The stakes could hardly be higher. Methane is the second most significant anthropogenic greenhouse gas after carbon dioxide and is responsible for roughly one third of the rise in global mean surface air temperature between 1750 and 2019. Because it has a short atmospheric lifetime of about nine years but a global warming potential nearly 82 times that of carbon dioxide over a 20-year period, cutting methane emissions is one of the most powerful levers available for slowing near-term climate change. More than 150 countries have now committed, through the Global Methane Pledge and their Nationally Determined Contributions, to reduce methane emissions across all sectors by at least 30 percent below 2020 levels by 2030. Every credible pathway to limiting warming to 1.5 degrees Celsius depends on achieving that reduction.

Yet cities, where more than half of humanity already lives and where the proportion is projected to reach 70 percent by 2050, remain stubbornly difficult to audit. Urban methane comes from a tangle of sources: natural gas consumption, leaks from aging pipeline networks, landfills and wastewater treatment systems. Field studies have repeatedly found that official bottom-up inventories, which rely on activity data and emission factors, can underestimate urban methane emissions by a factor of two to three. Aircraft surveys, ground-based mobile monitors and tower networks can deliver accurate snapshots of a city’s emissions, but they cannot provide the continuous, repeatable monitoring needed to track progress year after year, and their spatial sensitivity is often limited.

Satellites promise to close that gap. The TROPOspheric Monitoring Instrument, or TROPOMI, flies aboard the polar sun-synchronous Sentinel-5 Precursor satellite, launched in October 2017 at an altitude of 824 kilometers with a consistent 13:30 local overpass time. The instrument retrieves the methane column by detecting solar backscattered light in the shortwave infrared absorption band at 2.3 micrometers, achieving daily global coverage at a spatial resolution of 7 by 5.5 square kilometers at nadir since August 2019. Previous studies have used TROPOMI to constrain urban methane inventories across 61 cities worldwide and to conduct a preliminary national analysis of urban emissions across North America, but a transparent, widely applicable framework with detailed error accounting has remained elusive.

The Manchester team, led by Huihui Long together with Maria Tsivlidou, Hugo Ricketts and Grant Allen, built their method on the concept of mass balance, refining the established source pixel approach. First, they remapped the heterogeneous Level-2 satellite pixels onto a regular 0.1 by 0.1 degree grid, treating each observation footprint as a surface polygon and calculating the fractional area of overlap with each grid cell, weighted by pixel area and retrieval error. This regularization allows orbit-level emission estimates that are directly comparable with gridded bottom-up inventories over the same spatial domain. Strict quality filters followed: only cloud-free retrievals with quality assurance values above 0.5, solar zenith angles below 70 degrees, smooth topography, low aerosol optical thickness and retrieval precisions better than 10 parts per billion were retained.

Two innovations distinguish the new framework. The first concerns wind. Conventional source pixel methods rely on 10-meter near-surface winds bundled with the satellite retrievals, but methane emitted at the surface is ventilated by winds throughout the planetary boundary layer, the turbulent lowest kilometer or so of the atmosphere where surface emissions mix and are advected across urban scales. The team therefore computed pressure-weighted average wind speed and direction across all ERA5 reanalysis pressure levels within the boundary layer, using hourly-resolved meteorology. The difference matters: over Los Angeles, the near-surface winds pointed from the north-northeast to east-northeast at speeds below 6 meters per second, while the boundary-layer-weighted winds blew predominantly from the west-southwest to south-southwest at up to about 10 meters per second. Since emission rate is conceptually proportional to mean boundary-layer wind speed, and wind direction determines where the upwind background lies, these discrepancies directly reshape the calculated emissions.

The second innovation is an upwind-based approach to defining the background methane concentration. Instead of assuming that the air surrounding a city is clean, the method defines a background region of equal area located immediately adjacent to the city along the boundary-layer-pressure-weighted mean upwind direction, with both boxes oriented so their sides lie perpendicular to the wind. Emissions in the background are implicitly accounted for and subtracted, so the background region need not be free of methane sources. The methane enhancement over the city is then the difference between the mean column mixing ratio over the source region and that over the upwind background, and the emission rate follows from a mass balance equation combining that enhancement, the mean boundary-layer wind, surface pressure and the dimensions of the source box. Coverage thresholds, requiring at least 25 percent valid data over the source region and at least 10 background observations, guard against statistically unreliable single-orbit estimates.

