estimating [air] emissions at the individual farm level that reflects the different animal production methods that are commonly used at commercial scale operations.” The approach is intended to provide estimates of total annual air emissions from animal feeding operations (AFOs) for defined geographic areas by kind of animal and manure handling practices for each of eight kinds of emissions. It does this with a model farm construct that provides estimates of average annual emissions per animal unit (AU) for twenty-three model farms (two for beef, eight for dairy, two for poultry-broilers, two for poultry-layers, two for poultry-turkeys, five for swine, and two for veal; Appendix D). Each model is defined by three variable elements that describe manure management practices for typical large AFOs: (1) confinement and manure collection system, (2) manure management system, and (3) land application. The manure management system was further subdivided into solids separation and manure storage activities. Insofar as combinations of these elements are regionally distinctive, the model farms also reflect regional variations in air emissions.
Model farms, as used by EPA (2001a), are a useful device for aggregating emission rates across diverse sets of AFOs. A model farm can be used to represent the average emissions across some geographic area over some period of time per unit capacity of a class of farms (e.g., all pig farms in the United States that use an enclosed house with pit recharge and irrigation of supernatant onto forage land; model farm S2 in Appendix D).
The applicability and use of a model farm construct of the kind used by EPA (2001a) depends on:
defining models in which the dependent variable, the amount of an air emission per unit of time, is closely related to independent variables that accurately depict real feeding operations, and that can explain a substantial share of the variation in the dependent variable;
providing accurate estimates of the relationship between the dependent and independent variables in the model farm construct; and
having estimates of the relationship between dependent and independent variables that clearly distinguish among the kinds of AFOs being modeled.
A critical data requirement for estimating the appropriate emission factors is a statistically representative survey of emissions from the class of AFOs over several iterations of the time period to be represented. The size of the sample required to estimate the mean emission rate with a given degree of statistical significance increases with the variability of the factor to be measured (dependent variable) across the set of variables (independent variables) that affect it. Independent variables that have been discussed include animal type and age, diet, local climate, building type, land application method, and management. To the extent that some of these variables change over time (e.g.,