2. Uncertainties

Agricultural emissions are much more difficult to calculate than those in other sectors and far less certain. Governments and the private sector keep precise data about the amount of coal, oil, and gas used, which can be used to accurately determine the amount of CO₂ entering the atmosphere.

All emission calculations involve some uncertainty due to challenges with collecting accurate and representative data, selecting appropriate model parameters, and simplifying complex natural processes into a series of equations. Experts calculating emissions can determine how model results vary according to a range of likely inputs, and thus can establish what is known as a “95% confidence interval”—the range of values surrounding the estimate for which there is a 95% likelihood that the true value lies between.

Many emission sources within the energy and industrial sectors are associated with precise mean or central estimates. For example, the 95% confidence interval for total CO₂ emissions from energy-related fossil fuel combustion narrows this estimate to within 2%-4% of the mean estimate.⁶¹ In contrast, estimates for CH₄ and N₂O emissions from agriculture come with broader uncertainties across the board,⁶² and several of the largest agricultural greenhouse

55

gas emission sources, including soil N₂O emissions⁶³ and enteric fermentation,⁶⁴ have extremely wide confidence intervals. The confidence interval for the agricultural sector is between 451 and 847 MMT CO₂ eq., or from 7% to 13% of total U.S. greenhouse gas emissions, around the mean estimate of 10% of emissions. This broad range of uncertainty between the upper and lower bounds for U.S.

These wide uncertainties are partly attributed to fundamental differences in estimating agricultural emissions compared to other sectors. For example, to determine enteric emissions of methane from cattle, EPA uses U.S. Department of Agriculture (USDA) data on the age, weight, and location of different varieties of animals. Emissions from each subpopulation are then modeled based on parameters reflecting diet characteristics in the region and the CH₄ conversion rate, or fraction of calories converted to CH₄. A similar but coarser approach is used for non-cattle livestock. In addition to uncertainties associated with the demographic data on animal subpopulations, the cattle diet estimates are relatively speculative. EPA uses similarly complex models for manure emissions that incorporate the production rates of solid waste, CH₄ conversion factors, and N₂O emission factors, among other estimates, resulting in a 95% confidence interval from 18% below to 24% above the given figures.⁶⁶

A recent paper suggests an additional substantial underestimate of modeled emissions from CAFOs.⁶⁷ The authors compared atmospheric measurements taken above and downwind of animal production regions to standard EPA and other models and found that the measurements showed animal CH₄ emissions 39%–90% higher than model estimates of animal CH₄. They note that “bottom up” models based on data on animal inventory and characteristics underpredict enteric CH₄ emissions for multiple animal species, potentially in part due to the prevalence of diseased animals with higher rates of enteric emissions than predicted from models with healthy herds. Additionally, they note that manure emission estimates from these bottom-up models, which use parameters based on laboratory experiments within controlled test chambers, “appear to routinely underpredict emissions from manure….

56

is measured outside of the lab, in the air directly above manure tanks, pits, and piles, emissions tend to be greater than models predict, sometimes by more than 300%.” These findings suggest even greater attention must be paid to ways to reduce CH₄ emissions from CAFOs.

The greatest uncertainties in EPA’s greenhouse gas inventory are attributed to estimating N₂O emissions from agricultural soils. The calculation must include each of the five different ways N₂O is released, including (1) emissions from the application of synthetic fertilizers and other inputs; (2) emissions following the breakdown of organic matter; (3) emissions following soil drainage; (4) emissions following livestock manure deposits; and (5) indirect emissions following leaching or volatilization. Even with reasonably good data on nitrogen application activities, there are many uncertainties since the model must use intricate biogeochemical interactions in soil that vary with the weather, inputs, and other environmental conditions. As a result of this complexity, EPA indicates that the true N₂O emissions from direct and indirect sources could be between 37% below to 50% above the given figure,⁶⁸ which encompasses a range of 292 MMT CO₂ eq., itself an amount equal to almost half of the given figure for total U.S. agricultural greenhouse gas emissions.

As a result of all these uncertainties, demonstrated in Figure 11, EPA’s estimate for agricultural greenhouse gas emissions must be understood as simply one point in a wide range of possible figures. These uncertainties also point to a major challenge in developing policies to mitigate agricultural emissions and promote sequestration. When regulating emissions from other sectors, the government can identify emissions trends with minimal uncertainty, closely monitor emissions sources, and even compensate for emission reductions with precision. In contrast, agricultural emissions are diffuse. Monitoring and measuring emissions is often difficult or impossible, and model calculations are relatively uncertain. These factors make it challenging to disentangle trends or detect the impact of specific policies on total emissions relative to wide uncertainties. Fortunately, there is ample evidence that many climate-friendly practices do significantly reduce emissions or increase sequestration, and policymakers can craft programs that address the unavoidable uncertainty.

57

58