Urban methane under watch: Sentinel‑5P/TROPOMI and LSTM reveal hotspot dynamics over Tehran (2019–2025)

Urban methane under watch: Sentinel‑5P/TROPOMI and LSTM reveal hotspot dynamics over Tehran (2019–2025) | Research Square window.SnipcartSettings = { analytics: { enabled: false } }; (function() { var accessVector = localStorage.getItem('access_vector') || ''; window.dataLayer = window.dataLayer || []; if (accessVector) { window.dataLayer.push({ user: { profile: { profileInfo: { snid: accessVector } } } }); } })(); (function(w,d,s,l,i){w[l]=w[l]||[];w[l].push({'gtm.start':new Date().getTime(),event:'gtm.js'});var f=d.getElementsByTagName(s)[0],j=d.createElement(s),dl=l!='dataLayer'?'&l='+l:'';j.async=true;j.src='https://www.googletagmanager.com/gtm.js?id='+i+dl;f.parentNode.insertBefore(j,f);})(window,document,'script','dataLayer','GTM-K279D39R'); Browse Preprints In Review Journals COVID-19 Preprints AJE Video Bytes Research Tools Research Promotion AJE Professional Editing AJE Rubriq About Preprint Platform In Review Editorial Policies Our Team Advisory Board Help Center Sign In Submit a Preprint Cite Share Download PDF Research Article Urban methane under watch: Sentinel‑5P/TROPOMI and LSTM reveal hotspot dynamics over Tehran (2019–2025) Saeed Motesaddi Zarandi, Khashayar Partovi, pedram rastegary This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-8955243/v1 This work is licensed under a CC BY 4.0 License Status: Posted Version 1 posted You are reading this latest preprint version Abstract Rapidly rising atmospheric methane threatens near-term climate goals, yet the behaviour of urban methane over Middle Eastern megacities remains poorly constrained. Using Sentinel-5P/TROPOMI XCH₄ retrievals processed in Google Earth Engine, we assemble a daily record of column methane over Tehran for 2019–2025 and derive seasonal and annual means at the city scale. The annual mean XCH₄ increases from about 1906 ppb in 2019 to nearly 1978 ppb in 2025, corresponding to a trend of roughly 10 ppb yr⁻¹ and indicating a persistent strengthening of the urban methane burden. Gridded annual maps and a district-level aggregation show that the highest mean XCH₄ values occur persistently over the northern and north-eastern districts of Tehran, while elevated enhancements also appear over southern industrial and landfill-influenced areas; these hotspot polygons are quantified in terms of area and mean XCH₄ for each year.​ To link column enhancements with discrete emitters, we compile a multi-year inventory of satellite-detected methane plumes around Tehran, including source locations and annual emission rates (Q, ton yr⁻¹), and examine how plume occurrence and intensity co-vary with the hotspot fields. Building on the daily XCH₄ time series, we train a long short-term memory (LSTM) network to forecast next-day column methane over the city; on the independent test period the model achieves a correlation of about 0.7 with observations, a mean absolute error near 20 ppb, a root-mean-square error around 26 ppb, and a small positive mean bias of ~ 3 ppb. By jointly analysing daily, seasonal and annual XCH₄, district-scale hotspots and plume emissions, and by demonstrating a deep-learning forecast of urban methane, this study provides a comprehensive satellite-based picture of methane dynamics over Tehran and a readily transferable framework for monitoring methane hotspots in rapidly growing cities across the Middle East and developing regions. Figures Figure 1 Figure 2 Figure 3 Figure 4 Figure 5 Figure 6 Figure 7 Figure 8 Figure 9 1. Introduction Methane (CH₄) is the second most important anthropogenic greenhouse gas after carbon dioxide and is responsible for a substantial fraction of present-day radiative forcing and near-term global warming. Multiple assessments of the global methane budget show that atmospheric CH₄ has increased rapidly since the early 2000s, with an acceleration after 2007 that current emission inventories and sink estimates cannot fully reconcile. Because methane has an atmospheric lifetime of roughly a decade, reducing CH₄ emissions is regarded as one of the most effective options for slowing the rate of climate change in the next few decades and for helping to meet the temperature targets of the Paris Agreement( 1 ).​ Satellite remote sensing has transformed the monitoring of atmospheric methane from the global scale down to regional and local hotspots. The TROPOspheric Monitoring Instrument (TROPOMI) onboard the Sentinel-5 Precursor (Sentinel-5P) platform provides daily global observations of column-averaged dry-air mole fractions of methane (XCH₄) at high spatial resolution, enabling detection of regional enhancements, urban plumes and, under favourable conditions, strong point sources. These data products, which can be accessed and analysed efficiently through cloud-based platforms such as Google Earth Engine (GEE), have been widely used to constrain national and sectoral methane emissions, to derive global and zonal CH₄ trends, and to detect super-emitting facilities in the fossil-fuel, waste and agricultural sectors( 2 , 3 ).​ Nevertheless, detailed satellite-based assessments of urban methane over individual megacities in the Middle East are still limited, even though such cities are growing rapidly and are expected to play an increasing role in regional greenhouse-gas budgets. Tehran, the capital of Iran, is a high-elevation megacity with more than eight million inhabitants and extensive surrounding industrial and suburban areas. Its methane budget is influenced by multiple anthropogenic sources, including municipal solid-waste landfills, wastewater treatment plants, natural-gas production and distribution infrastructure, and nearby agricultural activities. Existing atmospheric studies on Tehran have mainly focused on conventional air pollutants and carbon dioxide, whereas the temporal evolution, spatial distribution and climate-relevant trends of XCH₄ over the city have not yet been systematically quantified using Sentinel-5P/TROPOMI observations( 4 ).​ In parallel, data-driven forecasting approaches based on deep learning have shown strong performance for predicting atmospheric variables from observational time series. Long short-term memory (LSTM) networks, in particular, can represent nonlinear and long-range temporal dependencies that are difficult to capture with traditional statistical models and have been successfully applied to forecast ozone, PM₂․₅ and composite air-quality indices. Extending such techniques to satellite-derived XCH₄ over urban areas offers the potential to characterise short-term methane variability, support early detection of anomalous emission events, and complement long-term climate-oriented trend analyses( 2 ).​ This study combines Sentinel-5P/TROPOMI observations, GEE-based processing and deep-learning time-series modelling to characterise atmospheric methane over Tehran for the period 2019–2025 in a way that aligns with the remote-sensing and methodological focus of the journal. First, Level-2 XCH₄ retrievals, filtered for data quality and cloud contamination, are used to construct daily, seasonal and annual mean time series for the Tehran urban area and to quantify their long-term trends in the context of regional and global methane increases. Second, annual mean XCH₄ maps on a regular grid are generated, methane hotspots are identified through a consistent thresholding scheme, and their extent and intensity are analysed relative to the administrative districts of Tehran, highlighting areas with persistently elevated methane that are most relevant for climate-mitigation efforts. Third, an LSTM model is developed and evaluated for forecasting daily XCH₄ over the city using the satellite-derived time series, with skill assessed through standard error metrics and residual analysis. By integrating temporal trend analysis, hotspot mapping and data-driven forecasting within a single satellite-based framework, the work demonstrates how Sentinel-5P/TROPOMI and modern machine-learning techniques can jointly support the monitoring of urban methane and its climate-relevant dynamics in rapidly developing regions( 1 , 2 , 5 ).Methane is not only a potent greenhouse gas but also a key component of the urban atmospheric environment, where it coexists with other air pollutants and reflects underlying energy and waste-management practices. Improved quantification and short-term prediction of urban methane therefore support both air-quality management and greenhouse-gas mitigation planning in megacities such as Tehran. 2. Materials and Methods 2.1 Study area Tehran is a high-elevation megacity in northern Iran with more than eight million inhabitants and a heterogeneous mix of residential, commercial, industrial and peri-urban agricultural land uses. The municipality is divided into 22 administrative districts that differ strongly in population density, infrastructure and emission sources relevant to methane. Major anthropogenic CH₄ emitters in and around the city include municipal solid-waste landfills, wastewater treatment plants, natural-gas distribution infrastructure and nearby agricultural areas( 6 )​. 2.2 Satellite methane data This study uses column-averaged dry-air mole fractions of methane (XCH₄) from the Sentinel-5 Precursor (Sentinel-5P) TROPOspheric Monitoring Instrument (TROPOMI) for the period 2019–2025. We used the Sentinel-5P TROPOMI OFFL XCH₄ Level-2 product (version 2.x) downloaded from the Copernicus Sentinel-5P data hub. The operational Level-2 XCH₄ product provides daily global coverage with a pixel size of about 7 × 7 km² (5.5 × 7 km² in the later mission phase) and has been widely applied to map regional and urban methane enhancements. Level-2 files covering Iran were downloaded from the official data distribution service, and all further processing steps were carried out locally( 7 , 8 ).​ Only retrievals with valid quality information were retained, excluding pixels flagged for cloud contamination, large viewing angles or retrieval artefacts. Pixels with qa_value 0.2 were discarded, and observations with solar zenith angle > 70° or extreme viewing geometry (across-track index outside ± 30) were also excluded. The remaining pixels were re-projected to a common geographic coordinate system and clipped to a polygon representing the Tehran metropolitan area, which was constructed from the municipal boundary shapefile. Obvious outliers were removed using simple range checks and temporal consistency checks applied to individual grid cells( 9 ).​ No additional bias correction was applied beyond the standard retrieval algorithm, which has been validated against ground-based TCCON and NDACC stations in previous studies. 2.3 Temporal aggregation For each day, all quality-filtered XCH₄ pixels within the Tehran mask were averaged to obtain a single city-mean XCH₄ value. Annual means were computed from the available daily values without gap-filling. The annual trend analysis was performed only for years with sufficient temporal coverage, which in our case includes all years from 2019 to 2025. Days without valid observations, mainly due to cloud cover, were left as missing to avoid introducing artificial variability. Seasonal means were computed for winter (December–February), spring (March–May), summer (June–August) and autumn (September–November) by averaging the available daily means within each season and year( 10 ).​ Annual means were calculated by averaging all valid daily means for each calendar year between 2019 and 2025, both for the full urban area and later for the hotspot region. Linear trends in annual XCH₄ were estimated using ordinary least-squares regression, and the slope and its standard error were used to quantify the long-term increase in urban methane over Tehran( 10 , 11 ).​ 2.4 Gridded maps and hotspot extraction To characterise spatial patterns, the quality-filtered Level-2 XCH₄ pixels were aggregated onto a regular grid covering Tehran and its surroundings. A grid resolution comparable to the native TROPOMI footprint was selected so that each cell contained a sufficient number of observations while still resolving intra-urban structure. For each year, all valid XCH₄ values falling into a given grid cell were averaged to obtain annual mean XCH₄ maps( 12 ).​ For each year, we computed the mean (µ) and standard deviation (σ) of XCH₄ across all urban grid cells. Methane hotspots were then defined as grid cells where the annual mean XCH₄ exceeded µ + 1σ, i.e. one standard deviation above the city-wide mean. Methane hotspots were identified from these maps using a thresholding approach. For each year, the mean and standard deviation of XCH₄ across all grid cells within the urban mask were calculated, and cells with XCH₄ above a chosen threshold (e.g. the city-wide mean plus one standard deviation) were marked as hotspot cells. Neighbouring hotspot cells were merged into contiguous polygons, and for each polygon the area, mean XCH₄ and maximum XCH₄ were computed to form a multi-year record of hotspot properties( 12 ).