Simulation of Maize Growth Under the Applications of Brackish Water in Northwest China

Abstract

The objective of this study is to assess the suitability of the AquaCrop model for growing maize using brackish water irrigation in Northwest China. Additionally, this study aims to examine how maize utilizes water in various soil layers when irrigated with varying water qualities. The AquaCrop model was calibrated and verified using experimental data from the years 2022 and 2023 in this research. (1) The findings indicated that the AquaCrop model effectively simulated the canopy cover, biomass, and yield of maize when irrigated with brackish water. The validation year’s R², MAPE, and RMSE values for canopy cover, biomass, and yield of maize were 0.95, 5.36%, and 4.77%, respectively. For biomass, the R², MAPE, and RMSE values were 0.91, 16.61%, and 2.12 t·hm⁻², respectively. For yield, the R², MAPE, and RMSE values were 0.84, 3.62%, and 0.42 t·hm⁻², respectively. (2) Irrigation with water of high mineral content, measured at 1.6 ds/m, as well as with fresh water over the whole reproductive period, resulted in an increased reliance on groundwater for maize cultivation. There was no notable disparity in the usage of various soil layers between the irrigation with alternating freshwater and brackish water. (3) The AquaCrop model simulated the effects of seven different irrigation water quality treatments. It was shown that using water with mineralization levels of 0.5 and 0.8 ds/m resulted in decreased freshwater use without causing a substantial decrease in maize yield and biomass.

1. Introduction

Food serves as the fundamental basis for maintaining global peace and promoting social advancement. Given the fast rise of the worldwide population, it is imperative to enhance food production in order to ensure food security [1]. Maize is a significant staple crop, comprising around 30% of global grain output [2]. The Inner Mongolia Autonomous Region (IMAR) is the primary region in China for maize production and serves as a significant commercial grain base. Approximately 40% of the country’s maize output comes from this region [3,4]. Nevertheless, maize output in the irrigation region on the south bank of the Yellow River in the Inner Mongolia Autonomous Region is constrained by reasons such as limited and irregular rainfall distribution, low temperature, and cold damage [5].

The worldwide shortage of freshwater resources and limited availability of water resources need the development of innovative ways to save water and energy. These strategies should primarily aim to enhance water usage efficiency [6]. Brackish water irrigation is an efficient technique for maximizing water use in a cost-effective manner [7,8,9,10]. Nevertheless, this approach may significantly affect maize cultivation, leading to decreased crop productivity and compromised maize quality [11,12,13].

Crop models may be used to forecast crop yields by considering various irrigation systems and unique soil–climate variables [14]. Non-crop-specific models such as CropSyst [15], STICS [16], and the maize simulation model CERES-maize [17] need the calibration of several parameters to facilitate the simulation process. Furthermore, the scientific approach used in these models does not endorse their utilization by farmers and technicians. The AquaCrop model aims to provide a harmonious combination of precision, straightforwardness, and resilience, with the purpose of offering technicians and managers in irrigated regions an effective tool [18,19]. The model assesses the progression of crop growth and development, including canopy cover and accumulated biomass. It also evaluates soil moisture levels and crop evapotranspiration. Nevertheless, the model lacks the complexity required to accurately predict soil erosion, carbon dioxide emissions to the atmosphere, and nutrient balances. The AquaCrop model is used to mimic the development of many crops, such as cotton [20], maize [21,22], wheat [23], barley [24], quinoa [25], and sunflower [26]. The AquaCrop handbook provides a collection of calibration settings. The model has been widely applied in recent years due to its distinctive simplicity and accuracy. Previous studies on the AquaCrop model primarily focused on the Hetao irrigation area in the Inner Mongolia Autonomous Region. However, there have been fewer studies on optimizing maize irrigation in the irrigation area located on the south bank of the Yellow River in the Inner Mongolia Autonomous Region.

Soil is essential in the water cycle [27]. Water in the soil environment has several crucial functions in the hydrological cycle, including providing nutrients for plants, transporting dissolved and solid substances, and acting as a reservoir [28]. Simultaneously, soil water serves as a vital water supply in dry regions. Additionally, it plays a crucial function in preserving the equilibrium of nearby water resources [29]. Crucially, it serves as the connection between surface water and groundwater. A portion of the soil moisture is released back into the atmosphere by evapotranspiration, while another portion is used to refill the groundwater [30]. The majority of information pertaining to surface hydrological processes may be clarified by the examination of soil water. The phenomenon affects the ratios of rainfall infiltration and runoff, regulates the distribution of water and energy [31], and provides insights into water inputs, storage, and losses. Gaining knowledge about the mechanisms and origins of soil water movement is crucial for comprehending hydrological phenomena and promoting ecological preservation. Soil moisture may vary owing to the combined influence of precipitation, evaporation, terrain, and plant cover. This makes it challenging to accurately determine the quantity and spatial arrangement of soil moisture at various scales [32].

