Assessment of the AquaCrop Model for Use in Simulation of Irrigated Winter Wheat Canopy Cover, Biomass, and Grain Yield in the North China Plain

Improving winter wheat water use efficiency in the North China Plain (NCP), China is essential in light of current irrigation water shortages. In this study, the AquaCrop model was used to calibrate, and validate winter wheat crop performance under various planting dates and irrigation application rates. All experiments were conducted at the Xiaotangshan experimental site in Beijing, China, during seasons of 2008/2009, 2009/2010, 2010/2011 and 2011/2012. This model was first calibrated using data from 2008/2009 and 2009/2010, and subsequently validated using data from 2010/2011 and 2011/2012. The results showed that the simulated canopy cover (CC), biomass yield (BY) and grain yield (GY) were consistent with the measured CC, BY and GY, with corresponding coefficients of determination (R2) of 0.93, 0.91 and 0.93, respectively. In addition, relationships between BY, GY and transpiration (T), (R2 = 0.57 and 0.71, respectively) was observed. These results suggest that frequent irrigation with a small amount of water significantly improved BY and GY. Collectively, these results indicate that the AquaCrop model can be used in the evaluation of various winter wheat irrigation strategies. The AquaCrop model predicted winter wheat CC, BY and GY with acceptable accuracy. Therefore, we concluded that AquaCrop is a useful decision-making tool for use in efforts to optimize wheat winter planting dates, and irrigation strategies.


Introduction
Winter wheat (Triticum aestivum L.) is an important staple food crop for the majority of the North China Plain (NCP) population [1]. However, increasing industrial, and domestic water use has resulted in a reduction in water available for irrigation of these crops. Thus there is a growing need for improvement to this region's agriculture water resources management, especially given increasing food demands of the region's increasing population.
It is widely known that well-timed irrigation can substantially increase water use efficiency (WUE) [2,3], providing an optimal growth environment throughout the season [4,5]. In fact, various studies have described several such irrigation strategies for use by farmers [6][7][8][9][10][11][12][13][14][15][16]. Since the mid-1960s, the relationship between water and crop yield has been described with both empirical and mechanistic models [17][18][19][20]. For example, De Wit [21] proposed that a linear relationship between yield and water consumption exists. In contrast, Downey [22], via deficit irrigation studies, suggested that there exists a nonlinear relationship between water and yield. Based on the above studies, the Minhas model [23], Rao model [18], Blank model [24], and the Stewart model [25] were developed. More recently, Wang and Sun [26] showed that a quadratic relationship between crop yield and crop water consumption did in fact exist. Their work was followed by Kang et al. [27] in which a multiple and synergistic model (developed under deficit irrigation conditions) was proposed. At present, the simulation of the soil-plant-atmosphere continuum remains an important part of such research, especially with regard to expansion of the application range of resulting models to a wider array of cropping systems. Therefore, the Food and Agricultural Organization (FAO) developed the AquaCrop model in an effort to meet this need in 2009. This model was originated from the ''yield response to water'' data of Doorenbos and Kassam [28], and evolved to a normalized crop water productivity (NCWP) concept [29]. Compared with other models, AquaCrop is relatively simple to operate by those with little, or no research experience, and allows for simulation of crop performance in multiple scenarios. In addition to a high level of accuracy, this robust model requires a limited set of input parameters, most of which are relatively easy to acquire [29,30]. The AquaCrop model is also capable of predicting crop productivity, water requirements, and water use efficiency under water-limiting conditions [31]. To date, this  model has been successfully tested for cotton [32,33], maize [30,[34][35][36][37][38], wheat [39][40][41][42], sugar beet [37], sunflower [37,43], groundnut [44], potato [45,46], quinoa [16], Teff [47], barley [48,49], green onion [50] and tomato [51] under a wide-range of environments. Previous studies have demonstrated that the AquaCrop model accurately simulates crop canopy cover (CC), biomass yield (BY) and grain yield (GY) under both regular, and deficit irrigation, and in low soil fertility conditions. In such unreliable water-limited environments as the NCP, the AquaCrop model is a potentially valuable tool for use in efforts to maximize this region's winter wheat yield. Therefore, the objective of this study was to validate this model in simulating the effects of planting date, and multiple irrigation scenarios on: (1) canopy cover, (2) biomass yield, (3) grain yield, and (4) water use efficiency of winter wheat in the NCP. These data will provide some guidelines for efforts to optimize irrigation management for winter wheat crops in this region.