​ To link hotspots to the urban fabric, the annual hotspot polygons were intersected with the vector layer of the 22 municipal districts in QGIS. This overlay provided, for each district and year, the fraction of its area classified as hotspot and the corresponding district-averaged XCH₄, enabling identification of districts with persistently elevated methane levels( 10 ).​ 2.5 Methane plume inventory A multi-year inventory of discrete methane plumes was compiled from satellite-based plume detections around Tehran between 2019 and 2025. For each detected plume, the source location (latitude and longitude) and emission rate Q (ton yr⁻¹) were estimated and stored in a tabular file. Effective wind speed (Ueff) was derived from ERA5 reanalysis 10 m wind fields at 0.25° × 0.25° resolution, bilinearly interpolated to the plume location and Sentinel-5P overpass time. For sources with multiple detected plumes within a year, annual Q was obtained by averaging the individual Q estimates, assuming continuous operation during the period of interest. The plume locations were imported into QGIS and overlaid on the hotspot maps and district boundaries to diagnose the spatial relationship between strong point sources and persistent column enhancements. Distances from plume sources to the nearest hotspot boundary were calculated to quantify the degree of co-location between plumes and hotspot regions( 7 , 8 , 11 , 12 ).​ Following the error analysis of Varon et al. (2018), we adopt a relative uncertainty of ± 50% on individual Q estimates to account for combined errors in IME, Ueff and plume geometry, and propagate this uncertainty to the annual totals. 2.6 LSTM model for daily XCH₄ forecasting A forecasting model based on a long short-term memory (LSTM) neural network was developed to predict next-day city-mean XCH₄ using the daily time series described above. The input to the model consists of sliding windows of consecutive daily XCH₄ values of fixed length (sequence length), and the target is the XCH₄ value on the following day. Before training, the time series was normalised using the mean and standard deviation of the training period, and the record was split chronologically into training, validation and test subsets( 11 , 13 ).​ The final LSTM architecture consisted of one or more stacked LSTM layers followed by a fully connected output layer with a single neuron representing the forecasted XCH₄. Input sequences provided several weeks of methane history for each prediction. Dropout regularisation was applied to the recurrent layers, and the network was trained using the Adam optimiser with a small learning rate, mini-batch training and a mean-squared-error loss function. Model parameters such as the number of units, learning rate, batch size and number of training epochs were tuned using the validation set, and training was stopped when the validation loss stopped improving, implementing an early-stopping criterion to reduce overfitting. Performance on the independent test set was evaluated using the Pearson correlation coefficient between predicted and observed XCH₄, the mean absolute error (MAE), the root-mean-square error (RMSE) and the mean bias, showing that the model can reproduce day-to-day variability with useful accuracy( 11 ).​ 2.7 Software environment All mapping, polygon operations, spatial overlays and production of hotspot and district maps were carried out in QGIS, using the Tehran district shapefile and raster layers derived from Sentinel-5P XCH₄. Time-series processing, statistical analysis and implementation of the LSTM model were performed in the Anaconda Python environment on a personal computer, using standard scientific and machine-learning libraries available within the distribution( 6 – 15 ). 3. Results 3.1 Annual and seasonal evolution of XCH₄ over Tehran The city‑wide annual mean XCH₄ increased steadily from about 1906 ppb in 2019 to nearly 1978 ppb in 2025, corresponding to a linear trend of roughly 10 ppb yr⁻¹ over the study period (Figure 3a). Inter‑annual variability is modest, with a slight dip in 2022 relative to 2021, but the overall evolution indicates a persistent strengthening of the urban methane burden.(1) Seasonal means reveal a consistent pattern in which winter and autumn exhibit the highest XCH₄ values, while summer shows the lowest concentrations (Figure 3b). This behaviour likely reflects a combination of increased emissions and reduced boundary‑layer ventilation during the cold season together with more efficient vertical mixing in summer(1,5). 3.2 Spatial distribution and hotspot dynamics Annual mean CH₄ maps show that methane is not uniformly distributed over Tehran but is concentrated in distinct hotspot regions that persist throughout 2019–2025 (Figure 4a–g). The strongest and most persistent enhancements occur over the northern and north‑eastern districts, while additional elevated areas are found over southern industrial and landfill‑influenced zones(15). Quantitatively, the mean CH₄ within the main hotspot increases from ≈1907 ppb in 2019 to ≈1972 ppb in 2025, with intermediate values of about 1931–1958 ppb in 2020–2024. The area of the hotspot region also expands over time, especially after 2023, indicating that not only the intensity but also the spatial extent of elevated methane has grown(16,17). Sensitivity tests using alternative thresholds (μ + 0.5σ and μ + 1.5σ) produced similar spatial patterns and trends in hotspot extent (not shown), indicating that our main conclusions are robust to the choice of threshold. 3.3 District‑scale variability Aggregating XCH₄ over the 22 administrative districts reveals pronounced spatial variability within the city. Northern and north‑eastern districts (e.g. Districts 1, 3, 4 and parts of 2 and 5) systematically exhibit the highest mean XCH₄ values, whereas some central districts remain closer to the city‑wide average(10). "Southern districts affected by industrial activity and waste management facilities also show elevated methane, although typically weaker than the northern hotspot (Table 2). The fraction of each district covered by hotspot pixels increases over time in several northern and southern districts, highlighting areas where mitigation efforts may yield the largest benefits. These districts represent priority zones for targeted methane abatement and infrastructure inspection campaigns(18). 3.4 Satellite‑detected methane plumes The plume inventory for 2019–2025 documents multiple discrete methane emitters in the wider Tehran region, with annual source strengths spanning approximately Q = 1236–61200 ton yr⁻¹ (Table 1). Even when accounting for the ±50% uncertainty on Q, the largest source still exceeds O(3 × 10⁴) ton CH₄ yr⁻¹, confirming its role as a major ultra‑emitter . Most plumes are located near waste‑management facilities, gas infrastructure and industrial sites on the urban periphery(16,17,19). Many of the stronger plumes occur within or adjacent to the satellite‑derived hotspot polygons, particularly in the north‑east and south‑west of the city. This co‑location suggests that a limited number of high‑emitting point sources contribute disproportionately to the persistent column enhancements observed over Tehran(7,10). Table 1 . Annual methane plume characteristics year IME L Ueff Q_kg_s Q_kg_h Q_ton_y Q_CO2eq_ton_y 2019 3771.806604 9299.701 4.781921107 1.94,6990.0 6990 61200 1713600 2020 163.0249519 16148.928 2.125417915 0.0215 77.4 678 18984 2021 235.4568572 8190.075 3.284403985 0.0945 340 829 23212 2022 0.786463339 8306.699 3.155152031 0.0002989 1.076 9.43 264.04 2023 356.4435 8326.073 4.105666243 0.176 635 1540 43120 2024 712.0095627 24593.923 3.417371225 0.099 356 868 24304 2025 1060.130978 41813.398 1.54666552 0.0392 141.17 1236.65 34626.2 3.5 Performance of the LSTM forecast The LSTM model trained on the daily XCH₄ time series reproduces much of the observed day‑to‑day variability over Tehran. On the independent test period, the model achieves a correlation of about 0.7 with the observations, a mean absolute error of roughly 20 ppb and a root‑mean‑square error close to 26 ppb, with a small positive bias of approximately 3 ppb(13). Time‑series plots show that the LSTM captures the timing and magnitude of most seasonal peaks and troughs, although extreme methane episodes tend to be slightly under‑predicted (Figure 6). These results demonstrate that a relatively simple data‑driven approach can provide useful short‑term forecasts of urban column methane, which may support early detection of anomalous events and planning of targeted measurements(13,17,20,21). Table 2 summarises the performance metrics, and Figure X shows the distribution of residuals (forecast minus observed XCH₄), which is approximately symmetric with no strong seasonal dependence. Table 2 .Metrics: R, MAE, RMSE, bias Bias 3.07 ppb Correlation R 0.706 RSME 26.25 ppb MAE 19.87 ppb Std. deviation of residuals 23.52 ppb 4. Discussion The multi-year Sentinel-5P record reveals that Tehran’s methane burden is not only elevated but continues to rise at a rate that is climatically significant. Annual mean XCH₄ over the city increases by roughly 10 ppb yr⁻¹ between 2019 and 2025, implying a cumulative enhancement of about 70 ppb relative to the beginning of the study period. This trend is broadly consistent with the global post-2007 methane acceleration but appears stronger than typical hemispheric-scale increases, suggesting that local and regional emission changes play a substantial role in shaping Tehran’s column methane( 1 ). "Similar urban methane hotspots with comparable or lower enhancement rates have been reported for Los Angeles (Cusworth et al., 2021) and Madrid (González-Eguino et al., 2020), indicating that Tehran's methane burden and growth trajectory place it among the world's most methane-intensive cities. The pronounced winter–autumn maxima and summer minima in XCH₄ further demonstrate how seasonal variability in atmospheric mixing and emissions modulates the steadily rising background, with stable boundary layers and enhanced combustion-related activity likely amplifying cold-season enhancements( 16 , 18 , 21 ).​​ The hotspot analysis shows that the increase in methane is not spatially uniform but is concentrated over specific districts that persist as elevated regions throughout the study period. The northern and north-eastern districts, which host dense residential areas, major road corridors and parts of the gas distribution network, consistently exhibit the highest XCH₄ values, while additional hotspots emerge over southern industrial zones and landfill-influenced areas. The expansion of hotspot area and intensity after 2023 indicates that methane-rich air masses are increasingly covering a larger fraction of the city, which may reflect growth in natural-gas consumption, intensification of waste-management activities or changes in ventilation associated with urban development. District-level statistics underline this pattern by showing that a subset of northern and southern districts repeatedly account for a disproportionate share of hotspot coverage, identifying clear geographic priorities for mitigation( 6 , 7 , 15 ).​​ The plume inventory provides a complementary perspective by linking column enhancements to discrete high-emitting sources. Annual plume emission rates span more than four orders of magnitude, from less than 10 ton CH₄ yr⁻¹ in 2022 to about 6.1 × 10⁴ ton CH₄ yr⁻¹ in 2019, corresponding to up to 1.7 × 10⁶ ton CO₂-equivalent yr⁻¹ for the largest observed plume. Many of these plumes are located near landfills, wastewater treatment facilities and gas-related infrastructure on the urban periphery, and a substantial fraction lie within or adjacent to the hotspot polygons delineated from XCH₄ maps. This co-location indicates that a limited number of strong point sources can strongly influence the column-integrated methane signal, even when averaged at the several-kilometre resolution of TROPOMI, and it underscores the value of combining hotspot mapping with targeted plume detection for identifying super-emitting facilities. Targeted abatement interventions at these super-emitting facilities, particularly the landfill and wastewater treatment sites, could substantially reduce Tehran's methane footprint with geographically limited but high-impact mitigation efforts. At the same time, the presence of extensive hotspot regions without clearly associated plumes suggests that diffuse emissions from urban leakage, small sources and regional background advection also make important contributions to Tehran’s methane burden( 13 , 17 , 20 , 22 ).