Assess uncertainty of livestock inventories

Sadie Shelton

The following methods are applicable for MRV of mitigation actions and national inventories.

Analysis of uncertainty in an inventory can serve to guide decisions on choice of methodological tier and prioritize national efforts for inventory improvement (IPCC 2006 Vol 1, Ch 3). The IPCC provides general guidance on uncertainty assessment in GHG inventories (IPCC 2000, IPCC 2006 Vol 1 Ch 3). For livestock GHG sources, additional general guidance is given in IPCC (2006 Vol 4, Ch 10). The IPCC recommends that category-specific estimates of uncertainty at the 95% confidence interval are developed for inventory categories. Ideally, these uncertainty estimates are developed using category-specific data, but in the absence of such estimates default values for uncertainty are provided. For example, enteric fermentation emission factors estimated using a Tier 1 approach are assumed to have an uncertainty range of between ±30% and ±50%, while Tier 2 emission factors are assumed to have an uncertainty range of ±20% IPCC (2006 Vol 4, Ch 10). Beyond the use of IPCC default values, possible methods for uncertainty assessment include model validation, inter-model comparison, error propagation, Monte Carlo simulation and expert judgement (IPCC 2006 Vol, 1 Ch 3).

Previous analysis of developing countries’ livestock GHG inventories found that only about one third of countries reported any assessment of uncertainty in their national inventory (Wilkes et al. 2017). Of the 63 countries that use a Tier 2 approach in their livestock inventory, 7 countries did not report results of uncertainty assessment. Of the 56 countries that did, 49 reported a quantitative estimate of activity data uncertainty. Data sources used to derive activity data uncertainty estimates included reports of error ranges from statistical agencies, expert judgement and reference to values in other countries’ inventories. Six countries reported only an estimate for total uncertainty of livestock emissions (e.g. where Monte Carlo simulation had been applied), and 50 reported a specific estimate of emission factor uncertainty. Of these 50 countries, about 20 quantified emission factor uncertainty using the IPCC default values. Other methods used included error propagation, Monte Carlo analysis and expert judgement.

Text Box 7 – What difference does using a Tier 2 approach make to uncertainty in the inventory?
Comparison of uncertainty when using Tier 1 and Tier 2 approaches: 20 countries reported uncertainty of enteric fermentation or sub-categories (e.g. cattle or ‘dairy cattle’) emission estimates for earlier inventories using a Tier 1 approach and the initial Tier 2 inventory. Reported uncertainty decreased for 9 countries, remained the same for 8 countries and increased for 3 countries. In all cases, IPCC default values were used to estimate uncertainty. Whether adopting a Tier 2 approach reduces inventory uncertainty thus depends on whether data sources and methods for uncertainty estimation also change.
Trends in uncertainty of Tier 2 estimates over time: 36 countries reported uncertainty estimates for both the year of initial adoption of Tier 2 and the latest inventory submission. Reported uncertainty decreased for 20 countries, remained the same for 10 countries, and increased for 6 countries. All 6 countries reporting an increase in UNC(TOT) also changed the method used for uncertainty assessment between the two submissions assessed, replacing default uncertainty estimates with the results of error propagation or Monte Carlo analysis. The reported uncertainty values are therefore not strictly comparable.
These findings suggest that the effect of adopting Tier 2 on uncertainty of inventory estimates depends as much on improvement in methods for estimating and reporting uncertainty as it does on the benefits for uncertainty reduction of adopting a Tier 2 approach.

Analysis of livestock inventory uncertainty is provided in national inventory reports from Canada, Finland, New Zealand, and the UK. These examples show how uncertainty analysis can be used to identify the main parameters that are sources of uncertainty in a given inventory year, or in the trend in an inventory over time (Inventory practice: Analysis of uncertainty in Canada’s livestock inventory). They can also help inform decisions about the adoption of Tier 2 approaches (Inventory practice: Assessing sources of uncertainty in Finland’s livestock inventory), priorities for further research (Inventory practice: Uncertainty analysis to prioritize further research in New Zealand) and identify regional focuses within a country for reduction in uncertainty levels (Inventory practice: Assessing sources of uncertainty in the livestock inventory of the United Kingdom).