Methane is the primary component of natural gas and one of the most powerful greenhouse gases, with atmospheric concentrations rising rapidly enough to raise fears of climate tipping points. Because natural gas is a fossil fuel whose extraction, distribution, and combustion all release methane and carbon dioxide, the gas supply chain carries enormous mitigation potential for cities trying to meet regional and global climate goals. Urban studies have repeatedly shown that methane emissions from city gas systems are underestimated, and as renewable electricity displaces fossil power, scientists argue that electrifying end uses such as stoves, water heaters, and vehicles will be essential. Gas streetlights, it turns out, represent a small but glaringly inefficient pocket of that urban gas infrastructure, one that has survived more than a century of technological change largely on the strength of nostalgia.

The interdisciplinary team, led by Amy Townsend-Small of the University of Cincinnati’s School of Environment and Sustainability, together with Sacha Brewer and historian David Stradling, combined archival research with field measurement. Using historical records, satellite imagery, and on-the-ground verification, the researchers constructed the first publicly available map of gas streetlights in Cincinnati, documenting 1,126 lamps still in operation across the city. That number is a shadow of the peak: Cincinnati’s gaslight era began in 1843 with the purchase of 100 cast-iron streetlamps, and by 1892 the city boasted more than 9,500 of them. The decline slowed dramatically in recent decades, and the remaining lamps cluster in older, wealthier neighborhoods such as Clifton, Hyde Park, Avondale, College Hill, Kennedy Heights/Pleasant Ridge, Oakley, O’Bryonville, and Roselawn, with a few stragglers elsewhere. Most are “Boulevard” style fixtures, about eight feet tall, with two mantles flanking a metal pipe beneath a glass or plastic globe, and nearly all have burned around the clock since 1956, when the city decided that maintaining automatic timers and patrol crews had grown too expensive.

To quantify the emissions, the team adapted a high-flow sampling method well established in natural gas leak studies. They first screened each lamp with a Bascom Turner Gas Rover to detect methane enhancements, then fitted a custom aluminum flux chamber with a foil wind skirt over the top vent of the light and connected it to an Indaco High Flow Sampler. Of 82 gaslights tested, nearly all were emitting methane, with a mean emission rate of 1.7 grams of methane per hour, an order of magnitude higher than typical natural gas appliances. Storage and tankless hot water heaters, by comparison, emit roughly 0.2 to 0.3 grams per hour, and household stoves, averaged over their on-and-off usage patterns, emit about 0.07 grams per hour. The distribution was strongly skewed, a pattern familiar throughout the oil and gas supply chain: most lamps emitted less than 5 grams per hour, but six exceeded that figure, and the largest emissions came from lamps that were partially or completely unlit. The single highest emitter, an unlit gaslight, released 21.0 plus or minus 5.0 grams of methane per hour, approaching the total hourly gas consumption of an average lamp.

That consumption figure is central to the study’s most striking finding. According to city records, all 1,126 gaslights consume roughly 20.9 million cubic feet of natural gas per year, which works out to about 41 grams of methane per hour per lamp, assuming the gas is 94.9 percent methane at standard temperature and pressure. Set against the measured average emission rate of 1.7 grams per hour, that yields a loss rate of 4.2 percent of all gas delivered to the streetlights. For perspective, distribution pipelines and meters typically lose between 0.1 and 0.2 percent of the gas flowing through them, and even newly drilled unconventional wells, the largest emitting segment of the supply chain, show loss rates between 0.01 and 1.2 percent. In other words, Cincinnati’s gaslights lose a larger proportion of their fuel than most of the natural gas system that feeds them.

The lamps also emit carbon monoxide, sometimes at concerning levels. Using a Testo 300 Combustion Analyzer at six gaslights, the team measured ambient concentrations inside the glass enclosures and at the vents, and calculated emission rates from methane-to-carbon-monoxide ratios. While the outdoor sites are well ventilated and the city remains in attainment with EPA carbon monoxide standards, the researchers observed some streetlights positioned close to residential and commercial windows, and because the lamps burn continuously, they could contribute hazardous air pollutants to nearby occupants. The greatest risk, the authors suggest, likely falls on repair workers who might unknowingly service a high-emitting fixture. Excluding one anomalous lamp, the team found an approximate two-to-one ratio of carbon monoxide to methane emissions, a metric that could prove useful for estimating combustion efficiency or extrapolating carbon monoxide releases in other cities.

In regional context, the gaslights are a modest source compared with Cincinnati’s enormous landfill, one of the largest in the United States, which emits an estimated 3,100 kilograms of methane per hour. But they outpace the city’s underground sewer pipes and leaking gas distribution lines, which previous studies found emit up to only about 0.01 grams per hour at street level. The researchers emphasize that a comprehensive methane budget for Cincinnati has yet to be assembled and that future work should combine bottom-up inventories with top-down atmospheric measurements to place these sources in full perspective.

