Regional Atmospheric Circulation and Surface Temperatures Predicting Cotton Yields in the Southeastern USA.
Guillermo Baigorria1, James Hansen2, Neil Ward2, James W. Jones1, and James J. O'Brien3. (1) Univ. of Florida, Frazier Rogers Hall, Agricultural and Biological Engineering Dept., Gainesville, FL 32601, (2) Int'l Res. Inst. for Clim. & Societ, Monell Bldg., Palisades, NY 10964-8000, (3) Center for Ocean-Atmospheric Prediction Studies, The Florida State Univ., Tallahassee, FL 32306
Research has shown strong relationships between ENSO phase and climate in the southeastern USA during the boreal winter. Crop yields in this region are significantly affected by ENSO phase due to predictable patterns of climate during this time of the year. However, both climate during the boreal summer months and cotton yields in this region show little or no association with ENSO phase. With a goal of improving prediction of cotton yields at a long lead-time in the SE USA, we identify regional atmospheric variables that are related to historic boreal summer rainfall and cotton yields, and evaluate the use of predictions of those variables from a global circulation model (GCM) for forecasting cotton yields. We analyzed de-trended cotton yields (1970-2004) from 48 counties in Alabama and Georgia, monthly rainfall from 53 weather stations, monthly estimates of 850 and 200 hPa winds at and surface temperatures over the SE USA region from reanalysis data, and monthly predictions of the same variables from the ECHAM 4.5 GCM. Meridional wind fields and surface temperature around SE USA were correlated with cotton yield and with rainfall, especially during April and July, over most of the region. The tendency for cotton yields to be lower during years with atmospheric circulation patterns that favor higher humidity and rainfall is consistent with increased incidence of disease during flowering and harvest periods under wet conditions. Cross-validated yield predictions based on ECHAM hindcasts of wind and temperature fields forced by observed SSTs showed significant skill (55% and 60% hit skills based on terciles and averages respectively). Mean square errors varied from 3 to 10% over all locations and from 0 to 15% over all years. We conclude that there is potential to increase the skill of cotton yield forecasters using variables that are determined by numerical climate models.