To evaluate the performance of DSSAT and AquaCrop models in the estimation of soybean and grain maize yield under water stress conditions in a semi-arid region.
Area of study:
Kermanshah, Iran.
Material and methods:
AquaCrop and DSSAT were assessed to simulate soybean and maize. Both models were calibrated using field data. Field experiments were performed in a randomized complete block design with eight and four irrigation treatments for soybeans and maize, respectively with three replications. Measures of Normalized Root Mean Square Error (nRMSE) and Nash-Sutcliffe Model Efficiency were used to evaluate the accuracy of the models. For this purpose, simulated values of leaf area index / green crop canopy, grain yield, biomass, and soil moisture were compared with measured data.
Main results:
Results indicated that the CROPGRO-Soybean in DSSAT software simulated more accurate crop growth of soybean than AquaCrop. The average nRMSE of the DSSAT model for estimating soil moisture, leaf area index, grain yield, and biomass were 6%, 14%, 16% and 20%, respectively. For maize, AquaCrop simulated crop growth more reliably than CERES-maize. The average nRMSE of 3%, 10%, 13% and 27% of the Aquacrop model in simulating the parameters of soil moisture, green crop canopy, biomass, and grain yield.
Research highlights:
Considering the better performance of AquaCrop for maize and DSSAT for soybean in the study area, it is not possible to propose a specific model to simulate the growth of all crops in a region.
CERES-Maize;CROPGRO-Soybean;crop growth models;crop yield;Kermanshah;soil moisture;DSSAT
(Decision Support System for Agrotechnology Transfer)DW
(dry weight)EF
(Nash-Sutcliffe Model Efficiency)ETa
(actual crop evapotranspiration)ETc
(potential crop evapotranspiration)ETo
(reference crop evapotranspiration)HI
(harvest index)LAI
(leaf area index)nRMSE
(normalized root mean square error)
(Tables S1-S3) accompanies the paper on SJAR’s website.
The authors received no specific funding for this work.Data availability statement: The data that support the findings of this study are available from the corresponding author, upon reasonable request.Introduction
Water is one of the most important crop growth requirements, which is limited in many parts of the world. In conditions of water scarcity, crop water use efficiency must be increased to increase food production. Deficit irrigation is one of the irrigation management methods, especially in the situation of water scarcity, which is also associated with the reduction of crop yield. The incidence of water stress at any stage of crop growth can decrease crop yield, but the severity and magnitude of yield loss depend on the time and severity of the water stress (Brevedan & Egli, 2003). Therefore, it is necessary to study crop yield and water requirements under water stress conditions. These can be studied through field experiments. Since investigating the effects of limiting factors (such as water) on crop growth requires many expensive experiments in different regions, providing a solution to reduce the number of these tests is very important. With advances in modeling science, several crop growth simulation models (e.g., WOFOST, DSSAT, CropSyst, AquaCrop, …) have been developed. Considering the variety of existing models, choosing the most appropriate crop growth simulation model for each crop in each region is essential. Evaluation of irrigation management, especially in water scarcity conditions, is one of the issues that have been studied using crop growth simulation models. Since in this study, the CERES-Maize (Jones & Kiniry, 1986) and CROPGRO (Boote et al., 2018) models in the DSSAT software package (Jones et al., 2003) and the AquaCrop (Raes et al., 2009) have been used, the literature review has been limited to these models.
Decision Support System for Agrotechnology Transfer (DSSAT) is a software package, which was developed as the result of an international effort to facilitate the use of models in agricultural research. Crop growth simulation models such as CERES-Maize, CERES-Wheat, and CROPGRO are at the center of this collection (Jones et al., 2003). DSSAT has been used worldwide for different purposes (e.g., determination of optimum crop management practices, irrigation management, precision agriculture, the impact of climate change on crop yield, etc.) at different spatial and temporal scales (Soltani & Hoogenboom, 2007). Many studies have shown that the CERES-Maize model can reliably simulate phenological development, biomass accumulation, crop yield, nitrate leaching, and nitrogen uptake (Abedinpour & Sarangi, 2018; Malik et al., 2019). Abedinpour & Sarangi (2018) assessed the CERES-Maize model in estimating maize yield using different levels of water and nitrogen. Their results showed that although this model simulated corn yield with acceptable accuracy, it was less accurate in higher water stress treatments. Hammad et al. (2018) studied the accuracy of CERES-Maize in Faisalabad, Pakistan, and concluded that this model accurately simulated phenological development, grain yield, and final biomass production of maize under various water regimes and nitrogen rates.
The acceptable accuracy of the CROPGRO-Soybean model in the simulation of soybean phenological stages and yield has been confirmed in various studies (Kumar et al., 2008; Ovando et al., 2018; Teixeira et al., 2019). CROPGRO-Soybean also was used to assess the impacts of water stress on soybean yield. Sharda et al. (2019), after calibration and validation of CROPGRO-Soybean in their study area, have used the calibrated model to investigate the effect of irrigation depth on crop yield and water productivity. Although several studies show satisfactory results of CROPGRO-Soybean under water stress conditions, examples of model inefficiencies in these conditions have also been reported (Dogan et al., 2007).
AquaCrop is a crop-water productivity model to simulate yield response to water developed by FAO (Food and Agriculture Organization of the United Nations). Concepts and basic principles of the AquaCrop model were described by Steduto et al. (2009), and its main algorithms and software description were presented by Raes et al. (2009). This model is derived from the equation of Doorenbos & Kassam (1979). The ease of use of the AquaCrop model due to the need for low input parameters and sufficient simulation accuracy has made this model a valuable tool for simulating crop growth under different irrigation scenarios (Heng et al., 2009). AquaCrop was developed in 2009 and since then it has been used worldwide in different agro-ecological conditions to simulate the yield response of various crops to water such as wheat (Andarzian et al., 2011; Mkhabela & Bullock, 2012), barley (Hellal et al., 2019), cotton (García-Vila et al., 2009), sugar beet (Malik et al., 2017), saffron (Mirsafi et al., 2016) and many other crops (Araya et al., 2010). This model is also used for the simulation of soybean and maize. The use of AquaCrop in the simulation of maize growth and yield in Uganda (Mibulo & Kiggundu, 2018) and maize responses to different nitrogen stresses under the semi-arid climate in Iran (Ranjbar et al., 2019) have also been reported. Paredes et al. (2015) used the AquaCrop model to simulate soybean growth in North China. Based on the results, they claimed that the model would predict yield and biomass well if adequately parameterized. Adeboye et al. (2019), based on their study, concluded that the AquaCrop model is adequate in simulating canopy cover, dry above-ground biomass, and grain yield of rainfed soybean in Ile-Ife, Nigeria. Heng et al. (2009) used the AquaCrop model to simulate maize growth in three regions with different conditions (Spain, Texas, and Florida). Their results showed that the AquaCrop model could satisfactorily simulate crop water consumption under high evapotranspiration and wind speed. Although the accuracy of this model has been reported to be adequate in many cases, there are examples of inadequate accuracy of this model in the literature, especially in moderate to severe water stress conditions. Sandhu & Irmak (2019) used the AquaCrop model to simulate maize evapotranspiration, growth, yield, and also water use efficiency, on different planting dates under irrigated, and rainfed conditions in Nebraska, USA. Based on their results, the accuracy of the model was lower under arid climate conditions; the model overestimated crop yield and underestimated potential evapotranspiration; as a result, water use efficiency was overestimated. Ahmadi et al. (2015) reported that the accuracy of AquaCrop is insufficient to simulate maize yield under moderate to severe water stress conditions.
