Impact of irrigation management on performance of soybean

An experiment was conducted at the University of Florida in 1978 to study the impact of irrigation on soybean growth, development and yield (Wilkerson et al., 1983). The experiment included two treatments, i.e. rainfed and irrigated with 21 irrigation applications for a total of 206 mm of supplemental irrigation. The cultivar Bragg was planted June 29, 1978, at a plant density of 29.9 plants/m2. The crop was well managed. Phenology was observed non-destructively, and crop growth was sampled at regular intervals. The CSM-CROPGRO-Soybean model was previously calibrated for cv. Bragg. Days to flowering and to physiological maturity were predicted well, and yield and yield components were simulated very well (RMSE for Tops Weight: 218 kg/ha; RMSE for yield: 157 kg/ha). Early during the growing season there was no difference in above ground biomass and LAI between the two treatments (Figure 9) due to the high amount of rainfall received during this period (Figure 10). Around 60 days after planting, rainfall reduced significantly, with a decrease in extractable soil moisture (Figure 10). This resulted in water deficits for the rainfed treatment (Figure 10), reducing the increase in tops weight, seed weight, and LAI for the rainfed treatment compared to the irrigated treatment (Figure 9).

The experiment was then used as the basis of a strategy analysis scenario, using the same input conditions each year of 30-years. Historical weather data were not available for the entire period, so the internal weather generator (WGEN) was used. For the irrigation management, the automatic irrigation option was selected with different threshold values to determine when to irrigate. When the soil moisture content in the top 30 cm of the soil profile dropped below a given threshold value, an irrigation event was triggered. The thresholds were varied from 10% to 99% (remaining soil water at which to irrigate), providing 10 irrigation scenarios and one rainfed scenario. Final results were analyzed either as box and whisker plots, cumulative probability graphs, or a mean variance graph (Figure 6).

The results showed clearly that yield increased with an increase in the threshold value while the variance and variability were reduced (Figure 6). However, the amount of supplemental irrigation required also increased to over 300 mm for the highest threshold value with more than 30 irrigation applications (Figure 11). In contrast, the amount of water applied up to a threshold value of 50% was less than 100 mm. Water use efficiency was highest for the irrigation treatments that had the smallest threshold value, but the variability was very high (Figure 12). The scenario with a threshold value of 50% showed the best water use efficiency taking into consideration the associated uncertainty as well as yield (Figure 6) and total water use (Figure 11). If ground or surface water are limited due to governmental restrictions, water rights, or a drought, the model can be used to help determine the best scenario that maximizes yield while at the same time optimizing water use for irrigation.