Computational evaluations of HYDRUS simulations of drip irrigation in 2D and 3D domains (ii-subsurface emitters)

Highlights

• This is the 2nd part of a surface and subsurface drip simulation two-part study.

• Studied the effect of geometries, shapes, discharges & textures on simulations.

• The mass balance errors of subsurface simulations are less than 0.9%, unlike surface ones.

• We found that the 2D domains mostly underestimate the 3D slices.

• To simulate subsurface drip use 2D or 3D. For surface simulations, use only 3D.

Abstract

This is the second part of the companion papers assessing HYDRUS simulations of drip irrigation. While the first part focused on surface drip simulations, this part focused on the subsurface simulations and how they differ from the surface simulations. We evaluated the effect of different domain geometries, emission element shapes, coordinate systems, emitter discharges, and soil textures on the accuracy and stability of the simulation. The results showed that all the 2D and 3D simulations were done with a mass balance error of less than 0.9%, in contrast to the surface 2D cases that have large error values. The flux-conductivity ratio for subsurface simulations does not have the same influence on the mass balance error as in the surface simulations. We found extreme differences in the simulation speeds that were attributed to the emitter’s location (the subsurface is faster), the domain axes (axisymmetric is faster than Cartesian), soil texture (the lighter is faster), the elapsed time or stage (the redistribution is hugely faster than redistribution). For the 2D-3D comparisons, we found that the 2D domains mostly underestimate the 3D slices. Finally, we found that the flux profiles of the 2D domains perfectly match the flux profiles of the 3D domains, in contrast to the surface comparisons. Hence, we can use 2D simulations reliably for subsurface drip, but not for the surface drip.

Introduction

The popularity of the drip irrigation system encourages more research on optimizing its design and management. The wetting pattern under the dripper is one of the main features that benchmark the goodness of the design and management practices; it reflects the state of irrigation uniformity and water conservation. Simulating the wetting pattern for each design and management scenario saves time and effort, in addition to its widely acceptable results (Kandelous and Šimůnek, 2010, Singh et al., 2006, Yao et al., 2011). Water flow through subsurface drip sources were simulated by several researchers (e.g., Arbat et al., 2013, Cote et al., 2003, Saefuddin et al., 2019). Some models were analytical (e.g., Cote et al., 2003, Schwartzman and Zur, 1986), and some were numerical (e.g., Elmaloglou and Diamantopoulos, 2009, Ismail et al., 2006a, Ismail et al., 2006b). Each category has its advantages and disadvantages, the analytical models are easier to understand and implement, but they lack accuracy due to the approximation, while on the other hand, the numerical models are more accurate but harder to understand and implement. Among several modeling efforts, HYDRUS (Šimůnek et al., 2016) gained a lot of interest and confidence to simulate the wetting pattern of drip irrigation (e.g., Elnesr et al., 2013, Khezri and Irandoust, 2016, Mguidiche et al., 2015, Skaggs et al., 2004). This model uses flux instead of discharge to imitate emitters’ flow rate, which is highly affected by the shape of the emitting element. This work, along with the companion paper (Elnesr and Alazba, 2019) aims to assess the effect of geometries, coordinate systems, and emitting elements on the simulation results of drip irrigation on different soils. This paper processes the subsurface drip simulations, while the companion paper processes the surface drip simulations.