Study Site
This field experiments were conducted in the 2008/2009, 2009/2010, 2010/2011 and 2011/2012 growing seasons at the Xiaotangshan experimental site (44.17u N, 116.433uE), Beijing, PR China. This area is representative of the overall soil and crop management practices in this region. The soil is fine-loamy, with a nitrate Nitrogen (NO 3 -N) content of 3.16-14.82 mg kg 21 , an ammonium Nitrogen (NH 3 -N) content of 10.20-12.32 mg kg 21 , an Olsen P of 3.14-21.18 mg kg 21 , an exchangeable K of 86.83-120.62 mg kg 21 , and an organic matter content of 15.84-20.24g kg 21 with in the uppermost 0-30 cm layer. Beijing is characterized by a typical continental climate, with maximum temperatures of 26.1uC in the summer and minimum temperatures of 24.7uC in the winter. Throughout all seasons, the temperature fluctuated daily with significant differences between night and day. During this experimental period, the average annual precipitation was 650 mm, and the frost-free period was on average 180 days.
Local winter wheat cultivars and planting dates are shown in Table 1. Each plot area is 100 m 2 in 2008, 2009 and 2010, and 300 m 2 in 2011. The experiment was designed as a 2-way factorial arrangement of treatments in a randomized complete block design, with three replications for each treatment. Plot management followed local standard practices (weed control, pest management and fertilizer application) for wheat production in this region. The Xiaotangshan Experimental Site belongs to the National Engineering Research Center for Information Technology in Agriculture. It gives some permission for us to study relative agriculture research within this area. We confirm that the field studies did not involve endangered or protected species.

Climate Data Collection and Analysis
Climate data for the experimental site was obtained from the local Xiaotangshan meteorological station. The daily reference evapotranspiration (ET o ) for the growing season from 2008 to 2012 was calculated based on the FAO Penman-Monteith method as described in Allen et al. [52], and the ET o calculator (FAO, 2009) [53]. Daily maximum and minimum temperature, relative humidity, wind speed, rainfall, and total sunshine hours were recorded directly at the Xiaotangshan experimental site. The total rainfall, from sowing to harvest was 199, 208, 145 and 168 mm in    Table 2).

Soil Data of the Experimental Site
The soil at the Xiaotangshan experimental site represents the major soil type (fine-loamy) on which winter wheat is grown in NCP. The soil was at maximum field capacity (27.3% at 0.0-0.1 m, 27.3% at 0.1-0.2 m and 34.8% at 0.2-0.3 m) during sowing and early establishment. The physical soil characteristics were measured directly in the field and used for input into AquaCrop (Table 3). Leaf area index (LAI) was estimated using the following two methods:

Field Experiments and Crop Data Collection
(1) By multiplying the plant population by the leaf area per plant as described in Kar et al. [54]. Area of the leaf was measured manually from 20 plants using a straightedge. Counting of plant populations was conducted manually from a 0.1 m 2 area. The LAI equation is as follows: Where r is plant density, m is the number of measured plants, L ij is leaf length, B ij is the maximum leaf width, and n is the number of leaves of the nth plant.
where CC is canopy cover, and LAI is the leaf area index. Grain yield was measured following maturation from samples obtained from a 1.5 m 2 area in each plot, with three replications for each treatment. Collected grain was dried and weighed on an electronic scale (60.01 g). As there were no significant differences Figure 3. Relationship between the measured and simulated canopy cover (CC) in winter across 4 years. Note: x represents the simulated CC, y represents the measured CC. The intercept represents the relative error between the simulated CC and the measured CC. The slope represents the consistency between the simulated CC and the measured CC. doi:10.1371/journal.pone.0086938.g003 between winter wheat varieties in many of the measured characteristics (e.g. phenological development, canopy cover, etc.), the average grain yield of the different varieties was considered in model simulations.

Water Use Efficiency (WUE)
Water use efficiency (WUE) is defined as the grain yield per unit amount of water consumed [55]. In this study grain water use efficiency (Grain-WUE), and biomass water use efficiency (Biomass-WUE) were calculated using Eq. (3) and Eq. (4), respectively, as in Araya et al. (2010b) [48]: Biomass-WUE~B Where GY is the grain yield kg ha 21 (measured), T is the transpiration as determined using AquaCrop model, and BY is the total final aboveground biomass yield in kg ha 21 (measured).