The performance of crop growth simulation models was compared in several studies. Battisti et al. (2017) assessed the accuracy of five crop growth simulation models (including AquaCrop and DSSAT) in the simulation of soybean. Their results showed that most of the models still have difficulties in simulating soybean yield accurately under severe water deficit conditions. Babel et al. (2019) evaluated the performance of AquaCrop and DSSAT-CERES for maize in the higher Himalayan region of India. Although their results showed that both models were suitable, they recommended the AquaCrop model because of its low data requirement. Castañeda-Vera et al. (2015) assessed the usefulness of Aquacrop, CERES-Wheat, CropSyst, and WOFOST in wheat simulation. They concluded that CERES-Wheat and CropSyst are preferable for growing wheat in semi-arid conditions.
In this study, the crops of soybean (Glycine max L.) and grain maize (Zea mays L.) were studied. According to the statistics published on the Information and Communication Technology Center website of the Iranian Ministry of Agriculture Jahad (2016-2022), corn and soybeans rank first to third among Iran’s agricultural products (https://amar.maj.ir/page-amar/FA/65/form/pId28832). More than 90% of edible oil consumed is imported into Iran. Soybean, which is one of the primary sources of high-quality oil and protein, is commonly used in human food consumption (Sinclair et al., 1991). Maize is one of the most widely used agricultural products in Iran. Its production does not meet the domestic demand. Maize is a lucrative and favorite crop for farmers in the study area, which of course has high crop water consumption. Due to the water shortage in recent years, the area underplanting of maize has decreased. In some areas of Kermanshah province, farmers have been directed to plant soybean instead of maize. Although AquaCrop and DSSAT models have been used in different parts of the world to estimate the yield of corn and soybean under water stress, the performance of these models has not been evaluated in the studied area. This study aimed to evaluate the performance of a specific model (packages of CERES-Maize and CROPGRO-Soybean in DSSAT) and a generic model (AquaCrop) in the estimation of soybean and grain maize yield under water stress conditions in a semi-arid region in the west of Iran.
Material and methodsCrop growth simulation models
Crop growth simulation models differ in terms of data requirements, the type of plants that can be simulated, and the level of complexity. Some models are developed to simulate a specific crop, while others can be used for a variety of crops. In this study, a specific model (DSSAT) and a generic model (AquaCrop) were used to simulate the growth of soybean and maize under water-limited conditions. Although these models have been successfully used in different parts of the world to study the crop yield response to water stress (Adeboye et al., 2017; Malik et al., 2017; Babel et al., 2019; Hellal et al., 2019; Sandhu & Irmak, 2019), they have not yet been investigated in the study area. The basic principles and manual of these models are available on the website of these software developers as well as in the literature (Jones et al., 2003; Steduto et al., 2009; Hoogenboom et al., 2019). Therefore, in this article, only a brief explanation of these models has been given.
AquaCrop model
AquaCrop is a crop growth simulation model which simulates crop biomass and harvest regarding water availability (Steduto et al., 2009). The basic equation of this model is based on Doorenbos & Kassam’s (1979) approach in the FAO 33 report (Eq. 1).
where ETa: actual crop evapotranspiration (mm), ETc: potential crop evapotranspiration (mm), Ya: actual crop yield (corresponding to ETa) (kg/ha), Ym: maximum (theoretical) crop yield (corresponding to ETc) (kg/ha) and Ky: crop yield response factor to water deficit (-). AquaCrop estimates daily biomass production using Eq. 2:
where Bi is the daily aboveground biomass, Tri is the daily crop transpiration, EToi is the daily reference evapotranspiration, and WP* is the water productivity of the crop species normalized for both evaporative demand and atmospheric CO2.
Crop yield was obtained by multiplying aboveground biomass and specified harvest index (HI) using Eq. 3 (Steduto et al., 2009):
where Y is the main yield (kg/ha), B is aboveground biomass (kg/ha), and HI is the harvest index (%). The main components of the AquaCrop model and their relationships are presented in Steduto et al. (2009).
DSSAT model
DSSAT is a software application program that can be used for the simulation of more than 40 crops, including maize and soybean. The DSSAT is a collection of independent programs that operate together; crop simulation models are at its center. In this software package, there is a specialized model for each crop. CERES-Maize and CROPGRO-Soybean are examples of these models, respectively, for growth simulation of maize and soybean. The crop simulation models simulate growth, development, and yield as a function of the soil-plant-atmosphere dynamics on a daily basis. The tools include database management programs for soil, weather, crop management and experimental data, utilities, and application programs (Jones et al., 2003).
Study area
The field experiments were conducted in two research farms. The study of 2011 was conducted at the research farm of the Agricultural Organization in Mahidasht, and the studies of 2012 and 2015 were conducted at the research farm of Razi University in Kermanshah, Iran (Fig. 1). The geographical location of the research fields and the average annual rainfall and temperature of Kermanshah and Mahidasht are presented in Table S1. Average monthly weather parameters during the implementation of research projects are presented in Table S2.
Location of the experimental fields at Kermanshah province, Iran.
The physical characteristics of the soil of the research fields are presented in Table S3. The texture of the soil on the farm of Razi University was finer than that of the Mahidasht Agricultural Organization. The water holding capacity in the research farm of Razi University and the Agricultural Organization was 144 and 104 mm/m, respectively.
