implications are equally crucial and deserve parallel consideration. Moreover, the development of security measures against potential digital threats to these intelligent systems is of paramount importance. Given these findings, future research should focus on bridging the identified knowledge gaps, including the development of real-time data-based monitoring plans, validation and continuous optimization of models, and the formulation of supportive frameworks that take into account policy and management considerations. As society continues to grapple with urban development, climate change, and aging infrastructure, the intersection of IoT and GSI emerges as an innovative solution that promises not only efficient stormwater regulation but also the enhancement of daily life experiences and cultural ecosystem services. Therefore, continuous effort, investment, and cross-disciplinary collaboration are encouraged to explore this promising domain, paving the way for smarter and greener urban environments.