The isotopic composition of soil water may provide valuable insights into several hydrological processes occurring in the soil, such as infiltration, evaporation, transpiration, and percolation. These activities are challenging to study using other methods [33]. The isotopic fingerprints of water in various sections of the soil profile vary as a result of precipitation and groundwater recharge [34]. Stable isotopes of hydrogen and oxygen may provide valuable insights into hydrological processes such as water penetration, evaporation, and plant transpiration, without causing any damage to the soil’s natural structure. This is in contrast to typical hydrological research approaches [35].

This work used the AquaCrop model and hydroxide isotope data to create a brackish water irrigation system for enhancing maize fertility in the irrigation region located on the south bank of the Yellow River in Inner Mongolia. The findings of this study can provide theoretical guidance for the use of brackish water in Northwest China, promoting sustainable agricultural development.

This study combines the AquaCrop model with hydrogen–oxygen isotope tracing technology, offering an innovative approach to simulate and optimize irrigation systems. This method enables a more accurate assessment of the impact of brackish water irrigation on maize growth and water use efficiency. The primary activities we conducted were as follows: (1) Investigation of the impact of brackish water and freshwater irrigation on the soil water and salt content in the root zone of maize. (2) Analysis of the water uptake ratio of maize irrigated with brackish water on various soil layers using hydroxide isotope data in order to determine the trend. (3) Utilization of the AquaCrop model to design an effective brackish water irrigation system specifically for the maize reproductive phase.

2. Materials and Methods

2.1. Overview of the Pilot Area

The trials were conducted in 2022 and 2023 at the East Haixin Irrigation Experimental Station (40°29′32″ N, 109°53′10″ E) in the Dalateqi sector of the south bank of the Yellow River Irrigation District, located in Northwest China (Figure 1). The area has a moderate continental monsoon climate, characterized by an average annual sunlight duration of 3230 h and an average temperature of 6.8 °C. The average rainfall ranges between 260 mm and 340 mm. The mean annual precipitation ranges from 260 mm to 340 mm, but the rate of evaporation is about seven times greater than that of rainfall (Figure 2). The soil layer at a depth from 0 to 100 cm has an average bulk density of 1.426 g per cubic centimeter and a total salt content of 2.85 g per kilogram (

Table 1. Basic physical and chemical properties of the test soil from 0 to 100 cm.

Soil Depth/cm	Type of Soil	Bulk Density/(g·cm−3)	Organic Matter Content/(g·kg−1)	Nitrate Nitrogen Content/(mg·kg−1)	Ammonium Nitrogen Content/(mg·kg−1)	Total Salt Content/(g·kg−1)
0~20	loam	1.441	12.1	6.2	3.5	3.68
20~40	loam	1.446	12.4	6.8	3.6	2.86
40~60	loam	1.428	13.2	7.4	3.1	2.64
60~80	loam	1.42	11.8	6.6	2.8	2.59
80~100	sandy loam	1.396	11.0	5.3	3.0	2.48

). The water table’s depth in this location has consistently stayed between 1.4 m and 1.8 m for an extended period of time.

2.2. Experimental Design

We developed four irrigation strategies: (1) exclusive use of freshwater for irrigation (FW), (2) alternating between one brackish and one freshwater irrigation (1B1F), (3) two consecutive brackish water irrigations followed by one freshwater irrigation (2B1F), and (4) exclusive use of brackish water for irrigation (BW). The mineralization values were 0.3 ds/m for freshwater and 1.6 ds/m for brackish water. Each treatment received the same irrigation frequency and volume.

Table 2. Water allocation throughout the maize growing season.

Year of Experiment	Treatment	Irrigation Time	Irrigation Quota (mm)	Total Irrigation Quota (mm)	Use of Brackish Water	Freshwater Usage
2022	FW	25/Jun, 18/Jul, 31/Jul, 18/Aug, 08/Sep, 18/Sep	30	180	0	180
	1B1F	25/Jun, 18/Jul, 31/Jul, 18/Aug, 08/Sep, 18/Sep	30	180	90	90
	2B1F	25/Jun, 18/Jul, 31/Jul, 18/Aug, 08/Sep, 18/Sep	30	180	120	60
	BW	25/Jun, 18/Jul, 31/Jul, 18/Aug, 08/Sep, 18/Sep	30	180	180	0
2023	FW	28/Jun, 20/Jul, 01/Aug, 19/Aug, 07/Sep, 20/Sep	30	180	0	180
	1B1F	28/Jun, 20/Jul, 01/Aug, 19/Aug, 07/Sep, 20/Sep	30	180	90	90
	2B1F	28/Jun, 20/Jul, 01/Aug, 19/Aug, 07/Sep, 20/Sep	30	180	120	60
	BW	28/Jun, 20/Jul, 01/Aug, 19/Aug, 07/Sep, 20/Sep	30	180	180	0

displays the comprehensive watering schedule. The experimental technique included conducting three replications for each treatment in a totally randomized block design consisting of 12 fields. The dimensions of each treatment area were a length of 25 m and a width of 10 m.

