Crop yield, soil temperature and sensitivity of EPIC under central-eastern Canadian conditions

• Published in: Canadian journal of soil science 1998-08, Vol.78 (3), p.431-439
• Main Authors: Roloff, G, De Jong, R, Nolin, M.C
• Format: Article
• Language: English
• Subjects: crop yield; environmental modelling; environmental policy integrated climate model; errors; evaluation; Glycine max; leaching; nitrates; percolation; performance tests; prediction; runoff; sensitivity analysis; simulation models; site factors; soil temperature; water quality; Zea mays
• ISSN: 0008-4271; 1918-1841
• DOI: 10.4141/S97-087

The Environmental Policy Integrated Climate (EPIC) model can be an important tool for agricultural and environmental management. Application of EPIC in central-eastern Canada requires testing to evaluate its performance and to indicate eventual changes that ultimately can lead to a more accurate model. We compared crop yields and soil temperatures estimated using EPIC (version 5300) with measured data from two sites — Barrhaven, Ontario and St. Antoine, Quebec — where corn (Zea mays L.) and soybean (Glycine max L.) were grown during 4 yr. The sensitivity of several EPIC outputs to variations in selected input parameters was also determined. EPIC was run with either the Penman-Monteith (PM) or the Baier-Robertson (BR) potential evapotranspiration (PET) method. Mean estimated and measured soybean yields were not different, independent of PET method and location. Mean corn yields were underestimated at Barrhaven with the PM method, whereas they were overestimated by the BR method at St. Antoine. Using the PM method resulted in no difference between estimated and measured yields in 2 out of 5 corn years, while the BR method ensued in no difference in 3 out of 5 years. For soybean, only the BR method resulted in 1 yr of yield different than measured. Performance of the model in relation to experimental error and efficiency of the model varied with crop and location, and indicated that EPIC caused less error and was more efficient in estimating soybean than corn yields. Comparing limited measured soil temperature data with EPIC estimates reveled that EPIC tended to underestimate soil temperature; however no effect on estimated yield was observed. The relative sensitivity of EPIC outputs was least for yield and greatest for leached nitrate-N. The latter also displayed the highest variability in sensitivity. EPIC studies focusing on leached nitrate-N should require the most accurate crop, soil, hydrological and weather data in order to minimize errors. Key words: Environmental modelling, soybean yield, corn yield, sensitivity analysis