Field and laboratory experiments
For each crop, two years of field data were used to calibrate and validate the crop growth simulation models (Table 1). The models were calibrated based on field data from two M.Sc. research projects (Mirzaee, 2013; Esmaili, 2014) and validated based on field data from an M.Sc. (Ahmadpour, 2013) and a Ph.D. research project (Dehghan Moroozeh, 2019), which all were carried out under the supervision of the corresponding author of this paper. The four studies were conducted in the form of a randomized complete block design with a certain number of irrigation treatments and replications. To eliminate marginal effects, a distance of 3 m was considered between the plots. Except for Mirzaee’s study, which was conducted in Mahidasht, 25 km away from Razi’s research farm, three other experiments were conducted on the research farm of Razi University, Kermanshah, Iran. In the following, the studies of Mirzaee (2013) and Esmaili (2014) will be described briefly, and the studies of Ahmadpour (2013) and Dehghan Moroozeh (2019) will be described in detail.
Specifications of field experiments used for calibration and validation of models.
Crop
Year of experiment
No. of treatments
Measured parameters
Reference
Application in this study
Maize
2011
3
DW of leaves, stems, and storage organ and LAI
Mirzaee (2013)
Calibration
2012
4
DW of leaves, stems, and storage organs, LAI, crop canopy cover, and soil moisture
Ahmadpour (2013)
Validation
Soybean
2012
4
DW of leaves, stems, and storage organs, LAI and crop canopy cover
Esmaili (2014)
Calibration
2015
8
DW of leaves, stems, and storage organs, LAI, crop canopy cover, and soil moisture
Dehghan Moroozeh (2019)
Validation
DW: dry weight. LAI: leaf area index.
Mirzaee (2013) conducted a field experiment in 2011 to calibrate crop parameters of the WOFOST (de Wit et al., 2019) model for grain maize. In this experiment, maize was cultivated in three irrigation treatments (100%, 80% and 60% of irrigation requirement) with three replications. This experiment was conducted in the research farm of the agricultural organization in Mahidasht, Kermanshah, Iran, located 25 km away from the Faculty of Agriculture, where other studies were conducted. The altitude of this station is 1365 masl. The weather parameters in this study were given from a weather station which was located in the region. During the growth period, the dry weight (DW) of leaves, stems, and storage organs and leaf area index (LAI) were measured weekly in each plot. The average biomass, grain, and LAI in each irrigation treatment were used for model calibration.
Esmaeili (2014) conducted a field experiment in 2012 to calibrate the crop parameters of AquaCrop for soybean. This experiment was conducted in four irrigation treatments (120%, 100%, 80%, and 60% of irrigation requirement) with four replications. In this experiment, the DW of leaves, stems, and storage organs, LAI, and green canopy cover were measured every ten days. The average of crop parameters in each irrigation treatment was used for calibration of crop models of AquaCrop and DSSAT for soybean.
The experiment conducted in Ahmedpour’s study (Ahmadpour, 2013) was performed in four irrigation treatments with three replications for maize (Table 2). In this experiment, 12 plots with a size of 5 × 6 m were cultivated with a row distance of 75 cm (Fig. 2a).
Schematic map of experimental treatments for a) maize and b) soybean.
Specifications of the irrigation treatments in the studies of Ahmadpour (2013) and Dehghan Moorozeh (2019).
Crop
Treatment
Irrigation depth (% of crop water requirement)
Period of deficit irrigation
Total depth of applied water (mm)
Reference
Soybean
T1
100
---
842
Dehghan Moorozeh (2019)
T2
120
---
960
T3
80
Whole period
724
T4
60
Whole period
605
T5
80
Vegetative phase
827
T6
60
Vegetative phase
797
T7
80
Reproductive phase
746
T8
60
Reproductive phase
650
Maize
T1
100
---
828
Ahmadpour (2013)
Maize
T2
120
---
946
T3
80
Whole period
710
T4
60
Whole period
593
Dehghan Moorozeh (2019) studied the impacts of deficit irrigation on soybean yield and crop water productivity based on his field experiments in 2015. In this study, soybean was cultivated in 24 plots with the size of 4 × 4 m with a row spacing of 50 cm. The number of irrigation treatments and replications was 8 and 3, respectively (Table 2 and Fig. 2b).
During the growth period, crop parameters required for calibration and validation of crop growth simulation models were measured. Green canopy cover, LAI, DW of above-ground biomass, and grain were measured every seven (only in Mirzaee’s study) or ten days. In each sampling, three plants from each plot were harvested at predetermined locations and transported to the laboratory. In the laboratory, different parts of the plant (leaf, stem, and grain) were separated to measure each part separately. The leaves were scanned using a scanner (Epson Perfection V39 Scanner) with a resolution of 4800 dpi. The area of green leaves was calculated using image analysis. The LAI was calculated as the ratio of green leaf area per area allocated to each plant. Leaves, stems, and grains of the harvested plants were dried in the oven at 70 °C for 48 h and then weighed. The sum of the DW of leaves, stems and grain was considered as total above-ground biomass. In this paper, the measured and simulated crop yield (biomass and grain) are reported in terms of DW. The percentage of green canopy cover was estimated based on the analysis of photos taken over the top of the crop canopy.
Soil moisture is a key output of water-related crop growth simulation models. Unfortunately, this parameter was not measured in the studies of Mirzaee and Esmaeili. But, this parameter was measured in the studies of Ahmadpour and Dehghan Moorozeh which were used for validation of models. The soil moisture was measured one day before irrigation using the gravimetric method (Johnson, 1962) at depths of 15 and 45 cm for maize (Ahmadpour, 2013) and at depths of 15, 45, and 60 cm for soybean (Dehghan Moroozeh, 2019). Due to laboratory limitations, soil moisture was measured only in T1, T2, T3, and T4 treatments for soybean and T1, T3, and T4 for maize. Moreover, phenological growth stages (e.g., dates of crop emergence, flowering, the start of seed filling, the start of senescence, and maturity) were determined based on daily observations.
Irrigation treatments
Irrigation treatments were defined as a certain percentage of crop water requirement. For this purpose, first reference crop evapotranspiration (ETo) was calculated based on the FAO Penman-Monteith equation (Allen et al., 1998) (Eq. 4 ).
where ETo is the reference crop evapotranspiration (mm/d); ∆ is the slope of saturation vapor pressure (kPa/°C); Rn is the daily net radiation (MJ/m2·d); G is the soil heat flux (MJ/m2·d); γ is the psychometric constant (kPa/°C); Ta is the mean air temperature at 2 m height (°C); U2 is the daily mean of wind speed at 2 m height (m/s); es is the saturation vapor pressure (kPa); and ea is the actual vapor pressure (kPa).