The JiaNong 3168 variety of maize seed was planted on May 10th and May 9th, and the maize was subsequently harvested on October 1st and October 2nd of the years 2022 and 2023, respectively. We employed under-membrane drip irrigation, utilizing one membrane and one tube (drip irrigation tape). The planting method involved two rows, with a spacing of 10 cm for the drip irrigation tape laying. The film width was 70 cm. Maize was planted in both wide and narrow rows, with the wide rows measuring 60 cm and the narrow rows measuring 40 cm. The plant spacing was set at 26 cm. For the inner sickle patch drip irrigation tape, we used a field drip irrigation tape (Runtian Water Saving Co., Ltd., Shijiazhuang, China). The tube diameter was 16 mm, with a flow rate of 1.60 L·h⁻¹. The drip head spacing was 30 cm (Figure 3).

2.3. Data Observation and Computation

2.3.1. Meteorological Data and Groundwater Level Data

Meteorological data were collected using a Campbell automatic weather station, with a sampling frequency of once every 30 min. The meteorological data included temperature, precipitation, wind speed, relative humidity, air pressure, sunlight duration, solar radiation, and net radiation. Groundwater data were collected using a HOBO water level logger, with a sampling frequency of once every 4 h. The groundwater data comprises measurements of temperature and the depth of the groundwater.

2.3.2. Soil Salinity and Moisture Content

Simultaneous measurements of soil moisture and salinity were obtained. After collecting each soil sample, it was separated into two pieces. One of the sections was taken back to the laboratory to analyze the moisture content of the soil mass. The other part was used for measuring the electrical conductivity of the soil. The soil total salt content was converted using an empirical formula derived from multi-year data for the region:

$$\begin{array}{c}y=5.1949x+0.0685 R^2=0.93\end{array}$$

(1)

where y is the total soil salt (g·kg⁻¹); x is EC_1:5 (ds·m⁻¹).

2.3.3. Leaf Area, Biomass and Yield of Maize

In each experimental plot, five maize plants displaying consistent and typical growth were chosen for identification. The length of all the green leaves was measured using a steel ruler with an accuracy of 1 mm, and this measurement was taken every 15 days. The leaf area index was determined using the following method:

$$\begin{array}{c}LAI=0.75\rho {\displaystyle \frac{{\sum }_{j=1}^m{\sum }_{i=1}^n{L}_{ij}{B}_{ij}}{m}}\end{array}$$

(2)

where LAI—Leaf area index; ρ—maize plant density, plant/hm²; m—number of marker determination plants; n—number of leaves per maize plant, piece/plant; L_ij—maximum leaf length, m; B_ij—maximum leaf blade width, m.

Canopy cover (CC) is a crucial parameter used by the AquaCrop model to mimic the development of crops. It is determined by the following calculation:

$$\begin{array}{c}CC=1.005{\left(1-e^{-0.6LAI}\right)}^{1.2}\end{array}$$

(3)

Once maize reached the seedling stage, samples of the above-ground biomass were collected every 15 days. Five plants exhibiting typical growth were collected from each experimental plot. The underground roots were removed and subjected to a temperature of 105 °C for 30 min to ensure their destruction. Subsequently, the plants were dried at 75 °C until their mass remained constant. The dried plants were then weighed to determine their biomass. At the final stage of maize maturity, three random sample squares of equal area were chosen from each experimental plot. The number of plants within these squares was recorded, and the maize was weighed to calculate the hundred kernel mass and overall maize yield.

2.3.4. Collection of Hydroxide Isotope Samples

The soil samples were obtained by utilizing an iron auger with a 5 cm diameter. The samples were taken at a depth of 0 to 100 cm, divided into five layers of 20 cm each, in each test plot. The soil samples were promptly enclosed in brown glass vials upon collection and stored in a freezer. The soil samples were collected at a frequency of once every 7 days.

Plant rhizome samples were collected by extracting the subterranean section of maize rhizomes. The outer skin was eliminated, but the inner core was preserved. Subsequently, it was promptly enclosed inside a brown glass container and stored in the freezer. The collecting time for soil samples was the same.

Groundwater samples may be collected immediately from groundwater collection wells located around 10 m from the test area. The duration required for groundwater collection is equivalent to that of soil the samples.

The δ²H and δ¹⁸O measurements were conducted by extracting water from the soil using a low-temperature condensation vacuum extraction equipment (LI-2000, Beijing Riga United Science and Technology Co., Ltd., Beijing, China). This technology ensures thorough extraction with exceptional accuracy [36]. The water samples underwent analysis for δ¹⁸O and δ²H using laser absorption spectroscopy. This analysis was conducted at the Hydrogen and Oxygen Isotope Laboratory, which is part of the Institute of Pastoral Water Resources Science under the Ministry of Water Resources. The specific instrument used for the analysis was the liquid water isotope analyzer, Los Gatos Research DEL-100, located in California, USA.