​​ The LSTM-based forecasting experiment demonstrates that data-driven models can exploit the temporal structure of satellite-derived XCH₄ to provide useful short-term predictions. With a correlation of about 0.7, a mean absolute error near 20 ppb and a root-mean-square error around 26 ppb on the independent test period, the model successfully reproduces much of the observed day-to-day variability and the timing of seasonal peaks, while maintaining a small positive bias of roughly 3 ppb. These skill scores are comparable to or better than those reported for LSTM applications to other atmospheric pollutants, despite the relatively coarse temporal sampling and the absence of auxiliary meteorological predictors. The systematic underestimation of some extreme enhancement events suggests that incorporating additional input features, such as boundary-layer height, wind fields, synoptic indices or proxies for urban activity, could further improve forecast performance and help capture episodic emission surges associated with specific industrial or waste-management operations( 22 , 23 ).​​ Overall, the integrated picture that emerges from trends, hotspots, plume emissions and LSTM forecasts highlights both challenges and opportunities for methane mitigation in Tehran. The persistent and expanding hotspots, together with very large emissions from a few plume sources, imply that targeted interventions at waste facilities, gas infrastructure and selected districts could yield substantial reductions in the city’s methane footprint. At the same time, the demonstrated capability of Sentinel-5P to track urban-scale XCH₄ and of LSTM to anticipate short-term variability points to a practical framework for operational monitoring: routine satellite-based hotspot maps and plume detections could guide on-the-ground inspections, while daily forecasts could support early warning of anomalous methane episodes and inform dynamic emission-control strategies. Because the methods used here rely on globally available satellite data and widely accessible open-source tools, they are readily transferable to other rapidly growing cities in the Middle East and beyond, where robust methane monitoring remains scarce but is urgently needed to meet near-term climate goals( 16 , 19 , 21 ). From an air-quality and emission-control perspective, the identified methane hotspots and super-emitting facilities highlight concrete targets for inspection and mitigation, complementing existing efforts focused on conventional air pollutants. Integrating routine satellite-based hotspot mapping, plume detection and short-term LSTM forecasts into urban air-quality management frameworks could help authorities prioritise interventions, verify mitigation actions and respond more rapidly to anomalous emission episodes over Tehran. 5. Conclusions and Outlook This study provides a multi-year satellite-based assessment of urban methane over Tehran, combining Sentinel-5P/TROPOMI XCH₄ observations, hotspot mapping, plume characterization and data-driven forecasting. The city-wide annual mean XCH₄ increases from about 1906 ppb in 2019 to nearly 1978 ppb in 2025, corresponding to a statistically robust trend of roughly 10 ppb yr⁻¹ and indicating a persistent strengthening of the urban methane burden. The seasonal cycle exhibits pronounced winter and autumn maxima, consistent with reduced boundary-layer ventilation and enhanced cold-season emissions, superimposed on this long-term upward trend. Spatially, Tehran’s methane field is dominated by persistent hotspot regions that are centred over northern and north-eastern districts and extend into southern industrial and landfill-influenced zones. The mean XCH₄ within the principal hotspot rises by about 65 ppb between 2019 and 2025, and the hotspot area expands notably after 2023, showing that both the intensity and spatial extent of elevated methane are growing. District-scale aggregation reveals that a subset of northern and southern districts repeatedly accounts for a large fraction of hotspot coverage, providing a clear geographic focus for targeted mitigation measures. The plume inventory links these column enhancements to a small number of strong point sources with annual emission rates spanning several orders of magnitude. The largest plume emits on the order of 6 × 10⁴ ton CH₄ yr⁻¹, equivalent to more than 1.7 × 10⁶ ton CO₂-eq yr⁻¹, while many smaller plumes cluster near waste-management facilities, gas infrastructure and industrial sites on the urban periphery. The frequent co-location of plumes and hotspot polygons demonstrates that super-emitters can strongly shape the urban methane distribution and that satellite-based plume detection is an efficient tool for identifying priority facilities for inspection and repair. The LSTM model trained on the daily XCH₄ series achieves a correlation of about 0.7 with observations and errors near 20–26 ppb, showing that simple recurrent architectures can provide useful next-day forecasts of urban column methane. Such forecasts could support early warning of anomalous methane episodes, guide the timing of in-situ measurement campaigns and complement longer-term emission-inventory assessments. Incorporating meteorological predictors and activity proxies represents a promising avenue for further improving forecast skill and for disentangling the relative roles of emissions and transport in driving day-to-day variability. Taken together, the results highlight Tehran as an emerging methane hotspot at the regional scale and illustrate how routinely available satellite data and modern machine-learning tools can be combined into an operationally relevant monitoring framework. Extending this framework to other rapidly growing cities in Iran, the broader Middle East and beyond would help fill critical gaps in urban methane observations, support verification of national climate mitigation pledges and contribute to near-term climate-change mitigation by enabling faster detection and abatement of major methane sources. The demonstrated capability of satellite data and LSTM forecasting to guide operational monitoring offers a practical pathway toward routine, cost-effective methane hotspot surveillance in cities where ground-based emission inventories remain limited. Declarations Funding The authors received no financial support for the research, authorship, and/or publication of this article. Conflicts of Interest / Competing Interests The authors declare that they have no conflicts of interest. Ethics Approval and Consent to Participate Not applicable. Consent for Publication Not applicable. Availability of Data and Materials / Data Availability Data sharing is not applicable to this article as no datasets were generated or analyzed during the current study. Code Availability Not applicable. Authors’ Contributions Saeed Motesaddi Zarandi conceived and designed the study. Khashayar Partovi performed the data analysis and interpretation and was a major contributor in writing the manuscript. Pedram Rastegary contributed to the methodology and critically revised the manuscript. All authors read and approved the final manuscript. References Saunois M, Stavert AR, Poulter B, Bousquet P, Canadell JG, Jackson RB, et al. The Global Methane Budget 2000--2017. Earth Syst Sci Data [Internet]. 2020;12(3):1561–623. Available from: https://essd.copernicus.org/articles/12/1561/2020/ Lan X, Nisbet EG, Dlugokencky EJ, Michel SE. What do we know about the global methane budget? Results from four decades of atmospheric CH4 observations and the way forward. Philos Trans R Soc A Math Phys Eng Sci [Internet]. 2021 Sep 27;379(2210):20200440. Available from: https://doi.org/10.1098/rsta.2020.0440 Turner AJ, Frankenberg C, Kort EA. Interpreting contemporary trends in atmospheric methane. Proc Natl Acad Sci [Internet]. 2019;116(8):2805–13. Available from: https://www.pnas.org/doi/abs/10.1073/pnas.1814297116 Jackson RB, Saunois M, Bousquet P, Canadell JG, Poulter B, Stavert AR, et al. Increasing anthropogenic methane emissions arise equally from agricultural and fossil fuel sources. Environ Res Lett [Internet]. 2020 Jul;15(7):71002. Available from: https://doi.org/10.1088/1748-9326/ab9ed2 Nisbet EG, Manning MR, Dlugokencky EJ, Fisher RE, Lowry D, Michel SE, et al. Very Strong Atmospheric Methane Growth in the 4 Years 2014–2017: Implications for the Paris Agreement. Global Biogeochem Cycles [Internet]. 2019;33(3):318–42. Available from: https://agupubs.onlinelibrary.wiley.com/doi/abs/10.1029/2018GB006009 Varon DJ, Jacob DJ, McKeever J, Jervis D, Durak BOA, Xia Y, et al. Quantifying methane point sources from fine-scale satellite observations of atmospheric methane plumes. Atmos Meas Tech [Internet]. 2018;11(10):5673–86. Available from: https://amt.copernicus.org/articles/11/5673/2018/ Lauvaux T, Giron C, Mazzolini M, d’Aspremont A, Duren R, Cusworth D, et al. Global assessment of oil and gas methane ultra-emitters. Science. 2022 Feb;375(6580):557–61. Chen Z, Jacob DJ, Nesser H, Sulprizio MP, Lorente A, Varon DJ, et al. Methane emissions from China: a high-resolution inversion of TROPOMI satellite observations. Atmos Chem Phys [Internet]. 2022;22(16):10809–26. Available from: https://acp.copernicus.org/articles/22/10809/2022/ Li T, Cheng X. Estimating daily full-coverage surface ozone concentration using satellite observations and a spatiotemporally embedded deep learning approach. Int J Appl Earth Obs Geoinf [Internet]. 2021;101:102356. Available from: https://www.sciencedirect.com/science/article/pii/S0303243421000635 Varon DJ, McKeever J, Jervis D, Maasakkers JD, Pandey S, Houweling S, et al. Satellite Discovery of Anomalously Large Methane Point Sources From Oil/Gas Production. Geophys Res Lett [Internet]. 2019;46(22):13507–16. Available from: https://agupubs.onlinelibrary.wiley.com/doi/abs/10.1029/2019GL083798 Sha MK, Langerock B, Blavier J-FL, Blumenstock T, Borsdorff T, Buschmann M, et al. Validation of methane and carbon monoxide from Sentinel-5 Precursor using TCCON and NDACC-IRWG stations. Atmos Meas Tech [Internet]. 2021;14(9):6249–304. Available from: https://amt.copernicus.org/articles/14/6249/2021/ Lorente A, Borsdorff T, Butz A, Hasekamp O, aan de Brugh J, Schneider A, et al. Methane retrieved from TROPOMI: improvement of the data product and validation of the first 2 years of measurements. Atmos Meas Tech [Internet]. 2021;14(1):665–84. Available from: https://amt.copernicus.org/articles/14/665/2021/ Hochreiter S, Schmidhuber J. Long short-term memory. Neural Comput. 1997 Nov;9(8):1735–80. Psaropa MX, Kontogiannis S, Lolis CJ, Hatzianastassiou N, Pikridas C. A Proposed Deep Learning Framework for Air Quality Forecasts, Combining Localized Particle Concentration Measurements and Meteorological Data. Appl Sci [Internet]. 2025;15(13). Available from: https://www.mdpi.com/2076-3417/15/13/7432 Irakulis-Loitxate I, Guanter L, Liu Y-N, Varon DJ, Maasakkers JD, Zhang Y, et al. Satellite-based survey of extreme methane emissions in the Permian basin. Sci Adv. 2021 Jun;7(27). Hedelius JK, Liu J, Oda T, Maksyutov S, Roehl CM, Iraci LT, et al. Southern California megacity \chem{CO_{2}}, \chem{CH_{4}}, and CO flux estimates using ground- and space-based remote sensing and a Lagrangian model. Atmos Chem Phys [Internet]. 2018;18(22):16271–91. Available from: https://acp.copernicus.org/articles/18/16271/2018/ Kalaiselvi S, Anitha V, Manimaran V, Lawrence TS. Air quality prediction using multi-source remote sensing data integration with hybrid deep learning framework. Sci Rep. 2025 Dec; Chen Z, Jacob DJ, Nesser H, Sulprizio MP, Lorente A, Varon DJ, et al. Methane emissions from China: a high-resolution inversion of TROPOMI satellite observations. Atmos Chem Phys. 2022;22(16):10809–26. Maasakkers JD, Jacob DJ, Sulprizio MP, Scarpelli TR, Nesser H, Sheng J-X, et al. Global distribution of methane emissions, emission trends, and OH concentrations and trends inferred from an inversion of GOSAT satellite data for 2010--2015. Atmos Chem Phys [Internet]. 2019;19(11):7859–81. Available from: https://acp.copernicus.org/articles/19/7859/2019/ Zhou S, Wang W, Zhu L, Qiao Q, Kang Y. Deep-learning architecture for PM(2.5) concentration prediction: A review. Environ Sci ecotechnology. 2024 Sep;21:100400. Turner AJ, Jacob DJ, Benmergui J, Wofsy SC, Maasakkers JD, Butz A, et al. A large increase in U.S. methane emissions over the past decade inferred from satellite data and surface observations. Geophys Res Lett [Internet]. 