Evaluation of the SALTMED model to simulate potato yield and water productivity under different irrigation water management in the Shahrekord

Introduction
Potato is one of the most important food sources in the world and is considered one of the basic sources in different countries. Iran is the 13^th producer of this product with the production of about five million tons of potatoes. The yield of potatoes is high compared to other crops, and for this reason, it has a higher water efficiency than other crops. However, potato yield is strongly dependent on the amount of irrigation water. This issue has caused special attention to be paid to the amount of irrigation water and many studies have been conducted on determining the optimal irrigation water. However, providing the right amount of irrigation water in each climate and irrigation system requires many experiments. These tests require spending a lot of time and money, which cannot be done in research centers in the current situation. To solve this problem, various plant models have been presented. The SALTMED model is such tool, accommodate different plants, soils, irrigation systems, irrigation management solutions, water qualities, and environmental stresses.
 
Materials and Methods
To carry out this research, the data collected from two research projects were conducted (the first project in 2012 and 2013, the second project in 2017 and 2018) in Shahrekord's research station. In the research projects, to manage irrigation in potato cultivation, two irrigation factors were investigated: the first factor includes different irrigation methods (S: drip, Su: subsurface drip, and F: furrow), and quantitative management of irrigation water as the second factor (FI: providing 100% of water needs, RDI80: providing 80% of water needs and RDI65: providing 65% of water needs) considering three repetitions. Data collected from the first year were used to calibrate the SALTMED model. The area of the plots was 40 m^-2. Within each plot, four rows were planted with a row spacing of 75 cm and a seed spacing of 20 cm. The subplots were separated by one meter. In the surface drip irrigation method,  16mm drip pipes with an average flow rate of 1.75 L h^-1 were used for each crop row. In subsurface drip irrigation, 16mm drip pipes with an average flow rate of 1.85 L h^-1 were installed at a depth of 20 cm in the soil for each crop row. To evaluate the SALTMED model at this stage, the statistics of root mean square error (RMSE), normalized root mean square error (NRMSE), mean bias error (MBE), model efficiency (EF), agreement index (d), and determination of coefficient (R²) were used. Validation of this model was done using second-year data.
 
Results and Discussion
The highest and lowest differences between the observed and simulated values were 4.2 and 1.4 tons per hectare, respectively. The average difference between the observed and simulated values was 2.7 tons per hectare. The differences between the observed and simulated yields for drip, subsurface drip, and furrow irrigation methods were respectively 3, 1.9, and 3.1 tons per hectare. In addition, the accuracy of this model for simulating yield in the furrow irrigation method was lower than the other two methods. The highest and lowest difference between observed and simulated water productivity were 0.3 and 1.3 kg.m^-3, respectively. The average of this difference was determined to be 0.7 kg.m^-3. The difference between observed and simulated water productivity for drip irrigation, subsurface drip, and furrow irrigation methods was 0.8, 0.5, and 0.8, respectively. Based on the values of MBE statistics, the SALTMED model had an underestimation error in simulating yield and water productivity. The results of the NRMSE statistic showed that the accuracy of the SALTMED model for yield simulation was in the excellent category. The efficiency of the SALTMED model for yield simulation was acceptable based on two statistics EF and d. The R² statistic for the simulation of potato yield by this plant model ranged from 0.88 to 0.99.
 
Conclusion
This research evaluated the SALTMED model for simulating potatoes' yield and water productivity under three irrigation methods: furrow, drip, and subsurface drip. Using the recalibrated SALTMED model, a simulation of yield and water productivity was done to meet the water needs of 90, 55, and 45 % of potatoes for all three irrigation methods. A downward trend was observed between the reduction of water supply and potato yield in all three irrigation methods. The yield differences between the 100 and 90 %, and 90 and 80 % water supply in surface irrigation were 4.8 and 5.7 %, respectively. These values were respectively 18.6 and 8.7 % for drip irrigation and 13.3% and 1.7% for subsurface drip irrigation. Therefore, the slope of yield reduction until providing 80% of the water requirement was low in all three irrigation methods. The results showed that the SALTMED model had an underestimation error for simulating both parameters (MBE<0). However, the obtained error is negligible and the accuracy of this plant model was in the excellent category (NRMSE<0.1). Based on the results of EF (<0.88) and d (<0.99) statistics, the effectiveness of the SALTMED model was favorable for simulating both yield and water productivity. To determine the optimal amount of irrigation water, the recalibrated SALTMED model was used. The results showed that providing 80% of the potato's water requirement led to the achievement of an optimal yield and high water productivity.