Predicting the Future of Soy in South America
By Elisabeth Gawthrop, Climate and Society ’13
During the 20th century, southeastern South America experienced a soybean boom due in part to increased summer rainfall. Soybean prices have been so high that farmers are planting this crop in areas that were traditionally considered marginal. Why rainfall increased remains unclear. Human-induced climate change, stratospheric ozone depletion and natural climate fluctuations are all possible culprits.
Having a better understanding what drove the changes as well as the range of possible future shifts in rainfall can help soy growers, ministries of agriculture in the region and global commodities markets better plan for the future. This is especially important as soy production in Brazil, Argentina and Paraguay accounted for a projected 51 percent of overall global production. The International Research Institute for Climate and Society’s Arthur M. Greene and collaborators are attempting to improve climate and agricultural modeling for this region through a new stochastic simulation of future climate. Find out more about their efforts in the Q&A with Greene below and stop by his talk at AGU.
Can you explain “stochastic simulation of regional climate sequences?”
Sequences are simply lengths of climate data (“sequences” because the data is ordered in time, not just a bunch of numbers). These data have a random (“stochastic”) character but are produced in such as way as to reflect our expectations about the evolution of future climate in southeastern South America.
When did the trend in precipitation start changing and has it leveled off?
The trend has been more or less continually upward throughout the twentieth century, with perhaps a couple of downward wiggles near the beginning where the data is less certain. Since about 2000, however, the trend has been declining, for reasons that are unclear.
What goes into creating and developing the simulation model? Does the model vary according to region?
The simulation model synthesizes climate information from past observations, global climate modelprojections, and from relevant current research, to arrive at a holistic characterization of the range of plausible future climate trajectories for a particular region, including possible modulation of these trajectories by natural climate variability.
In developing the model, concepts from classical time series analysis were fused with knowledge and expectations about physical climate processes, in this case the influences of global warming and ozone variations on southeast South American rainfall, to design the simulation model.
Simulation models must differ from region to region because climate processes differ among regions. For example, South Polar ozone recovery isn’t likely to play the same role in the tropics’ climate as it does in southern high latitudes. Therefore it is less likely to be included in tropical simulations. The model for this study is adapted from one we recently developed for the Western Cape region of South Africa, because certain modeling strategies are effective for both climates.
What do the simulations show? What is the range of probable scenarios?
There is a significant dependence in the range of future trends on the degree to which ozone recovery plays a role in southeastern South America’s precipitation variability. Global warming, due to increasing greenhouse gases, also plays a role. The simulated trend reflects a balance between these two factors, which tend to drive future precipitation in opposite directions (ozone induces a drying tendency, greenhouse warming allows for more moisture). How this balance will actually play out in the future is not yet understood. The simulations reflect this uncertainty by providing the end user with a choice, between ozone- and greenhouse gas-dominated trends, or a “compromise trend” between these two extremes. Low-frequency climate variations that happen over decades or longer have the potential to amplify, or possibly reduce, effects of these trends for limited periods of time depending on their phase.
How can (or is) this information being used in planning for the region’s future?
The idea is to generate simulations that explore the range of likely future climate “trajectories” for the region. These will be processed and run in a crop model to translate the climate trajectories into terms of agricultural production, and eventually, economic impacts. We don’t yet know whether the recent downward trend is a harbinger of the future, but if it is, it is likely that production will decrease with the passage of time.