Experimental investigation and modelling of rainfall-induced erosion of semi-arid region soil under various vegetative land covers

Recent decades have witnessed expansion in urbanization associated with an increase in global population and a higher demand for water for different uses, especially for agriculture and drinking. Studies focus on sustainable water management, e.g., watershed management, water harvesting, and flood management. Soil erosion is considered a peristaltic disaster due to the loss of soil nutrients (An et al. 2013) and sequential problems such as water pollution and increased sediment in water bodies. Moreover, two crucial processes: soil detachment and sediment transport, occur during the surface flow. The change in land use due to human activities, especially on steep slopes, is another cause of the increase in erosion rates (Pereira et al. 2015; Cerdà et al. 2016; Prosdocimi et al. 2016). The rate of rainfall-induced soil erosion increases with an increase in total rainfall, with 400 mm/year of rainfall as a threshold for erosion threat (Kirkby 1984).

Land use and land cover (including vegetative covers) have a major role in soil and water conservation management, as they directly affect the processes in the hydrological cycle (surface runoff, evaporation, filtration, etc.), water quantity (Wang et al. 2014), and groundwater recharges (Kader et al. 2017). Soil and water conservation have been studied by many researchers based on field and laboratory scales. However, in-situ soil erosion requires at least 20–25 years of observation to achieve reliable data to study the effect of changing land due to varying intensities of natural rainfall (Wischmeier & Smith 1978). Scanlon et al. (2005) studied the effect of natural vegetation on soil conservation, groundwater recharge, and water quality in the southwestern part of the USA. Disposing agricultural waste such as straw and dry leaves is also a serious problem in water-scarce areas. Farmers are left with no other option but to burn these wastes leading to air pollution. An alternative to using this waste material could be to use it as bare soil covers, which eventually can solve the problem of air pollution, conserve soil nutrients, and at the same time be an effective solution for its disposal. Adekalu et al. (2007) reported that covering soil using agricultural waste effectively conserves water by reducing surface runoff and increasing the infiltration rate.

Studies on surface runoff in arid to semi-arid regions need to cover wide areas due to consideration of several variables (e.g., rain, natural rain and its intensity, raindrop size and energy, and its spatio-temporal distribution). The rainfall simulator provides controlled rainfall in the laboratory and at a field scale. Marston (1952) found that vegetation cover had reduced more than 65% of surface runoff and increased the infiltration into the soil as a source of groundwater recharge in an experimental setup in the laboratory. Wang et al. (2014) used rainfall simulators to study the impact of different parameters on soil moisture after a storm event. They found that the moisture yield in soil with vegetation cover is greater than in bare soil. This moisture yield is considered important for determining the recharge coefficient.

Rainwater quality is another very important component which is generally ignored. Recent studies by John et al. (2021a, 2021b) and Satyanaga et al. (2021) found that the soil carrying capacity may significantly modify if the rainwater carries sediment from the atmospheric deposition. The impurities of the rainwater will choke the pores of the vegetative covers and thus will alter the runoff, infiltration, and soil erosion.

Furthermore, recent studies have reported that small-scale plots at the laboratory can be used to estimate the impact on large-scale plots under similar rainfall conditions (Schindewolf & Schmidt 2012). This information can be useful in comparing various soil conditions, rainfall, and environments (Martínez-Murillo et al. 2013). Nevertheless, a rainfall simulator requires calibration to estimate the accurate runoff as natural rainfall drop size and its energies vary considerably. Meyer (1965) listed the required rainfall characteristics (fall velocity, drop-size distribution, kinetic energy, uniformity, and intensity) based on wide applications of rainfall simulators. Cooper & Tait (2010), Pu et al. (2017, 2021), and Pu (2021) carried out a series of studies and reported that accurate representation of the soil surface profile is important to recapture the natural soil condition in the laboratory environment. Due to changes in the topography, the shear stress will change, leading to significantly varying soil erosion quantiles.

Fragile environmental conditions and limited economic and technical mitigation alternatives have contributed to land degradation caused by water erosion in arid regions (Sharaiha & Ziadat 2008). Semi-arid regions are defined with the help of the aridity index that lies between 0.2 and 0.5 (Lal 1994). The aridity index relates annual rainfall to the potential evaporation in a catchment, as most of the southeast and middle-east countries are under the semi-arid region.

