GPFARM modeling of corn yield and residual soil nitrate-N

• Published in: Computers and electronics in agriculture Vol. 43; no. 2; pp. 87 - 107
• Main Authors: Shaffer, M.J, Bartling, P.N.S, McMaster, G.S
• Format: Journal Article
• Language: English
• Published: Amsterdam Elsevier B.V 01.05.2004; Elsevier
• Subjects: Agronomy. Soil science and plant productions; Biological and medical sciences; crop models; Decision support systems; Environmental quality; farm management; Fundamental and applied biological sciences. Psychology; Model validation; Soil–crop models; Whole-farm management; Decision support systems; Soil–crop models; Whole-farm management; Environmental quality; Model validation; Validation; Monocotyledones; Soil-crop models; Zea mays; Decision support system; Nitrates; Modeling; Soils; Gramineae; Angiospermae; Spermatophyta; Yield; Environment quality
• ISSN: 0168-1699; 1872-7107
• DOI: 10.1016/j.compag.2003.11.001

US agriculture is facing low commodity prices to farmers because of foreign competition, environmental concerns, and weather fluctuations such as droughts. Producers need to quickly evaluate the marketplace and select appropriate management systems for their farms and ranches. The Great Plains Framework for Agricultural Resource Management (GPFARM) decision support system was developed to assist farmers and ranchers with key strategic management decisions, but requires additional testing before general adoption by the agricultural community. This paper evaluated GPFARM simulation of continuous corn ( Zea mays L.) yields and soil residual nitrates under irrigated and partially irrigated conditions, fertilized and non-fertilized applications, and high and low planting densities. Validation results for a 3-year field study indicated the model could simulate corn yields and soil residual NO 3-N without bias at the P<0.05 level with R 2 values for predicted versus observed corn yields and soil residual NO 3-N of 0.830 and 0.383, respectively. Mean extended modeling error (EME), a measure of modeling error extending outside the error range of the validation measurements was 168 and 25.2 kg/ha for corn grain yields and soil residual NO 3-N, respectively. The EME results further showed that the scatter around the simulated versus observed 1:1 lines for soil residual NO 3-N versus corn yields was 53.5 and 19.9% of the mean sum of the absolute residuals, respectively, suggesting higher modeling error with the residual NO 3-N. The EME method also effectively separated modeling error from error that could be accounted for by uncertainty in the experimental validation data set. Agricultural producers, consultants, and action agencies should consider these validation results and potential errors when using the model to predict corn yields and related soil NO 3-N estimates in strategic management planning and environmental assessment studies.