Developing a Subsurface Drip Irrigation Scheduling Mode Based on Water Evaporation: Impacts Studies on Cucumbers Planted in a Greenhouse in the North China Plain

Abstract

China is a country short of water resources, and improving the water use efficiency (WUE) in agriculture has become the only way to ensure sustainable development. In this article, subsurface drip irrigation (SDI) experiments of cucumber were implemented with a randomized block design comprising two factors and three levels, and the two factors were depth of drip belt buried and irrigation amount, which were determined by a 20 cm diameter pan’s water evaporation times its coefficient (K_p). The effects of schedule of SDI on soil evaporation (E_s), evapotranspiration (ET), K_p, root dry matters, yield, and WUE of cucumber were studied. The results indicated that the E_s and the ET decreased along with increasing depth of drip belt buried or decreasing amount of irrigation water applied. The relationships between ET and its total irrigation amount were significant linear positive correlations. Proportions of plant transpiration accounting for the ET were about 41~69% in two years, and it became bigger along with the increase of ET within a year. The K_p became smaller along with the reducing of ET. The roots of cucumber were mainly concentrated in the soil layer of 0~20 cm, and the two factors could only influence root dry weights of 0~60 cm soil layers significantly. The relationships between yield and ET were quadratic polynomial correlations. At last, an ultimate SDI scheduling mode based on water evaporation was established.

1. Introduction

China is a country short of water resources, and there exists an inhomogeneity of regional, spatial, and temporal distribution. The per capita water resources are less than 1/3 of the world average, and in Northern China it would be even lower. Agricultural water accounts for nearly 60% of China’s total annual water consumption, most of which is used in irrigation. Therefore, improving efficiency of agricultural water use has become the only way to ensure sustainable development of agriculture in China. In order to achieve this objective, the application of advanced water-saving irrigation technology is the first consideration method [1,2].

Subsurface drip irrigation (SDI) has been proved to be a high-performance irrigation method, possessing virtues like increasing yields; decreasing evaporative losses, deep percolation, and weeds’ growth; and reducing greenhouse gases’ emissions [3,4,5,6,7,8,9,10], and the effect of SDI on controlling soil salinity was validated as well [11]. Compared with other irrigation methods, SDI offered the highest water use efficiency (WUE) and fertilizer use efficiency (FUE) [12,13]. Two factors affecting the normal use of the SDI system are appropriate burial depth of the drip belt and irrigation water amount. If the drip belt was buried too shallow, there would be no difference in soil water evaporation between surface and subsurface drip irrigation [14], and if it was buried too deep, the emitters on the drip belt would miss the main zone of roots living in the early living stage of the plant, which did not benefit the roots’ water uptake, and the proper depth of the drip belt buried should be ascertained according to crop type and soil texture. Meanwhile, too much irrigation water would lead to roots being in a waterlogging and hypoxia condition, and too little would not match the crop water requirement. Thus, appropriate irrigation regimes should be established through experiments.

In reference to the existing literature, irrigation regimes could be scheduled by the Penman–Monteith equation [15,16,17]. However, Zhang (2010) [18] pointed out that the Penman–Monteith reference evapotranspiration (ET₀) equation cannot be used directly under solar greenhouse microclimatic conditions because of the neglect of aerodynamics, and the Hargreaves equation and the radiation methods were recommended for the calculation of greenhouse ET₀ [19]. Some experts used soil moisture sensors or tension meters to schedule irrigation regimes [20,21,22]. These methods of irrigation schedule making are effective, but because of the variability of soil moisture, the number of sensors should be based on a certain amount, which increases monitoring costs. Furthermore, the soil moisture sensors or the tension meters could only determine whether the soil is short of water, but the amount of irrigation is still needed to calculate through the soil water balance method [23,24]. Compared with these methods, irrigation schemes based on a 20 cm water evaporation pan obtained a good method due to the advantages of easy use, low investment cost, and easy application [25,26,27,28], and pan evaporation (E_p) has been proven to have a good relationship with daily evapotranspiration (ET) measured by the weighing method in a greenhouse [18]; thus, the precision irrigation amount could be calculated based on E_p in practice. In order to acquire a precision irrigation regime, K_p needed to be changed to adapt to different evaporative conditions or living stages of crops to meet the crop water requirement and improve WUE. Then, the precision irrigation water amount could be calculated based on the products of three parameters such as K_p, E_p, and irrigation area. Now that the irrigation time and amount could be made sure based on the water evaporation pan, a smart irrigation system could be developed by using modern electronic information technology.

For the combination of different depths of drip belt and irrigation amounts, the soil wet patterns were different, which influenced ET, roots’ growth, and ultimate yields of crops [15,16,17]. The soil wet patterns were mainly influenced by the trickle discharge rate and the soil texture. An increase in the trickle discharge rate would result in an increase in the horizontal wetted area and a decrease in the soil wetted depth, and the decrease in the soil particle size would lead to a similar effect as the increase of the trickle discharge rate did on the soil wet patterns [29]. ET contains two parts, plant transpiration and soil evaporation, and the reduction of soil evaporation is the main contribution of WUE improvement of SDI compared with other irrigation methods. However, existing studies seldom possessed the accurate soil evaporation data and its proportion accounting for ET. It is known that SDI would benefit root growth in deep depth [30], but do the influences of SDI schedules on different depths of below ground matters share the same mechanism? It is not clear. Thus, these scientific problems still need to be clarified through experiment.

