GC076-12
Using emulation to explore effects of crop model uncertainty on projected land use and management
Friday, 11 December 2020: 06:03
Virtual
Sam S. Rabin, Karlsruhe Institute of Technology, Institute of Meteorology and Climate Research / Atmospheric Environmental Research, Karlsruhe, Germany, James A Franke, University of Chicago, Department of the Geophysical Sciences, Chicago, IL, United States; University of Chicago, Center for Robust Decision-making on Climate and Energy Policy, Chicago, IL, United States, Christoph Müller, Potsdam Institute for Climate Impact Research, Potsdam, Germany, Peter Alexander, University of Edinburgh, School of Geosciences, Edinburgh, United Kingdom; University of Edinburgh, The Royal (Dick) School of Veterinary Studies, Edinburgh, United Kingdom and Almut Arneth, Karlsruhe Institute of Technology, Institute of Meteorology and Climate Research / Atmospheric Environmental Research, Garmisch-Partenkirchen, Germany
Abstract:
As climate change and rising CO
2 concentrations affect crop growing conditions alongside changing demand patterns, the future may see significant changes in cropland and pasture area, crop species distribution, and the use of irrigation and fertilizer. Coupled models that account for both agricultural productivity and economics in a global change context are an important tool for exploring the potential magnitude and effects of these changes, but they inherently contain several sources of uncertainty. These include uncertainty stemming from choice of future scenario and climate model, as well as parameter uncertainty within sub-models—the magnitude of which are both commonly tested in the literature. However, uncertainty related to model structure and process inclusion also exists but is rarely explored because of technical complexity.
To test how much of a difference the choice of crop model makes in terms of projected land use area, ideally one would couple a given economic model with several different crop models. In practice, most coupled models (and their sub-models) are not designed to be modular in this way, and a research group using a coupled model system is likely not familiar with running other crop models. Thus, a substantial learning curve and technical work would need to be overcome in order to enable such an analysis. Emulation—the use of an empirically-derived simulator of a crop model that can sufficiently reproduce the model’s outputs—offers a simpler alternative.
Here, we couple emulators built for crop models participating in the Global Gridded Crop Model Intercomparison project (GGCMI) with the Parsimonious Land Use Model (PLUM) to explore the effects of crop model uncertainty on projections of future land use areas and management. This uncertainty is compared with that resulting from choice of climate-socioeconomic scenario and global climate model. We additionally review lessons learned from this work, which are applicable more broadly than just agricultural modeling.
