HarvestChoice Spatial Production Allocation Model (SPAM) 2000

Authors
Ulrike Wood-Sichra
Publisher
International Food Policy Research Institute

SPAM2000 (any version) is based on a large set of data which centers as much as possible on the year 2000: land cover / land use (Boston University’s MODIS-derived land cover 2000 and JRC’s GLC2000), crop suitability (Global AEZ Zones 2000 by FAO and IIASA), irrigated areas (FAO and CESR of University of Kassel), population density (CIESIN), and, most importantly, national and sub-national crops statistics for 2000.

In order to generate SPAM results for a more recent year, all the above data sets need to center on that particular year. IFPRI and Harvest Choice have started the process to acquire, adapt and include more recent data into SPAM, and later this year (2012) new results should be available.

In the meantime, a "shortcut" has been devised to produce crop production allocation data at pixel level for the year 2005 by scaling the current 2000 results for area harvested and production (and therefore yield) to FAO country totals reported for 2005 (FAOSTAT, downloaded February 2012).

Our method of scaling in 5 steps (calculations and numbers are per country, per crop and per technology):

1. Calculate area_scaled: For each pixel, multiply the current area harvested (from SPAM2000) by the area_ratio:

area_ratio = FAO2005 total harvested area/sum of the harvested area in all relevant pixels from SPAM2000

First scale each crop x technology value, then add all technologies of one crop (repeat for production below).

2. Calculate production_scaled: For each pixel multiply area_scaled by SPAM2000 yield, resulting in new_prod. Then multiply by prod_ratio:

prod_ratio = FAO2005 total production/sum of the new_prod in all relevant pixels

3. Calculate yield_scaled: For each pixel divide production_scaled by area_scaled, and limit results to yield_min <= yield_scaled <= yield_max; i.e., resulting yields are adjusted to fall within crop specific minima and maxima, which are currently defined per technology:

  • crop_irrigated_max = 5 * crop_total
  • crop_rainfed_high_max = 3 * crop_total
  • crop_rainfed_low_max = 2 * crop_total
  • crop_subsistence_max = 2 * crop_total
  • crop_any_min = 0.1 * crop_total

Remember which crop and technology needed adjustments of yields due to minima and maxima constraints.

4. Calculate again production_scaled:

production = area_scaled * yield_scaled * prod_ratio (new)

Only apply this prod_ratio (new) to pixels which were not corrected for min/max constraints.

5. Repeat steps 3 and 4 up to five times, thus arriving at scaled production and yield, where yields do not violate the min/max constraints.

Besides assuming that the relative land use/land cover pattern, irrigated areas and population densities have not changed significantly over five years, this method does not take account of possible changes in crop growing patterns at subnational levels. Neither does it consider the possibility that FAO country statistics may have classified some crops differently (e.g., moved beans to other pulses).

We also applied this scaling method to SPAM2000 as a last step after crop areas had been allocated to each pixel in order to create results which are as close as possible to FAO's 1999-2001 average for those crops.

Publication date
Source / Citation
You, L., S. Crespo, Z. Guo, J. Koo, K. Sebastian, M.T. Tenorio, S. Wood, U. Wood-Sichra. Spatial Production Allocation Model (SPAM) 2000 Version 3 Release 6.
Location
http://harvestchoice.org/labs/spatial-allocation-agricultural-production