Biofuels Induced Land Use Change Emissions: The Role of Implemented Land Use Emission Factors

2.1. Common Approach in Calcualting ILUC Emissions

In this formula, the index i represents the list of all types of land transitions (e.g., forest to cropland, forest to pasture, etc.), the index k shows spatial resolution (which could represent the national level, agro-ecological level, grid cell, or any other geographical resolution) within each country, and the index r indicates countries. The variables $∆L$, $LUEF$, T, and E are land conversions in hectares, land-use emission factors measured in gCO₂e per hectare, amortization time horizon in years, and annual energy produced by the pathway under study measured in megajoules (MJ), respectively. Therefore, an ILUC emissions value estimates emissions in gCO₂e/MJ.

2.2. Components and Sources of LUEFs

In calculating ILUC emissions values, one needs to determine the variable $LUE{F}_{i,k,r}$ for the i, k, and r indices, which is not a trivial task. In principle, this variable should capture all types of carbon fluxes associated with each type of land conversion. These fluxes are driven by changes in biological and mineral carbon pools, including soil organic carbon, carbon stock in above- and belowground live biomass, and dead organic matter and litter. Additionally, some carbon accounting frameworks include forgone carbon sequestration, emissions due to biomass burning through land clearing, and non-CO₂ emissions associated with the land use, land-use change, and forestry (LULUCF). Because of differences in background data and the included categories of emissions, alternative data sources provide widely varying estimates of $LUE{F}_{i,k,r}$ for the same land-use transition and location.

In addition to the required data on soil and vegetation carbon stocks, depending on the case under study, one may need additional information or use certain assumptions to mix and match $∆L_{i,k,r}$ and $LUE{F}_{i,k,r}$ variables. One may follow different approaches and assumptions to facilitate this process, which can cause significant variations in the resulting ILUC emissions values. The following three examples represent different approaches that the AEZ-EF and CCLUB models use to match the GTAP-BIO estimated land-use changes with their emission factors. Example 1: The GTAP-BIO model projects conversion of “cropland pasture” (a category of marginal land) to crop production due to biofuel shocks. In an ad hoc manner, the AEZ-EF model assumes that the soil carbon content for this type of land in each AEZ region is half of that of pasture land. On the other hand, the CCLUB model relies on a terrestrial-biogeochemical model (Century) to evaluate carbon content for this type of land by AEZ. Example 2: The AEZ-EF uses some assumptions and extends the original GTAP-BIO land conversions beyond the land conversions that this CGE model provides to match the land conversions with its emissions factors. For instance, the AEZ-EF model includes emission factors for converting forest or pasture to sugarcane. The GTAP-BIO model does not determine these land conversions. However, the AEZ-EF model uses some assumptions and determines these land conversions. The CCLUB model only uses the original GTAP-BIO land conversions. Example 3: The AEZ-EF uses some assumptions and assigns a portion of converted forest to cropland as forest on peat land, while the CCLUB uses more recent data with a different assumed portion of converted forest to cropland as forest on peat land.

Because the results of the GTAP-BIO model are used in this paper, we use two emissions accounting models that have been developed and used to convert the results of this model to ILUC emissions values. These two models are the AEZ-EF and CCLUB. The AEZ-EF model relies on IPCC, FAO, HWSD, and several other data sources to convert the GTAP-BIO results to ILUC emissions. This model follows the IPCC approach of using the differences in the biomass and soil organic carbon (SOC) pools between land-cover types as the emissions (or sequestration) values from land conversion.

In contrast, CCLUB provides users with Century simulated GHG emissions changes in US domestic land conversions to cropland and the option of using either the Winrock or Woods Hole data sources for international land conversions to simulate biomass and SOC changes between land-use categories over a period of time. As mentioned earlier, using the Century model, the CCLUB model also provides some assessments for the emission factors associated with the land category of “cropland pasture”. In conclusion, the AEZ-EF and CCLUB models use different sources of data on carbon pools and follow different assumptions to convert the results of the GTAP-BIO model to ILUC emissions, especially for US domestic land conversions.

To highlight uncertainties and variations in the data on $LUE{F}_{i,k,r}$, we first review four existing sets of emission factors for converting forest to cropland and pasture to cropland: AEZ-EF, TEM, Winrock, and Woods Hole. These datasets have been used in calculating ILUC values for various US biofuel pathways over the past 15 years, but are limited by their reliance on outdated data in assessing emission factors. For example, the AEZ-EF model uses the 2006 IPCC guidelines for national greenhouse gas inventories instead of the new guidelines published in 2019. To highlight the potential impacts of using outdated data, we compare changes in the reference values for SOC stocks obtained from the IPCC 2006 and 2019 guidelines.

2.3. A Short Review of GTAP-BIO Model and Implemented $∆L$ for the Examined SAF Pathways

In addition, the GTAP-BIO model represents land uses by the agricultural and forestry sectors and traces their changes due to changes in demands for foods and biofuels. The agricultural sectors in this model include crop producers (rice, wheat, coarse grains, soybeans, rapeseed, palm oil, other oilseeds, sugar crops, and other crops) and livestock producers (dairy farms, ruminants, and non-ruminants). The GTAP-BIO model divides the accessible land across three land-cover categories: forest, pasture/grassland, and cropland. It then allocates pasture land across livestock activities and cropland across crop producers. The model takes into account multiple cropping (producing more than one crop per year on the same cropland), allows the return of unused cropland to crop production if needed, and takes into account yield improvement due to higher crop profitability.