Mapping Global Carbon
NEW ESTIMATES OF CARBON STOCKS FOR ECONOMIC MODELS
We synthesized a range of geographically-explicit forest, grassland and cropland biomass and soil carbon input data sources and used geographic information systems (GIS) software to calculate new estimates of soil and biomass carbon stocks for use with global economic models. Our results quantify the average amount of carbon stored in soil and biomass in each of the 246 countries, stratified by agro-ecological zones. We also provide the data aggregated to the 134 regions defined for the GTAP 8.1 database both in spreadsheet form and in GTAP’s native binary file format. Finally, we provide an add-on to FlexAgg2 program to further aggregate the 134 regions as desired. Our analysis makes substantial refinements to the estimates of carbon stocks used for modeling carbon emissions from indirect land use change. The spatial detail of our analysis is a major advantage over previous databases because it provides estimates tailored to the regions of interest and better accounts for the variation of carbon stocks across the landscape, and between wetland and non-wetland regions.
In this research we asked the following question:
Can we improve upon existing global carbon stock datasets by creating a spatially explicit dataset for forest, pasture, and grassland biomass and soil carbon?
My role was doing the literature review and do all the spatial data collection and analysis. I used ArcGIS software to estimate the soil and biomass carbon stocks for forest, grazing land and cropland by overlaying national and AEZ boundaries on a range of geographically-explicit data sources to produce a database with values for 246 nations, stratified by 18 AEZs. We aggregated the national data to match the 134 regions defined in the GTAP database by totaling the carbon in each combined region-AEZ combination and dividing these values by total area representing each carbon pool. The national-level data were then further aggregate to the 19 regions used in the GTAP model to determine the final carbon estimates. We created 203 regions4 by combining the two maps, but because the resolution is coarse, there are several extremely small regions that could be integrated into nearby regions in the future.
We compared our updated values with the WHRC values used in Hertel et al. (2010) and Tyner et al. (2010) (Figure 16). Our values were higher around the Amazon basin, humid tropical Africa, and insular Southeast Asia. WHRC values were substantially higher (50+ Mg C / ha) in the US, Canada, Europe, and Russia. (Table A2).
Lastly, we compared our sources of forest biomass estimates with those used by Harris et al. (2009) for the US EPA in Table 6. Overall, our approach relied on more recently published data sources, particularly for the tropics, and improved upon the framework established by Harris in some instances. A comparison between the Harris values produced at the state level and our values estimated at the GTAP-Region-AEZ level is difficult because different scales are involved. However, general patterns can be 21observed. For example, our values are higher across most of the tropics but lower in Brazil. Our values were lower in Australia, Europe and Asia but mixed in the United States. Winrock used the HWSD to estimate soil carbon stocks as we did. Note that the Winrock estimates for soil carbon are for forests only, while we provided estimates for forest, cropland and pasture separately.
While we have provided the best available estimates for forest biomass carbon stocks available, it is important to note that uncertainty remains. The estimates we provide may under- or overestimate the values on the ground because of spatial variability. The science of mapping forest carbon stocks has improved considerably, but more attention has been focused on estimating changes in forest areas rather than their carbon stocks. In addition, our approach used a weighted average of forest carbon stocks within a region, and the actual value of any given forest may be higher or lower than the average. The HWSD soil database does not account for land use history in many cases, and this can have a great impact on soil carbon content.