Simulating Corn Yield Response to Irrigation Timings: Validation of the Epic Model

Abstract

Models simulating crop growth offer an opportunity to fill important data gaps from experimental plots. Examples include incorporation of many weather conditions as well as alternative management strategies. The simulated data must be reliable, however. This study examined the application of the EPIC growth model to simulate corn (Zea mays L.) yield on the Southern High Plains when water stress was imposed on various dates in the growing season. The EPIC model produced simulated yield distributions with means not significantly (P = 0.05) different from those of the actual data in 2 of the 3 yr considered, and standard deviations of simulated yields similar to those of the actual yields for all 3 yr. Linear regressions of simulated yields on actual yields resulted in slope coefficients not significantly (P = 0.05) different from 1.0 for 2 of the 3 yr. Simulated yields explained 83, 86, and 72% of the variance in actual yields for the 3 yr of measured data. The EPIC model appears useful for further studies of irrigation scheduling for corn.

Historical Development and Applications of the EPIC and APEX models

“…Bryant et al (1992) found that EPIC accurately predicted mean and annual corn yields measured for 38 irrigation stress experiments conducted during the 1975-77 period at Bushland, Texas, after the effects of a hail storm were accounted for in 1976. Gray et al (1997) reported r 2 values of 0.82 and 0.85 for corn yield production functions developed from EPIC simulations that were performed for irrigation timing experiments conducted at Bushland, Texas, during the years 1990-93.…”

“…Warner et al (1997a) report that annual yields estimated by EPIC did not correlate well with the measured yields and that the model exhibited a bias toward overprediction of yields for corn grown over a total of 15 treatment years (five nitrogen fertilizer rates for three years each) at Storrs, Connecticut. However, some studies noted that EPIC tended to underestimate peak yields and overestimate low yields (Bryant et al 1992;Touré, Major, and Lindwall 1994). Martin, Nearing, and Bruce (1993) further noted that differences in the predicted EPIC yields between erosion classes were less those measured.…”

“…Martin, Nearing, and Bruce (1993) concluded that EPIC could be useful for performing relative comparisons between different erosion class soils. Bryant et al (1992) pointed out that EPIC (and other simulation models) are best used to generate simulated yield distributions that are similar to measured yield distributions, rather than trying to match measured yields in each single year. Similar conclusions can be drawn from the majority of the other studies reviewed here.…”

“…Bryant et al (1992) found that EPIC can be used to assess the impacts of different irrigation amounts and timings, based on validation of the model for the 38 different irrigation stress experiments at Bushland, Texas. Rinaldi (2001) used EPIC, following calibration by Ventrella and Rinaldi (1999) and others, to assess over 60 45-year scenarios of irrigation timing and amount strategies for sunflowers at Foggia in southern Italy.…”

Calibration of a crop model to irrigated water use using a genetic algorithm

“…Because its original focus was erosion-related, EPIC can simulate decade-scale or longer intervals. These features have suited EPIC to a broad range of applications, including plant nutrition studies (Cole et al, 1987;Dautrebande et al, 1999); national and international assessments of agroecological change impact (Brown and Rosenberg, 1999;Brown et al, 2000;Bernardos et al, 2001), including the High Plains Aquifer (Easterling III et al, 1993); irrigation planning and water use (Bryant et al, 1992;Ellis et al, 1993;Evers et al, 1998;Guerra et al, 2005); and regional studies (Geleta et al, 1994;Cabelguenne et al, 1995;Fortin and Moon, 2000).…”