Using Rainfall Simulators to Design and Assess the Post-Mining Erosional Stability

Abstract

The mining industry is crucial for global economic growth but faces environmental challenges, especially in designing stable rehabilitated landforms. To tackle these issues, rainfall simulators have been recognized for their value in providing data for erosion modeling and analysis, aiding the development of effective land cover systems for long-term stability. This chapter provides an overview of the theory, specifications, and design principles of rainfall simulators. It explores the detailed design and construction of a well-known model, along with its calibration process ensuring accurate rainfall production and distribution. The chapter also discusses raindrop size distribution and associated kinetic energy calculations. Calibration results demonstrate satisfactory outcomes with Christiansen’s uniformity coefficient exceeding 85% and a median raindrop size of 2.15 mm. The device successfully generates desired kinetic energy for simulated rainstorms, crucial for studying soil erosion. Examples highlight the application of rainfall simulators in evaluating erosion stability in Queensland mines. Efforts to construct a soil erosion database for 34 open-cut mines in Queensland using a similar portable rainfall simulator are highlighted. This database contributes to developing user-friendly MINErosion models, providing estimates of soil erosion/deposition at different scales to support the Australian mining sector.

1. Introduction

The mining industry plays a crucial role in global economic development, but it often faces environmental challenges, particularly related to rehabilitated landform design and its erosional stability. Constant efforts must be made to develop and build an appropriate landform and cover system that effectively serves its intended purposes, such as supporting vegetation growth and preventing the infiltration of water and oxygen into reactive mine waste. The success of the soil cover system depends on the external surface of the landform remaining intact. If the surface erodes, the cover system’s functions are compromised, and it is likely to fail. To evaluate the long-term erosion stability of a constructed landform, erosion rate thresholds must be established to ensure the landform is acceptably resistant to erosion, or “stable.”

While the defensible erosion rate thresholds should come close to the term “tolerable or acceptable erosion” rate which was first proposed by Browning et al. [1], and a more thorough review of tolerable values was carried out by the Soil Science Society of America in 1979; there is still no broad agreement on what we could consider as a tolerable or “acceptable” rate of erosion on a rehabilitated landform.

In general, US soil conservationists have consistently based tolerable soil loss values largely on the natural soil forming rate and on consideration of the maintenance of soil productivity, although accepting that other factors may be important in some situations. The US soil conservation agencies have typically used tolerable erosion values of < 11.2 t/ha/y for deep fertile soils, and < 4.5 t/ha/y for shallow agricultural soils. However, the fact that these values have been quantified under the American agricultural soil conditions must be kept in mind; that makes it doubtful to blindly accept these values to judge or establish defensible erosion rate thresholds under the Australian mining conditions. Therefore, the Queensland Department of Mines and Energy previously used a range of 12–40 t/ha/year as a target erosion rate for rehabilitated mine sites [2, 3], which was also impacted by the fact that it should be manageable to levels that do not compromise post-mining land use.

It is also worth mentioning that there are a few Australian studies that have attempted to quantify erosion rates. Lu et al. [4] utilized spatial modeling methods (remotely sensed data) to predict the sheetwash (interrill) and rill erosion all over the Australian continent. They estimated that the average erosion rate is 4.1 ton/ha/year over the Australian continent; however, they stated that “Soil erosion is naturally highly variable. This needs to be recognized when comparing current rates of erosion from one place to another and when the erosion control policies are set. It should be expressed in relation to spatially variability, rather than referring to absolute rates alone or using a single benchmark applied across diverse landscapes.”

Since it seems difficult to agree on a specific value for erosion rate threshold, the determination of this threshold value must be done for each mining site independently, to make it conceivable to achieve the required erosion stability; and that which decision makers and governmental regulators work on this matter will accept. The concept that the soil erosion rate threshold for specific mining site should be equal or close to the erosion rate values of the surrounding areas in similar climatic and environmental conditions deserves support and should be applied; the value of this threshold should be also manageable to levels that do not compromise post-mining land use agreed upon with the local community and by the PRC plan.

