Evaluation of the SF6 tracer technique for estimating methane emission rates with reference to dairy cows using a mechanistic model.

A dynamic, mechanistic model of the sulfur hexafluoride (SF6) tracer technique, used for estimating methane (CH4) emission rates from ruminants, was constructed to evaluate the accuracy of the technique. The model consists of six state variables and six zero-pools representing the quantities of SF6 and CH4 in rumen and hindgut fluid, in rumen and hindgut headspace, and in blood and collection canister. The model simulates flows of CH4 and SF6 through the body, subsequent eructation and exhalation and accumulation in a collection canister. The model predicts CH4 emission by multiplying the SF6 release rate of a permeation device in the rumen by the ratio of CH4:SF6 in collected air. This prediction is compared with the actual CH4 production rate, assumed to be continuous and used as a driving variable in the model. A sensitivity analysis was conducted to evaluate the effect of changes in several parameters. The predicted CH4 emission appeared sensitive to parameters affected by the difference in CH4:SF6 ratio in exhaled and eructed air respectively, viz., hindgut fractional passage rate and hindgut CH4 production. This is caused by the difference in solubility of CH4 and SF6 and by hindgut CH4 production. In addition, the predicted CH4 emission rate appeared sensitive to factors that affect proportions of exhaled and eructed air sampled, i.e., eructation time fraction, exhalation time fraction, and distance from sampling point to mouth/nostrils. Changes in rumen fractional passage rate, eructation rate, SF6 release rate, background values and air sampling rate did not noticeably affect the predicted CH4 emission. Simulations with (13)CH4 as an alternative tracer show that the differences and sensitivity to parameters greatly disappear. The model is considered a useful tool to evaluate critical points in the SF6 technique. Data from in vivo experiments are needed to further evaluate model simulations.