Climate Change Impacts on US Agriculture and the Benefits of Greenhouse Gas Mitigation

Thursday, 18 December 2014
Erwan Monier1, Ian Sue Wing2 and Ari Stern2, (1)MIT, Center for Global Change Science, Cambridge, MA, United States, (2)Boston University, Boston, MA, United States
As contributors to the US EPA’s Climate Impacts and Risk Assessment (CIRA) project, we present empirically-based projections of climate change impacts on the yields of five major US crops. Our analysis uses a 15-member ensemble of climate simulations using the MIT Integrated Global System Model (IGSM) linked to the NCAR Community Atmosphere Model (CAM), forced by 3 emissions scenarios (a "business as usual" reference scenario and two stabilization scenarios at 4.5W/m2 and 3.7 W/m2 by 2100), quantify the agricultural impacts avoided due to greenhouse gas emission reductions.

Our innovation is the coupling of climate model outputs with empirical estimates of the long-run relationship between crop yields and temperature, precipitation and soil moisture derived from the co-variation between yields and weather across US counties over the last 50 years. Our identifying assumption is that since farmers’ planting, management and harvesting decisions are based on land quality and expectations of weather, yields and meteorological variables share a long-run equilibrium relationship. In any given year, weather shocks cause yields to diverge from their expected long-run values, prompting farmers to revise their long-run expectations. We specify a dynamic panel error correction model (ECM) that statistically distinguishes these two processes.

The ECM is estimated for maize, wheat, soybeans, sorghum and cotton using longitudinal data on production and harvested area for ~1,100 counties from 1948-2010, in conjunction with spatial fields of 3-hourly temperature, precipitation and soil moisture from the Global Land Data Assimilation System (GLDAS) forcing and output files, binned into annual counts of exposure over the growing season and mapped to county centroids. For scenarios of future warming the identical method was used to calculate counties’ current (1986-2010) and future (2036-65 and 2086-2110) distributions of simulated 3-hourly growing season temperature, precipitation and soil moisture. Finally, we combine these variables with the fitted long-run response to obtain county-level yields under current average climate and projected future climate under our three warming scenarios. We close our presentation with a discussion of the implications for mitigation and adaptation decisions.