Initialization of agricultural land
The discrepancies between agricultural statistics used from the Food and Agriculture Organization of the United Nations and geographic patterns for the distribution of forests, crops and agricultural production have made it impossible to consistently simulate land-cover shifts under climate change. Therefore a new global land-cover database was constructed to provide a realistic initialization for 1970 for simulations of global land-cover change with IMAGE 2.4. Important criteria used in the land cover initialization are:
- the geographical patterns of agricultural land should be consistent with the 1970 national land-use statistics from the Food and Agriculture Organization of the United Nations
- the database should allow assessment of vegetation response to climate change
- distributions of individual crops and potential natural vegetation should resemble the geographical patterns simulated in the terrestrial vegetation model.
Meeting these criteria guarantees a smooth transition between observed and simulated future land-cover patterns in IMAGE 2.4 runs. Computing initial patterns of agricultural land for the land-cover database consists of several steps. Agricultural land is defined here as land where food, grass and fodder or biofuel crops are cultivated.
- The Digitized Soil Map of the World is converted to a 5 x 5 minutes grid classified by the dominant soil type.
- The resulting 5 x 5 minutes soil map is aggregated to a 0.5 by 0.5 degree grid, each cell having its own frequency for different soil types.
- Each 0.5 by 0.5 degree cell containing at least one soil type, which is not water, is included in the land-cover database. Other cells are excluded. Soil types are used in the terrestrial vegetation model (TVM) to compute agricultural (potential) productivity (see also step 8).
- Cells for Greenland and Spitsbergen are added from the Zobler soil data.
- Cells are prioritized to become classified as agricultural land in the initialization process. This is done by ranking cells within a country according to their likelihood to be related to agriculture. For this ranking we use the following data sources.
- The DIScover database, adopted by the International Geosphere-Biosphere Programme Data and Information System office (IGBP-DIS) to fill its requirement for a global 1-km land-cover data set. The original database can be found at the USGS NASA Distributed Active Archive Center. For each 0.5 by 0.5 degree cell in IMAGE 2, fractions of all the original Global Land Cover Legend items are calculated on the basis of the number of original 1 by 1 km cells belonging to a certain Land Cover Legend item in a 0.5 by 0.5 degree cell in IMAGE. The different fractions in the 0.5 by 0.5 degree cells are weighted for determining priority in becoming agricultural land (the highest weight is assigned to the fractions 'croplands' and 'urban and built-up', a lower weight is given to 'cropland/natural vegetation mosaic', the lowest weight to 'savannas' and 'grasslands' and no weighting is applied to the remaining cells).
- The map of irrigated areas. The fraction of irrigated area in a cell determines the priority to become agricultural land.
- One by one, grid cells within a country are allocated to agricultural land until the total agricultural area within a country agrees with the estimate reported in FAOSTAT. Cells that are classified as protected bioreserves are excluded from the allocation algorithm.
- The area within particular agricultural cells is assigned to specific crops. Mixed cropping systems in single cells are allowed. The assignment of specific crops is based on:
- FAOSTAT estimates for the national coverage of harvested areas of specific crops and pastures
- the map of irrigated areas
- the potential productivity of crops simulated by the Terrestrial Vegetation model
- information on animal densities.
- The cells classified as agricultural cells are divided into two land-cover categories:
- agricultural land, defined as cells where potential productivity is more than 25% of the theoretical maximum productivity, i.e. the productivity under (crop-specific) optimal climate and soil conditions; and
- extensive grassland, defined as cells where potential productivity is less than 25% of the theoretical maximum productivity.
- Finally, all remaining cells are assigned the potential vegetation, as simulated by the terrestrial vegetation model.
The above procedure has led to IMAGE 2.4 's grid-specific global distribution of agricultural land for 1970, which allows it to be used for initializing global land-cover simulations (as dealt with in the Land Cover model); it is also consistent with Food and Agriculture Organization of the United Nations data for 1970. The resulting initial data set consists of the gridded crop and pasture distribution and crop yield, with explicit entries for each crop.