The methods for quantifying uncertainty in the GHG fluxes reported in COMET-Farm come from the USDA report Quantifying Greenhouse Fluxes in Agriculture and Forestry: Methods for Entity-Scale Inventory. Uncertainty quantification in entity-scale GHG prediction is the formal process of describing the likelihood of different possible emissions, given what is known and what is unknown at the entity scale. Uncertainty in GHG emissions is quantified using a probability density function (PDF). PDFs for the factors are propagated through the method to determine the final PDF of the GHG emission for the entity. The approach that is used is a Monte Carlo analysis. The values reported in COMET-Farm are the PDFs' median values after the uncertainty propagation. Source category totals, field totals, project totals, and the change in emissions/removals from the selected scenarios to the baseline are aggregated from the PDFs after uncertainty propagation. Because the PDFs have different distributions and the values reported are the 50th percentile (median), totals of sums and differences may not match the sum or difference of the propagated medians. This means that if you simply summed or took the difference of the source categories reported on the reports page, they will not match the totals presented in the report due to the aggregation of the PDFs, Median(X+Y)≠Median(X)+Median(Y).
For example, in the Report below, while arithmetic would show a difference of -20.6 between the baseline and scenario (27.9-7.3=20.6), the "Change" column shows a difference of -21.8. This is because the "Change" column is taking into account the aggregation of PDFs, so this is an accurate representation of the difference between the baseline and scenario.
See the figure below for a visual representation of the histograms displaying the medians and confidence intervals from the field totals in the Croplands Demo Project.