2016;43(5):2218–24. Available from: https://agupubs.onlinelibrary.wiley.com/doi/abs/10.1002/2016GL067987 Requia WJ, Di Q, Silvern R, Kelly JT, Koutrakis P, Mickley LJ, et al. An Ensemble Learning Approach for Estimating High Spatiotemporal Resolution of Ground-Level Ozone in the Contiguous United States. Environ Sci Technol. 2020 Sep;54(18):11037–47. Chen Z, Jacob DJ, Gautam R, Omara M, Stavins RN, Stowe RC, et al. Satellite quantification of methane emissions and oil--gas methane intensities from individual countries in the Middle East and North Africa: implications for climate action. Atmos Chem Phys [Internet]. 2023;23(10):5945–67. Available from: https://acp.copernicus.org/articles/23/5945/2023/ Additional Declarations No competing interests reported. Cite Share Download PDF Status: Posted Version 1 posted You are reading this latest preprint version Research Square lets you share your work early, gain feedback from the community, and start making changes to your manuscript prior to peer review in a journal. As a division of Research Square Company, we’re committed to making research communication faster, fairer, and more useful. We do this by developing innovative software and high quality services for the global research community. Our growing team is made up of researchers and industry professionals working together to solve the most critical problems facing scientific publishing. Also discoverable on Platform About Our Team In Review Editorial Policies Advisory Board Help Center Resources Author Services Accessibility API Access RSS feed Manage Cookie Preferences © Research Square 2026 | ISSN 2693-5015 (online) Privacy Policy Terms of Service Do Not Sell My Personal Information {"props":{"pageProps":{"initialData":{"identity":"rs-8955243","acceptedTermsAndConditions":true,"allowDirectSubmit":true,"archivedVersions":[],"articleType":"Research Article","associatedPublications":[],"authors":[{"id":600706321,"identity":"06ff1ff9-e3de-4b6a-a158-78b9c75de33d","order_by":0,"name":"Saeed Motesaddi Zarandi","email":"","orcid":"","institution":"","correspondingAuthor":false,"prefix":"","firstName":"Saeed","middleName":"Motesaddi","lastName":"Zarandi","suffix":""},{"id":600706322,"identity":"481382ee-256e-4c40-93fc-3c5168b33ad6","order_by":1,"name":"Khashayar Partovi","email":"data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAZAAAAAyAQMAAABI0h/eAAAABlBMVEX///8AAABVwtN+AAAACXBIWXMAAA7EAAAOxAGVKw4bAAABD0lEQVRIie3PsUrDQBzH8b8UOl2Jk/zDBfMKCYF0MNJXOSk0SwZBKBmkFALt5vyX+BYF58hBpnuAgA6C0MkhEOgonqVCh6Slm+B9p+P4fTgOwGT6gyEC9PYvmAXQP4EUmtjzUwl4xRFi59lHc3sv3WGerZs6lU5QjdfvdQqudVG0Eu6UIadS+k9vZUCFkiysJkOfFPiPuWgllyjCHuvLM0IRwMtiookI+WABwnvtInHTsC85In3YkoDizUHCMfH0QN4QJj+vRMzD5PArNiVTPniIx+SoO1AqYqg+pzYp7PwLVvGqYZura+LLFaQpjqxl/Ix1GrkWbye7Mo1/z+fbJXZud832NlZxbG0ymUz/rG/911s5L91azwAAAABJRU5ErkJggg==","orcid":"","institution":"","correspondingAuthor":true,"prefix":"","firstName":"Khashayar","middleName":"","lastName":"Partovi","suffix":""},{"id":600706323,"identity":"1fff741c-1012-4dd9-b93c-2da172ad24b8","order_by":2,"name":"pedram rastegary","email":"","orcid":"","institution":"","correspondingAuthor":false,"prefix":"","firstName":"pedram","middleName":"","lastName":"rastegary","suffix":""}],"badges":[],"createdAt":"2026-02-24 09:12:04","currentVersionCode":1,"declarations":"","doi":"10.21203/rs.3.rs-8955243/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-8955243/v1","draftVersion":[],"editorialEvents":[],"editorialNote":"","failedWorkflow":false,"files":[{"id":104081976,"identity":"6388c110-bb39-4562-a0cd-c4d557ae0bf9","added_by":"auto","created_at":"2026-03-06 14:35:28","extension":"png","order_by":1,"title":"Figure 1","display":"","copyAsset":false,"role":"figure","size":53293,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eAnnual mean CH₄ over Tehran\u003c/strong\u003e\u003c/p\u003e","description":"","filename":"1.png","url":"https://assets-eu.researchsquare.com/files/rs-8955243/v1/58cb5768384887124f912775.png"},{"id":104081982,"identity":"72ca3210-6872-41a9-b1da-011a8f1bd900","added_by":"auto","created_at":"2026-03-06 14:35:29","extension":"png","order_by":2,"title":"Figure 2","display":"","copyAsset":false,"role":"figure","size":32953,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eSeasonal mean CH4 over Tehran\u003c/strong\u003e\u003c/p\u003e","description":"","filename":"2.png","url":"https://assets-eu.researchsquare.com/files/rs-8955243/v1/1f1fd1f893b3aa38c2d5f093.png"},{"id":104081974,"identity":"07003efd-86dc-4666-9773-f6b0a10b4c23","added_by":"auto","created_at":"2026-03-06 14:35:28","extension":"png","order_by":3,"title":"Figure 3","display":"","copyAsset":false,"role":"figure","size":28296,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eSeasonal mean CH4 over Tehran\u003c/strong\u003e\u003c/p\u003e","description":"","filename":"3.png","url":"https://assets-eu.researchsquare.com/files/rs-8955243/v1/14dbbad2cca2923983227460.png"},{"id":104403332,"identity":"1199e2d6-54f8-441a-b19d-e0bc670a509c","added_by":"auto","created_at":"2026-03-11 12:18:04","extension":"png","order_by":4,"title":"Figure 4","display":"","copyAsset":false,"role":"figure","size":29463,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eSeasonal mean CH4 over Tehran\u003c/strong\u003e\u003c/p\u003e","description":"","filename":"4.png","url":"https://assets-eu.researchsquare.com/files/rs-8955243/v1/3374f74ebeb88eac5f8b3cfe.png"},{"id":104403722,"identity":"904709e8-8d6b-4a31-b438-25d88a73de9b","added_by":"auto","created_at":"2026-03-11 12:18:54","extension":"png","order_by":5,"title":"Figure 5","display":"","copyAsset":false,"role":"figure","size":395098,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eTehran Hotspots 2019-2025\u003c/strong\u003e\u003c/p\u003e","description":"","filename":"5.png","url":"https://assets-eu.researchsquare.com/files/rs-8955243/v1/63aab777e2603fc1c93ede7a.png"},{"id":104403277,"identity":"351515b4-a77a-4749-954f-dceef34db860","added_by":"auto","created_at":"2026-03-11 12:17:54","extension":"png","order_by":6,"title":"Figure 6","display":"","copyAsset":false,"role":"figure","size":73052,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eAnnual mean CH₄ over Tehran methane hotspot\u003c/strong\u003e\u003c/p\u003e","description":"","filename":"6.png","url":"https://assets-eu.researchsquare.com/files/rs-8955243/v1/42f18d2bee8ad677f48eb1b6.png"},{"id":104081975,"identity":"e5a13321-aedb-46d3-90fc-d813daf918ac","added_by":"auto","created_at":"2026-03-06 14:35:28","extension":"png","order_by":7,"title":"Figure 7","display":"","copyAsset":false,"role":"figure","size":44662,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eCH4 2019-2025 by District\u003c/strong\u003e\u003c/p\u003e","description":"","filename":"7.png","url":"https://assets-eu.researchsquare.com/files/rs-8955243/v1/8e028893dcb1b6910d79e8f3.png"},{"id":104403138,"identity":"3595e52f-b11e-4d0d-8149-bb0bfab7f4bd","added_by":"auto","created_at":"2026-03-11 12:17:34","extension":"png","order_by":8,"title":"Figure 8","display":"","copyAsset":false,"role":"figure","size":31430,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eMethane plume\u003c/strong\u003e\u003c/p\u003e","description":"","filename":"8.png","url":"https://assets-eu.researchsquare.com/files/rs-8955243/v1/98e32c4913a2e527382ecc2a.png"},{"id":104081980,"identity":"56203353-ed61-4944-9546-c7a51ea4121f","added_by":"auto","created_at":"2026-03-06 14:35:29","extension":"png","order_by":9,"title":"Figure 9","display":"","copyAsset":false,"role":"figure","size":78179,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eObserved vs LSTM-forecast mean CH₄\u003c/strong\u003e\u003c/p\u003e","description":"","filename":"9.png","url":"https://assets-eu.researchsquare.com/files/rs-8955243/v1/ac97b7ddacb3ed8d55c1efef.png"},{"id":108395978,"identity":"f67ed2b6-5624-4cac-adc7-87f5e0cb9025","added_by":"auto","created_at":"2026-05-04 07:56:15","extension":"pdf","order_by":0,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":843489,"visible":true,"origin":"","legend":"","description":"","filename":"manuscript.pdf","url":"https://assets-eu.researchsquare.com/files/rs-8955243/v1/5955c049-0443-4601-9b19-0f038bfdba05.pdf"}],"financialInterests":"No competing interests reported.","formattedTitle":"Urban methane under watch: Sentinel‑5P/TROPOMI and LSTM reveal hotspot dynamics over Tehran (2019–2025)","fulltext":[{"header":"1. Introduction","content":"\u003cp\u003eMethane (CH₄) is the second most important anthropogenic greenhouse gas after carbon dioxide and is responsible for a substantial fraction of present-day radiative forcing and near-term global warming. Multiple assessments of the global methane budget show that atmospheric CH₄ has increased rapidly since the early 2000s, with an acceleration after 2007 that current emission inventories and sink estimates cannot fully reconcile. Because methane has an atmospheric lifetime of roughly a decade, reducing CH₄ emissions is regarded as one of the most effective options for slowing the rate of climate change in the next few decades and for helping to meet the temperature targets of the Paris Agreement(\u003cspan citationid=\"CR1\" class=\"CitationRef\"\u003e1\u003c/span\u003e).​\u003c/p\u003e \u003cp\u003eSatellite remote sensing has transformed the monitoring of atmospheric methane from the global scale down to regional and local hotspots. The TROPOspheric Monitoring Instrument (TROPOMI) onboard the Sentinel-5 Precursor (Sentinel-5P) platform provides daily global observations of column-averaged dry-air mole fractions of methane (XCH₄) at high spatial resolution, enabling detection of regional enhancements, urban plumes and, under favourable conditions, strong point sources. These data products, which can be accessed and analysed efficiently through cloud-based platforms such as Google Earth Engine (GEE), have been widely used to constrain national and sectoral methane emissions, to derive global and zonal CH₄ trends, and to detect super-emitting facilities in the fossil-fuel, waste and agricultural sectors(\u003cspan citationid=\"CR2\" class=\"CitationRef\"\u003e2\u003c/span\u003e, \u003cspan citationid=\"CR3\" class=\"CitationRef\"\u003e3\u003c/span\u003e).​\u003c/p\u003e \u003cp\u003eNevertheless, detailed satellite-based assessments of urban methane over individual megacities in the Middle East are still limited, even though such cities are growing rapidly and are expected to play an increasing role in regional greenhouse-gas budgets. Tehran, the capital of Iran, is a high-elevation megacity with more than eight million inhabitants and extensive surrounding industrial and suburban areas. Its methane budget is influenced by multiple anthropogenic sources, including municipal solid-waste landfills, wastewater treatment plants, natural-gas production and distribution infrastructure, and nearby agricultural activities. Existing atmospheric studies on Tehran have mainly focused on conventional air pollutants and carbon dioxide, whereas the temporal evolution, spatial distribution and climate-relevant trends of XCH₄ over the city have not yet been systematically quantified using Sentinel-5P/TROPOMI observations(\u003cspan citationid=\"CR4\" class=\"CitationRef\"\u003e4\u003c/span\u003e).​\u003c/p\u003e \u003cp\u003eIn parallel, data-driven forecasting approaches based on deep learning have shown strong performance for predicting atmospheric variables from observational time series. Long short-term memory (LSTM) networks, in particular, can represent nonlinear and long-range temporal dependencies that are difficult to capture with traditional statistical models and have been successfully applied to forecast ozone, PM₂․₅ and composite air-quality indices. Extending such techniques to satellite-derived XCH₄ over urban areas offers the potential to characterise short-term methane variability, support early detection of anomalous emission events, and complement long-term climate-oriented trend analyses(\u003cspan citationid=\"CR2\" class=\"CitationRef\"\u003e2\u003c/span\u003e).