IRI@AGU: The Climate Scenarios behind Ag Models
This post is the fourth in a series of Q&As with scientists from the International Research Institute for Climate and Society who will be presenting their work at the annual meeting of the American Geophysical Union in San Francisco December 9 to 13.
IRI’s Arthur Greene develops methods for characterizing climate trends on “near-term” time scales, i.e., out to a few decades in the future. His latest research has been the creation of models that simulate climate on these scales for particular regions. In an interview last year, Greene explained how he developed simulation models in southeastern South America using information from past observations, global climate model projections of the future, and knowledge about the physical processes that affect climate. In South America, for example, global warming and recovery of the ozone layer are thought to influence precipitation oppositely, with warming allowing for more moisture and ozone decreasing precipitation. Combining this knowledge, scientists can run different scenarios with the model, allowing them to simulate a range of possible future climate conditions. Read the Q&A below to learn more, and attend Greene’s presentation if you’re going to AGU.
How has your research developed since last year?
We have added a range of refinements to the simulation model, including more sophisticated treatment of non-rainy-season periods and better reproduction of point-to-point correlations in seasonal climate variations. We have also incorporated covariation in long-range trends, which allows us to better account for how variables interact with each other. Since climate models that project wetter future conditions also tend to project smaller temperature increases, and vice versa, this effect is potentially quite important for crop yields. Finally, we have extended the simulation methodology to new regions.
What are the applications of the climate simulations?
A primary application is the driving of agricultural models. It is not sufficient, for the accurate prediction of yields, to say only that average temperatures and/or rainfall will change by some amount. Rather, crop models require the detail of daily, sequential data, including precipitation and minimum and maximum temperatures, in order to generate realistic yields. The simulation models I have been working on provide such data. Simulated yields can become inputs to economic models, which can help us to understand the effects of climate variations on economies, as mediated by agricultural responses.
Importantly, these models attempt to account not only for expected climate changes, but also the decade-to-decade variations that may act to either enhance or attenuate those changes. These effects are potentially important, but not often considered.
Which regions have you studied, and why did you select them for your research?
We have applied these methods to the Western Cape region of South Africa, southeastern South America, the West African Sahel and equatorial East Africa. We are currently working on an application in South Asia, on the Indian subcontinent. A number of these simulations are being produced in conjunction with the Agricultural Model Intercomparison and Improvement Project (AgMIP), and in the near future we will be focusing on AgMIP regions of interest. All of these regions are places where climate impacts on agriculture may have significant human consequences.
You are not predicting what you think necessarily will happen in, say, the next decade, but rather providing a range of possibilities of what climate trends may be over that time period. How is this kind of information beneficial to those modeling agriculture and economics?
Although skillful decadal predictions may eventually be realized, this does not seem to be the case so far. Having in hand a range of possible climatic futures, whose likelihoods can be quantified to some degree, can help to bridge the gap between information that is desired and that which can realistically be provided at this point in the evolution of climate modeling.
We have global climate models that predict climate through the end of the 21st century, and some even longer. Why can’t you predict at a more near-term, decadal scale more easily?
A good deal of energy is now being expended in an effort to improve decadal prediction, but this is still what might be described as a nascent science. Skillful prediction is hampered, first, by a lack of understanding of the processes involved in the generation of decadal variability within the climate system. It is also rendered difficult because the processes accounting for climate system “memory” are reasonably believed to reside in the ocean, with its great thermal inertia and relatively sluggish movements, yet the regions where predictions are needed are on land. Information is lost in the transmission of oceanic signals to the climate of terrestrial regions.
What questions are you planning to explore next in your research?
We are continually seeking to refine both our models and our understanding of the regional processes that influence the climates we seek to model. It is only in bringing these elements together that the goal of reliable regional climate simulations can be achieved.