Rainfall simulators are distinguished research tools that enable us to measure erosion rates for any mine site and its surrounding area in the laboratory or in the field with high accuracy and efficiency; the erodibility values of soil and materials that will be used as land cover can also be measured accurately. Therefore, the rainfall simulators can provide us with some important data necessary for analysis and erosion modeling processes to help us design the best land cover system that could achieve long-term erosional stability.

2. Rainfall simulator: assessing soil erodibility for rehabilitation sites

Rainfall simulators are commonly used as research tools to study soil erosion [5, 6, 7, 8, 9, 10, 11, 12], as well as, soil infiltration [13, 14], water runoff and sediment transport [15, 16, 17], geotechnical studies especially slope stability and landslides [18, 19, 20, 21, 22, 23], and many other related research areas. Rainfall simulators can produce unique data that are vital for calibration and validation purposes of empirical, conceptual, or process-based rainfall-runoff-sediment transport mathematical models [24].

The credit for designing and building the first rainfall simulator goes to Nichols and Sexton [25] who used their first rainfall simulator (spray-bar rainfall simulator) to study soil erosion and measure the infiltration rate. By the mid-1950s of the last century, the use of rain simulators expanded steadily in experiments related to soil erosion, and the rapid technological development contributed to the introduction of many improvements to its first designs to avoid many issues that affect the performance, results, credibility, and the feasibility of these machines [26].

The main purpose of a rainfall simulator (RFS) is to generate and create an artificial rainstorm with precise specifications in terms of the duration and intensity of the rainfall, as well as in some way the size distribution of the droplets, and its kinetic energy. The ability to control the physical characteristics of the generated rainstorm makes it possible to keep the climatic factor (rain erosivity) constant while studying other factors that may affect the process of erosion such as soil erodibility, slope factor, or vegetation cover. It can be argued that any success of the RFS design depends entirely on how closely it is simulating natural precipitation conditions with respect to spatial uniformity, raindrop size, raindrop terminal velocity, and kinetic energy.

Although rainfall simulators are frequently utilized in soil erosion experiments, their ability to replicate natural rainfall conditions with precision has been a topic of concern. Many studies have been carried out to evaluate the reliability of results obtained from rainfall simulators and their usefulness in modeling soil erosion processes [27, 28, 29, 30, 31]. Among the important initiatives in this domain was the PLPEWC “Post-mining Landscape Parameters for Erosion and Water Quality Control” project, which was financially supported by ACARP (the Australian Coal Association Research Project) between 1992 and 1998.

The project performed a range of experiments, including laboratory rainfall simulation, field plots, and catchments (Figure 1).

Figure 1.

Range of experimental approaches adopted to determine soil/spoil erodibility in a previous ACARP (the Australian coal association research project).

The experimental approaches adopted were designed to measure the basic erosion parameters at the different scales. A large amount of data has been collected on 34 spoil and soil materials from 16 mines in Central Queensland, as well as 9 years of field plot and field catchment data [7, 32]. The data collected from those different experimental approaches/scales studies proved that although the need for field plots and catchment flumes studies still exists, the results obtained from laboratory rainfall simulators showed reliability so that their results can be used in modeling soil erosion with a high degree of accuracy. Moreover, laboratory experiments using rainfall simulators are more manageable than field experiments, because the data on runoff and soil loss can be obtained without waiting for natural rain to happen.