​\u003c/p\u003e \u003cp\u003eThis study combines Sentinel-5P/TROPOMI observations, GEE-based processing and deep-learning time-series modelling to characterise atmospheric methane over Tehran for the period 2019\u0026ndash;2025 in a way that aligns with the remote-sensing and methodological focus of the journal. First, Level-2 XCH₄ retrievals, filtered for data quality and cloud contamination, are used to construct daily, seasonal and annual mean time series for the Tehran urban area and to quantify their long-term trends in the context of regional and global methane increases. Second, annual mean XCH₄ maps on a regular grid are generated, methane hotspots are identified through a consistent thresholding scheme, and their extent and intensity are analysed relative to the administrative districts of Tehran, highlighting areas with persistently elevated methane that are most relevant for climate-mitigation efforts. Third, an LSTM model is developed and evaluated for forecasting daily XCH₄ over the city using the satellite-derived time series, with skill assessed through standard error metrics and residual analysis. By integrating temporal trend analysis, hotspot mapping and data-driven forecasting within a single satellite-based framework, the work demonstrates how Sentinel-5P/TROPOMI and modern machine-learning techniques can jointly support the monitoring of urban methane and its climate-relevant dynamics in rapidly developing regions(\u003cspan citationid=\"CR1\" class=\"CitationRef\"\u003e1\u003c/span\u003e, \u003cspan citationid=\"CR2\" class=\"CitationRef\"\u003e2\u003c/span\u003e, \u003cspan citationid=\"CR5\" class=\"CitationRef\"\u003e5\u003c/span\u003e).Methane is not only a potent greenhouse gas but also a key component of the urban atmospheric environment, where it coexists with other air pollutants and reflects underlying energy and waste-management practices. Improved quantification and short-term prediction of urban methane therefore support both air-quality management and greenhouse-gas mitigation planning in megacities such as Tehran.\u003c/p\u003e"},{"header":"2. Materials and Methods","content":"\u003cdiv id=\"Sec3\" class=\"Section2\"\u003e \u003ch2\u003e2.1 Study area\u003c/h2\u003e \u003cp\u003eTehran is a high-elevation megacity in northern Iran with more than eight million inhabitants and a heterogeneous mix of residential, commercial, industrial and peri-urban agricultural land uses. The municipality is divided into 22 administrative districts that differ strongly in population density, infrastructure and emission sources relevant to methane. Major anthropogenic CH₄ emitters in and around the city include municipal solid-waste landfills, wastewater treatment plants, natural-gas distribution infrastructure and nearby agricultural areas(\u003cspan citationid=\"CR6\" class=\"CitationRef\"\u003e6\u003c/span\u003e)​.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec4\" class=\"Section2\"\u003e \u003ch2\u003e2.2 Satellite methane data\u003c/h2\u003e \u003cp\u003eThis study uses column-averaged dry-air mole fractions of methane (XCH₄) from the Sentinel-5 Precursor (Sentinel-5P) TROPOspheric Monitoring Instrument (TROPOMI) for the period 2019\u0026ndash;2025.\u003c/p\u003e \u003cp\u003eWe used the Sentinel-5P TROPOMI OFFL XCH₄ Level-2 product (version 2.x) downloaded from the Copernicus Sentinel-5P data hub. The operational Level-2 XCH₄ product provides daily global coverage with a pixel size of about 7 \u0026times; 7 km\u0026sup2; (5.5 \u0026times; 7 km\u0026sup2; in the later mission phase) and has been widely applied to map regional and urban methane enhancements. Level-2 files covering Iran were downloaded from the official data distribution service, and all further processing steps were carried out locally(\u003cspan citationid=\"CR7\" class=\"CitationRef\"\u003e7\u003c/span\u003e, \u003cspan citationid=\"CR8\" class=\"CitationRef\"\u003e8\u003c/span\u003e).​\u003c/p\u003e \u003cp\u003eOnly retrievals with valid quality information were retained, excluding pixels flagged for cloud contamination, large viewing angles or retrieval artefacts. Pixels with qa_value\u0026thinsp;\u0026lt;\u0026thinsp;0.5 and cloud fraction\u0026thinsp;\u0026gt;\u0026thinsp;0.2 were discarded, and observations with solar zenith angle\u0026thinsp;\u0026gt;\u0026thinsp;70\u0026deg; or extreme viewing geometry (across-track index outside \u0026plusmn;\u0026thinsp;30) were also excluded. The remaining pixels were re-projected to a common geographic coordinate system and clipped to a polygon representing the Tehran metropolitan area, which was constructed from the municipal boundary shapefile. Obvious outliers were removed using simple range checks and temporal consistency checks applied to individual grid cells(\u003cspan citationid=\"CR9\" class=\"CitationRef\"\u003e9\u003c/span\u003e).​ No additional bias correction was applied beyond the standard retrieval algorithm, which has been validated against ground-based TCCON and NDACC stations in previous studies.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec5\" class=\"Section2\"\u003e \u003ch2\u003e2.3 Temporal aggregation\u003c/h2\u003e \u003cp\u003eFor each day, all quality-filtered XCH₄ pixels within the Tehran mask were averaged to obtain a single city-mean XCH₄ value. Annual means were computed from the available daily values without gap-filling. The annual trend analysis was performed only for years with sufficient temporal coverage, which in our case includes all years from 2019 to 2025. Days without valid observations, mainly due to cloud cover, were left as missing to avoid introducing artificial variability. Seasonal means were computed for winter (December\u0026ndash;February), spring (March\u0026ndash;May), summer (June\u0026ndash;August) and autumn (September\u0026ndash;November) by averaging the available daily means within each season and year(\u003cspan citationid=\"CR10\" class=\"CitationRef\"\u003e10\u003c/span\u003e).​\u003c/p\u003e \u003cp\u003eAnnual means were calculated by averaging all valid daily means for each calendar year between 2019 and 2025, both for the full urban area and later for the hotspot region. Linear trends in annual XCH₄ were estimated using ordinary least-squares regression, and the slope and its standard error were used to quantify the long-term increase in urban methane over Tehran(\u003cspan citationid=\"CR10\" class=\"CitationRef\"\u003e10\u003c/span\u003e, \u003cspan citationid=\"CR11\" class=\"CitationRef\"\u003e11\u003c/span\u003e).​\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec6\" class=\"Section2\"\u003e \u003ch2\u003e2.4 Gridded maps and hotspot extraction\u003c/h2\u003e \u003cp\u003eTo characterise spatial patterns, the quality-filtered Level-2 XCH₄ pixels were aggregated onto a regular grid covering Tehran and its surroundings. A grid resolution comparable to the native TROPOMI footprint was selected so that each cell contained a sufficient number of observations while still resolving intra-urban structure. For each year, all valid XCH₄ values falling into a given grid cell were averaged to obtain annual mean XCH₄ maps(\u003cspan citationid=\"CR12\" class=\"CitationRef\"\u003e12\u003c/span\u003e).​ For each year, we computed the mean (\u0026micro;) and standard deviation (σ) of XCH₄ across all urban grid cells. Methane hotspots were then defined as grid cells where the annual mean XCH₄ exceeded \u0026micro;\u0026thinsp;+\u0026thinsp;1σ, i.e. one standard deviation above the city-wide mean.\u003c/p\u003e \u003cp\u003eMethane hotspots were identified from these maps using a thresholding approach. For each year, the mean and standard deviation of XCH₄ across all grid cells within the urban mask were calculated, and cells with XCH₄ above a chosen threshold (e.g. the city-wide mean plus one standard deviation) were marked as hotspot cells. Neighbouring hotspot cells were merged into contiguous polygons, and for each polygon the area, mean XCH₄ and maximum XCH₄ were computed to form a multi-year record of hotspot properties(\u003cspan citationid=\"CR12\" class=\"CitationRef\"\u003e12\u003c/span\u003e).​\u003c/p\u003e \u003cp\u003eTo link hotspots to the urban fabric, the annual hotspot polygons were intersected with the vector layer of the 22 municipal districts in QGIS. This overlay provided, for each district and year, the fraction of its area classified as hotspot and the corresponding district-averaged XCH₄, enabling identification of districts with persistently elevated methane levels(\u003cspan citationid=\"CR10\" class=\"CitationRef\"\u003e10\u003c/span\u003e).​\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec7\" class=\"Section2\"\u003e \u003ch2\u003e2.5 Methane plume inventory\u003c/h2\u003e \u003cp\u003eA multi-year inventory of discrete methane plumes was compiled from satellite-based plume detections around Tehran between 2019 and 2025. For each detected plume, the source location (latitude and longitude) and emission rate Q (ton yr⁻\u0026sup1;) were estimated and stored in a tabular file. Effective wind speed (Ueff) was derived from ERA5 reanalysis 10 m wind fields at 0.25\u0026deg; \u0026times; 0.25\u0026deg; resolution, bilinearly interpolated to the plume location and Sentinel-5P overpass time. For sources with multiple detected plumes within a year, annual Q was obtained by averaging the individual Q estimates, assuming continuous operation during the period of interest. The plume locations were imported into QGIS and overlaid on the hotspot maps and district boundaries to diagnose the spatial relationship between strong point sources and persistent column enhancements. Distances from plume sources to the nearest hotspot boundary were calculated to quantify the degree of co-location between plumes and hotspot regions(\u003cspan citationid=\"CR7\" class=\"CitationRef\"\u003e7\u003c/span\u003e, \u003cspan citationid=\"CR8\" class=\"CitationRef\"\u003e8\u003c/span\u003e, \u003cspan citationid=\"CR11\" class=\"CitationRef\"\u003e11\u003c/span\u003e, \u003cspan citationid=\"CR12\" class=\"CitationRef\"\u003e12\u003c/span\u003e).​ Following the error analysis of Varon et al. (2018), we adopt a relative uncertainty of \u0026plusmn;\u0026thinsp;50% on individual Q estimates to account for combined errors in IME, Ueff and plume geometry, and propagate this uncertainty to the annual totals.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec8\" class=\"Section2\"\u003e \u003ch2\u003e2.6 LSTM model for daily XCH₄ forecasting\u003c/h2\u003e \u003cp\u003eA forecasting model based on a long short-term memory (LSTM) neural network was developed to predict next-day city-mean XCH₄ using the daily time series described above. The input to the model consists of sliding windows of consecutive daily XCH₄ values of fixed length (sequence length), and the target is the XCH₄ value on the following day. Before training, the time series was normalised using the mean and standard deviation of the training period, and the record was split chronologically into training, validation and test subsets(\u003cspan citationid=\"CR11\" class=\"CitationRef\"\u003e11\u003c/span\u003e, \u003cspan citationid=\"CR13\" class=\"CitationRef\"\u003e13\u003c/span\u003e).​\u003c/p\u003e \u003cp\u003eThe final LSTM architecture consisted of one or more stacked LSTM layers followed by a fully connected output layer with a single neuron representing the forecasted XCH₄. Input sequences provided several weeks of methane history for each prediction. Dropout regularisation was applied to the recurrent layers, and the network was trained using the Adam optimiser with a small learning rate, mini-batch training and a mean-squared-error loss function. Model parameters such as the number of units, learning rate, batch size and number of training epochs were tuned using the validation set, and training was stopped when the validation loss stopped improving, implementing an early-stopping criterion to reduce overfitting. Performance on the independent test set was evaluated using the Pearson correlation coefficient between predicted and observed XCH₄, the mean absolute error (MAE), the root-mean-square error (RMSE) and the mean bias, showing that the model can reproduce day-to-day variability with useful accuracy(\u003cspan citationid=\"CR11\" class=\"CitationRef\"\u003e11\u003c/span\u003e).​\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec9\" class=\"Section2\"\u003e \u003ch2\u003e2.7 Software environment\u003c/h2\u003e \u003cp\u003eAll mapping, polygon operations, spatial overlays and production of hotspot and district maps were carried out in QGIS, using the Tehran district shapefile and raster layers derived from Sentinel-5P XCH₄. Time-series processing, statistical analysis and implementation of the LSTM model were performed in the Anaconda Python environment on a personal computer, using standard scientific and machine-learning libraries available within the distribution(\u003cspan additionalcitationids=\"CR7 CR8 CR9 CR10 CR11 CR12 CR13 CR14\" citationid=\"CR6\" class=\"CitationRef\"\u003e6\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR15\" class=\"CitationRef\"\u003e15\u003c/span\u003e).