Abhinav Anand

Exposure to high levels of air pollution and extreme heat stress has been individually linked to significant health and economic problems in India. However, not much is known about the threats resulting from the co-occurrence of these hazards, which have common meteorological drivers. In this study, we perform a country-wide joint risk analysis at the district level, examining the risks of chronic exposure to the joint extremes of particulate matter pollution (PM _2.5 , mass of particles smaller than 2.5 µm in diameter) and heat stress (wet bulb globe temperature, WBGT) in the presence of geophysical and socioeconomic vulnerability. We find that northern, central, and eastern India are at alarming levels of joint risk, potentially exposing 61% (~842 million) of the population and 56% (~1.84 million km²) of land area. Spatially differentiated drivers of joint risk are high levels of PM _2.5 in northern India, high heat hazards in northern and eastern India, and high vulnerability in the central and eastern states of Chhattisgarh, Madhya Pradesh, Bihar, West Bengal, and Odisha. Region-specific response actions for long-term risk mitigation include reducing emissions from residential and agricultural biomass combustion and industrial coal-burning sources as well as addressing the challenges of housing, healthcare delivery, and economic conditions in central and eastern India. A coordinated policy framework aimed at tackling individual hazards is vital to alleviating these joint risks.

Targeting urban air quality improvements in India, the National Clean Air Program (NCAP) was launched in 2019 to reduce PM_2.5 concentrations by 20%–30% in 122 initial cities over a 7 year period (2017–2024). However, considering the regional nature of air pollution nationwide and the significant emission load from rural areas, a potentially large fraction of urban PM_2.5 might originate from emissions outside of a city’s boundary, i.e. transboundary emissions. Here, we couple top-down (STILT-PM_2.5) and bottom-up (WRF-Chem) modeling approaches with a new, nationwide, monthly-resolved, and fine-scale (5 km × 5 km) anthropogenic emission inventory to assess the impact of transboundary emissions to urban PM_2.5 concentrations across 143 cities (122 NCAP and 67 million plus -cities with population >1 million, among which 46 cities are also NCAP cities). We find that, on average, ∼85% (STILT-PM_2.5: 82% [95% CI: 80–85%; IQR: 77%–94%] & WRF-Chem: 89% [95% CI: 87%–91%; IQR: 88%–96%]) of urban PM_2.5 across the 143 cities originates from transboundary emissions, with domestic biomass burning (32%), energy generation (16%) and industry (15%) being the leading average emission sources to the transboundary contribution. In addition, 107 of the 122 NCAP cities from both modeling approaches have annual transboundary PM_2.5 contributions exceeding 80%, indicating that an entire mitigation of within-boundary emissions alone in these cities will not achieve the most conservative targets outlined as part of NCAP. Our findings underscore the need for multi-scale, regional action planning and implementation to achieve PM_2.5-air quality targets throughout India.

Quantifying methane emissions is essential for meeting near-term climate goals and is typically carried out using methane concentrations measured downwind of the source. One major source of methane that is important to observe and promptly remediate is fugitive emissions from oil and gas production sites but installing methane sensors at the thousands of sites within a production basin is expensive. In recent years, relatively inexpensive metal oxide sensors have been used to measure methane concentrations at production sites. Current methods used to calibrate metal oxide sensors have been shown to have significant shortcomings, resulting in limited confidence in methane concentrations generated by these sensors. To address this, we investigate using machine learning (ML) to generate a model that converts metal oxide sensor output to methane mixing ratios. To generate test data, two metal oxide sensors, TGS2600 and TGS2611, were collocated with a trace methane analyzer downwind of controlled methane releases. Over the duration of the measurements, the trace gas analyzer’s average methane mixing ratio was 2.40 ppm with a maximum of 147.6 ppm. The average calculated methane mixing ratios for the TGS2600 and TGS2611 using the ML algorithm were 2.42 ppm and 2.40 ppm, with maximum values of 117.5 ppm and 106.3 ppm, respectively. A comparison of histograms generated using the analyzer and metal oxide sensors mixing ratios shows overlap coefficients of 0.95 and 0.94 for the TGS2600 and TGS2611, respectively. Overall, our results showed there was a good agreement between the ML-derived metal oxide sensors’ mixing ratios and those generated using the more accurate trace gas analyzer. This suggests that the response of lower-cost sensors calibrated using ML could be used to generate mixing ratios with precision and accuracy comparable to higher priced trace methane analyzers. This would improve confidence in low-cost sensors’ response, reduce the cost of sensor deployment, and allow for timely and accurate tracking of methane emissions.