Therefore, rill and interrill erodibilities and slope adjustment factors were measured for these 34 soil/overburden materials on a portable rainfall simulator tilting flume (3 m long × 0.8 m wide, slope adjustable from 0 to 50%) at the University of Queensland Erosion Processes Laboratory. Each material was subjected to 100 mm.h⁻¹ rainstorm for 30 min (equivalent to a 1-in-20-year event in Central Queensland) at 20% slope, followed by slopes of 5, 10, 15, and 30% for 15 min each. At these simulated rainfall intensities, a steady state was quickly formed. The data from these measurements and the derived parameters were used to develop the MINErosion V3.x model [33], which successfully estimates field scale erosion rates on simple linear hillslopes with various combinations of slope gradients and lengths. MINErosion 3 can also be used effectively to simulate multiple field plot experiments on a computer, based on a few measurements made on a tilting flume-rainfall simulator facility in the laboratory. MINErosion 3.4 cannot be used to predict sediment yield from a watershed with complex topography in terms of slope steepness and flow pathways. However, it is necessary and desirable to be able to estimate off-site sediment discharges from these rehabilitated post-mining landscapes. For this purpose, MINErosion 4 was developed, which combines the MINErosion 3.4 model and a geographic information system (GIS) package (ESRI ArcGIS 10.3 or the freeware QGIS 3.16), to estimate erosion rates and sediment movement and delivery from these constructed post-mining landscapes. Both MINErosion 3 and 4 demonstrated the opportunities and the value of using the rainfall simulators at the mining sites to model and assess the erosional stability, which should be proven achievable under the given circumstances as it is one of the main considerations of the landform design report within the progressive rehabilitation and closure (PRC) Plan.

Subsequently, the effectiveness of using rainfall simulation as a method for obtaining erodibility information for other soil erosion models such as Areal Nonpoint Source Watershed Environmental Response Simulation (ANSWERS), Chemicals, Runoff, and Erosion from Agricultural Management Systems (CREAMS), and Water Erosion Prediction Project (WEPP) was recognized. Loch, Silburn [34], Silburn and Connolly [35], and Silburn and Loch [36] achieved accurate predictions of erosion under field conditions by utilizing parameters derived from rainfall simulation using these models. Nevertheless, Silburn and Loch [36] emphasized the significance of ensuring that the erosion processes happening on rainfall simulator plots were identical to those occurring in field areas in order to obtain reliable predictions.

3. Rainfall simulator: specifications and design

Although there are hundreds of different designs of rain simulators [6, 7, 15, 17, 22, 24, 37, 38, 39, 40, 41, 42, 43, 44, 45, 46, 47, 48, 49, 50, 51, 52, 53], each striving to get closer and closer to simulating the distribution of natural rain with varying degrees of success, rain simulators can be divided into two main groups [54] based on the way raindrops are generated:

• Non-pressurized rainfall simulators.

• Pressurized rainfall simulators (nozzle type).

The earliest versions of the rainfall simulators belong to the non-pressurized category, where drop-forming mechanisms completely rely on passing the water through a perforated pipe, hanging yarns, or an array of syringe needles which form the droplets, then the droplets are left free to fall under the impact of gravity from a height of not less than 9.1 m to ensure the droplets reach the required terminal velocity, which should be almost equal to the one of the natural rainfall droplets. This type of rainfall simulator has always suffered, due to the long distance between the raindrop generator and the flume surface, from the effect of wind on the falling droplets which make it a necessity to install a huge wind shield (Figure 2). The huge structure (the frame and wind shield) makes it impossible to use this type of rainfall simulator in field experiments (lack of portability). Later, the second type of rainfall simulator appeared, which depended on nozzles and pressurized water flow system, and it achieved a widespread popularity—at the expense of the first type—due to it being more portable and usable in field experiments as it is smaller in size (no more than 2–3 m high, Figure 3) and less expensive to build and run. Nevertheless, due to the pressurized nature of the simulator which produced a high-intensity spray, rainfall intensity is usually controlled by applying the water intermittently.

Figure 2.

The Griffith University tilting flume simulated rainfall (GUTSR). The left picture shows the rainfall generator above the flume. The right picture shows a side view of the sloping flume and the wind shield.

Figure 3.

The Griffith University portable rainfall simulator (port-RFS) installation and operation in the field.