All the weather parameters needed to calculate ETo were obtained from nearby weather stations. Crop evapotranspiration (ETc) was calculated by multiplying ETo by the crop coefficient (Kc) provided in the National Water Document (Agri-PERI, 2017) for the study area.
Since there was no rainfall during the period of application of the treatments, the crop water requirement was considered equal to ETc. Considering the soil physical parameters of the research farm and crop water requirement, the irrigation interval was considered seven days for both crops. In all four studies, deficit irrigation was not applied in the early stages of crop growth to ensure crop establishment. For this purpose, in the first four irrigations, the irrigation interval was considered three days, and the amount of applied water was equal for all treatments. So, the plots were irrigated weekly except in the crop establishment period, in which the irrigation interval was three days. The volume of required water in the control treatment (T1) was calculated based on accumulated crop water requirements since the last irrigation and plot size. The irrigation volume of other treatments was calculated based on the control treatment. The applied water in the treatments of T1, T2, T3, and T4 was 100%, 120%, 80% and 60% of the crop water requirement, respectively. In treatments of T5 and T6, which were applied for soybean, 20% and 40% deficit irrigation was implemented only in the vegetative phase (no water stress was applied in the reproductive stage). In treatments of T7 and T8, deficit irrigation was implemented only in the reproductive stage (Table 2). The reason for choosing T2 treatment (120% of crop water requirement) was the uncertainty of the estimated ETo by the Penman-Monteith formula in the study area (Esmaeili et al., 2015; Ahmadpour et al., 2017). The required amount of water was applied using a hose connected to a flowmeter with an accuracy of 0.1 liters (Furrow irrigation). The crop management (e.g., land preparation, fertilizer, weed & pest management) was implemented under the supervision of an agronomist.
Calibration and validation
Crop, weather/climate, soil, and management files are required as model inputs for both models. Soil files were created based on the soil physical properties of the research farm (Table S3). Daily weather data in the study area was used to prepare the weather/climate files. For each treatment, a management file that contains the time and depth of irrigation was also created. Crop files/parameters are required to run the models. Some of the crop parameters are variety-specific and should be determined or calibrated.
In the calibration process, first, the phenological dates were set based on the full irrigation treatment. Then the crop parameters were changed by trial and error so that there was a slight difference between the simulated parameters and the measured average in all treatments. Three parameters were used to calibrate the models. The average grain yield, biomass, and crop canopy cover in each treatment and sampling day were used for AquaCrop calibration. For DSSAT, the LAI was used instead of crop canopy cover.
Field data collected by Ahmadpour (2013) and Dehghan Moroozeh (2019) were used to validate the calibrated models. According to the measurement of soil moisture in the latest studies, soil moisture was used in addition to the mentioned parameters in the validation process. The simulated biomass, grain yield, and LAI/green canopy cover at sampling dates were compared with measured data graphically and statistically. Although graphical comparisons were made for the three crop parameters, only graphs related to biomass are presented in the paper. Usually, statistical indices are used for the assessment of model validation. Yang et al. (2014) evaluated a group of deviation statistics that are commonly used for crop model calibration and validation. In this study, statistical indices of Normalized Root Mean Square Error (nRMSE) and Nash-Sutcliffe Model Efficiency (EF), which are among the suggested deviation statistics by Yang et al. (2014), were used to model evaluation (Eqs. 5 and 6):
where Oi is observed data, Pi is simulated data by the model, Oave is the mean of the observed data.
Lower values of nRMSE and higher values of EF indicate better performance of the model. The range of nRMSE is between 0 and 100%. Jamieson et al. (1991) rated model performance based on nRMSE as excellent (< 10%), good (between 10% and 20%), fair (between 20% and 30%), or poor (> 30%).
Results and discussion
DSSAT and Aquacrop models were calibrated based on the measured grain yield, biomass, and crop canopy/LAI in the studies of Mirzaee (2013) and Esmaeili (2014). First, the phenological parameters were determined based on field observations and weather data. Then, by trial and error, other crop parameters were determined to have the minimum difference between the measured and simulated grain yield, biomass, and crop canopy/LAI at the time of field measurements for all treatments. The calibrated crop parameters for both models and crops are reported in Tables 3, 4, and 5.
Calibrated crop parameters of soybean and maize (AquaCrop model).
Parameter
Unit
Soybean
Maize
Comments
Percentage of canopy cover decline rate
% per day
4
7
Calibrated
Percentage of canopy cover growth rate
% per day
9.4
13.4
Experimental
Upper threshold of soil available water that stresses canopy cover development
-
0.65
0.45
Calibrated
Lower threshold of soil available water that stops canopy cover development
-
0.15
0.15
Calibrated
Shape factor of stress on leaf development
-
1.5
3
Calibrated
Upper threshold of soil available water that stresses the crop by closing the stomata
-
0.27
0.20
Calibrated
Shape factor of stress after stomata closure
-
1.9
2
Calibrated
Upper threshold of soil available water that leads to stress or early senescence
-
0.35
0.28
Calibrated
Shape factor of stress after early senescence
-
1.77
6
Calibrated
Upper threshold of soil available water that leads to stress in flowering stage
-
0.78
0.76
Calibrated
Maximum crop coefficient
-
1
1.1
Calibrated
Base temperature below which crop development does not progress
°C
5.0
8.0
Default
Duration from cultivation to emergence
degree-day
90
112
Experimental
Duration from cultivation to flowering
degree-day
530
970
Experimental
Duration from cultivation to start of senescence
degree-day
870
1436
Experimental
Duration from cultivation to physiological maturity
degree-day
1160
1844
Experimental
Duration from cultivation to maximum canopy cover
degree-day
770
988
Experimental
Duration from cultivation to maximum root depth
degree-day
920
1436
Experimental
Duration of flowering stage
degree-day
480
421
Experimental
Maximum harvest index
%
24
38
Experimental
Calibrated genotype parameters of soybean (DSSAT model).
