Historical landcover change and wood harvest CO2 emissions


  • CO2 emissions weighted by population density

Startet den Datei-Downloadcarbon_emissions_landuse.nc (24MB)

  • CO2 emissions weighted by population density cut at 20persons/km2

Startet den Datei-Downloadcarbon_emissions_landuse_20person.nc (14MB)

Thomas Raddatz, MPI for Meteorology, Hamburg, Germany, June 2010


The datasets are constructed for the use in global climate-carbon cycle simulations of the period 1850-2005. The two files are in NetCDF format, each including four variables:

  1. carbon_emission: annual maps of anthropogenic carbon emissions due to land use change (including wood harvest). The unit is g(C)m-2s-1.
  2. population: annual maps of population density. The unit is person/km2.
  3. land-sea-mask: map indicating whether a grid cell is regarded as land (value 1) or ocean (value 0).
  4. area: map indicating the area ascribed to each grid cell for the weighting of the carbon emissions with the population density. The unit is m2/grid cell.

All maps have the same regular grid with a resolution of 0.5 degree.

The regional land use change emissions of Houghton (2008) are scaled, so that the sum of the ten regions exactly equals the global emissions. Within each region the emissions are weighted with the population densities also used in Klein Goldewijk (2000), which are linearly interpolated between the years 1850, 1900, 1910, 1920, 1930, 1940, 1950, 1960, 1970, 1980, and 1990. After year 1990 population density is assumed to stay constant.

Weighting fossil fuel emissions with population density is a common method (see e.g. Andres et al. 1996). Certainly, this approach is more problematic for landuse change emissions than for fossil fuel emissions, because in urban centers most of the land use change has already been done in former times. Therefore, the population density has been reduced to a maximum of 20 persons per km2 before constructing the second dataset (carbon_emissions_landuse_20person.nc). This is the only difference between the two datasets. It should be kept in mind that in reality there are considerable land use change emissions in areas with very low population density (e.g. tropical rainforests), which are not reflected in the datasets described here.

Overall, the weighting of land use change emissions within continental scale regions by gridded population data has the advantage to exclude large emissions from some areas, where there had been indeed no substantial emissions in reality (e.g. subtropical deserts, high northern latitudes). Nevertheless, the approach is not suited to provide realistic land use change emissions on a local basis. Therefore, the datasets should only be applied in global simulation (e.g. for a comparison with atmospheric CO2 concentrations at remote marine stations) and not in regional models.


Andres R.J. et al. 1996. Geographic patterns of carbon dioxide emissions from fossil-fuel burning, hydraulic cement production and gas production on a one degree by one degree grid cell basis: 1950 to 1990. ORNL/CDIAC-97, NDP058. Carbon Dioxide Information Analysis Center, Oak Ridge National Laboratory, Oak Ridge, Tennessee, U.S.A. 56 pp. doi: 10.3334/CDIAC/ffe.ndp058


Houghton, R.A. 2008. Carbon Flux to the Atmosphere from Land-Use Changes: 1850-2005. In TRENDS: A Compendium of Data on Global Change. Carbon Dioxide Information Analysis Center, Oak Ridge National Laboratory, U.S. Department of Energy, Oak Ridge, Tenn., U.S.A.


Klein Goldewijk, K., 2000. Estimating historical land use changes over the past 300 years: The HYDE database, version 2.0. Global Biogeochemical Cycles.