مقایسه مدل های Aquacrop و SWAP در شبیه سازی رشد و بهره وری آب فلفل دلمه ای تحت تنش شوری و کم آبی

Background and Objective: The use of simulation models is a strategy in agricultural water management and an effective method for predicting the impact of irrigation management and water quality on crop yield and water use efficiency. Given the existence of environmental stresses in each region, crop models must be evaluated and validated locally. Moreover, since investigating environmental and management factors such as irrigation water quantity and salinity, as well as soil amendments, on crop yield and water use efficiency requires costly and time-consuming field experiments, the effects of these factors can be simulated using crop models. In recent years, numerous models have been employed to study the interactions between water, soil, and plants. AquaCrop and SWAP are two examples of such models. These models must be calibrated and evaluated for each specific crop and region. The core of these models is the crop yield response to water use, and they simulate crop yield using climatic, plant, soil, and management variables. Although crop models are among the widely used tools for simulating crop performance under different field management practices, attention to their sensitivity to input parameters can enhance calibration and reduce simulation errors. Sensitivity analysis of variables helps researchers gain sufficient insight into the influence of each parameter and its variation during the calibration stage.
Materials and Methods: This study was conducted over two growing seasons, in the second half of February 2022 and 2023, at the research farm located in Zahak County Located in Sistan and Baluchestan Province. The experiment was laid out as a split-plot design in a randomized complete block with three replications. The treatments included three levels of irrigation water (I1, I2, I3 corresponding to 60%, 80%, and 100% of field capacity) and three salinity levels of irrigation water (S1, S2, S3 corresponding to 1, 3, and 5 dS/m, respectively).The land was initially plowed, disked, and plot arrangements were made. Each plot was 4 meters long and 3 meters wide. A spacing of 1 meter was maintained between subplots and 2 meters between main plots. Each plot contained five planting rows, with the two outer rows considered as borders, and data were collected from the middle rows. The spacing between rows was 75 cm, and between plants within each row was 30 cm. In both years, the first harvest of fruits was carried out in mid-May, and a total of eight harvests were performed. At the end of the harvest season, crop yield was measured, and water use efficiency was calculated. To simulate the effects of salinity and water deficit on yield and water productivity, two models—AquaCrop and SWAP—were used. Data from the first year were used for model calibration, and data from the second year were used for model validation. Model performance and accuracy were evaluated using statistical indices including Root Mean Square Error (RMSE), Mean Bias Error (MBE), Coefficient of Determination (R²), and Model Efficiency (EF). Additionally, sensitivity analysis was performed to determine the models' sensitivity to changes in input parameters and to identify essential data for calibration. For the sensitivity analysis, each input parameter was individually changed by 25% while keeping all other parameters constant, and the model was run under the new conditions. The results of the sensitivity analysis indicated that both models exhibited low to moderate sensitivity to the examined parameters.
Findings: Based on the comparison between measured and simulated values of quinoa yield and water use efficiency, and the calculated statistical evaluation indices, it can be stated that both AquaCrop and SWAP models were able to simulate yield and water use efficiency effectively under varying quantities and qualities of irrigation water. In both models, the discrepancies between the measured and simulated values of yield and water use efficiency were within 10%.The Root Mean Square Error (RMSE), Normalized RMSE (NRMSE), Mean Bias Error (MBE), Coefficient of Determination (R²), and Model Efficiency (EF) for yield during the validation stage using the AquaCrop model were 2.34, 0.16, –1.85, 0.93, and 0.87, respectively. For water use efficiency, these values were 0.36, 0.19, –0.42, 0.89, and 0.81, respectively. For the SWAP model during the validation stage, the RMSE, NRMSE, MBE, R², and EF for yield were 2.89, 0.17, –1.83, 0.92, and 0.84, respectively, and for water use efficiency were 0.43, 0.19, –0.38, 0.89, and 0.82, respectively. Given the lower error indices in the AquaCrop model, it can be concluded that AquaCrop simulated the yield and water use efficiency of bell pepper more accurately than the SWAP model under salinity and deficit irrigation stress conditions. The results of the sensitivity analysis indicated moderate to low sensitivity of both models to the parameters examined. The results of the sensitivity analysis indicated that both models exhibited low to moderate sensitivity to the examined parameters.
Conclusion: Based on the obtained results, it can be concluded that both AquaCrop and SWAP models can be used with acceptable confidence levels to simulate the yield and water use efficiency of bell pepper under various quantitative and qualitative irrigation water treatments. These models can serve as powerful and efficient tools to support farmers, designers, specialists, and agricultural managers in selecting optimal irrigation management strategies. However, the AquaCrop model demonstrated higher accuracy. Hence, given that the AquaCrop model requires relatively fewer input parameters than other crop models and possesses high accuracy and reliability in simulating plant growth, it can serve as an efficient tool for evaluating plant responses to changes in both the quantity and quality of irrigation water.