Parameter
Parameter description
Value
CSDL
Critical short day length below which reproductive development progresses with no daylength effect (for shortday plants) (hour)
13.4
PPSEN
Slope of the relative response of development to photoperiod with time (positive for shortday plants) (1/hour)
0.29
EM-FL
Time between plant emergence and flower appearance (R1) (photothermal days)
45
FL-SH
Time between first flower and first pod (R3) (photothermal days)
7
FL-SD
Time between first flower and first seed (R5) (photothermal days)
16
SD-PM
Time between first seed (R5) and physiological maturity (R7) (photothermal days)
34.5
FL-LF
Time between first flower (R1) and end of leaf expansion (photothermal days)
26
LFMAX
Maximum leaf photosynthesis rate at 30 C, 350 vpm CO2, and high light (mg CO2/m2-s)
1.03
SLAVR
Specific leaf area of crop under standard growth conditions (cm2/g)
375
SIZLF
Maximum size of full leaf (three leaflets) (cm2)
180
XFRT
Maximum fraction of daily growth that is partitioned to seed + shell
1
WTPSD
Maximum weight per seed (g)
0.11
SFDUR
Seed filling duration for pod cohort at standard growth conditions (photothermal days)
23
SDPDV
Average seed per pod under standard growing conditions (#/pod)
2.2
PODUR
Time required for crop to reach final pod load under optimal conditions (photothermal days)
10
THRSH
Threshing percentage. The maximum ratio of (seed/(seed+shell)) at maturity. Causes seeds to stop growing as their dry weight increases until shells are filled in a cohort.
77
SDPRO
Fraction protein in seeds (g(protein)/g(seed))
0.41
SDLIP
Fraction oil in seeds (g(oil)/g(seed))
0.21
Calibrated genotype parameters of maize (DSSAT model).
Parameter
Parameter description
Value
Variable
Description
Value
P1
Thermal time from seedling emergence to the end of the juvenile phase (expressed in degree days above a base temperature of 8 ℃) during which the plant is not responsive to changes in photoperiod
280
P2
Extent to which development (expressed as days) is delayed for each hour increase in photoperiod above the longest photoperiod at which development proceeds at a maximum rate (which is considered to be 12.5 h)
0.4
P5
Thermal time from silking to physiological maturity (expressed in degree days above a base temperature of 8 ℃)
850
G2
Maximum possible number of kernels on topmost ear
750
G3
Kernel filling rate during the linear grain filling stage and under optimum conditions (mg/day)
6.8
PHINT
Phylochron interval; the interval in thermal time (degree days) between successive leaf tip appearances
38.9
The results of the field experiments conducted in the studies of Ahmadpour (2013) and Dehghan Moroozeh (2019) were used to assess the accuracy of calibrated models under water stress conditions. Measured grain yield, biomass, canopy/LAI, and soil moisture at two (maize) or three (soybean) depths during the crop growth period were compared with data estimated by crop growth simulation models. Although comparisons were made for all the measured parameters, only graphical comparisons are presented for biomass. Of course, statistical comparisons are provided for all parameters.
The maximum biomass and grain yield of soybean in the T2 treatment (120% irrigation) were 11,180 and 2,140 kg/ha, respectively (Fig. 3), which is higher than in T1 (control treatment). It seems that the crop water requirement of soybean is underestimated by the FAO-Penman Monteith formula. Similar results have been reported by Ahmadpour et al. (2017) and Esmaeili et al. (2015) in the study area. Final biomass in treatments T1, T5, and T6 had differences lower than 10%. The reason is related to small differences between the amount of applied irrigation in these treatments (Table 2) as irrigation treatments were not implemented during the crop establishment period, which is part of the vegetative stage. The lowest yield was observed in treatment T4 (40% deficit irrigation in the whole crop growth period).
Measured final biomass and grain yield of soybean (kg/ha).
The measured and simulated soybean biomass for the studied treatments is shown in Fig. 4. Although both models express soybean biomass production trends relatively well, DSSAT shows a better estimation in all treatments than AquaCrop. In water stress treatments (especially in those with stress in the reproductive phase), the deviation between the measured soybean biomass and the simulated by the AquaCrop model increased. Giménez et al. (2017) reported poor accuracy of AquaCrop in the simulation of grain yield of soybean. Adeboye et al. (2017) showed that the AquaCrop model performed better in simulating the dry biomass of soybean for full irrigation than for deficit irrigation.
Measured and simulated soybean biomass with DSSAT and AquaCrop models.
Statistical indices of nRMSE and EF were calculated for soybean crop parameters for all treatments (Table 6). The average of statistical indices indicated the better performance of DSSAT in the simulation of soybean growth than AquaCrop. The average of nRMSE indicates good performance (Jamieson et al., 1991) of the DSSAT model in the simulation of LAI, biomass, and grain (Table 6). The same results are obtained based on the EF index. The average EF for DSSAT and AquaCrop models were in the range of 0.92–0.97 and 0.54–0.75, respectively. Considering statistical analysis, it can be concluded that DSSAT performed better than AquaCrop in the simulation of soybean growth. Similar results were also reported by Battisti et al. (2017), who found a better performance of DSSAT than AquaCrop in terms of soybean grain estimation. The result of this study is also in agreement with Kumar et al. (2008) and Ovando et al. (2018), who reported acceptable accuracy of CROPGRO in the simulation of soybean crop growth and yield. The performance of the models was lower in T4, where the crop was under severe water stress.
Statistical indicators in estimating soybean crop parameters by DSSAT and AquaCrop models.
Statistical index
Treatment
Biomass
Grain
LAI
Crop canopy
DSSAT
AquaCrop
DSSAT
AquaCrop
DSSAT
AquaCrop
nRMSE
T1
10.7
17.2
15.0
34.4
14.9
17.1
T2
21.1
24.8
8.5
26.2
11.9
17.5
T3
21.7
13.8
24.7
27.9
16.0
23.0
T4
25.3
62.3
28.3
45.9
14.4
18.4
T5
22.2
35.6
11.7
41.9
13.6
32.7
T6
17.4
24.3
8.4
31.5
7.7
29.6
T7
23.4
26.4
7.5
28.1
15.7
17.4
T8
19.7
38.6
23.8
35.6
17.2
22.5
Average
20.2
30.4
16.0
33.9
13.9
22.3
EF
T1
0.97
0.89
0.97
0.48
0.94
0.75
T2
0.92
0.86
0.98
0.72
0.96
0.76
T3
0.94
0.87
0.96
0.54
0.94
0.59
T4
0.88
0.45
0.96
0.45
0.94
0.69
T5
0.92
0.83
0.98
0.68
0.95
0.55
T6
0.95
0.88
0.98
0.6
0.99
0.41
T7
0.91
0.79
0.98
0.44
0.92
0.79
T8
0.93
0.40
0.97
0.41
0.92
0.46
Average
0.93
0.75
0.97
0.54
0.95
0.63
nRMSE: normalized root mean square error. EF: Nash-Sutcliffe model efficiency. LAI: leaf area index.
