January 12, 2016
by Rob Mitchum, Computation Institute
One of the most important consequences of climate change will be its effect upon global agriculture and food supply. In worst case scenarios, increased temperatures and more frequent droughts will create food scarcity and dramatic shifts in the types of crops different regions of the world can grow. But in order to better prepare for these changes, more nuanced forecasts about climate and agriculture are needed. This week, one of the most ambitious projects in this area announced a new phase in creating these important computational tools.
Since 2012, an international group of scientists have worked on the Global Gridded Crop Model Intercomparison (GGCMI) Project, an effort to assess climate impacts on agriculture at continental and global scales and compare and improve existing crop models. The ultimate goal is to create powerful new models that can help decision-makers at the United Nations, the Intergovernmental Panel on Climate Change, and governments around the world manage their food production under a changing climate. But to get to that important final goal, several intermediate steps are required.
Recently, GGCMI published the results of their first phase: a comparison of different crop models each using the same historical weather datasets to "hindcast" agricultural statistics and determine how accurately they "predicted the past.” These trials, run in part on the Midway supercomputing cluster, allowed a team of GGCMI researchers co-led by CI Fellow and RDCEP research scientist Joshua Elliott to validate and examine the similarities and differences of the models by comparing how they performed against actual agricultural data.
If Phase 1 of the project was the “historical” phase, Elliott described phase 2 as the “inward” phase -- a series of tests to determine the sensitivity of the models under various conditions. Researchers will examine how the models respond to different values for carbon, temperature, nitrogen, and water, determined by historical data and the results of Phase 1 simulations. This fine-tuning will provide valuable information on how crops respond to these factors in different regions of the world and under different future scenarios.
The second phase will also be useful for developing climate impact emulators, which help researchers in other fields use the output of these crop models for their own studies. For example, if economists want to study the effect of climate change and its agricultural impacts upon the global economy, they won’t need to run the full complex climate and crop models themselves, but can use a simplified model that provides the input data they need for their own forecasts.
Another exciting outcome from GGCMI Phase 2 will be a first-of-its-kind analysis of model responses at daily time scales for the entire globe. Researchers will gather data from the models at this high resolution to examine in more detail the biophysical processes underlying the results, Elliott explained.
“This effort will be groundbreaking at any scale, but the fact that we are attempting it globally for a bunch of crops and models is pretty serious,” Elliott said.
The AgMIP community is open to expanding the GGCMI network to any interested modeling groups. Gridded crop modelers, who are not already involved but wish to participate in the very exciting Phase 2, please contact Joshua Elliott and Christoph Müller at email@example.com.
See the original article at ci.uchicago.edu/blog/ggcmi-crop-simulation-study-enters-new-phase.