2.3.5. Isotopic Identification of Water Sources in Different Soil Layers

This research used the MixSIAR model to determine the depth of water absorption by maize and the quantity of water absorbed from each soil layer. The objective was to elucidate the respective contribution of each soil layer to the overall water content. The MixSIAR model was implemented using the R programming language, allowing for the incorporation of uncertainty arising from various sources of water, isotopic signatures, and isotopic fractionation [37]. Using this approach, we successfully evaluated the precise depth of water absorption by the plant’s root system in various soil layers and determined the individual contribution of each layer to the overall water uptake. This technique serves as a foundation for comprehensive examination of soil moisture dynamics and offers useful data for water management and predicting plant development. The procedure is as follows: First, install the MixSIAR package on the R 4.1.1 software platform. Then, import the Mixture, Source, Discrimination (TDF) data and the Markov Chain Monte Carlo (MCMC). Next, choose a long run length and pick “Residualonly” for the error structure. Visualize the data by plotting them. Execute the model by choosing the Model Structure option and result option. Finally, assess whether the model has converged and evaluate the result using Diagnostics.

2.4. AquaCrop Model

This is an agricultural water productivity model designed to replicate the relationship between crop output and water availability (Figure 4). The model necessitates a sequence of data inputs, including climatic factor, crop, soil and field, and irrigation management data [38]. Conversely, the model includes a collection of input parameters that are chosen and modified for various soil or crop varieties.

2.4.1. Climate Data

Data on meteorological variables such as air temperature, rainfall, air humidity, solar radiation, atmospheric pressure, and wind speed were gathered at hourly intervals utilizing automated weather stations in the experimental region. The Penman–Monteith formula was used to compute crop evapotranspiration (ET₀).

2.4.2. Crop Parameters

The crop parameter file primarily encompasses crop growth, crop evapotranspiration, and crop production, as well as water, salinity, and temperature stress. The crop growth parameters such as initial and maximum canopy cover, flowering, senescence, and maturity can be determined through direct observation in the field. On the other hand, the water production index, crop harvest index, water stress response coefficient, salinity stress response coefficient, and temperature stress response coefficient in crop production can be estimated using benchmark parameters provided by the model. Additionally, the “trial and error” method can be employed to fine-tune these parameters. Simultaneously, the “trial and error method” was used to implement adjustments. The experiment used the measured data from each treatment in 2022 to fine-tune the model parameters. The measured data from 2023 were then employed to verify the model.

Table 3. AquaCrop model maize crop parameters.

Parameters	Default	Calibration Value
Maximum canopy cover (%)	96	99
Maximum Rooting depth (m)	2.3	1.0
Canopy Growth Coefficient (%/day)	16.3	14.3
Canopy Decline Coefficient (%/GDD)	1.06	1.21
Reference Harvest Index (%)	48	43
Normalized crop water productivity (g/m2)	33.7	37.5
Expansion stress coefficient (Pupper)	0.14	0.24
Time from sowing to emergence	8	9
Time from sowing to max canopy cover	70	78
Time from sowing to maximum root depth	90	97
Time from sowing to senescence	115	120
Time from sowing to maturity	140	146
Time from sowing to flowering	70	67
Length of flowering stage	20	14

displays the primary crop parameters used in the AquaCrop model.

2.4.3. Soil Parameters

The soil parameters in the AquaCrop model primarily consisted of wilting water content, saturation water content, field water holding capacity, number of soil layers, and bulk weight. Prior to seeding sunflowers, a random selection of five pilots was made on the field for the purpose of soil sample. Each stratum was defined as every 20 cm. The soil parameters were then integrated into the model in order to generate a soil data file (SOL).

2.4.4. Irrigation and Field Management Parameters

The irrigation component of the AquaCrop model incorporates factors such as irrigation duration, irrigation volume, and irrigation water characteristics. The field management component includes factors such as soil fertility, weed control, soil cover materials, and characteristics related to soil drainage.

2.5. Calibration and Validation of the Aquacrop Model

The model employs an iterative methodology to generate data values that more accurately replicate the primary factors of crop development, including canopy cover, biomass, crop yield, and water usage efficiency. The parameters were calibrated to optimize the correspondence between measured and simulated results. The factors associated with crop types, both conservative and non-conservative, were treated as fixed values. Parameters that are not conservative were modified according to measurements taken in the field. The crop growth coefficient (CGC) and crop senescence coefficient (CDC), together with normalized water productivity (WP), were calibrated using field sample findings to ensure their accuracy and consistency.

The model was executed using empirical data from the 2022 and 2023 growth seasons. Subsequently, the projected values were juxtaposed with the actual outcomes of the trials and assessed for model validation output statistics.

