Evaluation of SALTMED and AquaCrop models in simulating sweet maize yield under deficit irrigation and superabsorbent application

Introduction Climate conditions in Iran and high water consumption in agriculture have led to some solutions including deficit irrigation and the use of soil amendments such as superabsorbent polymers. However, application of such methods in long term is time consuming and expensive. For this reason, various plant models such as AquaCrop and SALTMED have been developed to overcome these problems. AquaCrop is the crop growth model developed by FAO and it is superior to other plant models because of its simplicity, the need for less data, user-friendliness and acceptable accuracy. AquaCrop simulates the yield response of crops to water and is particularly well suited to conditions in which water is a key limiting factor in crop production (Raes et al., 2012). The SALTMED model is a physically based model and can be used to simulate crop growth with an integrated approach considering water, crop, soil and field management. This model also is capable to simulate evapotranspiration processes, water flow and solute transport in soil and crop yield (Hirich et al., 2014; Ragab et al., 2010). A review of past studies has shown that so far few studies have been carried out using the SALTMED and AquaCrop models to simulate maize yield under irrigated and superabsorbent applications. Regarding this issue, the objective of this study was to evaluate SALTMED and AquaCrop models for simulating sweet maize yield under both deficit irrigation and superabsorbent applications. Materials and methods This research was carried out in the experimental farm in Ahwaz with Longitude of 48°32ʹ05ʺ and latitude of 31°15ʹ20ʺ and a height of 11m above sea level, in two spring and summer crop seasons in 2016. The experiments were carried out in a split plot design based on randomized complete block design with 12 treatments and three replicates. The designed treatments consisted of irrigation water quantity (at three levels of I1: 100%, I2: 75%, I3: 50% plant water requirement) and different levels of superabsorbent A300 (at four levels of S0: 0, S1: 0.3, S2: 0.6 and S3: 0.8 gr/kg of soil). In order to simulate the yield of Sweet maize, the spring crop dataset was used for calibration and the summer crop dataset was used for verification. Before the experiment, sampling of soil (at 0-30 and 30-60 cm depths) and irrigation water were performed and their physicochemical characteristics were measured. The seeds were then seeded in the amount of 78430 seeds per hectare. Up to four or five leaves stage, irrigation was performed based on 100% water requirement of the plant. From this stage, irrigation treatments were applied for each treatment. At the end of the growing season, to remove the marginal effect, plants were harvested in two square meters of the middle of each plot, and then grain yield and biomass of each treatment was measured. To evaluate the simulated yield of SALTMED and AquaCrop models with actual yield, the normalized root mean square error (NRMSE), root mean square error (RMSE), mean basin error (MBE), modeling efficiency (EF) and d-index (d) statistics were used. Results and discussion The minimum difference between simulated values with SALTMED model and actual yield was obtained in I3S2 and I2S2 treatments with 3.5 and 3.9 percents, respectively. The maximum difference between these values was observed in I2S0 and I1S2 treatments with 16 and 15 percents, respectively. Nasrollahi et al. (2016) reported the maximum difference of 4.8% between simulated values by SALTMED model and actual values. The average of simulation error in this study was 9%, which is acceptable according to the results published by other researchers. Similar results for the AquaCrop model showed that the maximum and minimum differences between the simulated and measured values belongs to I2S2 and I1S0 treatments, respectively, with 23 and 1.3 percents, respectively. Results of NRMSE, RMSE, MBE, EF and D statistics for SALTMED model were 0.126, 0.587, 0.061, 0.92 and 0.91, respectively and for AquaCrop model were 0.155, 0.721 ton/ha, -0.090, 0.88 and 0.91 respectively. NRMSE for the AquaCrop model was equal to 0.155, which showed a higher value than the SALTMED model and this result is consistent with the results of Hassanli, Afrasiab and Ebrahimian (2015). The R2 statistic for the SALTMED and AquaCrop models was 0.93 and 0.91, respectively. These results further showed that the SALTMED and AquaCrop models were, respectively, overestimated and underestimated the real conditions. Conclusions In this study, SALTMED and AquaCrop models were evaluated for simulating Sweet maize yield under deficit irrigation and superabsorbent applications. The results showed that although SALTMED model performed a better accuracy than AquaCrop model, but the efficiency of both models was fairly acceptable in simulating sweet maize yield under the two applied experimental treatments.