1. Introduction
Water scarcity is a great concern for irrigation agriculture worldwide. More than half of the farmland in the world exists in arid and semiarid regions [
1]. Irrigation is essential to crop production in arid regions and plays an important role in crop water demands throughout the world. The site-specific application of irrigation water within a field improves water use efficiency and reduces water usage for sustainable crop production, especially in arid and semiarid regions [
2].
Rational irrigation scheduling is essential to irrigation management, and many irrigation scheduling studies have been performed in arid regions [
3,
4,
5]. In general, one type of irrigation scheduling is based on evapotranspiration demand, and another is based on the soil water content of the root zone [
5]. Shallow groundwater exists in many areas of the world [
6], and some farmlands are irrigated with shallow groundwater in arid regions. Shallow groundwater exists in many areas of the world and plays a vital role in sustaining agricultural productivity in many irrigated areas [
3,
7,
8]. Irrational or intensive irrigation leads to a decline in the shallow groundwater table. The variation in a shallow groundwater table strongly influences the water balance [
9]. Water cycling in soils with shallow groundwater is complex due to the deep percolation and groundwater evapotranspiration (ET
g) that occur in arid regions [
10]. An improved irrigation schedule could reduce the amount of deep percolation [
8].
Several studies have used lysimeters to measure crop water consumption from a shallow groundwater table [
11,
12,
13,
14,
15]. The lysimeters were accurate, but their use is limited because of their high cost [
3]. Therefore, some studies have attempted to quantify the groundwater as a part of the SPAC (soil-plant-atmosphere continuum) at the field and regional scales using models such as SWAP [
3], EPIC [
16], and DSSAT [
17]; however, the models mentioned above usually oversimplified the influence of groundwater [
18]. Han et al. [
18] used the Hydrus-1D model coupled with a simplified crop growth model from SWAT to estimate the effect of groundwater on the water balance in the cotton root zone, but their study did not include a method for irrigation scheduling and was not implemented or systematically used by the majority of growers. Huo et al. [
3] simulated the various amounts of irrigation applied to soil with different water tables, but the irrigation schedule was fixed in their study. Therefore, information relative to the irrigation scheduling method is still limited.
Accurate estimations of evapotranspiration (ET) are urgently needed. ET includes soil evaporation (E) and crop transpiration (T); transpiration is considered a physiological process, whereas soil evaporation is a physical process. Soil evaporation is an important component of the total crop water consumption, and the E ratio is higher in the earlier growing season due to wet surface soils and low crop cover (CC) [
19]; however, evaporation may be lower in the later growing season when the surface soils are drier, and the CC is higher [
20]. Unkovich et al. [
20] reviewed the published field measurements from Australia and found an average of 38% crop water consumption due to soil evaporation. Kool et al. [
21] noted that the E/ET ratio exceeded 30% in 32 of the studies and noted that E usually constituted a large fraction of ET and should be independently considered. Numerous measurement methods and analytical models have been developed to estimate T and E separately, but large variability exists in ET partitioning, suggesting that obtaining accurate ET partitioning is relatively challenging [
21]. In general, T is controlled by atmospheric evaporative demand, leaf area index (LAI) or crop cover (CC), surface soil water content (SWC), and reference crop evapotranspiration (ET
0) [
22,
23,
24,
25]. These factors influence T and E, as well as ET partitioning. Therefore, most previous studies focused on the influence of crops on ET partitioning using regression functions between CC and T/ET or E/ET [
26,
27,
28,
29,
30]. Zhao et al. [
25] established a function between ET partitioning and control factors (CC and SWC), but the metrological factors were neglected, and the CC and SWC had to be acquired by field experiment observations. The establishment of simple functions that are correlated with the main control factors without field experiment observations was necessary. It will provide a better way to interpret evapotranspiration partitioning and to model water consumption in cotton fields.
Existing models include Hydrus-1D; however, E or SWC must be acquired by field experimental observations. For SWC, E is just an input parameter, leading to the interaction between SWC and E being neglected. However, this interaction effect is present. For example, a high soil water content could cause high soil evaporation, but meteorological factors also influence the SWC [
20,
21]. If the SWC or E are included as only input parameters, then the interaction effect will be neglected, and the real condition will not be reflected. Although the HYDRUS model can simulate irrigation and water consumption at a given groundwater table, the model can only set a fixed or initial groundwater table. In practice, changes in groundwater table are not entirely dependent on evaporation consumption from cropland (but lysimeter with impermeable bottom) but involve lateral recharge and groundwater consumption from surrounding areas. Our model takes into account an actual groundwater change (groundwater variation with time as an input parameter), which better reflects a real situation. Therefore, in this study, we developed a model to estimate crop water consumption and considered an actual groundwater table fluctuation and the interaction between SWC and E.
The objective of our research was to (1) develop a model based on the Richards equation to simultaneously estimate crop water consumption with shallower groundwater, (2) develop a new function for estimating soil evaporation considering the interaction between SWC and E, (3) validate this model using cotton field experiment observations, and (4) apply the new tool to estimate a suitable irrigation schedule.
4. Conclusions
A model was developed to quantify cotton water consumption and to estimate a suitable irrigation schedule according to the groundwater depth (1.0–4.0 m). In this study, two irrigation scheduling methods were considered. In Method A, irrigation was managed based on the soil water content; in Method B, irrigation was based on the crop water demand. The simulation results were verified with soil water storage measurements in the root zone (0–60 cm) and soil water contents at depths of 20 and 150 cm from the cotton field. Suitable agreement was presented between the simulation results and experimental data obtained from cotton field experiments, indicating that the model showed suitable performance.
A new function was established in correlation with the surface soil water content, crop cover, and reference crop evapotranspiration to calculate the soil evaporation and determine the evaporation partition. These factors were calculated using the days after sowing and did not require observations. The interaction effect between meteorological factors and surface soil water content was also considered and better reflected the real condition, thus providing a better way to interpret evapotranspiration partitioning.
With a deeper groundwater table depth from 1.0 to 4.0 m, the ratio between the soil evaporation and evapotranspiration showed a decreasing trend, but Method A showed a smaller ratio than Method B. In addition, for Method A, the groundwater could supply enough water to the soil root zone, and irrigation was not necessary when the groundwater table was less than at a depth of 1.5 m; however, the groundwater could not supply enough water to meet the crop water consumption demands even if the groundwater table was at a depth of 1.0 m, and the crops could not use the groundwater when the groundwater was deeper than 3.0 m in Method B. Therefore, we suggest that the irrigation schedule should be managed based on the soil water content because this irrigation schedule could use more groundwater and have a smaller soil evaporation to evapotranspiration ratio.