Developing revised emission factors for nitrous oxide emissions from agricultural pasture treated with nitrification inhibitors
Nitrous oxide emissions inventory for agricultural soils
For agricultural soils, the nitrous oxide (N2O) emissions inventory begins by determining a nitrogen (N) application rate. An emissions factor is then applied to account for the fraction of applied N that is emitted into the atmosphere. This is known as the direct emissions component of the inventory. Indirect emissions account for N2O that comes from the fraction of applied N that leaches through the soil, FracLEACH, with the New Zealand specific value equal to 0.07. The sum of direct and indirect N2O emissions yields total N2O emissions. Direct emissions comprise about 70 % of total N2O emissions.
To illustrate the computation of direct N2O emissions (FN2O) in a given year, we write a simplified equation comprised of average quantities
FN2O = {[an d (1/c) pn xN] + f} EF
where an is the number of grazing, farmed animals, d is the animal’s energy requirement (MJ per animal per year, MJ is one million joules), c is feed (hereafter, pasture) energy content (MJ per kg dry matter), pn is pasture nitrogen (N) content, xN is the fraction of N intake (N intake is equal to the product of an, d, (1/c) and pN) that is excreted as urine and dung onto soils, f is the mass quantity of N fertiliser applied to soils (according to sales records compiled annually by FertResearch, Hilton Furness, personal communication) and EF is an emissions factor (mass of nitrous oxide emitted per mass of N deposited on the soil). To use the equation, the units of FN2O must be converted from mass of N per year to mass of N2O per year by multiplication by 1.57 (ratio of the molecular weights = 44/28). The equation shows that the N loading rate onto soils includes urine and dung excreta deposited during grazing and fertiliser application.
The primary data for an comes from a survey sent by the Ministry of Agriculture and Forestry to around 40,000 farms annually that yields close to a 90 % response. During the year, an depends on a monthly population model developed by H. Clark to account for births, deaths and slaughter. Variable d is also determined monthly by the Australian feeding standards for grazing ruminants (CSIRO 1990), including industry-supplied animal weight and production data (e.g. milk production, fecundity rates, weights of animals at slaughter, etc.), according to Clark et al. (2003). Weight data are used to account for the maintenance component of variable d. Values for c vary monthly ranging from 9.6 – 12.6 MJ kg-1, however the average value over each year is assumed constant. For this report, pN was 3.7 % for dairy cattle and 3.0 % for beef cattle, sheep and deer. Also for this report, we assumed that DCD application did not change the variables an, d, c and pN. Scenarios according to changes in these variables following DCD application were reported earlier by Clough et al. (2006, 2007). For this report, values of xN for urine and dung were computed monthly according to the difference between N intake and N that went into product (such as milk and wool) and weight gain.
The direct EF for excreta is called EF3(PRP). The EF3(PRP) data are cumulative values of direct N2O emissions over 5 to 10 months following an excreta application (fraction of applied nitrogen emitted to the atmosphere as nitrous oxide) based on field chamber measurements of the NzOnet field trials (Barton et al., 2000; de Klein et al., 2003, 2004; Sherlock et al., 2003a,b). The data are analysed to compute a statistic known as the geometric average, a robust measure of the central tendency. As an example, for four hypothetical EF3(PRP) measurements, a geometric average of 0.008 may be calculated from the quantity [0.001*0.011*0.012*0.030] raised to a power of (1/4) where the four measurements are multiplied together in the square brackets and the power coefficient is equal to the inverse of the number of measurements. The arithmetic average is 0.014 or 75 % larger than the geometric average because the large value, 0.030, skewed this central tendency’s estimate from the small sample. The analysis of EF3(PRP) data involves small samples that can exhibit tremendous variance. This reflects the conditions producing nitrous oxide in soils, mostly attributable to high nitrogen and water contents that are generally shortlived. As illustrated above, unlike the arithmetic average, the geometric average is not prone to undue contamination by (low probability) outliers that sometimes occur.
De Klein (2006) reviewed 40 different studies, including those conducted in New Zealand, and recommended to the IPCC that grazing animal excreta inputs to N2O emissions inventories should be disaggregated into cattle and sheep excreta. For EF3(PRP), de Klein (2006) recommended values of 0.02 for cattle excreta and 0.01 for sheep excreta. To our knowledge, nevertheless, the IPCC default value for EF3(PRP) remains equal to 0.02. The New Zealand specific value for EF3(PRP) is equal to 0.01. For dairy cattle urine, 17 trials yielded a geometric average of 0.009 = 0.01 (Table A.1 in the Appendix). These data support the New Zealand specific value of EF3(PRP).
The direct EF for N fertiliser is called EF1. For urea, applied over one year in eight equal dressings of 50 kg N/ha (during March, April, June, July, September, October, November and December 1990) to imperfectly-drained Manawatu silt loam soil beneath pasture at Palmerston North, a 365 day long study yielded EF1 = 0.0130 (Ruz-Jerez et al. 1990). The NzOnet field trials included two measurements of EF1 at the Hamilton site (de Klein et al., 2004). As urea, on 23 August 2003, 50 kg N/ha was applied to adjacent freely-drained and poorly-drained soils (Horotiu and Te Kowhai, respectively) beneath pasture. The 146 day long study yielded EF1 values of 0.0200 for the Horotiu soil and 0.0270 for the Te Kohai soil, so the geometric average EF1 is 0.0232. Seasonal measurements of EF1 were recently reported by Luo et al. (2007). They also applied 50 kg N/ha as urea to Te Kowhai soil beneath pasture at a site located close to that used for the NzOnet trials. Application dates, number of days when direct emissions from the urea treated areas were greater than the controls, and EF1 values were 9 June 2003, 21 days and 0.0052, 20 August 2003, 20 days and 0.0127, 13 November 2003, 14 days and 0.004, 8 April 2004, 7 days and 0.0003, 1 July 2004, 30 days and 0.0156, 24 November 2004, 5 days and 0.001, 19 February 2005, zero days and 0.0000 and 12 July 2005, 23 days and 0.0059. The geometric average cannot be computed when there is a zero in the set of data. For Luo et al. (2007), excluding data from the 19 February 2005 application, the geometric average value of EF1 is 0.0036.
The New Zealand specific value for EF1 is equal to 0.01. The three trials, discussed above, yielded (geometric) average values of EF1 equal to 0.013, 0.0232 and 0.0036. The geometric average of these three values is 0.0103. Consequently, on average, these data support the New Zealand specific value for EF1.
Laegreid and Aasveit (2002) reviewed international data comprised of 880 measurements and concluded EF1 averaged 0.008. Independently, and based on their international data set of 846 measurements, Bouwman et al. (2002) concluded EF1 averaged 0.009. Based on an updated and most recent international data set of 1008 measurements for agricultural soils, Stehfest and Bouwman (2006) also concluded EF1 averaged 0.009. As a result of Stehfest and Bouwman (2006), the IPCC decided to recommend a new default value for EF1 = 0.01 (de Klein, 2006). We believe these international data support New Zealand’s country specific EF1 value of 0.01.
Contact for Enquiries
Sustainable Land Management and Climate Change
MAF
Pastoral House
25 The Terrace
PO Box 2526, Wellington
Tel: 0800 CLIMATE (254 628)
Contact this person