2.6. Model Evaluation Guidelines

The calibration and validation procedure assessed the simulated outcomes of water usage efficiency, biomass, yield, and canopy cover generated by the AquaCrop model. Error statistics were largely used to ascertain the consistency between simulated and actual data. In order to evaluate the model’s performance, the following statistics were used. The primary metrics used were mean absolute percentage error (MAPE), root mean square error (RMSE), and coefficient of determination (R²).

$$\begin{array}{c}MAPE={\displaystyle \frac{1}{n}}{\displaystyle \sum _{i=1}^n}\left|{\displaystyle \frac{M_i-S_i}{S_i}}\right|\times 100\%\end{array}$$

(4)

$$\begin{array}{c}RMSE=\sqrt{{\displaystyle \frac{1}{n}}{\displaystyle \sum _{i=1}^n}{\left(M_i-S_i\right)}^2}\end{array}$$

(5)

$$\begin{array}{c}{R}^2=1-{\displaystyle \frac{{\sum }_{i=1}^N\left|S_i-M_i\right|}{{\sum }_{i=1}^N\left|S_i-S_{avg}\right|}}\end{array}$$

(6)

where S_i is the ith actual measured value; M_i is the ith simulated value; S_avg is the average of the measured values.

3. Results

3.1. Changes in Soil Moisture and Salinity

Figure 5 illustrates the impact of several alternating saltwater–freshwater irrigation methods on soil moisture. The soil moisture content under the four irrigation treatments exhibited a “high-low-high” trend throughout the growth stages of maize. The soil water content in the lower layer was higher than in the upper layer owing to the impact of groundwater. During the maize seedling stage, the treatment with the highest average water content was the FW treatment, which had a water content of 23.64%. There were no significant differences (p > 0.05) in water content among the four treatments during this time, as shown in

Table 4. Comparison of variability of soil moisture and salinity under different irrigation water quality treatments at different fertility stages of maize (different lower-case letters indicate significant differences).

Soil Moisture and Salinity	Period of Fertility	Treatment
		FW	1F1B	1F2B	BW
SWC	seedling stage	a	a	a	a
	elongation stage	a	bc	c	b
	staminate period	a	b	c	a
	grouting period	a	b	c	ab
	maturity period	a	a	a	a
SSC	seedling stage	b	a	a	a
	elongation stage	b	a	a	a
	staminate period	a	a	a	a
	grouting period	c	bc	bc	a
	maturity period	d	a	b	c

. During the elongation stage of maize, there was no significant difference in water content between the BW and 1F2B treatments, with an average value of 22.08%. However, the BW treatment had the highest water content of 23.86% during this time. At the nodulation–male extraction stage of maize, the water content of the four treatments ranged from 17.25% to 21.83%, with the lowest value observed. During the stumping stage of maize growth, the average soil water content rose, reaching a maximum of 25.30% for the FW treatment. The mean water content for the 1F1B, 1F2B, and BW treatments over this time was 23.94%, 22.54%, and 24.92%, respectively. During the maize filling stage, the soil water content saw a considerable rise of 11.13% compared to the soil water content during the male pumping stage. The water content of the four treatments achieved an average of 27.14% at the maturity stage of maize. The most significant fluctuation in soil water content occurred in the soil layer between 0 and 20 cm.

Figure 5 shows the variations in soil salinity under various irrigation regimens. The subterranean brackish water had a salinity that was 2.8 g/kg more than that of the Yellow River water, which is classified as fresh water. Additionally, the soil salinity showed significant variation during each period. The values of 1F1B, 1F2B, and BW were raised by 10.46%, 34.39%, and 48.91%, respectively, as compared to the all-freshwater irrigation treatment. During the nodulation stage, the salt levels in each soil layer varied between 1.19 g/kg and 6.89 g/kg in the three treatments irrigated with brackish water, and between 1.11 g/kg and 1.80 g/kg in the treatments irrigated with complete freshwater. The salt concentration of treatments 1F1B, 1F2B, and BW was elevated by 11.01%, 22.53%, and 40.31%, respectively, in comparison to the FW treatments during the nodulation–irrigation stage. The mean salt concentration varied considerably (p < 0.05) during the whole lifespan of the maize plants, at a depth from 0 to 100 cm, for each treatment.

3.2. AquaCrop Model Simulation Results

3.2.1. Canopy Cover

The AquaCrop model relies on dynamic modeling of canopy cover to depict crop development processes. Precise simulation of canopy cover profiles is crucial for the model to accurately mimic soil evapotranspiration, crop transpiration, biomass, and yield. For our research, we fine-tuned the model by including data from experiments conducted in 2022, and then assessed its accuracy by comparing its predictions with data from 2023.

Table 5. Comparison of simulated and measured values of yield, biomass, and canopy cover for the AquaCrop model in the calibration year (2022).

Treatments	Yield			Biomass			CC
	R2	MAPE/%	RMSE/(t·hm−2)	R2	MAPE/%	RMSE/(t·hm−2)	R2	MAPE/%	RMSE/%
FW	0.90	3.5	0.31	0.91	13.25	1.29	0.86	7.78	4.99
1F1B	0.89	4.1	0.39	0.88	11.54	2.69	0.84	7.45	5.12
1F2B	0.86	3.9	0.64	0.90	12.69	3.12	0.87	6.99	4.32
BW	0.85	4.2	0.83	0.83	14.59	2.99	0.88	5.87	4.57

displays the outcomes of the calibration of the model simulation values. Figure 6 displays the results of the canopy cover validation.