Several qualitative and quantitative investigations have been conducted in Jordan to deal with the growing problem of soil erosion by flowing water. Quantifying various factors affecting soil erosion, such as slope, vegetation, and soil type is dealt with by several modelling or experimental approaches in the literature. These approaches often lack repeatability because either the modelling approach becomes data-intensive or experimental approaches are site-specific. In this regard, the novelty in the present study is the consideration of several factors affecting the soil erosion problem in the study area and their impact on the surface, sub-surface flow, and soil loss estimation under different land covers in an experimental setup. The choice of land covers is naturally occurring vegetative covers, otherwise considered waste materials. These include dry tree leaves, khus, and dry wheat straw. Because vegetation covers reduce surface runoff and soil erosion, the challenging aspect is finding out the appropriateness of the nature-based solutions for scientific application and filling the knowledge gap of their performances in contrast to the traditional approaches (Apollonio et al. 2021; Gonzalez-Ollauri 2022). Thus, the motivation behind using these nature-based vegetative covers was to assess their efficacy in controlling soil erosion, enhancing infiltration rate, and reducing surface runoff under different combinations of soil type, slope, and rainfall characteristics consistent with the actual field conditions of the study area. Thus, nature-based solutions to the soil erosion problem in semi-arid regions are required. The solution can further be important when it can also solve the groundwater recharge problem in water-scarce regions.

This study conducted laboratory-scale experiments with two locally available soil samples under three different organic land covers and four surface slope conditions. The main objective of this study was (i) to find the impact of various organic covers on surface runoff, soil erosion, infiltration, and groundwater recharge under laboratory conditions and (ii) to quantify the reduction in possible soil erosion using the most effective organic land cover.

The present study was carried out as a part of the hydrological study of the Wadi Shueib catchment in Jordan (Balasmeh 2020). It is expected that the knowledge derived from this study will be useful in determining the differences in soil loss and groundwater recharge using different land covers at a large field scale. In this study, the soil used was highly erodible due to its higher sand content. A similar type of high erodibility of soil was reported by Farhan & Anaba (2016) in southern Jordan. The ground slope and rainfall intensity were similar to the field conditions at the Wadi Shueib catchment, Jordan (Figure 3). The experimental findings in bare soil were also compared with the HYDRUS-1D groundwater model. Wells et al. (1986) reported that the infiltration could be better estimated using Richard's equation than Green-Ampt's model. Hence, only Richard's equation was used in the present study. Stomph et al. (2002) reported a strong scale effect of laboratory model results while applying to a field scale. However, a laboratory flume experiment is a good alternative to bypass the complex field-scale measurement for the initial study (Stomph et al. 2001; Huang et al. 2002). It is expected that the differences obtained in processes such as soil erosion, infiltration, and runoff between experiments conducted with soil covers and with bare soil will give a preliminary idea, which can either help design a large-scale experiment or calibrate the numerical model for groundwater and surface runoff.

In-situ soil density was measured in the field and the flume using an SDG200 Non-Nuclear Soil Density Gauge (make: Transtech systems). Soil water contents were also measured before the start of each experiment to ensure that soil moisture was similar in all the experiments.

The experiments were planned as a combination of three variables: (i) slope, (ii) land cover, and (iii) soil properties. Two graduated cylinders were used to collect the surface runoff and sub-surface flow simultaneously during the experiments. Soil eroded was collected over a piece of fine fabric to allow free water drainage. The rainfall intensity was measured during the experiments using a non-recording rain gauge placed inside the soil sample. Water pressure was adjusted to generate the required raindrop sizes (0.5–3.0 mm) and rainfall intensity based on nozzle diameter size.

At the start of each experiment, the flume was filled with 600 ± 10 kg of naturally air-dried soil. This soil was manually spread over the bottom of the flume and compacted such that the layer was not more than 100 mm in depth (Figure 2). Soil density gauge was used to ensure that its density was similar to in-situ soil bulk density (∼1,900 kg/m³) up to the top level. Table 2 shows soil characteristics and physical properties for each soil type used (Figure 4). The compaction process was carried out in four layers with a top layer of ∼30 mm. Soil moisture content was measured 24 h before the start of each experiment using the oven-dry method (at a temperature of ∼100 °C) and maintained at ∼17.5% for all the experiments. Additional moisture, if required, was added to the soil in the flume to get similar in-situ soil moisture content. Similar experiments were also conducted with bare soil, considered the base case in the present study. The experiments continued until the sub-surface flow became steady under uniform precipitation.

