Evaluating Hybrid-Maize model in rainfed conditions in northwestern Indiana

Salah F Issa, Purdue University

Abstract

Scientists face an unprecedented challenge of increasing food production by 100% within the next fifty years to meet demand while responding to the undetermined effect and impact of climate change. To respond to these challenges, scientists are turning to models to assist them in mapping out the complex interactions between environmental conditions, management strategies and crop genotype. The Hybrid-Maize (HM) model is a process-based model that predicts maize growth, grain yields and stover in rainfed or irrigated conditions. The HM model has been widely used to predict yield potential in diverse locations such as Indonesia, Vietnam, China and the United States. The overall goals of this study are to evaluate HM model in the wet, rainfed conditions representative of northern Indiana and compare its performance to Crop Environment Resource Synthesis-Maize (CERES-Maize) model available in the Decision Support System for Agrotechnology Transfer (DSSAT) program. The specific objectives were to: 1) model performance for simulating early growth, stover and grain yield; 2) quantify the impact of excess water stress and grain filling temperatures on gaps between predicted and actual yields; and 3) compare HM and DSSAT CERES-Maize (DSSAT-M) under optimal conditions predictions to grain yield contest winners in Indiana. Weather, soil and management inputs were carefully gathered to ensure accuracy in model input data. Predicted growth, yield and stover were compared against the measured data collected at Purdue's Agronomy Center of Research and Education (ACRE). The HM model was found to be a good predictor of grain yield (n-RMSE=18%), a fair predictor of stover biomass (n-RMSE=23%) but it significantly over-predicted early growth (RMSE=1.19 Mg dry matter ha-1). In contrast, DSSAT-M was found to be a good predictor of grain yield (n-RMSE=18%) and an excellent predictor of stover yield (n-RMSE=9%) and early growth (RMSE=0.31 Mg dry matter ha-1). Excess water stress was significantly correlated to gaps in biomass predictions for HM at growth (V6) and stover growth stages. However, overall excess water stress was not correlated with grain yield gaps for both models. Both HM and DSSAT-M grain simulated yields were significantly correlated with average grain filling temperatures. Years experiencing relatively colder grain filling temperatures had larger yield gaps. Lastly, HM was a better predictor of regional yields from commercial contests than DSSAT-M. In conclusion, HM is a good predictor of grain yield and yield potential, DSSAT-M a good predictor of early growth biomass, stover and grain yield. Future research directions include: 1) adjusting the temperature based empirical equations used in predicting grain yield by to take into account cooler weather conditions during grain fill for both models, 2) expanding HM water stress function to include excess water stress and 3) Identifying the management, environmental and genotype factors involved in current contest winner grain yield tends in Indiana as compared to Nebraska contest winner grain yield trends.

Degree

M.S.A.B.E.

Advisors

Brouder, Purdue University.

Subject Area

Agronomy|Agriculture

COinS