Description of AquaCrop Model
The AquaCrop model was proposed by the FAO in 2009, with a detailed description presented in Steduto et al. [29], and Raes et al. [31]. The model computes a daily water balance, and separates evapotranspiration into evaporation and transpiration components. Transpiration is correlated with canopy cover, which is proportional to the degree of soil cover, and evaporation is proportional to the area of soil not covered by vegetation. The crop's stomata conductance, canopy senescence, leaf growth, and yield response to water stress are modeled using four stress coefficients (stomata closure, leaf expansion, canopy senescence, and change in harvest index (HI)). The model subsequently estimates yield from the daily crop transpiration values.
In general, the normalized crop water productivity (NCWP) is considered constant for a given climate condition and crop (For crops not nutrient-limited, the model provides categories ranging from slight to severe deficiencies corresponding to lower water productivity (WP).) is applicable for using in different locations, seasons, and even future climates [29]. Depending on the crop, NCWP increases slightly with an increase in atmospheric CO 2 concentration [29]. NCWP is set between 13 and 20 g m 22 for C 3 crops. For example, NCWP is set at 15 g m 22 for the winter wheat according to the AquaCrop Manual (Annex I Section I.10 Wheat, Pages A39-A42) [56,57]. In our current study, we have not included any of the water stress study data; therefore NCWP remained at 15 g m 22 for the winter wheat. The crop's daily aboveground biomass is calculated using NCWP from the AquaCrop model [29,30]. Biomass yield (BY) is calculated by multiplying NCWP by the ratio of crop transpiration (T), and evapotranspiration (ET o ), following calculation of BY (its harvestable portion), and the grain yield (GY) is determined via harvest index (HI).
These changes are described by the following Eqs. (5) and (6): Where BY is biomass yield in kg ha 21 , T is crop transpiration in mm, ET o is evapotranspiration in mm, NCWP is the normalized crop water productivity in g m 22 , HI is harvest index, and GY is grain yield in kg ha 21 .

Data Analysis
Winter wheat canopy cover (CC), biomass yield (BY) and grain yield (GY) in AquaCrop were calibrated using the measured data sets of 2008/2009 and 2009/2010, and validated using the 2010/ 2011 and 2011/2012 measured data sets. The good fit regression equation between the simulated and observed values was corroborated using prediction error statistics. The coefficient of determination (R 2 ), root mean square error (RMSE), and model efficiency (E) were used as the error statistics to evaluate both calibration and validation results. The E and R 2 were used to access the predictive power of the model, and the RMSE indicated the error in model prediction. In this study, the prediction model output for CC, GY and BY during harvest was used for model evaluation. These statistical indices were used to compare measured and simulated values. Model performance was assessed using E (Nash and Sutcliffe, 1970) as follows: where S i and O i are predicted, and observed data, respectively.

AquaCrop Model Calibration and Validation Results
The crop parameters used to calibrate the AquaCrop model are presented in Table 4. Key stress parameters (e.g. canopy growth, canopy senescence stress coefficient) (P upper ) were adjusted as needed to simulate CC. There was a strong linear relationship between the simulated and the measured CC (R 2 Table 5).
The simulated aboveground BY was similar to that measured (Figs. 4a, 4b, 4c and 4d). The stress coefficients were also adjusted, and readjusted as needed to simulated aboveground biomass. There was a strong relationship between measured and simulated BY across the four years ( Fig. 5 and Table 6). The GY was also similar to the measured GY across all four years (R 2 , RMSE and E values of 0.93, 0.52 ton ha 21 and 0.92, respectively) (Fig. 6).
The R 2 , RMSE and E also showed good performance between the simulated and the measured values for CC (R 2

