Assessing Uncertainty and Bias in the U.S. Land Use, Land Use Change, and Forestry Greenhouse Gas Inventory

Every year, U.S. land use, land use change, and forestry (LULUCF) removes a net 600-800 million metric tons CO₂ from the atmosphere, significantly reducing economy-wide greenhouse gas (GHG) emissions. Yet LULUCF contributes over 70% of the total uncertainty in U.S. annual economy-wide GHG emissions estimates. This paper seeks to support ongoing improvements to the U.S. LULUCF GHG Inventory (Inventory) by (1) quantifying uncertainty from all equations, datasets, and major omitted GHG sources using statistical attribution methods and literature review, and (2) using this quantitative analysis to prioritize recommendations for research and data improvements.

This analysis covers the entire LULUCF chapter of the Inventory as well as soil management sections of the Agriculture chapter. We estimate omitted GHG fluxes as well as attribute sources of uncertainty, evaluating over 100 uncertainty elements and fluxes. For each omitted GHG flux, literature review is used to identify activity data and emission factors to generate first-order flux estimates. For uncertainty estimates, Monte Carlo and error propagation are used to perform contribution index analysis across key data inputs and parameters.

The largest source of uncertainty by far (32% of total uncertainty attribution) is from extrapolating forest carbon plot measurements to total U.S. forest area, or sampling error. Other large sources of uncertainty include the "gross to net sequestration ratio" parameter for urban tree carbon flux estimates, parameters for converting tree diameter to tree volume and tree biomass, and inputs to the DayCent soil carbon model for croplands and grasslands. Sources of uncertainty are relatively well-distributed across land types. The largest potential omission is carbon fluxes from non-forest landscapes in Alaska, followed by CO₂ emissions from urban mineral soils, and N₂O emissions from federal cropland and grassland.

Policy needs include additional field measurements in existing datasets, incorporating new data into existing or new datasets, improving models and methods, and new primary research to address scientific uncertainties. We further highlight the long term need for a more dedicated, comprehensive, and publicly-accessible system for estimating LULUCF GHG fluxes across all U.S. land types.