Emissions: CO2, methane leaks, NOx and air permits

At the smokestack, the comparison is not close. EIA reports that US utility-scale coal plants emitted 2.31 pounds of CO2 per kWh in 2023 and gas plants emitted 0.96. Gas puts out less than half the CO2 for the same electricity. The fight over gas and climate is almost entirely about what happens before the gas reaches the plant: how much methane escapes on the way.

Why gas burns cleaner

The reasons stack. Methane has four hydrogen atoms per carbon, so more of its energy comes from making water and less from making CO2. EIA's emissions coefficients put natural gas at 116.65 pounds of CO2 per million Btu, against 205.57 for bituminous coal and 216.64 for lignite. Then gas plants are more efficient: a combined-cycle plant needs about 7,500 Btu per kWh, while the average coal plant needs about 10,000 (lesson 3).

Gas also emits almost no sulfur dioxide, mercury or soot compared with coal. EIA notes that burning gas produces fewer air pollutants than coal or oil for the same energy. That part of the case for gas is not seriously disputed.

Methane: the part that is disputed

Methane is a far stronger greenhouse gas than CO2 while it lasts. EPA gives methane a global warming potential of 27 to 30 over 100 years and 81 to 83 over 20 years, and says it lasts about a decade on average in the air. Fossil methane carries the higher end of those ranges because it also becomes CO2 when it breaks down.

Methane escapes all along the chain: from wells and their equipment, from storage tanks, from processing plants, from compressor stations, from pipelines and from the meters and pipes under city streets. Some is intentional venting. Some is incomplete flaring. Some is leaks, and a handful of malfunctioning "super-emitter" sites can account for a large share of the total.

What the studies found

Study	Method	Finding
Howarth, Santoro and Ingraffea, Climatic Change, 2011	Estimates from limited early data	3.6% to 7.9% of shale gas methane lost over a well's life; shale gas footprint at least 20% worse than coal over 20 years, comparable over 100
Alvarez et al., Science, 2018 (synthesis of the EDF-coordinated studies)	Site-level ground and aircraft measurements across basins	About 13 million metric tons of methane a year from the US oil and gas supply chain, about 60% above EPA's inventory; commonly reported as about 2.3% of gross gas production
Sherwin et al., Nature, 2024	Nearly one million aerial site measurements, six US regions	Weighted average 2.95% of production; range 0.75% in a gas-rich, high-productivity region to 9.63% in a fast-growing, oil-focused region; roughly three times the government inventory

Measured leak rates have repeatedly come in above government inventories, which are built mostly from equipment counts and emission factors rather than direct measurement. The range across basins is huge. Gas from an oil-focused basin where operators flare or vent unwanted gas can carry a much bigger methane burden than gas from a dry-gas basin with modern equipment. The 2011 Howarth estimate was criticized at the time by other researchers for relying on sparse data and high assumptions; later measurement-based studies mostly landed lower, but the newest aerial work landed higher than the 2018 synthesis. The number is a moving target, and anyone quoting a single national leak rate should say which study and which years.

What a leak rate does to the comparison

Here is a simplified version of the math researchers do. It uses only the numbers in this lesson.

So where does the evidence point? At the leak rates most measurement studies report for the US average (roughly 2% to 3%), gas power still beats coal on both 20- and 100-year horizons, though the 20-year advantage shrinks a lot. At the high end of the basin-level range, gas from the leakiest fields can be worse than coal over 20 years. Ramon Alvarez's group argued as early as 2012, in PNAS, that new gas plants give immediate climate benefits over new coal plants only if total leakage stays below about 3.2%, a stricter line than this lesson's simplified 4.8% because their method weighs warming at every point in time. Most of the policy debate now is over how low leakage can go and who makes sure it does.

Federal methane rules are in flux. EPA's 2024 standards for new and existing oil and gas sources (known as OOOOb and OOOOc) had their compliance deadlines extended in 2025, and EPA finalized revisions loosening some flare and vent requirements in April 2026, now under court challenge. Congress repealed the rule implementing the methane Waste Emissions Charge in February 2025 and later barred EPA from collecting the charge until 2034, per Harvard's Environmental and Energy Law Program tracker. Check the current status before you cite it.

NOx and local air quality

Burning anything in air at high temperature makes nitrogen oxides. NOx contributes to ground-level ozone and fine particle pollution, which matter for asthma and heart and lung disease. Gas turbines control NOx two main ways:

• Dry low-NOx combustors mix fuel and air carefully before burning to keep flame temperatures down.
• Selective catalytic reduction (SCR) injects ammonia or urea into the exhaust ahead of a catalyst that turns NOx into nitrogen and water. Most new combined-cycle plants use SCR. It brings its own small issue, "ammonia slip," which permits limit.

Simple-cycle peakers and engines often have higher NOx per MWh than combined-cycle plants, and they run during hot afternoons, which is also when ozone forms. That is why peakers near dense neighborhoods have drawn environmental justice complaints in places like New York City, where older peakers were targeted for replacement.

Air permits for a new gas plant

A large new gas plant in an area meeting federal air standards needs a Prevention of Significant Deterioration (PSD) permit under the Clean Air Act, usually issued by the state air agency. EPA describes the requirements: Best Available Control Technology set case by case, an air quality analysis showing the plant won't cause violations of national standards, an additional impacts analysis (soils, vegetation, visibility) and public involvement. In areas that don't meet standards, a stricter "nonattainment" permit requires the lowest achievable emission rate and emissions offsets. Smaller plants may get a minor-source permit with less review. On-site engines at data centers often try to stay under major-source thresholds by limiting hours, which is worth checking.