In this article, SDI experiments of cucumber were implemented in an autumn–winter cycle greenhouse in the North China Plain, and the influences of schedules of SDI on soil evaporation, ET, K_p, root dry matters, ultimate yield, and WUE of cucumber were studied, and finally, an SDI scheduling mode based on water evaporation was established.

2. Materials and Methods

2.1. Study Site

The study was conducted in a greenhouse at Qiliying comprehensive test base of Farmland Irrigation Research Institute (FIRI), Chinese Academy of Agricultural Science (CAAS), located in Xinxiang City, Henan Province China (35°19′ N, 113°53′ E) at an elevation of 73.2 m. The map of the study area and its specific location is shown in Figure 1 [31]. The greenhouse used in the experiment is 55 m long and 9 m wide, facing south, and the north wall of the greenhouse is made up of brick and concrete structure, and the arched steel frame structure serves as the top of the greenhouse, covered with thin poly film and an insulation quilt.

The soil textural in the greenhouse was loamy soil, with a bulk density of 1.38 g·cm⁻³ and field capacity (FC) of 24% (accounting for dry soil weight, tested by the cutting ring method), and soil particle distribution and chemical characteristics in the plough layer were listed in

Table 1. Soil particle distribution and chemical characteristics in plough layer.

Depth (cm)	Clay (%)	Silt (%)	Sand (%)	Organic (g·kg−1)	NO3−N (mg·kg−1)	Available P (mg·kg−1)	Available K (mg·kg−1)	pH
0~20	6.23	65.2	28.57	15.60	43.21	20.76	93.21	7.01
20~40	6.78	66.9	26.32	8.43	17.82	6.32	92.70	7.24

.

2.2. Experiment Design

The studies were implemented in 2016 and 2017 during two autumn–winter cycle seasons. Zhongnong No. 20 was used in this experiment, which was a matrix cultivation, and then consistent and strong seedlings were transplanted to the experimental plots. The date of seedlings’ transplanting was on 26 August and the terminal date of research observation was on 30 November, with total 97 experimental days in both years.

The test plot was 8.0 m long and 1.1 m wide, and cucumber was double row planted, with a row spacing of 60 cm and plant spacing of 40 cm. Compound fertilizer (N, P₂O₅, and K₂O: 15% each, 400 kg/ha, Yuntianhua Corporation, Kunming, China) were applied to soil with tillage as a base fertilizer. At seedling, jointing, flowering, and fruiting stages of cucumber, carbamide (N 46%, Xinlianxin Corporation, Xinxiang, China) was fertilized with irrigation water four times after manuring, with 37.5 kg/ha nitrogen each time. A subsurface drip irrigation system was installed in the test plots before the seedling cucumber was transplanted, and each plot had its own independent valve control. One pressure-compensating drip belt (made in Netafim Corporation, Haifa, Israel, with 40 cm spacing, 1.6 L/h flow rate and 5% manufacturing coefficient of variation) was buried beneath each planting row of cucumber in advance.

A 2-factor, 3-level randomized block design was used in the trial. The two factors were the depth of drip belt buried (D) and the amount of irrigation water (W), and after carefully thinking about the similar scenarios reported by former studies, the three levels of depth of drip belt buried were 10 cm (D₁), 20 cm (D₂), and 30 cm (D₃). The three levels of irrigation water amount (W₁, W₂, and W₃) were scheduled by the water evaporation based on a 20 cm diameter pan and calculated by the following equation [25,26] with three different K_pi values (1):

W_i = A × E_p × K_pi/1000

(1)

where W_i is irrigation water amount (m³), A is irrigation area (m²), K_pi is the coefficient of the pan, E_p is accumulative water evaporation based on the pan (mm), and when the sum of daily E_p attained 20 mm, the irrigation valves of all plots would be turned on meanwhile for irrigation, and the end time of irrigation was determined by the individual flow meter of each plot, which could measure the irrigated water precisely. When the accumulated flow reached the designed irrigation amount, the valve would be closed to ensure the accuracy of treatments. Three K_pi levels (W₁, W₂, and W₃) were set with 1.2, 1.0, and 0.80 in 2016, but serious side infiltration of irrigation water beyond the crop planting area was found at the seedling stage of cucumber, which indicated excessive irrigation occurred, and so K_pi level was changed to 0.80, 0.60, and 0.45 in 2017. The detailed irrigation times and irrigation volumes of each treatment are shown in

Table 2. Irrigation amount (mm) of different treatments in both years. Since the cucumber seedlings of these two years were transplanted into the greenhouse at the end of August, only surface irrigation with the same amount of irrigation was applied. IF means irrigation frequency, and None indicates no irrigation in Nov. 2016.

Treat.	Aug.		Sep.		Oct.		Nov.		Total
	2016	2017	2016	2017	2016	2017	2016	2017	2016	2017
W1	53.00	53.00	71.59	48.86	47.73	16.29	None	32.58	172.32	150.73
W2	53.00	53.00	60.23	36.36	40.15	12.12	None	24.24	153.38	125.72
W3	53.00	53.00	48.86	27.27	32.58	9.09	None	18.18	134.44	107.54
IF	1	1	3	3	2	1	None	2	6	7

. The water evaporation pan was placed at one plot, and the height of placement changed according to canopy height of cucumber increasing. In order to compare differences between SDI and DI, one DI treatment with W₁ irrigation amount was set as CK in both years. Thus, the total treatments were ten, and each treatment had three duplicates. Before the experiment, the lottery method determined the trial position for each treatment to ensure the objective, and three duplicates of the same treatment were conjoint, and a 30 cm distance was left between two adjacent treatments to prevent side infiltration.