\u003c/p\u003e \u003c/div\u003e"},{"header":"3. Results","content":"\u003cp\u003e\u003cstrong\u003e3.1 Annual and seasonal evolution of XCH₄ over Tehran\u003cbr\u003e\u003c/strong\u003eThe city‑wide annual mean XCH₄ increased steadily from about\u0026nbsp;1906 ppb in 2019\u0026nbsp;to nearly\u0026nbsp;1978 ppb in 2025, corresponding to a linear trend of roughly\u0026nbsp;10 ppb yr⁻\u0026sup1;\u0026nbsp;over the study period (Figure 3a). Inter‑annual variability is modest, with a slight dip in 2022 relative to 2021, but the overall evolution indicates a persistent strengthening of the urban methane burden.(1)\u003c/p\u003e\n\u003cp\u003eSeasonal means reveal a consistent pattern in which\u0026nbsp;winter and autumn\u0026nbsp;exhibit the highest XCH₄ values, while\u0026nbsp;summer\u0026nbsp;shows the lowest concentrations (Figure 3b). This behaviour likely reflects a combination of increased emissions and reduced boundary‑layer ventilation during the cold season together with more efficient vertical mixing in summer(1,5).\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e3.2 Spatial distribution and hotspot dynamics\u003c/strong\u003e\u003cbr\u003e\u0026nbsp;Annual mean CH₄ maps show that methane is not uniformly distributed over Tehran but is concentrated in distinct hotspot regions that persist throughout 2019\u0026ndash;2025 (Figure 4a\u0026ndash;g). The strongest and most persistent enhancements occur over the northern and north‑eastern districts, while additional elevated areas are found over southern industrial and landfill‑influenced zones(15).\u003c/p\u003e\n\u003cp\u003eQuantitatively, the mean CH₄ within the main hotspot increases from \u0026asymp;1907 ppb in 2019 to \u0026asymp;1972 ppb in 2025, with intermediate values of about 1931\u0026ndash;1958 ppb in 2020\u0026ndash;2024. The area of the hotspot region also expands over time, especially after 2023, indicating that not only the intensity but also the spatial extent of elevated methane has grown(16,17). Sensitivity tests using alternative thresholds (\u0026mu; + 0.5\u0026sigma; and \u0026mu; + 1.5\u0026sigma;) produced similar spatial patterns and trends in hotspot extent (not shown), indicating that our main conclusions are robust to the choice of threshold.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e3.3 District‑scale variability\u003c/strong\u003e\u003cem\u003e\u003cbr\u003e\u003c/em\u003eAggregating XCH₄ over the 22 administrative districts reveals pronounced spatial variability within the city. Northern and north‑eastern districts (e.g. Districts 1, 3, 4 and parts of 2 and 5) systematically exhibit the highest mean XCH₄ values, whereas some central districts remain closer to the city‑wide average(10).\u003c/p\u003e\n\u003cp\u003e\u0026quot;Southern districts affected by industrial activity and waste management facilities also show elevated methane, although typically weaker than the northern hotspot (Table 2). The fraction of each district covered by hotspot pixels increases over time in several northern and southern districts, highlighting areas where mitigation efforts may yield the largest benefits. These districts represent priority zones for targeted methane abatement and infrastructure inspection campaigns(18).\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e3.4 Satellite‑detected methane plumes\u003c/strong\u003e\u003cem\u003e\u003cbr\u003e\u003c/em\u003eThe plume inventory for 2019\u0026ndash;2025 documents multiple discrete methane emitters in the wider Tehran region, with annual source strengths spanning approximately\u0026nbsp;Q = 1236\u0026ndash;61200 ton yr⁻\u0026sup1;\u0026nbsp;(Table 1).\u0026nbsp;Even when accounting for the \u0026plusmn;50% uncertainty on Q, the largest source still exceeds O(3 \u0026times; 10⁴) ton CH₄ yr⁻\u0026sup1;, confirming its role as a major ultra‑emitter\u003cem\u003e.\u003c/em\u003e Most plumes are located near waste‑management facilities, gas infrastructure and industrial sites on the urban periphery(16,17,19).\u003c/p\u003e\n\u003cp\u003eMany of the stronger plumes occur within or adjacent to the satellite‑derived hotspot polygons, particularly in the\u0026nbsp;north‑east and south‑west\u0026nbsp;of the city. This co‑location suggests that a limited number of high‑emitting point sources contribute disproportionately to the persistent column enhancements observed over Tehran(7,10).\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eTable\u0026nbsp;\u003c/strong\u003e\u003cstrong\u003e1\u003c/strong\u003e\u003cstrong\u003e\u003cspan dir=\"RTL\"\u003e.\u003c/span\u003e\u003c/strong\u003e\u003cstrong\u003eAnnual methane plume characteristics\u003c/strong\u003e\u003c/p\u003e\n\u003ctable border=\"1\" cellspacing=\"0\" cellpadding=\"0\"\u003e\n \u003ctbody\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 78px;\"\u003e\n \u003cp\u003eyear\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 78px;\"\u003e\n \u003cp\u003eIME\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 78px;\"\u003e\n \u003cp\u003eL\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 78px;\"\u003e\n \u003cp\u003eUeff\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 78px;\"\u003e\n \u003cp\u003eQ_kg_s\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 78px;\"\u003e\n \u003cp\u003eQ_kg_h\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 78px;\"\u003e\n \u003cp\u003eQ_ton_y\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 79px;\"\u003e\n \u003cp\u003eQ_CO2eq_ton_y\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 78px;\"\u003e\n \u003cp\u003e2019\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 78px;\"\u003e\n \u003cp\u003e3771.806604\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 78px;\"\u003e\n \u003cp\u003e9299.701\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 78px;\"\u003e\n \u003cp\u003e4.781921107\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 78px;\"\u003e\n \u003cp\u003e1.94,6990.0\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 78px;\"\u003e\n \u003cp\u003e6990\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 78px;\"\u003e\n \u003cp\u003e61200\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 79px;\"\u003e\n \u003cp\u003e1713600\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 78px;\"\u003e\n \u003cp\u003e2020\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 78px;\"\u003e\n \u003cp\u003e163.0249519\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 78px;\"\u003e\n \u003cp\u003e16148.928\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 78px;\"\u003e\n \u003cp\u003e2.125417915\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 78px;\"\u003e\n \u003cp\u003e0.0215\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 78px;\"\u003e\n \u003cp\u003e77.4\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 78px;\"\u003e\n \u003cp\u003e678\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 79px;\"\u003e\n \u003cp\u003e18984\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 78px;\"\u003e\n \u003cp\u003e2021\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 78px;\"\u003e\n \u003cp\u003e235.4568572\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 78px;\"\u003e\n \u003cp\u003e8190.075\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 78px;\"\u003e\n \u003cp\u003e3.284403985\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 78px;\"\u003e\n \u003cp\u003e0.0945\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 78px;\"\u003e\n \u003cp\u003e340\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 78px;\"\u003e\n \u003cp\u003e829\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 79px;\"\u003e\n \u003cp\u003e23212\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 78px;\"\u003e\n \u003cp\u003e2022\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 78px;\"\u003e\n \u003cp\u003e0.786463339\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 78px;\"\u003e\n \u003cp\u003e8306.699\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 78px;\"\u003e\n \u003cp\u003e3.155152031\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 78px;\"\u003e\n \u003cp\u003e0.0002989\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 78px;\"\u003e\n \u003cp\u003e1.076\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 78px;\"\u003e\n \u003cp\u003e9.43\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 79px;\"\u003e\n \u003cp\u003e264.04\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 78px;\"\u003e\n \u003cp\u003e2023\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 78px;\"\u003e\n \u003cp\u003e356.4435\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 78px;\"\u003e\n \u003cp\u003e8326.073\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 78px;\"\u003e\n \u003cp\u003e4.105666243\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 78px;\"\u003e\n \u003cp\u003e0.176\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 78px;\"\u003e\n \u003cp\u003e635\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 78px;\"\u003e\n \u003cp\u003e1540\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 79px;\"\u003e\n \u003cp\u003e43120\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 78px;\"\u003e\n \u003cp\u003e2024\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 78px;\"\u003e\n \u003cp\u003e712.0095627\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 78px;\"\u003e\n \u003cp\u003e24593.923\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 78px;\"\u003e\n \u003cp\u003e3.417371225\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 78px;\"\u003e\n \u003cp\u003e0.099\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 78px;\"\u003e\n \u003cp\u003e356\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 78px;\"\u003e\n \u003cp\u003e868\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 79px;\"\u003e\n \u003cp\u003e24304\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 78px;\"\u003e\n \u003cp\u003e2025\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 78px;\"\u003e\n \u003cp\u003e1060.130978\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 78px;\"\u003e\n \u003cp\u003e41813.398\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 78px;\"\u003e\n \u003cp\u003e1.54666552\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 78px;\"\u003e\n \u003cp\u003e0.0392\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 78px;\"\u003e\n \u003cp\u003e141.17\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 78px;\"\u003e\n \u003cp\u003e1236.65\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 79px;\"\u003e\n \u003cp\u003e34626.2\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003c/tbody\u003e\n\u003c/table\u003e\n\u003cp\u003e\u003cstrong\u003e3.5 Performance of the LSTM forecast\u003c/strong\u003e\u003cem\u003e\u003cbr\u003e\u003c/em\u003eThe LSTM model trained on the daily XCH₄ time series reproduces much of the observed day‑to‑day variability over Tehran. On the independent test period, the model achieves a\u0026nbsp;correlation of about 0.7\u0026nbsp;with the observations, a\u0026nbsp;mean absolute error of roughly 20 ppb\u0026nbsp;and a\u0026nbsp;root‑mean‑square error close to 26 ppb, with a small positive bias of approximately\u0026nbsp;3 ppb(13).\u003c/p\u003e\n\u003cp\u003eTime‑series plots show that the LSTM captures the timing and magnitude of most seasonal peaks and troughs, although extreme methane episodes tend to be slightly under‑predicted (Figure 6). These results demonstrate that a relatively simple data‑driven approach can provide useful short‑term forecasts of urban column methane, which may support early detection of anomalous events and planning of targeted measurements(13,17,20,21).\u003c/p\u003e\n\u003cp\u003eTable 2 summarises the performance metrics, and Figure X shows the distribution of residuals (forecast minus observed XCH₄), which is approximately symmetric with no strong seasonal dependence.