The historical analysis embedded in the study reveals why these lamps have survived so long, and it is here that the research acquires its broader political resonance. When the city proposed removing 600 of its remaining 1,700 gaslights in the early 1960s as a cost-saving measure, residents of affluent Clifton organized, petitioned, and appeared before city council, arguing that gas lamps gave their streets a distinctiveness that mercury vapor lamps could not match, and warning that the substitutes might even hasten neighborhood decline. The city relented, allowing residents to keep the lamps by majority vote with a special assessment to cover the extra cost. Some residents, like Alice Posten, who found her street too dark in winter to see curbs and puddles and worried that poor lighting bred crime, pushed back, but the forces of preservation prevailed. By 1976, Vice Mayor Jim Cissell declared that gas lights had become “a mark of charm and character,” and two years later the lamps earned a place on the National Register of Historic Places. Similar dynamics have played out globally: Berlin’s gaslight preservationists have catalogued more than 20,000 lamps, Düsseldorf activists have fought to save more than 14,000, London’s Gasketeers have secured heritage protection for some of the city’s roughly 1,300 gaslights, South Orange, New Jersey maintains about 2,500, Boston has about 2,800 with an electrification plan that preserves their “look and feel,” and San Diego even installed imitation gas lamps in the 1980s to manufacture a sense of history in its Gaslamp Quarter.

That emotional attachment, the authors argue, is precisely what makes gaslights a potential case study for understanding broader cultural resistance to electrification, whether of stoves, vehicles, or heating systems. During their fieldwork, residents frequently approached the sampling crew to say “we love our gaslights,” while only a few mentioned the environmental or safety implications of natural gas. Several residents could not distinguish gas lamps from gaslight-looking electric ones, and many conversations turned simply to how the streetlights were powered, a level of curiosity the researchers see as an opening for public education about natural gas and the climate benefits of electrification. The findings echo previous research showing that fire evokes positive emotional responses and resistance to change, as documented in Swedish studies of opposition to phasing out old wood stoves.

The study’s conclusion is both a warning and an invitation. Gas streetlights burning 24 hours a day for over seventy years, each leaking on average about 4 percent of the gas it consumes, are undeniably good candidates for methane mitigation, and their high visibility as symbols of the city makes them powerful teaching tools. But the researchers caution that efforts to reduce emissions will fail unless they grapple with the material and cultural factors that have sustained these lamps through more than a century of technological transition. As the authors ask, does the historic charm of dim, flickering light justify the continued emission of methane and carbon monoxide and a continued reliance on fossil fuel extraction? Future work, including interviews with residents about their emotional connections to the lamps, may help cities craft electrification strategies that honor heritage while retiring the infrastructure that fuels it.

The study, published in the esteemed journal Nature Water, details how the research team meticulously measured emissions from a diverse spectrum of wastewater treatment plants spread across the United States. Spearheaded by professors Mark Zondlo and Z. Jason Ren from Princeton University in collaboration with UC-Riverside’s Francesca Hopkins, the inquiry spanned 14 months and involved direct atmospheric monitoring of 96 plants. Collectively, these plants represent about 9 percent of the total wastewater treatment capacity in the country, offering a robust dataset that challenges prior national emission inventories.

Central to the success of their methodology was the use of the Princeton Atmospheric Chemistry Experiment, a custom-designed electric vehicle outfitted with advanced laser-based sensor systems. These technologies enabled real-time, sensitive detection of greenhouse gases such as methane and nitrous oxide as the mobile lab traversed roads encircling the plants. Unlike static measurements or extrapolations from limited samples, this dynamic approach captured a more comprehensive and nuanced picture of gas emissions, accounting for variables such as seasonality, weather, and operational conditions.

Results demonstrated that wastewater treatment facilities emit roughly 1.9 times the amount of nitrous oxide and 2.4 times the methane than previously recognized by EPA estimates. Given that methane and nitrous oxide are respectively 28 and over 250 times more potent than carbon dioxide in terms of global warming potential, these findings highlight a substantial additional source of climate forcing. The cumulative contribution of wastewater plants equates to approximately 2.5 percent of U.S. methane emissions and 8.1 percent of nitrous oxide emissions, a notable fraction considering the otherwise underexamined nature of these sources.

The variability inherent to biological wastewater treatment processes further complicates emissions assessment. Microbial populations responsible for degrading organic waste inevitably produce methane and nitrous oxide as metabolic byproducts, but reaction rates fluctuate widely. Factors such as wastewater composition, ambient temperature, precipitation events, and treatment technique diversity all influence emission profiles. The project’s comprehensive sampling regime, which entailed multiple visits per plant across differing environmental conditions, illuminated these dynamics with unprecedented clarity.