2.3. Measurements

Daily water evaporation (E_p) was measured using a graduated cylinder (accuracy 0.1 mm) at 8:00 a.m. every day. Daily soil evaporation (E_s) was measured by the micro lysimeters, and the structures and materials of the micro lysimeters and their test methods referred to the existing literature [28]. The radiation and temperature were collected by a meteorological station (HOBO U30, Onset Computer Corporation, Bourne, MA, USA) installed in the greenhouse with an acquisition frequency of half an hour. Evapotranspiration (ET) in the greenhouse was calculated using a classic water balance method:

ET = P + I_r − F + Q − S + ΔW

(2)

where ET is evapotranspiration (mm), P is precipitation (mm), I_r is irrigation water amount (mm), F is surface runoff (mm), Q is ground water recharge (mm), S is deep leakage (mm), and ΔW is change of soil water storage within 1 m depth (mm). Since the trial was conducted in a greenhouse and the irrigation method was drip irrigation, no precipitation, runoff, or deep leakage occurred during the experiment. The ground water depth was more than 10 m at the experiment location; thus, no ground water recharge happened. The classic water balance Equation (2) changed to Equation (3):

ET = I_r + ΔW

(3)

When the change of soil water storage within 1 m increased, the ΔW was a negative value, and it is a positive value otherwise.

A TDR (Time domain Reflectometry with Intelligent Micro Elements, TRIME-PICO-IPH, Bonn, Germany) was used to test changes of soil water content within 1 m, and measuring tubes with 42 mm diameter and 1.2 m long were buried vertically in advance in different plots and the tube mouth was 20 cm above the ground with 3 tubes in each plot. When testing, the sensor of TDR was put into each tube for different depths of soil water content with interval of 20 cm till to 1 m, and the acquisition data were transferred to a computer through Bluetooth.

Along with the cucumber fruits being mature, marketable yields of treatments were harvested and weighed timely, and the weighed yield of each time was summed up as the final yield of each treatment. Water use efficiency (WUE) and irrigation water use efficiency (IWUE) were calculated as follows,

WUE = Y/ET

(4)

IWUE = Y/I

(5)

where Y is the weight of final yield (kg) of each treatment, ET is the evapotranspiration (m³), and I is the total irrigation water amount (m³).

At the end of the experiment, one steel root drill with 10 cm diameter and 20 cm length at the top was used to take root from soil after the shoots of cucumber were cut off in 2017. A root drill was put above the below-ground remaining of cucumber, and then the drill was squeezed to 20 cm depth each layer till to 1 m. The different depths of below-ground matters were separated from soil through sieving and washing procedures. At last, roots were put in an oven for dry matters, and an auncel with 0.01 g accuracy was used to weigh these dry matters.

2.4. Statistical Analysis

Data used in this article are a mean of three duplicates. Main effects of experimental factors were analyzed by the procedure of ANOVA with Duncan’s multiple range test (SAS Institute, Ver8.1, Raleigh, NC, USA), and the differences between treatments with Least Significant Difference. All figures in this paper were created with EXCEL 2013 (Microsoft Office, Ver 2013, Redmond, WA, USA).

3. Results

3.1. Changes of Meterological Factors, Water Evaporation (E_p) in Greenhouse

The meteorological factors in the greenhouse including daily accumulated radiation (R) and maximum temperature (Tmax) are detailed in Figure 2. The R and Tmax ranged from 11.5 MJ to 30.6 MJ and 31.1 °C to 38.8 °C in 2016, from 11.8 MJ to 25.4 MJ and 31.2 °C to 36.7 °C in 2017. The biggest monthly average of R and Tmax existed in September in both years and declined in turn with October, November in 2016 and November, October in 2017, and the distribution of light and temperature in 2017 was even more than that in 2016 at which point they concentrated most in September. The total R and Tmax of the whole season were 1520.7 MJ and 3045.2 °C in 2016, 1553.4 MJ and 3058.4 °C in 2017; the differences of the total R and Tmax between 2016 and 2017 were tiny.

The changes of E_p in both years are presented in Figure 3. The E_p was tested from 31 August to 29 November with accumulated 92 days in 2016, and the monthly values of E_p in September, October, and November was 82.8 mm, 24.3 mm, and 19.7 mm, respectively, with seasonal accumulated E_p 126.8 mm. In 2017, the E_p was measured from 30 August to 3 December for a total of 96 days, and the monthly E_p in September, October, and November was 62.1 mm, 35.6 mm, and 38.8 mm, respectively, with seasonal accumulated E_p 136.5 mm. The changes of E_p both in 2016 and in 2017 abided by the changing laws of meteorological factors in the greenhouse, and the average E_p both in 2016 and 2017 almost shared the same value with 1.5 mm per day.

3.2. Soil Evaporations (E_s) and Evapotranspiration (ET) in the Greenhouse and Its Relationship with Irrigation Amount

Monthly and season total E_s of each treatment are shown in

Table 3. Soil evaporation (E_s) of cucumber of different treatments in one month. The total is a sum of E_s in each month within the same year.