Applied to the three megacities, the method produced three-year mean emissions of 7.78 plus or minus 4.84 tonnes per hour for London, 47.19 tonnes per hour for Los Angeles, and 42.94 tonnes per hour for New York. London showed only a modest urban enhancement, averaging between 1.47 and 1.98 parts per billion across the study years, which made the estimate highly sensitive to background variability; roughly 80 percent of the total emission uncertainty there was attributable to the methane enhancement, and an anomalously low October in 2021, when background concentrations exceeded urban ones, dragged the annual figure down until it was excluded as an outlier. Los Angeles, by contrast, displayed a pronounced and persistent hotspot, with emissions rising from 26.21 tonnes per hour in 2021 to 62.77 tonnes per hour in 2023, though the large uncertainties mean a previously reported declining trend cannot be ruled out. In New York, emissions ranged from 30.85 to 44.77 tonnes per hour, with a more balanced uncertainty structure in which wind variability contributed about 40 percent.

The comparison with official inventories is sobering. Over identical spatial domains, EDGAR reported a mean of 20.07 tonnes per hour for Los Angeles, where landfills account for more than 90 percent of inventory emissions, and just 4.38 tonnes per hour for New York, roughly 11 percent of the satellite-derived figure. The UK’s National Atmospheric Emission Inventory underestimated London’s emissions by up to 45 percent in 2022, while the US EPA inventory fell short by approximately 56 percent for Los Angeles and 84 percent for New York. Encouragingly, when the team rescaled estimates from six previous top-down studies to the same urban domains, most fell within their uncertainty bounds, suggesting the new method is consistent with independent measurement-led approaches while being far simpler to repeat.

The authors are candid about the limitations. Cloud cover, coastal albedo heterogeneity and aerosol scattering eliminated the majority of overpasses, leaving as few as five valid orbits over New York in 2021 and almost no usable summer data for Los Angeles, where persistent marine boundary-layer clouds dominate. Such uneven sampling can introduce seasonal and inter-annual biases that must be weighed when comparing annualized emissions with inventories. Nevertheless, the team argues that the trade-off is necessary to preserve statistical reliability, and that the framework’s transparent uncertainty budgeting, combining retrieval, albedo, aerosol and wind uncertainties in quadrature, offers a repeatable path forward. With the Global Methane Pledge’s 2030 deadline approaching, the ability to audit cities from space, anywhere on Earth including poorly observed regions such as India and China where most of the global population resides, may prove essential for verifying whether stated methane targets are actually being met.

Subject of Research: Satellite-based quantification of urban methane emissions using TROPOMI observations and an improved source pixel method

Article Title: Satellite-based global monitoring of urban-scale methane emissions

Article References: Long, H., Tsivlidou, M., Ricketts, H., & Allen, G. (2026). Satellite-based global monitoring of urban-scale methane emissions. Atmospheric Measurement Techniques, 19(18), 6171-6191. https://doi.org/10.5194/amt-19-6171-2026

Image Credits: AI Generated

DOI: 10.5194/amt-19-6171-2026

Keywords: methane, TROPOMI, satellite monitoring, urban emissions, greenhouse gases, London, Los Angeles, New York, planetary boundary layer, EDGAR inventory, climate change, Global Methane Pledge

The study arrives at a moment when wildfire has become an inescapable feature of Southern California life. Los Angeles sits in a landscape where fire ecology predates human settlement by a millennium; chaparral ecosystems historically burned roughly every thirty years, with extended droughts allowing dead vegetation to accumulate as fuel. Fire, in turn, rejuvenates chaparral seeds that inhibit germination until after a blaze, returning nutrients to the soil and enabling new growth. That natural cycle now collides with millions of residents, and the catastrophic fires of recent years have pushed questions about disaster and social order into public debate. Popular narratives often assume that disasters unleash looting and lawlessness, but decades of criminological research on hurricanes, floods, and tornadoes tell a more complicated story, and wildfires have remained a conspicuous gap in that literature.

To close that gap, the researchers assembled tract-level data for the City of Los Angeles and identified seven major wildfires that occurred inside the municipal boundaries during the 2016 to 2022 window. Their statistical strategy centered on random-effects negative binomial regression models, a choice tailored to the structure of crime data. Crime counts in small geographic units are overdispersed, meaning their variance exceeds their mean, and a negative binomial framework accommodates that distribution far better than ordinary linear regression. The random-effects specification, meanwhile, accounts for unobserved, stable characteristics of each census tract, such as entrenched disadvantage or built environment features, that could otherwise confound the estimated relationship between fire exposure and crime.

The team measured wildfire exposure in two distinct ways, and that distinction turned out to be analytically decisive. The first measure identified census tracts that experienced direct wildfire damage, that is, tracts where the fire itself burned. The second captured proximity, identifying tracts located within a quarter mile of burned areas even if no fire touched them directly. Separating these two forms of exposure allowed the authors to test whether the crime effects of wildfire are confined to physically damaged neighborhoods or spill over into surrounding communities, a question with direct implications for how police and emergency managers allocate resources after a disaster.

