Salt Distribution in a Soil Irrigated by Subsurface Emitter

The best design of subsurface trickle irrigation systems requires knowledge of water and salt distribution patterns around the emitters that match the root extraction and minimize water losses. The transient distribution of water and salt in a two-dimensional homogeneous Iraqi soil domain under subsurface trickle irrigation with different settings of an emitter is investigated numerically using 2D-HYDRUS software. Three types of Iraqi soil were selected. The effect of altering different values of water application rate and initial soil water content was investigated in the developed model. The coefficient of correlation (R) and the root-mean-square error (RMSE) was used to validate the predicted numerical result. This statistical analysis revealed that there was no much difference between the predicted numerical results, and the measured values. R varied from 0.75 to 0.93 and the (RMSE) from 0.079 to 0.116. The comparison confirms the accuracy of the developed model, and it shows that it can be used to simulate the front wetting patterns of water and salt distribution under subsurface trickle irrigation systems. The simulation outcome showed that as the distance from the emitter increased, soil salinity far from the emitter decreased. As expected, irrigation duration and amount affects the dimension of the solute distribution.


INTRODUCTION
Subsurface trickle irrigation system is increasingly used in regions with limited water resources to irrigate crops. The best design of subsurface trickle irrigation systems requires knowledge of soil salt and water distribution patterns around the emitters that match the root extraction and minimize water losses. Numerical models can be used to simulate the transport of solutes and the flow of water in the subsurface. HYDRUS-2D software is considered a useful tool for simulating water flow and salt distribution under drip irrigation. The model numerically solves the Richards governing equation in two-dimensional coordinates for variablysaturated soil water flow and convection dispersion equation for solute transport. Several studies have been conducted on soil water with salt allocation beneath subsurface drip irrigation (SDI) with saline irrigation water. (Roberts, et al., 2009) used the HYDRUS-2D model to predict pattern salt accumulation from an SDI system on the yield of cantaloupe and broccoli for various germination practices and salt for water between (1.5 -2.6 dS/m), and forecast salt allocation after a whole plant season. The soil is a sandy loam, and van Genuchten parameters were specified using Rosetta by entering the particle volume allocation and bulk density. Average initial soil profiles (saturated soil electrical conductivity ECe) to a deep of 30 cm was approximately 2.4 dS/m. The sink was used to calculate the water consumption by plant roots. The equation of Fickian-based convectiondispersion was used to described solute transport. Solute transport and water flow equations were solved by Galerkin linear finite elements numerically. Solutes were assumed that nonreactive and no net solubilization or dissolution. These assumptions allowed the salinity to be simulated based on the convection-dispersion equation for nonreactive solutes. The salts closing the roof may decrease the hydraulic conductivity for soil and cause extra runoff from the soil roof. Correlations show the model's capability to forecast ECe can be effective when the insert parameters like root allocation are noted. (Kandelous and Simunek, 2010) investigated simulated water movement in the soil from the subsurface emitter and estimated distance of the wetted area. A subsurface emitter irrigation system executed two groups of tests. One group tests were complete in the laboratory, and other group tests were executed in the field. The laboratory tests were executed on a lysimetric (200 ×100 ×1200 cm) full of a clay loam soil. Field tests were executed at the fieldwork on a clay loam soil, the depth of emitters were 5, 15, 25 cm in the soil like that in the laboratory. The method of Galerkin finite-element used HYDRUS-2D to solve the equation of water flow governing. The domain was rectangular (100 and 150 cm). A comparison of calculated and simulated data proved that HYDRUS-2D could be the best model to help in designing and control subsurface emitter irrigation systems to soils with various textures. (Kahlaoui, et al., 2011) used subsurface and surface emitter irrigation with water salinity (6.57 dS/m) on three tomato types. The goal was to develop water irrigation management beneath saline conditions and assess the reactions of three tomato types to many irrigation methods. The test was defined using drips buried at 30 cm deep in the subsurface and on the surface of the soil. Irrigation system owns a significant influence on the salt of soil ECe, and it was obviously salt higher in surface emitter than in subsurface emitter. The salts was increased because of raising the evaporation level close soil roof in the situation of the surface emitter. Salts can move down and extend to the root zone for the time of spread uniform of water and precipitation on the soil surface. This action may prevent nutrient uptake and water; therefore, it makes negative effects on yield and crop growth. (Selim, et al., 2013 a) investigated brackish irrigation water with salinity and soil water distribution under (surface drip irrigation DI). They used HYDRUS-2D simulation by supposing tomato plants with salinity soil. They found the influence of the irrigation system on moisture shape varies with the hydraulic soil characteristic, the influence of (initial soil moisture content θi) will vanish the next few days and don't have any important effect on distributions of soil salinity. In the sand, soil salinity was found near the emitter match to loamy sand and sandy loam. They used a 1.4 m drip line spacing and 35 cm drip spacing. When the first irrigation was ended, the wetted depth was smaller in sandy loam and loamy sand than in sand, because sand is distinguishing by less water-holding ability as compared to loamy sand and sandy loam. A region of high salinity in sand happened (0.4 -0.55 m) away from emitter while it happened (0.50 -0.65 m) for loamy sand and sandy loam, and it can be decreased by distributing mulch to reduce the evaporation rate. The main objective of this research was to study the effect of subsurface drip irrigation with emitter and emitters located at variable depths in the Iraqi soils and of applying different water qualities of the water pattern distribution and salinity distribution in the soil profile and analyzing a twodimensional water flow and solute distribution by using a numerical model (Hydrus-2D).