Treat.	Es (mm)
	Sep.		Oct.		Nov.		Total
	2016	2017	2016	2017	2016	2017	2016	2017
CK	44.85 a	25.67 ab	13.68 ab	10.01 a	10.14 a	8.45 b	68.67 a	44.13 ab
D1W1	44.87 a	24.34 bc	13.51 abc	9.95 ab	9.77 ab	8.33 b	68.15 a	42.62 b
D1W2	45.54 a	26.75 a	14.42 ab	10.49 a	10.92 a	10.29 a	70.88 a	47.53 a
D1W3	48.46 a	24.68 b	13.11 abc	9.42 abc	9.75 ab	8.08 bc	71.32 a	42.18 b
D2W1	38.75 bc	22.81 cd	14.54 a	9.68 abc	8.89 ab	9.61 a	62.18 b	42.11 b
D2W2	34.01 cd	22.08 de	12.10 abc	9.71 ab	8.29 b	7.98 bc	54.40 d	39.77 c
D2W3	34.25 cd	20.62 ef	11.26 c	8.59 c	9.92 ab	7.16 d	55.43 cd	36.37 d
D3W1	39.70 b	21.26 def	12.72 abc	9.18 bc	8.40 b	7.49 cd	60.82 bc	37.92 cd
D3W2	32.60 d	19.94 f	12.85 abc	8.84 bc	9.09 ab	8.05 bc	54.54 d	36.83 d
D3W3	26.97 e	21.82 de	11.85 bc	8.89 bc	8.20 b	6.75 d	47.02 e	37.46 d
Anova
D	**	**	ns	*	*	**	**	**
W	ns	ns	*	ns	ns	**	ns	**
D × W	*	*	*	ns	ns	**	**	**

Values followed by different letters within the same column are significantly different at p < 0.05 probability level. * represents significance at p < 0.05 level, ** represents extreme significance at p < 0.01 level, ns represents not significant to the main effect.

. In 2016, the total E_s of each treatment ranged from 47.02 mm to 71.32 mm. The monthly values were sorted from great to small as September, October, and November, which was consistent with the monthly sequence of E_p, and the biggest total E_s belonged to the treatment of D₁W₃ with 71.32 mm, but no significant differences were found among CK, D₁W₁, D₁W₂, and D₁W₃ treatments, and the smallest one was the treatment of D₃W₃ with 47.02 mm. In 2017, the total E_s of treatments changed between 36.37 mm and 47.53 mm, and the monthly values were sorted from great to small as September, October, and November too, but the differences of E_s between October and November were tiny. The biggest total E_s laid in the treatment of D₁W₂ with 47.53 mm, but no significant difference existed with CK. The smallest total E_s was the treatment of D₂W₃ with 36.37 mm, and no significant difference existed among D₂W₃, D₃W₂, and D₃W₃ treatments. Furthermore, the depth of drip belt buried and the amount of irrigation water applied could all have extremely significant (p < 0.01) effects on the total E_s in 2017, but only the depth of drip belt took an extremely significant effect on the total E_s in 2016, and the influence of the amount of irrigation water on the total E_s was not significant, and the interaction effects of the two factors were extremely significant in both years as well. The E_s decreased along with the deepening depth of drip belt buried or the lessening amount of irrigation water applied. In addition, one important clue should be noted: there were no significant differences in E_s between CK and D₁ treatments in both years.

Monthly and season total ET of each treatment is listed in

Table 4. Evapotranspiration (ET) of cucumber of different treatments in one month. The total is a sum of evapotranspiration in each month within the same year.

Treat.	ET (mm)
	Sep.		Oct.		Nov.		Total
	2016	2017	2016	2017	2016	2017	2016	2017
CK	83.60 a	62.78 a	42.01 a	39.17 a	31.65 a	40.97 a	157.26 a	142.92 a
D1W1	81.25 a	60.43 a	41.74 a	38.21 a	30.31 a	40.56 a	153.30 a	139.20 a
D1W2	74.38 b	56.38 ab	35.27 b	31.79 bc	25.25 ab	36.83 ab	134.90 c	125.00 b
D1W3	71.23 b	50.12 bc	30.26 b	26.73 cd	18.73 cd	30.21 bc	120.22 d	107.06 d
D2W1	73.09 b	57.23 ab	42.89 a	38.12 a	26.74 ab	39.21 ab	142.72 b	134.56 a
D2W2	63.54 c	51.23 bc	32.21 b	30.29 bc	23.36 bc	32.53 bc	119.11 d	114.05 c
D2W3	60.13 c	47.22 cd	30.88 b	24.21 cd	18.21 cd	28.29 cd	109.22 e	99.72 e
D3W1	60.41 c	51.24 bc	35.26 b	35.24 ab	26.71 ab	38.25 ab	122.38 d	124.73 b
D3W2	55.36 cd	48.32 cd	30.27 b	28.12 bc	22.21 bc	34.25 bc	107.84 e	110.69 cd
D3W3	50.28 d	44.36 d	22.36 c	22.23 d	10.34 d	25.61 d	82.98 f	92.20 f
Anova
D	**	**	ns	*	*	*	**	**
W	*	ns	ns	ns	*	ns	**	*
D × W	*	*	**	ns	*	*	**	**

Values followed by different letters within the same column are significantly different at p < 0.05 probability level. * represents significance at p < 0.05 level, ** represents extreme significance at p < 0.01 level, ns represents not significant to the main effect.

