Importance of Financially Intensive Climate Products in Large-Scale Crop Yield Estimates

Because agricultural yield is highly sensitive to climate variability, yield projections require high spatial and temporal resolution weather products. Reanalysis data, which combines model output with climate observations, is commonly used in historical impact studies, while general circulation models (GCMs) allow for future projections. The climate science community has recently invested considerable financial and computational resources into high-resolution regional climate model (RCM) simulations (coupled with GCM output) for the purpose of impact studies, but we know little about their improvement (if any) in large-scale crop yield estimates. Alternatively, insufficient investments in climate observations (particularly in the developing world) forces crop modelers to rely on reanalysis climate model output for historical analyses, leaving the importance of observational products largely unchecked. The computational demand of processing a high-resolution crop model on large scales restricts most studies to the local scale, limiting our understanding of large-scale agricultural impacts. We drive a parallelized version of the DSSAT crop model with a variety of climate products to evaluate the importance of financially intensive high-resolution climate observations and models in large scale yield estimates. By analyzing convergence properties as we aggregate high-resolution impact measures, we can assess the value of climate products at numerous scales relevant to decision-making, helping prioritize future investments into new climate products.