Biomass, Grain Yield and Water Use Efficiency
Winter wheat was planted on Sep. 25th (normal sowing), Sep. 28th (normal sowing), Oct. 5th (late sowing), Oct. 7th (late sowing), Oct. 15th (late sowing) and Oct. 20th (late sowing) during 2008/ 2011. Winter wheat that was planted on Sep. 25th and 28th had greater biomass and grain yield than did those planted on Oct. 5th, 7th, 15th, and 20th (Table 6). There was relatively more transpiration and biomass yield, yet lower grain yield in 2008 than in 2010, but there was relatively more transpiration (T), biomass yield, and even grain yield in 2008 than in 2009 ( Table 7). The highest biomass and grain yield was obtained in crops planted on Sep. 28th 2008 and on Sep. 25th 2011, respectively; the lowest biomass and grain yield was obtained in crops planted on Oct.  (Table 5). A good relationship did exist between GY, BY, and T (R 2 values of 0.57 and 0.71, respectively) (Fig. 7), thus suggesting that T might be used in estimating biomass and grain yield.  . Relationship between the measured and simulated biomass yield (BY) in winter wheat across 4 years. Note: x represents the simulated BY, y represents the measured BY. The intercept represents the relative error between the simulated biomass yield and the measured BY. The slope represents the consistency between the simulated BY and the measured BY. doi:10.1371/journal.pone.0086938.g005 Figure 6. Relationship between the measured and simulated grain yield (GY) in winter wheat across 4 years. Note: x represents the simulated GY, y represents the measured GY. The intercept represents the relative error between the simulated GY and the measured GY. The slope represents the consistency between the simulated GY and the measured GY. doi:10.1371/journal.pone.0086938.g006  (Table 7). This was likely due to the higher growing degree days (accumulated temperature) promoting CC growth, and BY, GY accumulation at the earlier planting date. The degree of CC affects the rate of transpiration and consequently BY and GY accumulation [32]. Therefore, the relatively higher BY and GY required relatively more temperature accumulation. Figure 7. Relationships between the measured grain yield (GY), biomass yield (BY) and transpiration (T) in winter wheat. Note: (a) x represents transpiration, y represents the measured GY. The intercept represents the relative estimation error between transpiration and the measured GY. The slope represents the estimation consistency between transpiration and the measured GY. (b) x represents transpiration, y represents the measured BY. The intercept represents the relative estimation error between transpiration and the measured BY. The slope represents the estimation consistency between transpiration and the measured BY. doi:10.1371/journal.pone.0086938.g007 Biomass, and grain yield WUE decreased with increasing transpiration amount for all four years. This is consistent with that presented in Farahani et al. [32], but is not with that presented in Hedge [58]. In this present study, grain yield WUE ranged from 1.70 to 2.34 kg m 23 , reaching its maximum value on Oct. 15th 2009. Wang et al. [21] reported that grain yield WUE for winter wheat was between 0.7 and 1.3 kg m 23 , and Li et al. [50] reported it to be between 0.93 and 1.51 kg m 23 in WUE. These results are; however, not consistent with Wang et al. [26], and Li et al. [50], in which grain yield WUE was reported to be much greater. It indicated that winter wheat varieties developed faster, resulting in greater yield, leading to improvement in WUE. However, our results are consistent with that found in Fang et al. [59], in which the grain yield WUE ranged from 1.71 to 2.21 kg m 23 . The BY and GY in 2010/2011 and 2011/2012 were higher than in 2008/ 2009 and 2009/2010. This suggested that the frequent irrigation could be used to increase BY, GY, and promote biomass yield WUE and grain yield WUE for winter wheat at drought stages. This might be one of the reasons that the effects of drought on BY and GY were reduced by the frequent irrigation, thereby improving WUE.

Discussion
This study demonstrated that the AquaCrop model could be used to evaluate different planting date and irrigation strategies for winter wheat in NCP. Many others have conducted AquaCrop model application studies under different crops and environment conditions [32,35,37,38,43,49]. These results indicated that AquaCrop model is stable and usable for different crops and environmental conditions. It is therefore plausible to use the AquaCrop model to improve irrigation management strategies that would maximize grain and biomass yield It is important to note that this study was limited to winter wheat in Xiaotangshan experimental site, Beijing, China. A subsequent study is focused on validating this model under deficit irrigation, thereby expanding the extent of this model's application in the future.

Conclusion
This paper demonstrated that the AquaCrop model adequately simulated the CC, BY, and GY of winter wheat under different planting dates and irrigation strategies. The simulated CC agreed well with the measured CC across all 4 years. The R 2 , RMSE, E of CC winter wheat ranged from 0.89 to 0.98, 3.18% to 7.19% and 0.90 to 0.96, respectively. The measured and simulated BY were also closely related. The AquaCrop model calibrated the BY with the prediction error statistics of 0.92, R 2 ,0.98, 1.12, RMSE ,1.84 ton ha 21 and 0.92, E,0.96. The simulated GY was also consistent with the measured GY with the R 2 , RMSE and E values of 0.93, 0.52 ton ha 21 and 0.92, respectively. The results demonstrated that frequent irrigation obviously improved BY, GY, biomass WUE and grain WUE for winter wheat in 2010/ 2011. These results suggest that the AquaCrop model could be used to predict CC, BY and GY of winter wheat with a high degree of reliability under various planting dates and irrigation strategies situations in the North China Plain (NCP).