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eTable\u0026nbsp;\u003c/strong\u003e\u003cstrong\u003e2\u003c/strong\u003e\u003cstrong\u003e.Metrics: R, MAE, RMSE, bias\u003c/strong\u003e\u003c/p\u003e\n\u003ctable border=\"1\" cellspacing=\"0\" cellpadding=\"0\" align=\"\"\u003e\n \u003ctbody\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 143px;\"\u003e\n \u003cp\u003eBias\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 143px;\"\u003e\n \u003cp\u003e3.07 ppb\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 143px;\"\u003e\n \u003cp\u003eCorrelation R\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 143px;\"\u003e\n \u003cp\u003e0.706\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 143px;\"\u003e\n \u003cp\u003eRSME\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 143px;\"\u003e\n \u003cp\u003e26.25 ppb\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 143px;\"\u003e\n \u003cp\u003eMAE\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 143px;\"\u003e\n \u003cp\u003e19.87 ppb\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 143px;\"\u003e\n \u003cp\u003eStd. deviation of residuals\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 143px;\"\u003e\n \u003cp\u003e23.52 ppb\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003c/tbody\u003e\n\u003c/table\u003e"},{"header":"4. Discussion","content":"\u003cp\u003eThe multi-year Sentinel-5P record reveals that Tehran\u0026rsquo;s methane burden is not only elevated but continues to rise at a rate that is climatically significant. Annual mean XCH₄ over the city increases by roughly 10 ppb yr⁻\u0026sup1; between 2019 and 2025, implying a cumulative enhancement of about 70 ppb relative to the beginning of the study period. This trend is broadly consistent with the global post-2007 methane acceleration but appears stronger than typical hemispheric-scale increases, suggesting that local and regional emission changes play a substantial role in shaping Tehran\u0026rsquo;s column methane(\u003cspan citationid=\"CR1\" class=\"CitationRef\"\u003e1\u003c/span\u003e).\u003c/p\u003e \u003cp\u003e\"Similar urban methane hotspots with comparable or lower enhancement rates have been reported for Los Angeles (Cusworth et al., 2021) and Madrid (Gonz\u0026aacute;lez-Eguino et al., 2020), indicating that Tehran's methane burden and growth trajectory place it among the world's most methane-intensive cities. The pronounced winter\u0026ndash;autumn maxima and summer minima in XCH₄ further demonstrate how seasonal variability in atmospheric mixing and emissions modulates the steadily rising background, with stable boundary layers and enhanced combustion-related activity likely amplifying cold-season enhancements(\u003cspan citationid=\"CR16\" class=\"CitationRef\"\u003e16\u003c/span\u003e, \u003cspan citationid=\"CR18\" class=\"CitationRef\"\u003e18\u003c/span\u003e, \u003cspan citationid=\"CR21\" class=\"CitationRef\"\u003e21\u003c/span\u003e).​​\u003c/p\u003e \u003cp\u003eThe hotspot analysis shows that the increase in methane is not spatially uniform but is concentrated over specific districts that persist as elevated regions throughout the study period. The northern and north-eastern districts, which host dense residential areas, major road corridors and parts of the gas distribution network, consistently exhibit the highest XCH₄ values, while additional hotspots emerge over southern industrial zones and landfill-influenced areas. The expansion of hotspot area and intensity after 2023 indicates that methane-rich air masses are increasingly covering a larger fraction of the city, which may reflect growth in natural-gas consumption, intensification of waste-management activities or changes in ventilation associated with urban development. District-level statistics underline this pattern by showing that a subset of northern and southern districts repeatedly account for a disproportionate share of hotspot coverage, identifying clear geographic priorities for mitigation(\u003cspan citationid=\"CR6\" class=\"CitationRef\"\u003e6\u003c/span\u003e, \u003cspan citationid=\"CR7\" class=\"CitationRef\"\u003e7\u003c/span\u003e, \u003cspan citationid=\"CR15\" class=\"CitationRef\"\u003e15\u003c/span\u003e).​​\u003c/p\u003e \u003cp\u003eThe plume inventory provides a complementary perspective by linking column enhancements to discrete high-emitting sources. Annual plume emission rates span more than four orders of magnitude, from less than 10 ton CH₄ yr⁻\u0026sup1; in 2022 to about 6.1 \u0026times; 10⁴ ton CH₄ yr⁻\u0026sup1; in 2019, corresponding to up to 1.7 \u0026times; 10⁶ ton CO₂-equivalent yr⁻\u0026sup1; for the largest observed plume. Many of these plumes are located near landfills, wastewater treatment facilities and gas-related infrastructure on the urban periphery, and a substantial fraction lie within or adjacent to the hotspot polygons delineated from XCH₄ maps. This co-location indicates that a limited number of strong point sources can strongly influence the column-integrated methane signal, even when averaged at the several-kilometre resolution of TROPOMI, and it underscores the value of combining hotspot mapping with targeted plume detection for identifying super-emitting facilities. Targeted abatement interventions at these super-emitting facilities, particularly the landfill and wastewater treatment sites, could substantially reduce Tehran's methane footprint with geographically limited but high-impact mitigation efforts. At the same time, the presence of extensive hotspot regions without clearly associated plumes suggests that diffuse emissions from urban leakage, small sources and regional background advection also make important contributions to Tehran\u0026rsquo;s methane burden(\u003cspan citationid=\"CR13\" class=\"CitationRef\"\u003e13\u003c/span\u003e, \u003cspan citationid=\"CR17\" class=\"CitationRef\"\u003e17\u003c/span\u003e, \u003cspan citationid=\"CR20\" class=\"CitationRef\"\u003e20\u003c/span\u003e, \u003cspan citationid=\"CR22\" class=\"CitationRef\"\u003e22\u003c/span\u003e).​​\u003c/p\u003e \u003cp\u003eThe LSTM-based forecasting experiment demonstrates that data-driven models can exploit the temporal structure of satellite-derived XCH₄ to provide useful short-term predictions. With a correlation of about 0.7, a mean absolute error near 20 ppb and a root-mean-square error around 26 ppb on the independent test period, the model successfully reproduces much of the observed day-to-day variability and the timing of seasonal peaks, while maintaining a small positive bias of roughly 3 ppb. These skill scores are comparable to or better than those reported for LSTM applications to other atmospheric pollutants, despite the relatively coarse temporal sampling and the absence of auxiliary meteorological predictors. The systematic underestimation of some extreme enhancement events suggests that incorporating additional input features, such as boundary-layer height, wind fields, synoptic indices or proxies for urban activity, could further improve forecast performance and help capture episodic emission surges associated with specific industrial or waste-management operations(\u003cspan citationid=\"CR22\" class=\"CitationRef\"\u003e22\u003c/span\u003e, \u003cspan citationid=\"CR23\" class=\"CitationRef\"\u003e23\u003c/span\u003e).​​\u003c/p\u003e \u003cp\u003eOverall, the integrated picture that emerges from trends, hotspots, plume emissions and LSTM forecasts highlights both challenges and opportunities for methane mitigation in Tehran. The persistent and expanding hotspots, together with very large emissions from a few plume sources, imply that targeted interventions at waste facilities, gas infrastructure and selected districts could yield substantial reductions in the city\u0026rsquo;s methane footprint. At the same time, the demonstrated capability of Sentinel-5P to track urban-scale XCH₄ and of LSTM to anticipate short-term variability points to a practical framework for operational monitoring: routine satellite-based hotspot maps and plume detections could guide on-the-ground inspections, while daily forecasts could support early warning of anomalous methane episodes and inform dynamic emission-control strategies. Because the methods used here rely on globally available satellite data and widely accessible open-source tools, they are readily transferable to other rapidly growing cities in the Middle East and beyond, where robust methane monitoring remains scarce but is urgently needed to meet near-term climate goals(\u003cspan citationid=\"CR16\" class=\"CitationRef\"\u003e16\u003c/span\u003e, \u003cspan citationid=\"CR19\" class=\"CitationRef\"\u003e19\u003c/span\u003e, \u003cspan citationid=\"CR21\" class=\"CitationRef\"\u003e21\u003c/span\u003e). From an air-quality and emission-control perspective, the identified methane hotspots and super-emitting facilities highlight concrete targets for inspection and mitigation, complementing existing efforts focused on conventional air pollutants. Integrating routine satellite-based hotspot mapping, plume detection and short-term LSTM forecasts into urban air-quality management frameworks could help authorities prioritise interventions, verify mitigation actions and respond more rapidly to anomalous emission episodes over Tehran.\u003c/p\u003e"},{"header":"5. Conclusions and Outlook","content":"\u003cp\u003eThis study provides a multi-year satellite-based assessment of urban methane over Tehran, combining Sentinel-5P/TROPOMI XCH₄ observations, hotspot mapping, plume characterization and data-driven forecasting. The city-wide annual mean XCH₄ increases from about 1906 ppb in 2019 to nearly 1978 ppb in 2025, corresponding to a statistically robust trend of roughly 10 ppb yr⁻\u0026sup1; and indicating a persistent strengthening of the urban methane burden. The seasonal cycle exhibits pronounced winter and autumn maxima, consistent with reduced boundary-layer ventilation and enhanced cold-season emissions, superimposed on this long-term upward trend.\u003c/p\u003e \u003cp\u003eSpatially, Tehran\u0026rsquo;s methane field is dominated by persistent hotspot regions that are centred over northern and north-eastern districts and extend into southern industrial and landfill-influenced zones. The mean XCH₄ within the principal hotspot rises by about 65 ppb between 2019 and 2025, and the hotspot area expands notably after 2023, showing that both the intensity and spatial extent of elevated methane are growing. District-scale aggregation reveals that a subset of northern and southern districts repeatedly accounts for a large fraction of hotspot coverage, providing a clear geographic focus for targeted mitigation measures.\u003c/p\u003e \u003cp\u003eThe plume inventory links these column enhancements to a small number of strong point sources with annual emission rates spanning several orders of magnitude. The largest plume emits on the order of 6 \u0026times; 10⁴ ton CH₄ yr⁻\u0026sup1;, equivalent to more than 1.7 \u0026times; 10⁶ ton CO₂-eq yr⁻\u0026sup1;, while many smaller plumes cluster near waste-management facilities, gas infrastructure and industrial sites on the urban periphery. The frequent co-location of plumes and hotspot polygons demonstrates that super-emitters can strongly shape the urban methane distribution and that satellite-based plume detection is an efficient tool for identifying priority facilities for inspection and repair.\u003c/p\u003e \u003cp\u003eThe LSTM model trained on the daily XCH₄ series achieves a correlation of about 0.7 with observations and errors near 20\u0026ndash;26 ppb, showing that simple recurrent architectures can provide useful next-day forecasts of urban column methane. Such forecasts could support early warning of anomalous methane episodes, guide the timing of in-situ measurement campaigns and complement longer-term emission-inventory assessments. Incorporating meteorological predictors and activity proxies represents a promising avenue for further improving forecast skill and for disentangling the relative roles of emissions and transport in driving day-to-day variability.\u003c/p\u003e \u003cp\u003eTaken together, the results highlight Tehran as an emerging methane hotspot at the regional scale and illustrate how routinely available satellite data and modern machine-learning tools can be combined into an operationally relevant monitoring framework. Extending this framework to other rapidly growing cities in Iran, the broader Middle East and beyond would help fill critical gaps in urban methane observations, support verification of national climate mitigation pledges and contribute to near-term climate-change mitigation by enabling faster detection and abatement of major methane sources. The demonstrated capability of satellite data and LSTM forecasting to guide operational monitoring offers a practical pathway toward routine, cost-effective methane hotspot surveillance in cities where ground-based emission inventories remain limited.\u003c/p\u003e"},{"header":"Declarations","content":"\u003cp\u003e\u003cstrong\u003eFunding\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe authors received no financial support for the research, authorship, and/or publication of this article.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eConflicts of Interest / Competing Interests\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe authors declare that they have no conflicts of interest.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eEthics Approval and Consent to Participate\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eNot applicable.