Quantifying methane emissions is essential for meeting near-term climate goals and typically done using methane concentrations measured downwind of the source. One major source of methane that is important to observe and promptly remediate is fugitive emissions from oil and gas productions sites but installing methane sensors at the thousands of sites within a production basin is expensive. In recent years, relatively inexpensive metal oxide sensors have been used to measure methane concentrations at production sites. Current methods used to calibrate metal oxide sensors have been shown to have significant shortcomings resulting in limited confidence in methane concentrations generated by these sensors. To address this, we investigate using machine learning (ML) to generate a model that converts metal oxide sensor output to methane mixing ratios. To generate test data, two metal oxide sensors, TGS2600 and TGS2611, were collocated with a trace methane analyzer downwind of controlled methane releases. Over the duration of the measurements, the trace gas analyzer’s average methane mixing ratio was 2.40 ppm with a maximum of 147.6 ppm. The average calculated methane mixing ratios for the TGS2600 and TGS2611 using the ML algorithm were 2.42 ppm and 2.40 ppm with maximum values of 105.9 ppm and 106.3 ppm, respectively. A comparison of histograms generated using the analyzer and metal oxide sensors mixing ratios show overlap coefficients of 0.95 and 0.94 for the TGS2600 and TGS2611, respectively. Overall, our results showed there was a good agreement between the ML derived metal oxide sensors’ mixing ratios and those generated using the more accurate trace gas analyzer. This suggests that the response of lower-cost sensors calibrated using ML could be used to generate mixing ratios with precision and accuracy comparable to higher price trace methane analyzers. This would improve confidence in low-cost sensors’ response, reduce the cost of sensor deploying, and allow for timely and accurate tracking of methane emissions.

Methane is a powerful greenhouse gas with a 25 times higher 100-year warming potential than carbon dioxide and is a target for mitigation to achieve climate goals. To control and curb methane emissions, estimates are required from the sources and sectors which are typically generated using bottom-up methods. However, recent studies have shown that national and international bottom-up approaches can significantly underestimate emissions. In this study, we present three bottom-up approaches used to estimate methane emissions from all emission sectors in the Denver-Julesburg basin, CO, USA. Our data show emissions generated from all three methods are lower than historic measurements. A Tier 1/2 approach using IPCC emission factors estimated 2022 methane emissions of 358 Gg (0.8% of produced methane lost by the energy sector), while a Tier 3 EPA-based approach estimated emissions of 269 Gg (0.2%). Using emission factors informed by contemporary and region-specific measurement studies, emissions of 212 Gg (0.2%) were calculated. The largest difference in emissions estimates were a result of using the Mechanistic Air Emissions Simulator (MAES) for the production and transport of oil and gas in the DJ basin. The MAES accounts for changes to regulatory practice in the DJ basin, which include comprehensive requirements for compressors, pneumatics, equipment leaks, and fugitive emissions, which were implemented to reduce emissions starting in 2014. The measurement revealed that normalized gas loss is predicted to have been reduced by a factor of 20 when compared to 10-year-old normalization loss measurements and a factor of 10 less than a nearby oil and production area (Delaware basin, TX); however, we suggest that more measurements should be made to ensure that the long-tail emission distribution has been captured by the modeling. This study suggests that regulations implemented by the Colorado Department of Public Health and Environment could have reduced emissions by a factor of 20, but contemporary regional measurements should be made to ensure these bottom-up calculations are realistic.

Quantifying and controlling fugitive methane emissions from oil and gas facilities remains essential for addressing climate goals, but the costs associated with monitoring millions of production sites remain prohibitively expensive. Current thinking, supported by measurement and simple dispersion modelling, assumes single-digit parts-per-million instrumentation is required. To investigate instrument response, the inlets of three trace-methane (sub-ppm) analyzers were collocated on a facility designed to release gas of known composition at known flow rates between 0.4 and 5.2 kg CH₄ h⁻¹ from simulated oil and gas infrastructure. Methane mixing ratios were measured by each instrument at 1 Hertz resolution over nine hours. While mixing ratios reported by a cavity ring-down spectrometer (CRDS)-based instrument were on average 10.0 ppm (range 1.8 to 83 ppm), a mid-infrared laser absorption spectroscopy (MIRA)-based instrument reported short-lived mixing ratios far larger than expected (range 1.8 to 779 ppm) with a similar nine-hour average to the CRDS (10.1 ppm). We suggest the peaks detected by the MIRA are likely caused by a micrometeorological phenomenon, where vortex shedding has resulted in heterogeneous methane plumes which only the MIRA can observe. Further analysis suggests an instrument like the MIRA (an optical-cavity-based instrument with cavity size ≤10 cm³ measuring at ≥2 Hz with air flow rates in the order of ≤0.3 slpm at distances of ≤20 m from the source) but with a higher detection limit (25 ppm) could detect enough of the high-concentration events to generate representative 20 min-average methane mixing ratios. Even though development of a lower-cost, high-precision, high-accuracy instrument with a 25 ppm detection threshold remains a significant problem, this has implications for the use of instrumentation with higher detection thresholds, resulting in the reduction in cost to measure methane emissions and providing a mechanism for the widespread deployment of effective leak detection and repair programs for all oil and gas infrastructure.