GOVERNING EQUATIONS 2.1. Water Equation
The Richards equation controlling water flow from a point source through saturated water stream in soils can be written in two-dimensional coordinates as follows (

Domain and Boundary Conditions
The domain to whole scenarios was classified to be axisymmetric about a perpendicular pivot. The water flows from a subsurface emitter is two-dimensional axisymmetric, "half the flow domain required to be simulated in HYDRUS -2D" (Kandelous, et al., 2011; El-Nesr and Alazba, 2017). The single subsurface emitter was placed at the upper left of the domain. Three emitter depths were used in this study (10, 20, and 30) cm. The simulation was carried out on a rectangle domain in this research. The domain was 80 cm in width and 100 cm in depth. The domain was selected large enough to avoid interference in the water content of near emitters. The boundary conditions considered in this research are: the top boundary of the domain was submitting to atmospheric state, while the down boundary became a free drain. The border at both perpendicular directions was specified as "no-flux". Emitters were executing in all status as half circles with a radius of 1 cm, situated on the left vertical boundary of the domain. The emitter was defining a "Variable Flux".  The boundary conditions performing salt as a third-type Cauchy boundary condition that prescribes the salt motion through determine irrigation periods. To run the model, it needed an initial soil moisture content and the hydraulic parameters (Ks, θs, θr, α, n). These hydraulic parameters of the soil were assessed applying the Rosetta Lite software (Schaap et al., 2001), and it was involved in the HYDRUS-2D model. Salinity for soil is a calculation of the concentration of whole dissoluble salts in water and soil, can ordinarily be presented as electrical conductivity (EC). The most commonly used EC units are deciSiemens per meter (dS/m). Soil salt and water wetting patterns from a single subsurface emitter were simulated by using three types of Iraqi soil texture, classified as loam, sandy loam, and silty clay loam soil, respectively. Table 1 illustrates the physical properties of the soils according to the laboratory test of the soil samples. Discharges of the emitter were shown in Table 2, were used in the simulation of the salt and water wetting patterns. In the simulation process, two initial volumetric soil moisture contents were used. The initial water content was limited between the field capacity and the welting point. The soil moisture content at field capacity, wilting point, and initial water content for Iraqi soil are shown in Table 3.

STATISTICAL ANALYSIS
Statistical analysis was used to evaluate a relationship between the predicted results from HYDRUS-2D/3D software and those observed from the field. These parameters include root mean square error (RMSE), the optimal value approaches to zero, and the coefficient of correlation (R 2 ), which has the maximum at 1 when predicted values perfectly match the observed ones (Shan, et al., 2011). N = is the total number of data points, Pi = is the predict data point, Oi = is the observed data and Ō = is the mean of observed.

