Sensitivity analysis of alternative model structures for an indicator of ammonia emissions from agriculture

Abstract

. 2007. Sensitivity analysis of alternative model structures for an indicator of ammonia emissions from agriculture. Can. J. Soil Sci. 87: 129-139. Ammonia (NH 3 ) emission from agriculture is an environmental and health concern in many nations, and has trans-border impacts. Direct toxicity, terrestrial eutrophication and production of inhalable aerosols (< 2.5 µm diameter) are the specific concerns. Canada, among other northern hemisphere nations, has computed a national inventory of NH 3 emissions, and a new emission inventory estimate is being prepared jointly under the National Agri-Environmental Health Analysis and Reporting Program (NAHARP) and National Agri-Environmental Standards Initiative (NAESI). However, there has been a rapid evolution in the models used, and a concomitant change in the NH 3 -specific data required. This paper compares several model structures and options using Monte Carlo simulation and sensitivity analysis methods. The results indicate the more recent models, that compute a mass balance of NH 3 from excretion to landspreading, have tended to focus uncertainty onto the dietary efficiency of animal N nutrition. After excretion, the total ammoniacal nitrogen (TAN) in the manure contributes to NH 3 emissions at each stage as the manure passes from the animal housing to storage and to landspreading. There are many variants of these processes because every farm is different, resulting in diminished sensitivity to any one NH 3 -loss mechanism after excretion. This finding suggests that although NH 3 -emission factors, the empirical data at the core of the models, are not well characterised (especially for Canadian conditions), it is at least as important to expend research effort on factors that influence TAN excretion. Results of this paper will also guide development of the NAHARP/NAESI model.

Monthly NH3 emissions from poultry in 12 Ecoregions of Canada

“…Previous sensitivity analysis (Sheppard et al 2007b) and examination of Eq. 1 indicated that the amount of TAN excreted is probably the single most sensitive input.…”

“…A previous analysis of two similar emission models (Sheppard et al 2007b) indicated that with nominal assumptions about the uncertainty in inputs, the estimates of total emission had coefficients of variation of 11 to 23%. This variation was minimally affected by judgmental data from farmers, such as timing of land applications and incorporation of manure.…”

Estimation of ammonia emission episodes for a national inventory using a farmer survey and probable number of field working days

“…Sensitivity of results from this model to perturbations in input data was investigated by Sheppard et al (2007) using Monte Carlo simulations. They showed that parameters related to excretion of TAN were most important, followed closely by emissions factors for landspreading.…”

“…The probable number of manure-spreading days for the remaining months was well represented with triangular PDFs. Other variable inputs were characterized as normally distributed, as discussed by Sheppard et al (2007), with truncations as appropriate to eliminate impossibilities (e.g., negative emission factors). A coefficient of variation (CV) of 10% was assigned to branching ratios and a CV of 25% to emission factors, as rationalized by Sheppard et al (2007).…”

“…Therefore, the model was run in a fully stochastic mode, with variability of each input as rationalized previously (Sheppard et al 2007). The results (Fig.…”