The following section describes the estimation of rainfall parameters such as rainfall intensity, raindrop size, rainfall uniformity, kinetic energy, and rainfall velocity. These parameters directly control the process of soil erosion. Therefore, to obtain consistent results, these parameters must be set following the procedures reported in the previous studies conducted by other researchers.

Rainfall intensity is the main parameter in any experimental study with a rainfall simulator mostly based on nozzles spray (Parsons & Stone 2006). Full cone spray nozzles with a circular spray pattern by Spraying Systems Co., India (www.sprayindia.com) were used here. The flow rate and rainfall intensity are affected by water pressure, although the nozzle spray angle remains the same. The rainfall intensity was measured using Symon's rain gauge (Mhaske et al. 2019).

The drop size of natural rainfall varies from 0.5 to 6 mm. A rainfall drop size larger than this is still possible for heavy rain. There are several methods to estimate the raindrop size (e.g., stain, flour pellet, photographic, and laser-optical methods (Mhaske et al. 2019). This study used the flour pellet method and digital image enhancement techniques to measure raindrop size.

The model can be simulated with a wide range of boundary conditions such as deep/free drainage, constant water content, atmospheric boundary conditions with surface runoff, constant/variable flux, seepage face, horizontal drains, variable pressure head, etc. In atmospheric boundary conditions with surface runoff, there is an option to assign unsteady precipitation boundary conditions. This feature is very useful in simulating surface runoff under variable rainfall conditions.

It is worth mentioning that the change in the rainfall intensity can completely change the runoff characteristics (Hu et al. 2021). In the present study, the objective was to find the appropriateness of the vegetative waste material to control soil erosion, reducing runoff with a rainfall intensity similar to the study area of Jordan. A separate study must be carried out to incorporate the effect of rainfall intensity on the runoff characteristics.

The time of initiation is required to understand the effectiveness of a cover to hold the rainwater fallen on the ground for a longer duration and gradually release it to the soil so that that soil can get sufficient time for infiltration. Figure 6 shows the variation of surface runoff with time under different soil, slope, and cover conditions. The surface runoff initiation time for bare soil with Soil-1 and 5% slope was ∼2 min, but with khus and leaves surface cover, it increased to ∼4 min. However, the initiation time was found to be ∼7 min (∼3 times of bare soil) with wheat straw cover. For experiments at a 10% slope, initiation time increased from 3 min for bare soil to 10 min for wheat straw (∼3 times for bare soil). Similar findings were also observed in flow initiation time with Soil-2, where an increase of ∼5 times was observed using wheat straw cover compared to bare soil. At a slope of 40%, no sub-surface flow was recorded even up to 30 min.

Soil erosion is a serious environmental concern for semi-arid to arid regions of the world. The quantification of the soil loss is dealt with through several interventions while considering the factors determining the soil loss. The present study sought a nature-based solution to control soil erosion in semi-arid regions under a laboratory-based experimental setup. A combination of 56 experiments with varying soil types, slopes and vegetation cover were studied to determine the overall efficacy in soil erosion control and enhancement in infiltration rates. Dry tree leaves, wheat straw, and khus were chosen as naturally available vegetation covers, and soil type, slope and rainfall intensities were determined, giving due consideration to the in-situ conditions in semi-arid regions since this study is a part of the hydrological study of the Wadi Shueib catchment in Jordan.

The study investigated three parameters: surface runoff, sub-surface flow and soil loss, and response time. The selected three land covers were compared to the bare soil scenario. As was expected, all three vegetative covers effectively controlled the soil loss, surface runoff, and enhanced infiltration rates. However, wheat straw showed a better ability to increase infiltration and was the most efficient organic land cover among other alternatives. The experimental results were validated by simulating the HYDRUS-1D model under similar conditions for a bare soil scenario.

The findings from the present study can be useful in choosing vegetation covers for soil erosion on arid to semi-arid soils. However, for the findings to be replicated and upscaled, large-scale field experimental results need to be further validated by numerical modelling. Farmer's perception in identifying the extent of soil erosion and its impact, choice and performance of nature-based solution to soil erosion and its validation through modelling will constitute a sustainable solution for dealing with issues related to soil erosion and enhancement of groundwater recharge in semi-arid to arid regions that are water deficient and lack efficient soil conservation plans.

Data cannot be made publicly available; readers should contact the corresponding author for details.

The authors declare there is no conflict.