. From the data point of view, the intensive water consumption period of cucumber was in September in both years, and ET in October or November were only half of that in September, which agreed with the changing laws of the E_p. The total ET of treatments ranged from 82.98 mm to 157.26 mm in 2016 and 92.20 mm to 142.92 mm in 2017, respectively. Total ET reduced with the increasing burial depth of drip belt or the decreasing irrigation water amount significantly or extreme significantly in both years, and the interaction effect of these two factors on total ET was extreme significant too. Thus, the maximum total ET was the CK with 157.26 mm in 2016 and 142.92 mm in 2017, respectively, for its surface drip irrigation method and the greatest irrigation amount amongst all treatments, and the least ET always belonged to the treatment of D₃W₃ with 82.98 mm and 92.20 mm for its deepest depth of drip belt buried and least irrigation water amount in both years. In addition, there were no significant differences on ET between CK and D₁W₁ treatment in both years.

Daily mean value of ET within a month in both years is illustrated in Figure 4. The daily mean value of ET ranged from 1.68 mm to 2.79 mm, 0.72 mm to 1.36 mm and 0.34 mm to 1.06 mm on September, October, and November, respectively, in 2016, and 1.48 mm to 2.09 mm, 0.72 mm to 1.26 mm and 0.85 mm to 1.37 mm on September, October, and November, respectively, in 2017.

ET contained two parts, E_s and plant transpiration (Tr). Tr is thought to be necessary for photosynthesis and dry matter accumulation in plants, but E_s is considered as an ineffective loss. Therefore, it is meaningful to study the proportion between them to improve the efficiency of agricultural water use. The proportion of E_s and Tr accounting for the total ET is illustrated in Figure 5, and Tr was calculated through ET subtracting E_s. The proportion of Tr ranged from 0.41~0.56 in 2016, and 0.59~0.69 in 2017 among treatments. The greatest proportion of Tr was attributed to CK treatment in both years, and the smallest one belonged to D₁W₃ treatment in 2016 and D₃W₃ treatment in 2017. Furthermore, it is clear that the proportion of Tr would get bigger along with the increase of irrigation water amount in both years, but the effect of the depth of drip belt buried was dim, and the surface drip irrigation (CK) was greater than the subsurface drip irrigation.

Relationships between ET and irrigation water amount (I_r) under different depths of drip belt buried are illustrated on Figure 6. The relationships between ET and I_r are extremely significant positive linear correlations, which means that ET would augment significantly along with the increasing amount of irrigation water applied within the scope of the trial. D₁ and D₂ treatments almost share the same slope of the formula, but that of the D₃ treatment is smaller than them and its regression line is flat. Furthermore, their decision coefficients of formulas decreased along with the depth of drip belt buried deepening, which indicated the correlation between ET and I_r got worse along with the depth of drip belt deepened.

3.3. Changes of K_p in Different Months of Cucumber

The K_p of different treatments, which is the quotient of ET and E_p, are listed in

Table 5. Pan coefficient (K_p) of each treatment in a month. Value is a quotient of evapotranspiration (ET) and water evaporation (E_p).

Treat.	Kp
	Sep.		Oct.		Nov.
	2016	2017	2016	2017	2016	2017
CK	1.01	1.01	1.73	1.10	1.61	1.06
D1W1	0.98	0.97	1.72	1.07	1.54	1.05
D1W2	0.90	0.91	1.45	0.89	1.28	0.95
D1W3	0.86	0.81	1.25	0.75	0.95	0.78
D2W1	0.88	0.92	1.77	1.07	1.36	1.01
D2W2	0.77	0.82	1.33	0.85	1.19	0.84
D2W3	0.73	0.76	1.27	0.68	0.92	0.73
D3W1	0.73	0.83	1.45	0.99	1.36	0.99
D3W2	0.67	0.78	1.25	0.79	1.13	0.88
D3W3	0.61	0.71	0.92	0.62	0.52	0.66

. Based on these data, one rule could be concluded: K_p would get smaller along with the increasing depth of drip belt buried or the decreased amount of irrigation water applied for the reason of reduced ET. Thus, the greatest K_p appeared in the treatment of CK, and the smallest one laid in D₃W₃ treatment, and the sequence of monthly K_p from great to small was October, November, and September. In addition, there were differences on the values of K_p between 2016 and 2017 within a month, and K_p values in October and November in 2016 were obviously bigger than that in 2017 for the obviously smaller E_p in the corresponding month.

3.4. Responses of Below-Ground Dry Matters of Cucumber

Below-ground dry matters of cucumber in 2017 were detailed in

Table 6. Root dry matters at different layers of soil of cucumber.