The statistical indices of nRMSE and EF for soil moisture at depths of 15, 45, and 60 cm in the soybean experiment are reported in Table 7. The lower nRMSE and higher EF in the DSSAT model indicate the better performance of this model in soil moisture estimation than the AquaCrop model. The results indicated a better estimate of soil moisture at a depth of 45 cm.
Statistical indices in estimating soil moisture (cm<sup>3</sup>/cm<sup>3</sup>) at depths of 15, 45, and 60 cm in the soybean.
Index
Treatment
DSSAT
AquaCrop
15
45
60
Average
15
45
60
Average
nRMSE
T1
5.0
8.1
8.2
7.1
5.3
10.1
10.1
8.5
T2
5.3
5.6
8.6
6.5
5.5
5.8
9.7
7.0
T3
4.7
5.0
7.8
5.8
10.1
10.1
10.1
10.1
T4
8.0
5.0
1.0
4.7
7.9
5.1
5.1
6.0
Average
5.75
5.9
6.4
6.0
7.2
7.7
8.8
7.9
EF
T1
0.72
0.61
0.64
0.66
0.69
0.21
0.46
0.45
T2
0.52
0.78
0.52
0.61
0.48
0.69
0.39
0.52
T3
0.88
0.77
0.73
0.79
0.47
0.66
0.55
0.56
T4
0.40
0.79
0.99
0.73
0.41
0.73
0.76
0.63
Average
0.63
0.74
0.72
0.70
0.51
0.57
0.54
0.54
nRMSE: normalized root mean square error. EF: Nash-Sutcliffe model efficiency.
The highest and lowest maize yield was observed respectively in T2 and T4. As in soybean, the maximum crop yield of maize was observed in T2 (120% irrigation). Considering the underestimation of crop water requirement by FAO-56 Penman Monteith equation for both studied crops (soybean and maize), the initial assumption regarding the definition of 120% irrigation treatment was reasonable. Therefore, it is suggested that in cases where crop water requirement is estimated using FAO-56 Penman Monteith equation, treatment of 20% of over-irrigation is recommended.
The measured and simulated maize biomass by DSSAT and AquaCrop is presented in Fig. 5. These graphs show the better simulation of maize biomass by the AquaCrop model. Although the biomass simulated by the DSSAT model is accurate in T2 treatment, the accuracy of this model is lower in other treatments.
Measured and simulated biomass of maize in the validation stage for treatments T1-T4 with DSSAT and AquaCrop models.
The statistical analysis between measured and simulated parameters also indicated the better overall performance of AquaCrop in the simulation of maize than CERES-Maize in the study area (Tables 8 and 9). These results are consistent with Ziaii et al. (2014) and Mibulo & Kiggundu (2018). The accuracy of both models in simulating maize grain yield was lower than biomass and LAI/crop canopy cover (Table 8), which is in conformity with the results of Liu et al. (2011) and is inconsistent with other studies such as Heng et al. (2009). The reason could be due to the measurement error of the grain in the early stage of the grain formation period. The comparison of EF model efficiency in Tables 8 and 9 shows the lower efficiency of both models in estimating soil moisture compared to biomass, grain yield, and LAI. Since soil moisture was not measured in the studies used for model calibration, the soil moisture results should be considered with caution (for both crops).
Statistical indicators in estimating maize crop parameters by DSSAT and AquaCrop models.
Statistical index
Treatment
Biomass
Grain
LAI
Crop canopy
DSSAT
AquaCrop
DSSAT
AquaCrop
DSSAT
AquaCrop
nRMSE
T1
22.4
9.3
15.8
33.8
20.8
7.8
T2
10.6
10.1
21.4
25.6
16.4
6.0
T3
37.3
18.4
20.0
18.7
22.2
12.1
T4
35.2
15.0
37.0
30.3
28.3
13.2
Average
26.4
13.2
23.6
27.1
21.9
9.8
EF
T1
0.94
0.99
0.94
0.72
0.91
0.99
T2
0.99
0.99
0.72
0.82
0.94
0.99
T3
0.76
0.94
0.88
0.90
0.88
0.96
T4
0.85
0.97
0.57
0.71
0.82
0.95
Average
0.88
0.97
0.86
0.94
0.89
0.97
nRMSE: normalized root mean square error. EF: Nash-Sutcliffe model efficiency. DW: dry weight. LAI: leaf area index.
Statistical indices in estimating soil moisture at depths of 15 and 45 cm in the maize field.
Index
Treatment
DSSAT
AquaCrop
15
45
Average
15
45
Average
nRMSE
T1
3.0
4.0
3.5
2.9
2.5
2.7
T3
5.1
2.1
3.6
4.7
2.0
3.3
T4
3.4
3.7
3.6
2.9
2.4
2.7
Average
3.8
3.2
3.5
3.5
2.3
2.9
EF
T1
0.72
0.65
0.7
0.73
0.45
0.6
T3
0.50
0.40
0.4
0.58
0.45
0.5
>
T4
0.52
0.31
0.4
0.65
0.65
0.7
Average
0.58
0.45
0.50
0.65
0.52
0.60
nRMSE: normalized root mean square error. EF: Nash-Sutcliffe model efficiency.
Conclusions
Although crop growth simulation models are useful tools for simulating the impact of environmental parameters on crop yield and water requirements, the use of these models is limited in developing countries such as Iran. The need to calibrate the (crop) parameters of these models is one of the limiting factors of their use. Due to the diversity of available models and disagreements about the appropriate model, the limited studies conducted in these areas are scattered and not very practical. In this study, crop parameters of AquaCrop and DSSAT crop growth simulation models were calibrated and validated based on two-year measured yield (biomass and seed), and leaf area/crop canopy index. It is better to use these data to evaluate the performance of other crop models (e.g. WOFOST). Two years of data could be the minimum data to calibrate and validate these models. Therefore, it is suggested that agricultural research centers have a program for the calibration and validation of crop growth simulation models and use multi-year data for this purpose.
In this study, the performance of AquaCrop and DSSAT crop growth simulation models was investigated under full and deficit irrigation conditions. The results showed that in the study area, the overall performance of AquaCrop in estimating crop yield and soil moisture for maize was better than DSSAT, while the performance of DSSAT for soybean was better than AquaCrop. The accuracy of the models in the condition of deficit irrigation was lower than that of full irrigation. Therefore, it is not possible to propose a specific model to simulate the growth of all crops in a region. Choosing the appropriate model depends on the crop type, environmental conditions, and the purpose of the study.