The findings obtained during the calibration year indicate that the canopy cover simulations for each treatment exhibited superior performance, with a minimum coefficient of determination (R²) of 0.84, a maximum mean absolute percentage error (MAPE) of 7.99%, and a maximum root mean square error (RMSE) of 5.12%. The model simulates better outcomes when a larger proportion of freshwater is used under varied irrigation water quality conditions. The FW treatment exhibited superior R², MAPE, and RMSE values compared to BW. The validation year findings indicated that the canopy cover simulation outcomes for the various irrigation treatments were superior to those of the calibration year. The coefficient of determination (R²) is more than or equal to 0.93, the mean absolute percentage error (MAPE) is less than or equal to 7.36%, and the root mean square error (RMSE) is less than or equal to 6.96%. The highest extent of foliage coverage occurred in the FW and 1F1B treatments. The simulated values of maximum canopy cover for the FW and 1F1B treatments occurred 5 to 10 days prior to the actual values. The simulated maximum canopy cover values for the 1F2B and BW treatments exhibited reduced disparity compared to the actual values. Furthermore, the simulation results outperformed those of the FW and 1F1B treatments.

3.2.2. Above-Ground Biomass and Production

In our study, we examined the biomass produced by the AquaCrop model. The model accurately predicted the growth pattern of above-ground biomass under various irrigation water quality treatments during a two-year experiment, with the exception of the FW and 1F1B treatments. In these cases, the simulated values were slightly greater than the measured values at the end of the maize growth period (Figure 7). When comparing the various irrigation treatments, it was found that the ultimate above-ground biomass simulated values were superior for totally brackish water irrigation and inferior for fully freshwater irrigation. During the calibration year of 2022, the irrigation simulation performed well for each irrigation treatment, as shown by R² values of 0.83 or higher, MAPE values of 14.59% or lower, and RMSE values of 3.12 t·hm⁻² or lower (Table 5). During the validation year of 2023, the completely freshwater irrigation (FW) simulation showed lower effectiveness compared to the other treatments, with an R² value of 0.89, a MAPE value of 19.19%, and an RMSE value of less than or equal to 2.29 t·hm⁻². The performance of fully brackish water irrigation (FW) was worse compared to the other treatments, as shown by an R² value of 0.89, a MAPE of 19.19%, and an RMSE of <2.29 t·hm⁻². The irrigation method using fully brackish water (BW) showed the highest level of accuracy in the simulation, with an R² value of 0.92, a MAPE of less than or equal to 15.19%, and an RMSE of less than or equal to 1.99 t·hm⁻².

When comparing the yield simulation results of the two years, it was found that the simulated values closely matched the measured values (Figure 8). The R², MAPE, and RMSE values were 0.88 and 0.84, 3.93% and 3.62%, and 0.54 t·hm⁻² and 0.42 t·hm⁻², respectively.

3.3. Percentage of Water Uptake by Maize in Different Soil Layers

A Bayesian isotope mixing model called MixSIAR was used to measure the amount of soil water and groundwater that contributes to the crop at various depths under different treatments and fertility stages of maize. Additionally, the model was used to identify the primary depth at which maize absorbs water via its roots at each fertility stage. The findings indicated that the average percentage of soil water and groundwater contribution to maize at different depths (0~20 cm, 20~40 cm, 40~60 cm, 60~80 cm, and 80~100 cm) was 29.41%, 19.32%, 17.43%, 14.70%, 12.39%, and 6.75%, respectively. As maize grows, the water use efficiency of the soil surface layer (0–20 cm) gradually decreases. During the reproductive stage of maize, there was a noticeable shift in the primary supply of water in the soil layer, as seen in Figure 9. The treatments during the maize seedling stage primarily absorbed water from the soil at depths of 0 to 60 cm, which accounted for over 90% of the total water absorption. The largest proportion of water absorption occurred in the top layer of soil (0 to 20 cm), with a contribution value ranging from 62.01% to 72.1%. During the nodulation stage, the four treatments resulted in an increased contribution of water from deeper soil layers (60 to 100 cm) and groundwater, ranging from 19.32% to 30.95%. During the male flowering stage of maize, the BW treatment resulted in a 10.51% decrease in water utilization between 0 and 60 cm. During the pollination stage, both the BW and FW treatments showed significant increases in groundwater utilization, with the BW treatment increasing by 84.04% and the FW treatment increasing by 148.7%. At the maturity stage, the BW treatment primarily utilized water from depths of 60 to 100 cm and groundwater. The 1F1B treatment mainly relied on soil water from depths of 20 to 100 cm, while the 1F2B and FW treatments primarily utilized soil water from depths of 20 to 60 cm and 80 to 80 cm, respectively. During the whole life cycle of maize, the BW treatment distributed water uniformly across all five soil levels. In contrast, the 1F1B and 1F2B treatments focused on the top and middle soil layers (0 to 40 cm) and accounted for about 40% of the total water use. Conversely, the FW treatment focused primarily on the absorption of water in the intermediate soil layer, namely between 20 and 60 cm.