Treat.	Root Dry Matter (10−2 g)
	0~20 cm	20~40 cm	40~60 cm	60~80 cm	80~100 cm
CK	48.00 d	2.10 d	1.59 c	0.70 a	0.20 bc
D1W1	49.03 d	2.12 d	1.59 c	0.72 a	0.21 bc
D1W2	67.10 bcd	3.72 c	1.55 c	0.72 a	1.09 a
D1W3	61.71 bcd	6.90 a	2.25 b	1.02 a	0.66 abc
D2W1	78.13 bc	3.96 c	1.24 d	0.73 a	0.66 abc
D2W2	59.01 cd	3.02 cd	4.48 a	0.93 a	0.33 bc
D2W3	64.45 bcd	5.19 b	1.99 bc	1.66 a	0.89 ab
D3W1	95.02 a	3.45 c	1.41 c	1.47 a	0.46 abc
D3W2	86.44 b	5.86 b	1.61 c	1.09 a	0.15 c
D3W3	57.03 cd	6.47 a	1.51 c	0.68 a	0.83 abc
Anova
D	**	*	*	ns	ns
W	*	*	*	ns	ns
D × W	**	*	*	ns	ns

Values followed by different letters within the same column are significantly different at p < 0.05 probability level. * represents significance at p < 0.05 level, ** represents extreme significance at p < 0.01 level, ns represents not significant to the main effect.

. As we can see, roots of cucumber were mainly concentrated in the soil layer of 0~20 cm, and with the increase of depth of soil layer, the root dry matters decreased sharply. The greatest root dry matter in 0~20 cm laid in the treatment of D₃W₁, and the smallest one belonged to the treatment CK. Depth of drip belt buried or irrigation water amount influenced the root dry weights of the 0~60 cm soil layers significantly, and their interaction effects also reached significance too, but no significant influence was found on those roots below 60 cm soil layers, and the influencing laws of the two factors on the roots growth of different soil layers were different yet. Root dry matters in 0~20 cm increased with the increasing of depth of drip belt buried and the amount of irrigation water significantly. Root dry matters in 20~40 cm increased with the increasing depth of drip belt buried as well, but decreased with the increasing irrigation amount. As to the root dry matter of 40~60 cm, the two factors could all play a significant role, but the influencing laws were more complex than above two layers. The effects of the drip belt depth were sorted from great to small as D₂, D₁, and D₃, and there were significant differences between the D₂ treatment and D₁, D₃ treatments but no significant difference occurred between the D₁ treatment and the D₃ treatment. Furthermore, the effect of irrigation amount from great to small was W₂, W₃, and W₁, and there were significant differences between the W₂, W₃ treatments and the W₁ treatment, but no significant difference existed between the W₂ treatment and the W₃ treatment.

3.5. Responses of Yield, Water Use Efficiency (WUE), and Irrigation Water Use Efficiency (IWUE) of Cucumber

Table 7. Yield, water use efficiency (WUE), and irrigation water use efficiency (IWUE) of cucumber of different treatments.

Treat	Yield (t·ha−1)		ET (mm)		WUE (kg·m−3)		Irrigation Amount (mm)		IWUE (kg·m−3)
	2016	2017	2016	2017	2016	2017	2016	2017	2016	2017
CK	65.70 ab	60.37 ab	157.26 a	142.92 a	41.78 bc	42.24 bc	172.32	150.73	38.13	40.05
D1W1	64.31 ab	59.03 ab	153.30 a	139.20 a	41.95 bc	42.41 bc	172.32	150.73	37.32	39.16
D1W2	60.25 b	54.22 b	134.90 c	125.00 b	44.66 bc	43.38 b	153.38	125.72	39.28	43.13
D1W3	69.21 a	53.39 bc	120.22 d	107.06 d	57.57 a	49.87 ab	134.44	107.54	51.48	49.65
D2W1	62.39 b	64.19 a	142.72 b	134.56 a	43.71 bc	47.71 ab	172.32	150.73	36.21	42.59
D2W2	70.34 a	54.61 b	119.11 d	114.05 c	59.05 a	47.88 ab	153.38	125.72	45.86	43.44
D2W3	63.54 b	50.73 c	109.22 e	99.72 e	58.18 a	50.87 a	134.44	107.54	47.26	47.17
D3W1	50.14 c	49.09 cd	122.38 d	124.73 b	40.97 c	39.36 c	172.32	150.73	29.10	32.57
D3W2	45.32 cd	45.59 d	107.84 e	110.69 cd	42.03 cd	41.19 bc	153.38	125.72	29.55	36.26
D3W3	40.28 d	36.99 e	82.98 f	92.20 f	48.54 b	40.12 bc	134.44	107.54	29.96	34.40
Anova
D	**	**	**	**	*	*	None	None	None	None
W	**	*	**	*	**	**	None	None	None	None
D × W	**	**	**	**	*	*	None	None	None	None

There were no statistical analyses for the irrigation amount and IWUE items. In order to make a comparison with former results, the unit of WUE and IWUE was converted to kg/m³.Values followed by different letters within the same column are significantly different at p < 0.05 probability level. * represents significance at p < 0.05 level, ** represents extreme significance at p < 0.01 level, ns represents not significant to the main effect.

detailed the data of yield, WUE, and IWUE of cucumber in this experiment. The two factors could all influence the yield of cucumber significantly, and its interaction effect on the yield of cucumber was significant as well. The effects of drip belt depth were sorted from great to small as D₂, D₁, and D₃, and there was significant difference between the D₂, D₁ treatments and the D₃ treatment but no significant difference existing between the D₂ treatment and D₁ treatment. The effects of irrigation amount were sorted from great to small as W₁, W₂, and W₃ in both years, and there were significant differences between the W₁, W₂ treatments and W₃ treatment but no significant difference existing between the W₁ treatment and W₂ treatment in 2016, and significant difference existed between the W₁ treatment and the W₂ treatment in 2017. Therefore, the greatest yield laid in the D₂W₂ treatment in 2016 and the D₂W₁ treatment in 2017, and the smallest one belonged to the D₃W₃ treatment in both years. In addition, there were no significant differences in the yields of cucumber among the CK, D₁W₁, D₁W₃, and D₂W₂ treatments in 2016 and the CK, D₁W₁, and D₂W₁ treatments in 2017, which indicated that some kind of transformations would occur under certain conditions.