Although determining the crop water requirement has been the focus of many studies for decades, the appropriate method for estimating ETo for many regions has not yet been identified. This issue can also be a limiting factor in the use of crop growth simulation models. In this study, the FAO Penman-Monteith formula, which many researchers believe is the most appropriate method for estimating ETo, was used to determine crop water requirements. The results showed that this formula underestimates the amount of ETo in the conditions of the study area. Therefore, we recommend treatments of 20% over-irrigation in similar studies.
Authors’ contributionsConceptualization: A. Dehghan Moroozeh, B. Farhadi Bansouleh, M. GhobadiData curation: A. Dehghan Moroozeh, A. AhmadpourFormal analysis: A. Dehghan MoroozehFunding acquisition: B. Farhadi BansoulehInvestigation: A. Dehghan Moroozeh, A. AhmadpourMethodology: A. Dehghan Moroozeh, B. Farhadi BansoulehProject administration: B. Farhadi Bansouleh, M. GhobadiResources: B. Farhadi Bansouleh, M. GhobadiSoftware: A. Dehghan Moroozeh, B. Farhadi Bansouleh, A. AhmadpourSupervision: B. Farhadi BansoulehValidation: A. Dehghan MoroozehVisualization: A. Dehghan Moroozeh, B. Farhadi BansoulehWriting – original draft: A. Dehghan MoroozehWriting – review & editing: B. Farhadi Bansouleh, M. GhobadiReferencesAbedinpourMSarangiA2018Evaluation of DSSAT-CERES model for maize under different water and nitrogen levels.Pertanika J Sci Technol26416051618AdeboyeOBSchultzBAdekaluKOPrasadK2017Modelling of response of the growth and yield of soybean to full and deficit irrigation by using Aquacrop.Irrig Drain662192205https://doi.org/10.1002/ird.2073AdeboyeOBSchultzBAdekaluKOPrasadKC2019Performance evaluation of AquaCrop in simulating soil water storage, yield, and water productivity of rainfed soybeans (Glycine max L. Merr) in Ile-Ife, Nigeria.Agr Water Manage21311301146https://doi.org/10.1016/j.agwat.2018.11.006Agri-PERI2017National Water Document. APERDRI, Tehran, Iran [in Persian].AhmadiSHMosallaeepourEKamgar-HaghighiAASepaskhahAR2015Modeling maize yield and soil water content with AquaCrop under full and deficit irrigation managements.Water Resour Manag2928372853https://doi.org/10.1007/s11269-015-0973-3AhmadpourA2013Estimation of maize crop yield under various irrigation management using WOFOST and Aqua Crop models in Kermanshah. MSc. Thesis, Razi University, Kermanshah, Iran. [in Persian].AhmadpourAFarhadi BansoulehBGhobadiM2017Effects of deficit irrigation on growth trend, quantity and quality characteristics of maize in Kermanshah.J Water Soil Resour Conserv6399112[in Persian].AllenRGPereiraLSRaesDSmithM1998Crop evapotranspiration-Guidelines for computing crop water requirements.FAO Irrig Drain Paper 56, FAO, Romep. 300AndarzianBBannayanMStedutoPMazraehHBaratiM2011Validation and testing of the AquaCrop model under full and deficit irrigated wheat production in Iran.Agr Water Manage100118https://doi.org/10.1016/j.agwat.2011.08.023ArayaAKeesstraSStroosnijderL2010Simulating yield response to water of teff (Eragrostis tef) with FAO's AquaCrop model.Field Crops Res1161-2196204https://doi.org/10.1016/j.fcr.2009.12.010BabelMSDebPSoniP2019Performance evaluation of AquaCrop and DSSAT-CERES for maize under different irrigation and manure application rates in the Himalayan region of India.Agric Res82207217https://doi.org/10.1007/s40003-018-0366-yBattistiRSentelhasPCBooteKJ2017Inter-comparison of performance of soybean crop simulation models and their ensemble in southern Brazil.Field Crops Res2002837https://doi.org/10.1016/j.fcr.2016.10.004BooteKJJonesJWHoogenboomG2018Simulation of crop growth: CROPGRO model, agricultural systems modeling and simulation.CRC Press651692https://doi.org/10.1201/9781482269765-18BrevedanREgliD2003Short periods of water stress during seed filling, leaf senescence, and yield of soybean.Crop Sci43620832088https://doi.org/10.2135/cropsci2003.2083Castañeda-VeraALeffelaarPAÁlvaro-FuentesJCantero-MartínezCMínguezM2015Selecting crop models for decision making in wheat insurance.Eur J Agron6897116https://doi.org/10.1016/j.eja.2015.04.008de WitABoogaardHFumagalliDJanssenSKnapenRvan KraalingenD201925 years of the WOFOST cropping systems model.Agric Syst168154167https://doi.org/10.1016/j.agsy.2018.06.018Dehghan MoroozehA2019Yield estimation and water requirements for soybean and maize under climate change by using crop growth simulation models in Kermanshah.Ph.D. Thesis, Razi University, Kermanshah, Iran[in Persian].DoganEKirnakHCopurO2007Deficit irrigations during soybean reproductive stages and CROPGRO-soybean simulations under semi-arid climatic conditions.Field Crops Res1032154159https://doi.org/10.1016/j.fcr.2007.05.009DoorenbosJKassamA1979Yield response to water.FAO Irrig Drain Paper No. 33, Rome257https://doi.org/10.1016/B978-0-08-025675-7.50021-2EsmailiM2014Estimation the effects of water deficit irrigation on soybean crop yield in Kermanshah under climate change scenarios using AquaCrop model.MSc. Thesis, Razi University, Kermanshah, Iran[in Persian].EsmaeiliMFarhadi BansoulehBGhobadiM2015Effects of deficit irrigation on quantity and quality of soybean crop yield in Kermanshah region.J Water Soil293551559[in Persian].García-VilaMFereresEMateosLOrgazFStedutoP2009Deficit irrigation optimization of cotton with AquaCrop.Agron J1013477487https://doi.org/10.2134/agronj2008.0179sGiménezLParedesPPereiraLS2017Water use and yield of soybean under various irrigation regimes and severe water stress. Application of AquaCrop and SIMDualKc