Land Allocation

Initial land cover, description

Current distribution of grassland and arable land

The current 5 by 5 minute-resolution global cropland and grassland maps were developed by Klein Goldewijk et al. (2006). These maps are based on satellite data and national land-use statistics (FAO, 2006)and sub-national land-use data for U.S.A. (USDA, 2006) and China (China National Bureau of Statistics, 2006a,b). Klein Goldewijk et al. (2006) used two satellite maps: 

1. the International Geosphere-Biosphere Programme (IGBP) map based on DISCover version 2 data using the IGBP classification map (Loveland et al., 2000)
2. the Global Land Cover (GLC) based on the Global Land Cover 2000 VEGA2000 data (Bartholome et al., 2002).                  

Statistics versus satellites

Using the classes in the IGBP map with dominant cropland and grassland did not yield a satisfactory match with FAOSTAT land-use data (FAO, 2006). Therefore, the underlying DISCover data from the seasonal land-cover regions were used to determine fractions of cropland and grassland within the various classes, with mosaics of natural vegetation and cropland and grassland from the IGBP map. No such data were available for the GLC map, so a trial and error approach was used. While neither of the two base maps produced in this manner completely matched the FAO country data, a combination of the IGBP and GLC maps resulted in a satisfactory match with FAO data for all countries. Apart from noise in the data, interpretation problems and uncertainties in the ancillary data used in interpretation, a major problem in allocating grassland is the broad definition used for grassland by FAOSTAT, allowing for important differences between countries in the type of grassland included in the statistics. A large area (~40%) of global grassland occurs in semi-deserts, deserts and sub-polar tundra, land regions with unfavourable climates with low productivity and low carrying capacity (Klein Goldewijk et al., 2006). Cropland occurs under more favourable climates. 

Methodology

It should be noted that for applying the base map in IMAGE land-cover simulations, the fractional 5 by 5 minute land-cover map was aggregated to larger 0.5 by 0.5 degree grid cells. This was done in three steps, reserving the fractions of grid cells covered by built-up (urban and infrastructure) and protected areas: 

1. step 1 involved a simple aggregation to the 0.5 by 0.5 degree cells. This aggregation implies a loss of spatial detail in regions where agricultural land occurs in mosaics with natural vegetation; 
2. in step 2, cropland and grassland were grouped into one class of agricultural land. The IMAGE model has no fractional land use, i.e. a grid cell is either covered by agricultural land or natural vegetation; 
3. therefore, in step 3, agricultural land was allocated (concentrated) to 0.5 by 0.5 degree grid cells with 100% agricultural coverage.          

This allocation was done by ranking, in which grid cells with the largest fraction of agricultural land after step 2 were classified as agricultural land first, until a complete match with FAO land-use data was achieved. Hence, although the general land-use patterns are the same in the 5 by 5 minute maps of Klein Goldewijk et al. (2006) and in IMAGE, the details on the spatial distribution of grassland and cropland are partly lost in steps 2 and 3.

• to  Urban area
• to  external references
• to  MNP references