Page 92 - Climate Change and Food Systems

climate change and food systems: global assessments and implications for food security and trade
crop simulation models. Each crop model includes generic parameters that are based on field experience and which describe the growth process of common genetic varieties of cereals, legumes and grasses. Recent applications of the DSSAT model to project the yield effects of alternative SRES scenarios in developing countries include Lal (2011), Thornton, et al., (2011) and Felkner et al. (2009).
Other crop growth models that have been applied in the study of climate change impacts
on developing countries include WOFOST (WOrld FOod STudies, documented in Ittersum, et al., 2003; Boorgaard et al., 2011), AquaCrop, developed by FAO (Steduto et al., 2008) and the Agricultural Production Systems Simulator (APSIM), a wiki-modelling framework.9
A key contribution of crop growth simulation models in research on climate change adaptation is their ability to simulate the effects of changes in farm management on biophysical plant growth. Analysts can exogenously change planting and harvesting dates, fertilizer use, irrigation, or
choice of crops or crop varieties and evaluate the effectiveness of these adaptations in offsetting any negative effects of climate change on crop yields. The idea that farmers do not automatically adapt to changing climate is a common, but outdated, criticism of these models. Many crop simulation models now incorporate automatic adjustments by farmers in planting dates, choice of cultivar and use of irrigation and fertilizer.
Two frequent criticisms of crop simulation models are that their data requirements are intensive and that they are applicable only at small spatial scales that are not readily usable in economic analyses, which are mostly conducted at the regional, national and global scales. The development of large area crop models, including a new feature of DSSAT and recently including PEGASUS 1.0 (Predicting Ecosystem Goods
and Services Using Scenarios) (Deryng, 2009; Deryng et al., 2011), and the Model to capture
the Crop-Weather relationship over a Large Area (MCWLA) (Tao et al., 2009a), address both critiques by using parsimonious representations of crop growth processes. This has both considerably reduced their data requirements and improved the models’ applicability at larger scales.
For example, PEGASUS 1.0 is a global
crop model designed to overcome the multiple problems of intensive data requirements, the small scale that typifies many crop yield simulation analyses, and the absence of an automatic farmer response to climate changes. The PEGASUS model, developed so far for soybean, maize and spring wheat, describes daily biophysical plant growth processes in response to climate and
to the farm management practices of irrigation, planting and harvesting dates, and fertilizer
use. Planting and harvesting dates and choice
of cultivars can be fixed or allowed to adjust automatically to changing climate conditions.
Data requirements for PEGASUS are relatively small and they are drawn from newly available global data bases on climate, soils, yields, harvested areas, cropping calendars and irrigation and fertilizer use. These data inputs are aggregated to match the 10’ longitude and 10’ latitude scales of the climate data. By aggregating model results, PEGASUS can describe the effects of climate variables on yields, planting dates and cultivar choice at the regional and global levels. Planting dates simulated by Derying et al. (2011) matched 74 percent of the observed planting dates in
the global cropping area of maize, 91 percent for soybeans and 75 percent for spring wheat. Correlations between simulated and actual yields are also strong. Running two climate change scenarios, SRES A1B and B1, with and without adaptive behavior, Deryng et al. project significant declines in global crop yields in 2050 relative to 2000, but find that 60-78 percent of the losses could be averted through adaptation of planting dates and cultivar choices.
A criticism of crop simulation models that
is made less frequently, but which is key for economic analysis, relates to uncertainties in the
9
APSIM is documented on its wiki website at:
http://www.apsim.info/Wiki/APSIM-and-the-APSIM- Initiative.ashx
72