models.Water96393https://doi.org/10.3390/w9060393HammadHMAbbasFAhmadAFarhadWAnothaiJHoogenboomG2018Predicting water and nitrogen requirements for maize under semi-arid conditions using the CSM-CERES-Maize model.Eur J Agron1005666https://doi.org/10.1016/j.eja.2017.10.008HellalFMansourHAbdel-HadyMEl-SayedSAbdellyC2019Assessment water productivity of barley varieties under water stress by AquaCrop model.AIMS Agric Food43501517https://doi.org/10.3934/agrfood.2019.3.501HengLKHsiaoTEvettSHowellTStedutoP2009Validating the FAO AquaCrop model for irrigated and water deficient field maize.Agron J1013488498https://doi.org/10.2134/agronj2008.0029xsHoogenboomGPorterCHBooteKJSheliaVWilkensPWSinghU2019The DSSAT crop modeling ecosystem. In: Advances in crop modelling for a sustainable agriculture.Burleigh Dodds Sci Publ173216JamiesonPPorterJWilsonD1991A test of the computer simulation model ARCWHEAT1 on wheat crops grown in New Zealand.Field Crops Res274337350https://doi.org/10.1016/0378-4290(91)90040-3JohnsonAI1962Methods of measuring soil moisture in the field.US Department of the Interior, US Geological SurveyJonesCKiniryJ1986A simulation model of maize growth and development.Texas A & M University Press, College StationJonesJWHoogenboomGPorterCHBooteKJBatchelorWDHuntLA2003The DSSAT cropping system model.Eur J Agron183-4235265https://doi.org/10.1016/S1161-0301(02)00107-7KumarAPandeyVShekhADixitSKumarM2008Evaluation of cropgro-soybean (Glycine max. L. Merrill) model under varying environment condition.Am-Euras J Agr123440LiuHYangJDruryCAReynoldsWTanCBaiY2011Using the DSSAT-CERES-Maize model to simulate crop yield and nitrogen cycling in fields under long-term continuous maize production.Nutr Cycl Agroecosyst89313328https://doi.org/10.1007/s10705-010-9396-yMalikAShakirASAjmalMKhanMJKhanTA2017Assessment of AquaCrop model in simulating sugar beet canopy cover, biomass and root yield under different irrigation and field management practices in semi-arid regions of Pakistan.Water Resour Manag3142754292https://doi.org/10.1007/s11269-017-1745-zMalikWIslaRDechmiF2019DSSAT-CERES-maize modelling to improve irrigation and nitrogen management practices under Mediterranean conditions.Agr Water Manage213298308https://doi.org/10.1016/j.agwat.2018.10.022MibuloTKiggunduN2018Evaluation of FAO AquaCrop model for simulating rainfed maize growth and yields in Uganda.Agronomy811238https://doi.org/10.3390/agronomy8110238MirsafiZSSepaskhahARAhmadiSHKamgar-HaghighiAA2016Assessment of AquaCrop model for simulating growth and yield of saffron (Crocus sativus L.).Sci Hortic211343351https://doi.org/10.1016/j.scienta.2016.09.020MirzaeeM2013Effects of deficit irrigation on maize crop yield in Mahidasht region using crop growing simulation models.Ms.C., Razi University, Kermansh, Iran.[in Persian]MkhabelaMSBullockPR2012Performance of the FAO AquaCrop model for wheat grain yield and soil moisture simulation in Western Canada.Agr Water Manage1101624https://doi.org/10.1016/j.agwat.2012.03.009OvandoGSayagoSBoccoM2018Evaluating accuracy of DSSAT model for soybean yield estimation using satellite weather data.ISPRS J Photogramm Remote Sens138208217https://doi.org/10.1016/j.isprsjprs.2018.02.015ParedesPWeiZLiuYXuDXinYZhangB2015Performance assessment of the FAO AquaCrop model for soil water, soil evaporation, biomass and yield of soybeans in North China Plain.Agr Water Manage1525771https://doi.org/10.1016/j.agwat.2014.12.007RaesDStedutoPHsiaoTCFereresE2009AquaCrop-the FAO crop model to simulate yield response to water: II. Main algorithms and software description.Agron J1013438447https://doi.org/10.2134/agronj2008.0140sRanjbarARahimikhoobAEbrahimianHVaravipourM2019Assessment of the AquaCrop model for simulating maize response to different nitrogen stresses under semi-arid climate.Commun Soil Sci Plant Anal502228992912https://doi.org/10.1080/00103624.2019.1689254SandhuRIrmakS2019Assessment of AquaCrop model in simulating maize canopy cover, soil-water, evapotranspiration, yield, and water productivity for different planting dates and densities under irrigated and rainfed conditions.Agr Water Manage224https://doi.org/10.1016/j.agwat.2019.105753ShardaVGowdaPHMarekGKisekkaIRayCAdhikariP2019Simulating the impacts of irrigation levels on soybean production in Texas high plains to manage diminishing groundwater levels.J Am Water Resour Assoc5515669https://doi.org/10.1111/1752-1688.12720SinclairTRKitaniSHinsonKBruniardJHorieT1991Soybean flowering date: linear and logistic models based on temperature and photoperiod.Crop Sci313786790https://doi.org/10.2135/cropsci1991.0011183X003100030049xSoltaniAHoogenboomG2007Assessing crop management options with crop simulation models based on generated weather data.Field Crops Res1033198207https://doi.org/10.1016/j.fcr.2007.06.003StedutoPHsiaoTCRaesDFereresE2009AquaCrop-The FAO crop model to simulate yield response to water: I. Concepts and underlying principles.Agron J1013426437https://doi.org/10.2134/agronj2008.0139sTeixeiraWWRBattistiRSentelhasPCde MoraesMFde Oliveira JuniorA2019Uncertainty assessment of soya bean yield gaps using DSSAT-CSM-CROPGRO-Soybean calibrated by cultivar maturity groups.J Agron Crop Sci2055533544https://doi.org/10.1111/jac.12343YangJMYangJYLiuSHoogenboomG2014An evaluation of the statistical methods for testing the performance of crop models with observed data.Agric Syst1278189https://doi.org/10.1016/j.agsy.2014.01.008ZiaiiGBabazadehHAbbasiFKavehF2014Evaluation of the AquaCrop and CERES-Maize models in assessment of soil water balance and maize yield.Iran J Soil Water Res454435445https://dorl.net/dor/20.1001.1.2008479.1393.45.4.8.1 [in Persian]