RESULTS AND ANALYSIS 4.1. Validation of Results
The transient water and salt movement through three types of homogeneous Iraqi soil from a subsurface emitter were simulated. The simulation was done numerically using HYDRUS-2D model to study the distribution of soil salinity and its relationship with soil texture and characterize water and salinity distribution at different depth of subsurface emitter. Equation (1) with the initial and boundary conditions as in Fig. 1 and equation (2) were solved numerically using the finite element method with the HYDRUS-2D model. The accuracy of the predicted numerical results was validated by a comparison made between the predicted numerical results and the measured values of (Shan and Wang, 2012). This comparison is illustrated in Fig. 2 at a different distance from the intersection of overlap between two emitters. The numerical solution is seen to be in good agreement with field measurements. Fig. 3 shows water wetting pattern for a subsurface emitter in silty clay loam, loam and sandy loam soil, "initial soil moisture content" i = 0.18 cm 3 /cm 3 , emitter discharge, Q = 0.4 L/h, EC= 8 dS/m and emitter depth =20 cm after end of simulation time 12 hr (irrigation time 3 hr). The change in water wetting pattern distribution and initial soil moisture content was small. The wetting width and depth, in sandy loam soil, was bigger than in loam and silty clay loam soil, likewise in loam was bigger than in silty clay loam soil. The water wetting was spreading longitudinally and laterally, but the vertical spreading was more significant than the horizontal due to the gravity. In the finer-textured soils, there was less available air-filled pore space that reduced the infiltration capacity causing a considerable fraction of the applied water to move laterally, and this forced water to move far from the emitter. Initial soil moisture content in silty clay loam was bigger than in loam and sandy loam soil, likewise in loam was bigger than in sandy loam soil. The reason was that because the characterized of the soils have lower hydraulic conductivity of sandy loam in addition to the higher water-holding capacity of sandy loam compared with loam and silty clay loam (Abou Lila, et al., 2013). Fig. 4 shows salt distribution for a subsurface emitter in sandy loam, loam and silty clay loam soil, i = 0.18 cm 3 /cm 3 , Q= 0.4 L/h, EC= 8 dS/m and emitter depth =20 cm after end of simulation time (12 hr) (irrigation time 3 hr). The change in salt pattern distribution and salt quantity was very small. The salt width, depth, and quantity in sandy loam soil were bigger than in loam and silty clay loam soil, likewise in loam was bigger than in silty clay loam soil, and that because characterized by the soils. The salt was spreading longitudinally and laterally, but the vertical spreading was larger than the horizontal by 49% due to the gravity. Fig. 5 shows the salt pattern for a subsurface emitter in sandy loam for two i (0.090 and 0.098 cm 3 /cm 3 ) and two Q (0.75 and 1 L/h), EC= 8 dS/m and emitter depth =20 cm after the end of simulation time 3 hr (irrigation time 3 hr). In all soil types, it was found that the wetted salt and wetting volume depended directly on the amount of irrigation water. Higher irrigation amounts initially produced higher salinity quantity and water content near the emitter. The larger lateral extension in sandy loam can be attributed to Journal of Engineering Volume 26 August 2020 Number 8 76 less available air-filled pore space that decreases the infiltration capacity. Besides, the lower soil hydraulic conductivity limits the vertical movement of water and enhances the possibility of water moving laterally (Selim, et al., 2013). Fig. 6 shows salt pattern for a subsurface emitter in silty clay loam soil, i = 0.3 cm 3 /cm 3 , Q= 0.4 L/h, EC= 8 dS/m and emitter depth =10, 20 and 30 cm after end of simulation time 3 hr (irrigation time 3 hr). The salt and water pattern reached the soil surface when the emitter depth is 10 cm, and that caused a rise in the losses of water by evaporation. The vertical spread of salt was larger than the lateral due to gravity; the salt quantity in a depth 10 cm is larger than 20 cm and 30 cm and has the bigger lateral spread. When the emitter depth was 30 cm, the salt and water wetting pattern may by penetrated to deep percolations. In the coarse-textured soil the perfect depth for emitter should be at deep depth (30 cm) and for medium-textured soil should be at a shallow depth (20 cm), noticing that the depth of the root of the plant to guarantee the transport of water straight to the roots (Abid, 2019). Fig. 7 shows salt pattern for a subsurface emitter in loam for two i (0.18 and 0.2 cm 3 /cm 3 ) and two Q (0.3 and 0.4 L/h), EC= 8 dS/m and emitter depth =20 cm after end of simulation time 3 hr (irrigation time 3 hr). It was found that the soil moisture content decreased by 46% as the distance from the emitter increased. When i decreased, the water wetting pattern and salinity pattern decrease too, but salinity quantity increase.

Effect of Irrigation Time on Salt and Water Wetted Soil
It was observed that when the duration of irrigation increases, the width and depth of salt and water pattern distribution increase largely, and almost all distributions pattern reaches the soil surface except when the emitter depth is 30 cm for two scenarios. It was noted that the maximum widths found when the emitter depth was 10 cm for two irrigation duration (12 and 24 hr), and the maximum depth found when emitter depth was 30 cm for two irrigation duration.

Effect of Salt on Crop Yield
In this simulation, the Cucumber plant was used with two values of EC= 8 and 4 dS/m in sandy loam soil. The results showed that irrigation with water has salt 8 dS/m, making the yield reduction =100%. A large amount of freshwater was needed to leach the accumulated salt in the soil, but when irrigating with water of 4 dS/m, the yield reduction became 25% to 50%, and little fresh water was used for leaching the salt.

CONCLUSIONS
1. The salt width, depth, and quantity in sandy loam soil have very little increase more than loam by 0.8%, 1.2%, and 0.3% and by 2.6%, 1.6%, and 0.55% from silty clay loam soil. Likewise, loam increased little more than silty clay loam soil by 1.8%, 0.5%, and 0.3%, as shown in Fig.  4.