3.4. Irrigation System Optimization

We enhanced the efficiency of simulating irrigation schedules for brackish water in field maize using the rate-determined AquaCrop model. We developed a system for irrigation that involves employing varying proportions of brackish and fresh water. This method aims to decrease the negative effects of salt stress caused by brackish water irrigation and enhance the yield and biomass of maize crops.

Table 6. Irrigation regimes for maize at different fresh and brackish water ratios (0.5, 0.8, and 1.2 denote values of conductivity of irrigation water in ds/m at different fresh and brackish water ratios).

Treatments	Period of Irrigation
	30 Days after Sowing/(ds·m−1)	45 Days after Sowing/(ds·m−1)	60 Days after Sowing/(ds·m−1)	75 Days after Sowing/(ds·m−1)	90 Days after Sowing/(ds·m−1)	110 Days after Sowing/(ds·m−1)
S1	0.5	0.5	0.5	0.5	0.5	0.5
S2	0.8	0.8	0.8	0.8	0.8	0.8
S3	1.2	1.2	1.2	1.2	1.2	1.2
S4	1.2	0.5	1.2	0.5	1.2	0.5
S5	1.2	1.2	0.5	1.2	0.5	0.5
S6	0.8	0.5	0.5	0.8	0.5	0.5
S7	0.8	0.8	0.5	0.8	0.8	0.5

provides a summary of the irrigation regime. S1–S7 represent the seven different irrigation water quality treatments. A volume of 30 mm of water was used for a solitary irrigation.

The AquaCrop model evaluated seven scenarios of irrigation water quality, and the results of the simulation are shown in Figure 10. The highest yield value, 12.31 t·hm⁻², was seen in S1, whereas the lowest yield value, 10.78 t·hm⁻², was observed in the S3 irrigation scenario. There were significant differences between these two scenarios (p < 0.01). The ranking of the seven treatments based on yield was as follows: S1 > S7 > S2 > S4 > S6 > S5 > S3.

The simulated findings of maize above-ground biomass exhibited a pattern that was parallel to the changes in yield. The S4 treatment had the highest maize above-ground biomass, measuring 23.01 t·hm⁻², whereas the S1 treatment had a slightly lower biomass of 22.98 t·hm⁻². There was no statistically significant variation (p > 0.05) seen among the seven treatments in relation to the above-ground biomass of maize.

According to the simulation findings, the irrigation strategies of S1, S2, and S7 treatments demonstrated superior performance in increasing the use of brackish water and decreasing the reliance on freshwater, while not causing a substantial decrease in maize yield and biomass.

4. Discussion

4.1. Analysis of Soil Water Utilization under Different Irrigation Water Quality

This research demonstrated that the water use of maize in various soil layers under alternating brackish–freshwater irrigation exhibited considerable variation. The BW and FW treatments primarily relied on groundwater, with BW being influenced by soil water content and soil salinity. BW absorbed water with lower salt content into deeper layers, while the FW treatment stimulated the growth of the maize root system, leading to increased utilization of groundwater. Zhang [39] demonstrated that the frequency of saline irrigation had a positive impact on the stress-induced inhibition of maize root development. As a result, the absorption of water from the lower layers of soil was decreased. The conclusion of the current research is somewhat incongruous with the data, mostly due to the limited depth of groundwater in the study location. It is more probable that the maize root system would rely on groundwater as an additional supply of irrigation water. In the Huaihe River Plain, China, Zhuang [40] carried out a maize experiment in which hydroxide isotopes were artificially increased in the root zone of maize from 0 to 100 cm. The enrichment of hydroxide isotopes increased as the soil layer became deeper. When subjected to brackish water irrigation, maize showed a preference for soil layers that were supplemented with water. This preference was particularly noticeable in the area where the groundwater was shallow. Among the four treatments, the use of complete freshwater irrigation resulted in the highest yield index, but the use of full brackish water was the least efficient. Murad [41] used a method of alternating brackish and freshwater irrigation and determined that this alternating approach resulted in a reduction in yields ranging from 1.2% to 31.0% when compared to the usage of just freshwater irrigation. The yield of maize declined as the use of brackish water increased.

The soil water potential fluctuates due to variations in irrigation water sources, leading to differential water absorption by the root system and exchange of groundwater with soil water [42]. Irrigation with high-mineral-content brackish water can effectively lower the groundwater level and reduce soil salinity in the short term. However, prolonged irrigation with such water can lead to increased soil salinization [43]. Therefore, when using brackish water for irrigation, it should be alternated appropriately with freshwater. Drip irrigation focuses on monitoring the vertical variations in soil moisture and salt inside the crop’s root zone. The distribution of soil salinity varies dramatically across different soil levels, namely the 0–20 cm, 20–40 cm, and 40–100 cm layers. Studying the dynamic changes in soil moisture and salinity, understanding how crops absorb water from different soil layers, and developing a strategy for alternating irrigation with brackish freshwater and the right proportion of brackish water use can significantly enhance crop yield and biomass while reducing the need for freshwater [44].

