Predicting agricultural impacts of large-scale drought: 2012 and the case for better modeling

Working Paper 111

Abstract

The 2012 growing season saw one of the worst droughts in a generation in much of the United States and cast a harsh light on the need for better analytic tools and a comprehensive approach to predicting and preparing for the effects of extreme weather on agriculture. We present an example of a simulation-based forecast for the 2012 US maize growing season produced as part of a high-resolution multi-scale predictive mechanistic modeling study designed for decision support, risk management, and counterfactual analysis. We estimate national average yields of 7.507 t/ha for 2012, 24.6% below the expected value based on increasing trend yield alone, with an interval based on resampled forecasts errors stretching from 5.586 to 8.967 t/ha. On average, the median yield simulations deviate from NASS observations by 8.3% from 1979 to 2011.

Joshua Elliot, Michael Glotter, Neil Best, Ken Boote, Jim Jones, Jerry Hatfield, Cynthia Rozenweig, Leonard A. Smith and Ian Foster