WUE and IWUE is an important and popular indicator for evaluating whether new irrigation technology could really work in water saving and yield improving in practice. The yield, ET, and irrigation amount data are given in Table 7. WUE and IWUE of treatments in this experiment could be calculated through Equations (4) and (5). The two factors could all play significant roles on WUE as well. The WUE increased with the decrease of irrigation amount, but the effect of depth of drip belt buried on WUE was more complex than irrigation amount, and the sequence of depth of drip belt buried on WUE from great to small was D₂, D₁, and D₃, and significant differences existed between any two of them. Therefore, the greatest WUE laid in D₂W₂ treatment in 2016 and D₂W₃ treatment in 2017 for the better combination between yield and ET, and the smallest one belonged to D₃W₁ treatment for its significant yield reduction in both years. In addition, because of the greatest ET among all treatments, CK got smaller WUE, which even had no significant difference with that of D₃W₁ treatment in this experiment. As to IWUE, the values of IWUE of D₃ treatments were smaller than that of WUE obviously, which meant that the irrigation water amounts in D₃ treatments were bigger than their crop water requirements in the experiment. Thus, the greatest IWUE belonged to the D₁W₃ treatment, which was somewhat different from that of WUE, and IWUE and WUE shared the same smallest treatment of D₃W₁.

A regression analysis was performed between yield and ET (Figure 7), and the yield of cucumber and its ET presented an extreme significant quadratic polynomial correlation in both years. The yield of cucumber would augment with the increasing of ET to a point of inflexion then decrease with the persistent increase of ET. Through optimization analysis, the optimum values of ET appeared at 139.3 mm in 2016 and 177.3 mm in 2017, respectively, at which point the greatest yield of cucumber planted in greenhouse during the autumn–winter cycle would be harvested.

4. Discussion

SDI systems make it possible to apply slow, steady, and uniform amounts of water and nutrients within the plant’s root zone, while minimizing deep percolation and maintaining high productivity levels, and the wetted area on the soil surface should be reduced in order to decrease soil evaporation [4,7]. Compared with DI and other irrigation methods, SDI possess the important virtue of reducing soil evaporation (E_s) just right. According to E_s data of this experiment, E_s of cucumber under SDI decreased along with the increasing depth of drip belt buried or the decreasing amount of irrigation water applied, but two important findings should be noted in advance.

Firstly, there were no significant differences on E_s between CK and D₁ treatments after two years’ studies, which meant the drip belt should be buried more than 10 cm deep to achieve the purpose of reducing E_s for the soil texture of loam. Usually, the smaller the soil particle size is, the deeper the drip irrigation belt should be buried. However, too deep depth would make the plant root water uptake unavailable in the early growing stage of the plant and result in dysplasia of the plant and ultimate yield penalty, and this had been confirmed by the yield loss of D₃ treatments in the experiment. Considering the combined effect of reducing water consumption and increasing yield of cucumber, the suitable depth of the drip belt should be buried 20cm for the loamy soil texture, and similar results were found in other studies [14,16].

Secondly, the influence of irrigation amount was not significant on the total E_s in 2016, which indicated that if irrigation water amount was applied too heavily, the E_s of irrigation water treatments under the same depth of drip belt buried would have no significant differences due to the oversize irrigation, and this finding needs to be verified through the experiment of soil hydrodynamics in the future to get the boundary to a particular soil texture.

The relationships between ET and I_r were extreme significant positive linear correlations, but yield of cucumber and ET presented a quadratic polynomial correlation in both years. Therefore, increasing cucumber yield simply by increasing irrigation is not feasible. The point of inflexion of ET at which the yield of cucumber would obtain the greatest should be taken into account when scheduling irrigation, and irrigation amount beyond the point of inflexion would lead to the reduction of yield of cucumber and the decline of WUE. For this possible reason, too much irrigation water beyond water requirement of crop would make roots under waterlogging conditions, which made roots in hypoxia restrict normal growth and development of the crop. Furthermore, heavy irrigation led to water deep percolation, which wasted electricity power used for irrigation system and decreased the IWUE.

It has been proved that the proportion of transpiration (Tr) of cucumber got bigger along with the increase of irrigation amount, but the yields of treatments did not always increase with Tr yet, which indicated a luxury transpiration might exist in the process of physiological metabolism of cucumber [17,32]. In addition, the point of inflexion was not always the same every year, and it was influenced by meteorological factors or cultivation methods in the greenhouse. That is the reason the ET point of inflexion in 2017 was bigger than that in 2016.