One surprising discovery was the transient and heterogeneous nature of emissions at certain sites. For instance, elevated nitrous oxide concentrations were sometimes detected near aeration tanks one week, only to drop to undetectable levels on subsequent visits. Such findings emphasize the complexity of microbial ecosystems within treatment plants and how operational or environmental changes can drastically alter emissions over short time frames. This temporal variability poses significant challenges to prior emission modeling efforts, which often relied on snapshot measurements from limited locations.

Historically, national greenhouse gas inventories relied on extrapolation from studies at a handful of treatment plants, typically focusing on ideal or laboratory conditions rather than real-world operational variability. The Princeton team’s large-scale, seasonally varied field data provides a more accurate foundation for recalibrating models and regulatory frameworks. The research highlights the necessity of monitoring full-facility emissions rather than isolated treatment stages or partial measurements, as plant infrastructure and processes have evolved considerably since many were originally constructed decades ago.

Despite the daunting scale of emissions, the study offers hope as a relatively small subset of facilities disproportionately contribute to total greenhouse gas outputs. Targeted interventions at these high-emission plants could achieve outsized reductions efficiently. The researchers advocate working closely with plant operators to characterize internal process emissions, operational inefficiencies, or aging equipment that may exacerbate gas release. Such insights would pave the way for tailored mitigation technologies that address both air quality and water treatment goals.

Moreover, the economic dimension of emissions management comes to the fore with the possibility of reclaiming methane as a renewable energy source. Wastewater facilities frequently generate methane, a compound traditionally viewed strictly as an environmental liability. Capture and utilization of this methane could yield not only greenhouse gas reductions but also provide a revenue stream or operational cost offset for utilities. This dual environmental and financial incentive underscores the integrative potential of emission control innovations.

The broader implications of this study reverberate through climate policy and urban infrastructure planning. The overlooked footprint of wastewater treatment architectures necessitates recalibrated national greenhouse gas accounting and incentivized emission reduction strategies. Enhancing transparency and empowering operators with better monitoring tools and guidance are essential next steps. As cities worldwide grapple with sustainability challenges, incorporating more accurate assessments of wastewater emissions can inform comprehensive climate action plans that bridge water and air quality considerations.

In summary, the pioneering work led by Princeton’s engineering team unveils a substantially underestimated source of climate-warming gases emanating from municipal wastewater plants. By deploying cutting-edge atmospheric measurement technology across hundreds of kilometers and seasons, the research presents a compelling case for overhauling traditional greenhouse gas inventories and targeting high-impact interventions at these facilities. These revelations are critical as urban centers aim to reconcile infrastructure demands with ambitious climate targets, emphasizing the importance of interdisciplinary collaboration and technological innovation in tackling complex environmental issues.

Subject of Research:
Not applicable

Article Title:
Comprehensive assessment of the contribution of wastewater treatment to urban greenhouse gas and ammonia emissions

News Publication Date:
8-Oct-2025

Web References:
http://dx.doi.org/10.1038/s44221-025-00490-z

Image Credits:
Nathan Li/Princeton University

Keywords:
Climatology, Climate change, Climate data, Atmosphere, Climate systems, Earth sciences, Atmospheric science, Atmospheric chemistry

The study reveals a rapidly expanding body of research dedicated to leveraging ML methods for climate mitigation in urban settings, underscoring the vibrancy and urgency with which the scientific community addresses this nexus. Yet, paradoxically, the adoption of cutting-edge AI paradigms such as generative artificial intelligence and large language models—a transformative force across many other urban application domains—remains surprisingly minimal within this crucial field. This disconnect suggests a technological gap between the potential of these advanced tools and their current utilization in climate-specific urban mitigation strategies.

Central to the study’s findings is the identification of 40 distinct application areas where ML contributes to urban climate mitigation. These areas predominantly align with mitigation strategies highlighted by the Intergovernmental Panel on Climate Change (IPCC) as having significant potential to reduce greenhouse gas emissions. The concentration of research efforts on these high-impact strategies indicates a targeted approach that prioritizes effective interventions. Still, it raises questions about the diversity and inclusivity of mitigation options being explored with AI technologies.

The researchers note a critical factor influencing this tendency: data availability. ML algorithms thrive on rich, high-quality datasets, and urban climate data’s accessibility dramatically shapes what mitigation strategies can be modeled and optimized. Commercial interests further skew research focus toward more lucrative or data-rich approaches, potentially at the expense of less straightforward but equally important climate interventions. This dynamic risks reinforcing existing geographic and socioeconomic inequities since cities and regions with robust data infrastructures attract more research investment, while others remain underrepresented.

Methodologically, the systematic map employed in the study is rigorous, leveraging bibliometric analysis and content coding to categorize and assess trends in ML applications for UCCM. This methodological framework not only quantifies research outputs but also highlights emergent patterns, technological gaps, and priority areas for future inquiry. By cataloging the development of the field, the study situates itself as an essential roadmap for both researchers and policymakers eager to maximize the environmental impact of AI-driven mitigation.

