Evaluation of the Erosion-Productivity Impact Calculator (EPIC) model from long-term rotations

Abstract

The Erosion-Productivity Impact Calculator (EPIC) model was used to predict spring wheat (Triticum aestivum) yields from long-term (1960 to 1989) crop rotations at the Agriculture Canada Research Station at Melfort. These predictions were compared to actual yields using the paired t test and linear regression analysis. Simulated mean yields of wheat for the different rotations were generally similar to actual yields, when daily maximum and minimum temperatures and precipitation were input into the model. The fertilized continuous wheat rotation gave the highest yield when averaged over the 30-yr period. EPIC was also satisfactory in predicting yield when weather was generated from long-term climate data. However, regression analyses indicated no significant relationship between simulated (using actual and generated weather data) and actual annual yields over the 30-yr period. Therefore, EPIC predicts long-term average yields with sufficient accuracy, but is not recommended for yield prediction for individual years. It can be a valuable decision-making tool for identifying crop rotations and management practices that offer potential yield benefits over a long time period.