The results paint a nuanced picture. Census tracts that experienced direct wildfire damage reported significantly lower levels of property crime. Violent crime was also marginally lower in those directly burned tracts, though the evidence there was weaker. When the researchers accounted for proximity to wildfire damage, however, the direct-damage effect on property crime was no longer statistically significant. What proved most robust was the proximity effect: tracts within a quarter mile of burned areas reported significantly lower levels of both violent and property crime. In other words, the strongest crime-suppressing signal appeared not in the neighborhoods the flames consumed, but in the neighborhoods standing just outside them.

The authors interpret these findings through the lens of two foundational criminological frameworks. Social disorganization theory holds that neighborhood crime is shaped by the capacity of local communities to exercise informal social control, a capacity rooted in stable networks, shared trust, and collective efficacy. Disasters can disrupt or reorganize those networks in unpredictable ways. Research after Hurricane Katrina, for example, documented both crime displacement and, in some domains, strengthened collective resources among repopulating residents. The routine activities approach offers a complementary mechanism: crime requires a motivated offender, a suitable target, and the absence of a capable guardian converging in time and space. Wildfires radically alter that convergence calculus. Evacuations empty neighborhoods of suitable targets, closed roads and fire-perimeter checkpoints supply guardianship, and burned or damaged structures simply cease to be burglarizable.

The proximity finding fits neatly into this routine-activities logic. A quarter-mile buffer around a burned area is precisely the zone where emergency access controls, evacuation orders, utility shutoffs, and heavy first-responder presence are most intense, even when the homes themselves remain intact. Fewer people on the street, restricted vehicular movement, and a saturated official guardianship presence can suppress both opportunistic property offenses and interpersonal violence. At the same time, the social disorganization perspective suggests that shared threat experiences can galvanize neighborly cohesion, with residents monitoring one another’s properties and coordinating informally during and after evacuations. The study cannot adjudicate between these mechanisms, but the spatial pattern it documents, with effects strongest near rather than within fire perimeters, is consistent with guardianship and mobility explanations operating at the edge of disaster zones.

The findings also connect to a broader and rapidly growing research program on environment, green space, and crime. Recent work has linked urban greenspace to lower crime risk across hundreds of major U.S. cities, examined park size and community safety, and used quasi-experimental designs to test whether urban greenways such as the Atlanta Beltline reduce offending. Related studies have explored how drought conditions relate to violent crime trends in California counties and how natural hazards reshape spatiotemporal crime patterns nationally. The Los Angeles wildfire analysis adds a distinctive dimension to this literature: rather than a gradual environmental amenity, wildfire is an acute, spatially bounded shock, and its crime effects appear to radiate outward from the burn scar in measurable ways.

For practitioners, the implications are concrete. If crime genuinely declines in and around burned neighborhoods, then post-fire policing strategies premised on looting panics may be misdirected, diverting officers from communities where they are actually needed and potentially exposing evacuated areas to over-policing. Emergency response planning, similarly, could incorporate the finding that a quarter-mile buffer around fire perimeters functions as a distinct criminological zone, one where the priorities may be public safety and recovery support rather than crime suppression. For neighborhood recovery, the results hint that the social fabric of fire-adjacent communities may be more resilient than disaster folklore assumes, though the authors caution that their analysis captures associations over a defined study period rather than the full arc of long-term recovery.

The authors are explicit about the limits of their design and the directions it opens. Their data cover seven wildfires within one city over seven years, and census tracts, while useful units of neighborhood analysis, aggregate individual behavior into coarse spatial bins. The proximity threshold of a quarter mile is one plausible buffer among many, and future work could test whether the effect decays smoothly with distance or exhibits spatial spillovers of the kind modeled in wildfire risk mitigation research. Longer follow-up periods could reveal whether suppressed crime rebounds as neighborhoods rebuild, whether reconstruction and gentrification dynamics, of the kind documented after the 1994 Northridge earthquake in Los Angeles, reshape local crime trajectories, and whether the patterns hold in other fire-prone regions. As climate change intensifies fire regimes across the American West, understanding how wildfires reshape not only landscapes but also the social order of the neighborhoods that survive them is becoming an urgent scientific question, and this study provides one of the clearest spatial maps yet of where that reshaping begins.

Subject of Research: The association between neighborhood wildfire exposure and violent and property crime patterns in Los Angeles

Article Title: We Didn’t Start the Fire: How Neighborhood Crime Patterns in the City of Los Angeles Were Associated with Wildfires

Article References: Barton, M. S., Valasik, M., LaPointe, R., & Lopez, S. (2026). We Didn’t Start the Fire: How Neighborhood Crime Patterns in the City of Los Angeles Were Associated with Wildfires. American Journal of Criminal Justice. https://doi.org/10.1007/s12103-026-09955-7

Image Credits: AI Generated

DOI: 10.1007/s12103-026-09955-7

Keywords: wildfires, crime, Los Angeles, violent crime, property crime, social disorganization theory, routine activities approach, disasters, criminology, neighborhoods, negative binomial regression, emergency response