One of the most striking revelations is the negligible presence of generative AI and large language models in the literature related to UCCM. These technologies, capable of producing novel data, optimizing complex systems, and interpreting vast textual datasets, have revolutionized other sectors such as urban planning, mobility, and public health. Their limited use in climate mitigation research prompts a reevaluation of potential barriers—be these technical, ethical, or infrastructural—that impede their application to mitigating climate impacts in urban environments.

Further emphasizing the socio-technical challenges, the study discusses how commercial incentives shape the trajectories of research and development in climate mitigation technologies. Market-driven priorities may unavoidably prioritize short-term gains or visible impacts, overlooking less lucrative but necessary climate solutions. This imbalance could hinder holistic urban decarbonization efforts, particularly in underserved or smaller urban contexts where commercial data ecosystems are weaker.

Moreover, the systematic map highlights the importance of interdisciplinary collaboration. Effective deployment of ML in climate mitigation demands convergence between climate science, urban planning, data science, and policy expertise. Advances in ML algorithms alone are insufficient without integrating domain-specific knowledge of urban systems and climate dynamics. The study calls for enhancing such collaborations to ensure ML methods are not only technically sound but also practically relevant and ethically guided.

Ethical considerations also feature prominently in the discourse. The potential for AI to perpetuate or even exacerbate existing inequalities underscores the necessity of careful, transparent, and inclusive ML model design. Equitable access to data, participatory modeling approaches, and sensitivity to vulnerable populations must be embedded within future research trajectories to avoid reinforcing systemic disparities under the guise of technological progress.

Importantly, the study serves as a call to action for urban stakeholders to consciously steer ML development toward impactful, inclusive, and data-enriched mitigation pathways. Cities, as experimental grounds for climate interventions, stand at the crossroads where AI and urban governance intersect. Strategic investments in data infrastructure, capacity building, and open data initiatives could catalyze broader and more equitable ML adoption, transforming urban climate mitigation from theory into widespread practice.

The study’s in-depth analysis also offers insights into promising yet under-explored mitigation domains. For instance, energy efficiency in urban buildings, climate-resilient transportation networks, and nature-based solutions represent fertile grounds for ML innovation. Targeted efforts here could harness the full spectrum of AI capabilities, from predictive analytics to optimization algorithms that balance environmental goals with socioeconomic objectives.

A vital takeaway from the systematic map is that technological innovation alone cannot drive urban climate change mitigation. The success of ML applications hinges on embedding technical solutions in comprehensive policy frameworks that incentivize sustainable behavior and infrastructure development. This interplay necessitates ongoing dialogue among AI developers, policymakers, urban planners, and affected communities to ensure solutions are responsive to local contexts and scalable across diverse urban environments.

As the climate crisis demands increasingly sophisticated and scalable responses, the integration of ML into urban mitigation strategies holds transformative promise. However, realizing this potential requires confronting the current gaps identified by the study—expanding data accessibility, fostering inclusion, aligning commercial incentives with public interest, and accelerating the adoption of emerging AI technologies. Only through coordinated, multidisciplinary efforts can ML become a pivotal tool in reshaping the future of urban sustainability.

In conclusion, this seminal systematic mapping study by Hintz, Milojevic-Dupont, Creutzig, and colleagues charts a critical frontier in understanding how AI and machine learning intersect with urban climate change mitigation. By illuminating the contours and constraints of existing research, their work not only benchmarks progress but lays the foundation for more deliberate, impactful, and equitable AI-driven climate strategies. For cities committed to beating back climate change, embracing this roadmap may be one of the most powerful steps forward in the digital age.

As urban populations soar and climate risks intensify, the urgency to deploy effective mitigation tools has never been greater. Machine learning offers unprecedented opportunities for precision, scale, and innovation in tackling emissions, but converting promise into practice demands strategic foresight and inclusive governance. This systematic overview serves as a beacon, guiding future research, policy, and technological development toward a resilient and climate-smart urban future.

The findings urge the global research community and urban policymakers to critically assess and address the socio-technical dimensions shaping ML’s role in climate mitigation. By bridging existing divides and stimulating technological advancements responsibly, cities worldwide can harness AI not just as a computational asset but as a catalyst for transformative climate action that leaves no community behind.

Subject of Research: Machine learning applications for urban climate change mitigation.

Article Title: A systematic map of machine learning for urban climate change mitigation.

Article References:
Hintz, M.J., Milojevic-Dupont, N., Creutzig, F. et al. A systematic map of machine learning for urban climate change mitigation. Nat Cities (2025). https://doi.org/10.1038/s44284-025-00328-5

Image Credits: AI Generated