David Lobell will be giving two talks this week at the AGU fall meeting:
Towards accurate models of global crop-climate interactions
This talk will provide a brief overview of the major links between climate and crops that are most in need of representation in global models. The talk will then focus on one of the key links - the effect of climate variation on crop productivity across a range of cropping systems and regions. I will present some recent work to use historical datasets to build statistical models at the scale of individual countries. The effects of different datasets and modeling assumption will be explored, to identify areas where statistical models provide robust representation of crop responses to climate. A sample application of these models - estimating the net regional and global impacts of recent trends in climate - will be presented.
Assessing the future of crop yield variability in the United States with downscaled climate projections
One aspect of climate change of particular concern to farmers and food markets is the potential for increased year-to-year variability in crop yields. Recent episodes of food price increases following the Australian drought or Russian heat wave have heightened this concern. Downscaled climate projections that properly capture the magnitude of daily and interannual variability of weather can be useful for projecting future yield variability. Here we examine the potential magnitude and cause of changes in variability of corn yields in the United States up to 2050. Using downscaled climate projections from multiple models, we estimate a distribution of changes in mean and variability of growing season average temperature and precipitation. These projections are then fed into a model of maize yield that explicitly factors in the effect of extremely warm days. Changes in yield variability can result from a shift in mean temperatures coupled with a nonlinear crop response, a shift in climate variability, or a combination of the two. The results are decomposed into these different causes, with implications for future research to reduce uncertainties in projections of future yield variability.