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eConsent for Publication\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eNot applicable.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eAvailability of Data and Materials / Data Availability\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eData sharing is not applicable to this article as no datasets were generated or analyzed during the current study.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eCode Availability\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eNot applicable.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eAuthors\u0026rsquo; Contributions\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eSaeed Motesaddi Zarandi conceived and designed the study. Khashayar Partovi performed the data analysis and interpretation and was a major contributor in writing the manuscript. Pedram Rastegary contributed to the methodology and critically revised the manuscript. All authors read and approved the final manuscript.\u003c/p\u003e"},{"header":"References","content":"\u003col\u003e\n\u003cli\u003eSaunois M, Stavert AR, Poulter B, Bousquet P, Canadell JG, Jackson RB, et al. The Global Methane Budget 2000--2017. Earth Syst Sci Data [Internet]. 2020;12(3):1561\u0026ndash;623. Available from: https://essd.copernicus.org/articles/12/1561/2020/\u003c/li\u003e\n\u003cli\u003eLan X, Nisbet EG, Dlugokencky EJ, Michel SE. What do we know about the global methane budget? Results from four decades of atmospheric CH4 observations and the way forward. Philos Trans R Soc A Math Phys Eng Sci [Internet]. 2021 Sep 27;379(2210):20200440. Available from: https://doi.org/10.1098/rsta.2020.0440\u003c/li\u003e\n\u003cli\u003eTurner AJ, Frankenberg C, Kort EA. Interpreting contemporary trends in atmospheric methane. Proc Natl Acad Sci [Internet]. 2019;116(8):2805\u0026ndash;13. Available from: https://www.pnas.org/doi/abs/10.1073/pnas.1814297116\u003c/li\u003e\n\u003cli\u003eJackson RB, Saunois M, Bousquet P, Canadell JG, Poulter B, Stavert AR, et al. Increasing anthropogenic methane emissions arise equally from agricultural and fossil fuel sources. Environ Res Lett [Internet]. 2020 Jul;15(7):71002. Available from: https://doi.org/10.1088/1748-9326/ab9ed2\u003c/li\u003e\n\u003cli\u003eNisbet EG, Manning MR, Dlugokencky EJ, Fisher RE, Lowry D, Michel SE, et al. Very Strong Atmospheric Methane Growth in the 4 Years 2014\u0026ndash;2017: Implications for the Paris Agreement. Global Biogeochem Cycles [Internet]. 2019;33(3):318\u0026ndash;42. Available from: https://agupubs.onlinelibrary.wiley.com/doi/abs/10.1029/2018GB006009\u003c/li\u003e\n\u003cli\u003eVaron DJ, Jacob DJ, McKeever J, Jervis D, Durak BOA, Xia Y, et al. Quantifying methane point sources from fine-scale satellite observations of atmospheric methane plumes. Atmos Meas Tech [Internet]. 2018;11(10):5673\u0026ndash;86. Available from: https://amt.copernicus.org/articles/11/5673/2018/\u003c/li\u003e\n\u003cli\u003eLauvaux T, Giron C, Mazzolini M, d\u0026rsquo;Aspremont A, Duren R, Cusworth D, et al. Global assessment of oil and gas methane ultra-emitters. Science. 2022 Feb;375(6580):557\u0026ndash;61. \u003c/li\u003e\n\u003cli\u003eChen Z, Jacob DJ, Nesser H, Sulprizio MP, Lorente A, Varon DJ, et al. Methane emissions from China: a high-resolution inversion of TROPOMI satellite observations. Atmos Chem Phys [Internet]. 2022;22(16):10809\u0026ndash;26. Available from: https://acp.copernicus.org/articles/22/10809/2022/\u003c/li\u003e\n\u003cli\u003eLi T, Cheng X. Estimating daily full-coverage surface ozone concentration using satellite observations and a spatiotemporally embedded deep learning approach. Int J Appl Earth Obs Geoinf [Internet]. 2021;101:102356. Available from: https://www.sciencedirect.com/science/article/pii/S0303243421000635\u003c/li\u003e\n\u003cli\u003eVaron DJ, McKeever J, Jervis D, Maasakkers JD, Pandey S, Houweling S, et al. Satellite Discovery of Anomalously Large Methane Point Sources From Oil/Gas Production. Geophys Res Lett [Internet]. 2019;46(22):13507\u0026ndash;16. Available from: https://agupubs.onlinelibrary.wiley.com/doi/abs/10.1029/2019GL083798\u003c/li\u003e\n\u003cli\u003eSha MK, Langerock B, Blavier J-FL, Blumenstock T, Borsdorff T, Buschmann M, et al. Validation of methane and carbon monoxide from Sentinel-5 Precursor using TCCON and NDACC-IRWG stations. Atmos Meas Tech [Internet]. 2021;14(9):6249\u0026ndash;304. Available from: https://amt.copernicus.org/articles/14/6249/2021/\u003c/li\u003e\n\u003cli\u003eLorente A, Borsdorff T, Butz A, Hasekamp O, aan de Brugh J, Schneider A, et al. Methane retrieved from TROPOMI: improvement of the data product and validation of the first 2 years of measurements. Atmos Meas Tech [Internet]. 2021;14(1):665\u0026ndash;84. Available from: https://amt.copernicus.org/articles/14/665/2021/\u003c/li\u003e\n\u003cli\u003eHochreiter S, Schmidhuber J. Long short-term memory. Neural Comput. 1997 Nov;9(8):1735\u0026ndash;80. \u003c/li\u003e\n\u003cli\u003ePsaropa MX, Kontogiannis S, Lolis CJ, Hatzianastassiou N, Pikridas C. A Proposed Deep Learning Framework for Air Quality Forecasts, Combining Localized Particle Concentration Measurements and Meteorological Data. Appl Sci [Internet]. 2025;15(13). Available from: https://www.mdpi.com/2076-3417/15/13/7432\u003c/li\u003e\n\u003cli\u003eIrakulis-Loitxate I, Guanter L, Liu Y-N, Varon DJ, Maasakkers JD, Zhang Y, et al. Satellite-based survey of extreme methane emissions in the Permian basin. Sci Adv. 2021 Jun;7(27). \u003c/li\u003e\n\u003cli\u003eHedelius JK, Liu J, Oda T, Maksyutov S, Roehl CM, Iraci LT, et al. Southern California megacity \\chem{CO_{2}}, \\chem{CH_{4}}, and CO flux estimates using ground- and space-based remote sensing and a Lagrangian model. Atmos Chem Phys [Internet]. 2018;18(22):16271\u0026ndash;91. Available from: https://acp.copernicus.org/articles/18/16271/2018/\u003c/li\u003e\n\u003cli\u003eKalaiselvi S, Anitha V, Manimaran V, Lawrence TS. Air quality prediction using multi-source remote sensing data integration with hybrid deep learning framework. Sci Rep. 2025 Dec; \u003c/li\u003e\n\u003cli\u003eChen Z, Jacob DJ, Nesser H, Sulprizio MP, Lorente A, Varon DJ, et al. Methane emissions from China: a high-resolution inversion of TROPOMI satellite observations. Atmos Chem Phys. 2022;22(16):10809\u0026ndash;26. \u003c/li\u003e\n\u003cli\u003eMaasakkers JD, Jacob DJ, Sulprizio MP, Scarpelli TR, Nesser H, Sheng J-X, et al. Global distribution of methane emissions, emission trends, and OH concentrations and trends inferred from an inversion of GOSAT satellite data for 2010--2015. Atmos Chem Phys [Internet]. 2019;19(11):7859\u0026ndash;81. Available from: https://acp.copernicus.org/articles/19/7859/2019/\u003c/li\u003e\n\u003cli\u003eZhou S, Wang W, Zhu L, Qiao Q, Kang Y. Deep-learning architecture for PM(2.5) concentration prediction: A review. Environ Sci ecotechnology. 2024 Sep;21:100400. \u003c/li\u003e\n\u003cli\u003eTurner AJ, Jacob DJ, Benmergui J, Wofsy SC, Maasakkers JD, Butz A, et al. A large increase in U.S. methane emissions over the past decade inferred from satellite data and surface observations. Geophys Res Lett [Internet]. 2016;43(5):2218\u0026ndash;24. Available from: https://agupubs.onlinelibrary.wiley.com/doi/abs/10.1002/2016GL067987\u003c/li\u003e\n\u003cli\u003eRequia WJ, Di Q, Silvern R, Kelly JT, Koutrakis P, Mickley LJ, et al. An Ensemble Learning Approach for Estimating High Spatiotemporal Resolution of Ground-Level Ozone in the Contiguous United States. Environ Sci Technol. 2020 Sep;54(18):11037\u0026ndash;47. \u003c/li\u003e\n\u003cli\u003eChen Z, Jacob DJ, Gautam R, Omara M, Stavins RN, Stowe RC, et al. Satellite quantification of methane emissions and oil--gas methane intensities from individual countries in the Middle East and North Africa: implications for climate action. Atmos Chem Phys [Internet]. 2023;23(10):5945\u0026ndash;67. Available from: https://acp.copernicus.org/articles/23/5945/2023/\u003c/li\u003e\n\u003c/ol\u003e"}],"fulltextSource":"","fullText":"","funders":[],"hasAdminPriorityOnWorkflow":false,"hasManuscriptDocX":true,"hasOptedInToPreprint":true,"hasPassedJournalQc":"","hasAnyPriority":false,"hideJournal":true,"highlight":"","institution":"","isAcceptedByJournal":false,"isAuthorSuppliedPdf":false,"isDeskRejected":"","isHiddenFromSearch":false,"isInQc":false,"isInWorkflow":false,"isPdf":false,"isPdfUpToDate":true,"isWithdrawnOrRetracted":false,"journal":{"display":true,"email":"[email protected]","identity":"researchsquare","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":true,"externalIdentity":"","sideBox":"","snPcode":"","submissionUrl":"/submission","title":"Research Square","twitterHandle":"researchsquare","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"","reportingPortfolio":"","inReviewEnabled":false,"inReviewRevisionsEnabled":true},"keywords":"","lastPublishedDoi":"10.21203/rs.3.rs-8955243/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-8955243/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"\u003cp\u003eRapidly rising atmospheric methane threatens near-term climate goals, yet the behaviour of urban methane over Middle Eastern megacities remains poorly constrained. Using Sentinel-5P/TROPOMI XCH₄ retrievals processed in Google Earth Engine, we assemble a daily record of column methane over Tehran for 2019\u0026ndash;2025 and derive seasonal and annual means at the city scale. The annual mean XCH₄ increases from about 1906 ppb in 2019 to nearly 1978 ppb in 2025, corresponding to a trend of roughly 10 ppb yr⁻\u0026sup1; and indicating a persistent strengthening of the urban methane burden. Gridded annual maps and a district-level aggregation show that the highest mean XCH₄ values occur persistently over the northern and north-eastern districts of Tehran, while elevated enhancements also appear over southern industrial and landfill-influenced areas; these hotspot polygons are quantified in terms of area and mean XCH₄ for each year.​\u003c/p\u003e \u003cp\u003eTo link column enhancements with discrete emitters, we compile a multi-year inventory of satellite-detected methane plumes around Tehran, including source locations and annual emission rates (Q, ton yr⁻\u0026sup1;), and examine how plume occurrence and intensity co-vary with the hotspot fields. Building on the daily XCH₄ time series, we train a long short-term memory (LSTM) network to forecast next-day column methane over the city; on the independent test period the model achieves a correlation of about 0.7 with observations, a mean absolute error near 20 ppb, a root-mean-square error around 26 ppb, and a small positive mean bias of ~\u0026thinsp;3 ppb. By jointly analysing daily, seasonal and annual XCH₄, district-scale hotspots and plume emissions, and by demonstrating a deep-learning forecast of urban methane, this study provides a comprehensive satellite-based picture of methane dynamics over Tehran and a readily transferable framework for monitoring methane hotspots in rapidly growing cities across the Middle East and developing regions.\u003c/p\u003e","manuscriptTitle":"Urban methane under watch: Sentinel‑5P/TROPOMI and LSTM reveal hotspot dynamics over Tehran (2019–2025)","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2026-03-06 14:35:12","doi":"10.21203/rs.3.rs-8955243/v1","editorialEvents":[{"type":"communityComments","content":0}],"status":"published","journal":{"display":true,"email":"[email protected]","identity":"researchsquare","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":true,"externalIdentity":"","sideBox":"","snPcode":"","submissionUrl":"/submission","title":"Research Square","twitterHandle":"researchsquare","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"","reportingPortfolio":"","inReviewEnabled":false,"inReviewRevisionsEnabled":true}}],"origin":"","ownerIdentity":"97245dad-2408-4f36-9a5e-c3737c1706a2","owner":[],"postedDate":"March 6th, 2026","published":true,"recentEditorialEvents":[{"type":"decision","content":"Rejected","date":"2026-05-04T07:48:46+00:00","index":"","fulltext":""},{"type":"editorInvitedReview","content":"","date":"2026-04-29T10:37:50+00:00","index":93,"fulltext":""}],"rejectedJournal":[],"revision":"","amendment":"","status":"posted","subjectAreas":[],"tags":[],"updatedAt":"2026-05-04T07:55:32+00:00","versionOfRecord":[],"versionCreatedAt":"2026-03-06 14:35:12","video":"","vorDoi":"","vorDoiUrl":"","workflowStages":[]},"version":"v1","identity":"rs-8955243","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-8955243","identity":"rs-8955243","version":["v1"]},"buildId":"XKTyCvWXoU3ODBz1xrDgd","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}