Root growth patterns are altered in response to changes in soil conditions [15,16,20]. According to the research results of below-ground dry matters of cucumber, the roots were significantly affected by the irrigation method, and the weight of root dry matter under SDI was greater than that under DI (CK), and the deeper the drip belts were buried, the greater the root dry weights were. For possible reasons, the increase of the depth of drip belt buried let plants not receive enough root water uptake in the early stage of cucumber, putting it under water stress, which instead promotes the development of cucumber roots. Klepper (1991) [33] pointed out that there was a significant loss of root material in the top meter profile and gain in the lower meter of cotton when water deficit occurred. Moreover, the drip irrigation is a kind of localized irrigation, and it could only work on those roots’ growth living in the plan wet layer of soil, which might change in response to the changes of depth of drip belt buried or irrigation amount. Thus, the influencing laws of the two experimental factors were not always the same on the root dry weights formed in different depths of soil, and the two factors only took effect on the root dry weights of 0~60 cm. In addition, the root taking method in this experiment is through root dill with only 10 cm diameter at the top, which cannot load all roots in soil, and some important information might be missed. In fact, irrigation method and regime could not only take effect on root dry weight but also root structure and type. Pisciotta (2018) [34] pointed out that SDI could result in a higher density of root contacts and incidence of fine roots compared to DI.

Using a 20 cm water evaporation pan to schedule irrigation regime, the pan coefficient (K_p) should be known first so that users could calculate accurate irrigation water amount using the product between K_p and E_p in practice. Through optimizing irrigation amount and timing, crop yield and WUE could be increased by improving canopy structure and microenvironment under irrigation agriculture. In this experiment, September is an important month to the cucumber planted in an autumn–winter cycle greenhouse when light and temperature conditions are enough in the greenhouse and the cucumbers go through the seedling stage to flower and fruit stage with plant height and leaf area fast expanding. Therefore, September is the greatest soil water consumption period during the whole growing season of cucumber, in which accounting for 50~60% total ET, with 2~3 mm daily soil water consumption in greenhouse, and the irrigation frequency was the highest meanwhile. The calculating values of K_p were smaller than 1 in September, which meant ET was smaller than E_p at this stage, and the soil evaporation accounted for a large proportion of ET. Along with the season changing from autumn to winter in the following two months, the evaporation conditions in the greenhouse such as temperature and radiation go down quickly, and the ET of cucumber declined sharply with only 1~1.5 mm daily water consumption, and the calculating values of K_p were bigger than 1, which meant the ET was greater than the E_p, and plant transpiration dominated at this period. Referring to the phase water consumption of the greatest yield treatment in both years, the changing K_p according to different months should be set to 0.8~0.9, 1.1~1.4, and 1.0~1.2 at September, October, and November, respectively, and the K_p values determined in this experiment partially agree with that of Zhang (2010) [18] derived from weighting lysimeters.

Although WUE and IWUE are important indicators for saving water agriculture, the ultimate pursuit is the greatest yield of crop, not the pure highest WUE. Based on the yield, WUE, and IWUE data of the experiment, the greatest yield of cucumber laid in the D₂W₂ treatment in 2016 and the D₂W₁ treatment in 2017, which yielded higher cucumber production at 7.06% in 2016 and 6.33% in 2017, and reduced ET 24.26% in 2016 and 5.85% in 2017, improved WUE 41.34% in 2016 and 12.95% in 2017 and IWUE 16.86% and 6.34% compared with CK. This proved the SDI was a better choice than DI in water saving and WUE improving in the North China Plain. The highest WUE in 2016 approached 59.05 kg/m³, which was higher than some of the existing studies [35,36].

However, one shortcoming should be noticed that emitter clogging by either roots or soil particles may compromise water delivery in SDI systems, so the addition of chemical drugs or the air injection at regular intervals to remove the emitter clogging is also necessary.

5. Conclusions

Based on the results of the 2-year studies, E_s and ET of cucumber under SDI decreased along with the increasing depth of drip belt buried or the decreasing amount of irrigation water applied, and the proportions of transpiration accounting for ET were about 41~69% in two years. K_p got smaller along with the increase of depth of drip belt buried or the decrease of irrigation water applied as well for the reduced ET. Roots of cucumber mainly concentrated in the soil layer of 0~20 cm, and the weights of below-ground dry matters under SDI were greater than that under DI. Both the depth of drip belt buried and amount of irrigation water applied affected yield and WUE of cucumber significantly. Optimized ET of cucumber planted in an autumn–winter cycle greenhouse would be 140~180 mm given the water evaporation reached 130~140 mm, and the appropriate WUE was about 55 kg/m³ at which the yield and ET would have a better combination.

Based on the data in this two-year experiment, an ultimate subsurface drip irrigation scheduling mode based on water evaporation could be summarized as irrigation start with the accumulated water evaporation attaining 20 mm, and the appropriate K_p is set to 0.8~0.9 in September, 1.1~1.4 in October, and 1.0~1.2 in November, respectively, and a single irrigation amount can be calculated through the product of K_p, E_p, and irrigation area, and the drip belts are buried 20 cm for the loamy soil texture. If the above conditions are met, cucumber planted in the autumn–winter greenhouse will obtain a higher yield and WUE in the North China Plain.

If the technology of subsurface drip irrigation scheduling based on water evaporation is combined with the automatic control and information technologies, it would not only realize the sustainable development of agriculture water resource but also offer an easy, low cost, and efficient method for the smart irrigation management, which will benefit the reduction of labor intensity and the lack of labor force in the future.