Wednesday, November 15, 2006

Using Empirical and Stochastic Models to Evaluate Corn Yield by Soil Series.

Raymond R. Struthers, Chris Johannsen, Gary Steinhardt, and Darrell Schulze. Purdue Univ, 914 W State, West Lafayette, IN 47907

The Dideriksen Soil Yield Model has been used by the National Resources Conservation Service (NRCS) in Indiana to determine the relative difference in corn yield for each soil series. This is an empirical model that has not been tested using modern agriculture technologies such as Global Positioning Systems (GPS) or Geographical Information Systems (GIS) that can, with a relatively high degree of accuracy, measure the observed yield. The Dideriksen Soil Yield Model was compared to the more stochastic Hybrid Maize Model which is based primarily on temperature and rainfall. Both models were tested by soil series, using observed corn yield data collected by a combine yield monitor at the Davis Purdue Agricultural Center in Randolph County Indiana from 1996 through 2004. The linear relationship between observed and predicted corn yield for both models was similar with a slight advantage to the Dideriksen Soil Yield Model. The Hybrid Maize Model was pushed beyond the intentions of its creators, but preformed remarkably well in the Eastern Corn Belt. Therefore, either crop model may be used to identify corn yield potential for soil series found in this study site with relative accuracy.