4.2. Evaluation of the Applicability of the Aquacrop Model

The AquaCrop model demonstrated superior accuracy in simulating canopy cover, as shown by its R² value of at least 0.83 and MAPE value of no more than 14.59% for both the calibration year (2022) and validation year (2023). During the calibration year, the simulation results showed that the canopy cover of FW (freshwater) and 1F1B (one part freshwater and one part brackish water) were of worse quality compared to 1F2B (one part freshwater and two parts brackish water) and BW (brackish water). However, the difference in canopy cover between FW and 1F1B was not statistically significant when compared to 1F2B and BW. Nevertheless, it can be provisionally inferred that the AquaCrop model diminishes the precision of simulating canopy cover in arid regions when the crop experiences water stress. The model improved the process of canopy senescence in response to water stress, leading to premature senescence in the simulated canopy cover. The recorded canopy cover exhibited a reduced growth rate when subjected to water stress during the first reproductive phase of maize. In a study conducted by Hengetal [45], similar findings were observed. They found that the model was able to accurately simulate the irrigation treatment of CC well (with a root mean square error (RMSE) ranging from 5% to 16% and an efficiency (EF) between 0.8 and 0.98) when deficit irrigated maize was grown under three different environmental conditions in Texas. However, the model did not perform well in simulating canopy cover under severe water stress conditions, with an RMSE of 34.5% and a negative EF of −2. In their study conducted in Florida and Spain, Sandhu (2019b) [46] identified the model’s difficulties in accurately simulating canopy cover during periods of severe water shortage. They highlighted that the model overestimates the inhibitory impact of water stress compared to the real conditions, resulting in a bias in the predicted CC values. The model successfully replicated canopy development in the majority of the four water quality irrigation treatments in this investigation.

The AquaCrop model yielded superior simulation results for both crop yield and biomass. The coefficient of determination (R²) for maize yield and biomass in the calibration and validation years was 0.85, 0.84, 0.83 t·hm⁻², 0.42 t·hm⁻² and 0.83, 0.89, 3.12 t·hm⁻², 2.29 t·hm⁻², respectively. The root mean square error (RMSE) values for maize yield and biomass in the calibration and validation years were 0.85, 0.84, 0.83 t·hm⁻², 0.42 t·hm⁻² and 0.83, 0.89, 3.12 t·hm⁻², 2.29 t·hm⁻², respectively. In general, the model’s simulated yield and biomass were somewhat greater than the actual values. The primary factor was the model’s failure to accurately estimate the impact of plastic film on increasing maize yield under certain moisture stress conditions. The substantial rise in maize productivity when mulch was applied under moderate moisture stress may be attributed to the various impacts of mulch cover on the hydrothermal conditions, which were not included in the model. This also suggests that only altering evapotranspiration is not enough to replicate the impact of mulch cover, since mulch may enhance the soil’s hydrothermal conditions and stimulate crop development [47,48]. Tan [49] discovered in their research that the yields were inaccurately assessed in over 30% of the experimental plots for treatments including mulch cover, with an error rate of 30% or higher. Guo [50] showed that the AquaCrop model correctly simulated millet without mulch cover. However, when millet had mulch cover, the simulation accuracy was significantly lowered. Hence, the incorporation of a cover component into the AquaCrop model may enhance the accuracy of yield forecasting for crops grown with mulch. Compared to DSSAT, CropSyst, SWAT, and other models, the AquaCrop model exhibits superior operational stability and simplicity [51]. Additionally, the AquaCrop model demonstrates better simulation accuracy, particularly in terms of crop yield and biomass, compared to the aforementioned models [52].

5. Conclusions

This paper focuses on calibrating and validating the AquaCrop model parameters using two years of measured maize field data. The study also examines the simulation effect and functional applicability of maize growth under different brackish and freshwater irrigation treatments in the irrigation area on the south bank of the Yellow River in Inner Mongolia. We conducted an analysis of how maize uses water in various soil layers, considering varying grades of irrigation water. This analysis was performed by utilizing the hydroxide isotope tracer technique. The primary outcomes of this study, which provide a theoretical basis for optimizing the irrigation system during maize fertility in this region, are as follows:

(1)

The AquaCrop model is capable of accurately simulating the development and yield of maize under varying irrigation water quality conditions throughout the fertile period. The model simulation’s accuracy will diminish when maize is subjected to water stress. Consequently, the model will result in a lower estimation of the predicted maize yield when mulching is taken into account.

(2)

Utilizing the hydroxide isotope tracer approach, it was determined that irrigating maize with both mineralized water quality of 1.6 ds/m and fresh water quality over the whole reproductive period would result in an enhanced percentage of groundwater usage. However, there was no notable disparity in the exploitation of various soil strata between the irrigation methods of alternating freshwater and brackish water.

(3)

Based on scenario simulation analysis using the AquaCrop model, it was determined that in order to achieve the goals of lowering freshwater consumption and increasing maize yield, it is advisable to irrigate the site with either 0.5 ds/m or 0.8 ds/m of water source irrigation for a total of 6 or 7 irrigation cycles.