In this section, we will describe the features, structure, and calibration processes of a modified version of Queensland Department of Primary Industries (QDPI) rainfall simulator (as depicted in Figure 3), The device has undergone testing by numerous researchers [7, 55]. Queensland Department of Primary Industries (QDPI) rainfall simulator has been constructed multiple times in various locations throughout Queensland and beyond, with minor variations between each iteration. As a result, it can be replicated with ease by others. The authors of this chapter constructed the Griffith University version of the model, known as the Port-RFS, which will be presented and discussed here. This rainfall simulator is characterized by the following specifications:

• High portability, which means it should be lightweight and easy to assemble, as it should be easy to transport from one site to another.

• Efficiency in water usage (as the available supplies of acceptable quality water were sometimes very limited in some locations).

• Reliable and ease of fixing and repairing on site; and

• Suitability for operation by a fieldwork team of three or four people.

As a pressurized rainfall simulator (nozzle type), the Port-RFS consists of a structural frame, the rain drop generator system, the water supply unit (tank), and the flume/soil container. The structural frame was constructed from aluminum tubing of 38 mm outside diameter (O.D.) and a 3 mm wall thickness. The bottom sections of some of the upright tubes are bent to form detachable legs to make it possible for the frame to be mounted on a hydraulic-tipper trailer when used inside a closed place/laboratory, so that it can use the slope mechanism of the trailer; as well as make it possible to be pegged to the ground, if the Port-RFS is used in the field (Figure 3). In order for the RFS to be transportable, the frame consists of a number of separate parts (12 pieces of metal tubes) that are connected together by nylon joiners that fit inside the end of aluminum tubes. The joiners are machined from solid nylon material and are about 180 mm in length. Locking pins ensure that the vital parts are securely interconnected.

The nozzle boom is made up of a 4 m long aluminum tubing with an outer diameter of 38 mm and a wall thickness of 1.6 mm. It can be shortened to 3 m to allow for mounting on a hydraulic-tipper trailer of the same length. The boom rotates in two plastic bushes (graphite-impregnated) that also prevent lateral movement of the boom. Four male threaded (1/2” BSPT (British Standard Pipe Taper)) unions are welded onto the boom 1 m apart. Check valves are threaded onto the bushes to prevent nozzle flow or dripping when the unit is not in use, and nozzles are fitted to the check valves. The water supply inlet is via a 1¼” BSPT tee fitting attached to the boom, opposite to the row of nozzles. A tapping from this fitting provides a connection for a pressure gauge, so that the operating pressure of the unit can be monitored. The rain drop generator system of Port-RFS consists of three to four VeeJet 80,100 nozzles (depending on the length of the boom) that are installed on the boom, 1 m apart, to cover (overlapped) the flume area (3 m × 1 m) underneath. A McLennan Unipolar Permanent Magnet Stepper Motor 1.8°, 3.8 Nm, 120 V dc, 4.3 A, eight wires with a programmable digital controller to drive and control the oscillating action of the nozzles’ boom. The water supply unit consists of Matrix 10-5 VFD: The Ward 10-5304 stainless steel horizontal multistage pump coupled to a 2.2 kW single-phase motor drawing 13 Amp full load current, with 40 mm female BSP inlet and 32 mm female BSP outlet. Controlled via the SteadyPress variable speed drive unit capable of a maximum flow rate of 200 L/min; the pump was used to feed the system with water from a 200-liter water tank. A solenoid valve and pressure gauge are attached to the water tank unit to help control the water flow rate.

The decision to use VeeJet 80,100 nozzles was made based on previous personal experience of the authors as well as the previous research of Bubenzer [56], Boulange, Malhat [15], and Loch, Robotham [7]. In pressurized rainfall simulators (nozzle type), it is important that the water flow rate provide enough pressure to allow the rain drops to have the capability to reach the required terminal velocity [46], which ensures that the kinetic energies of the generated storm satisfactorily resemble those of an intense natural rainstorm as well as the drop size distribution of erosive storm patterns. The operational sequence of this RFS relies on a continuous flow of water through the nozzles. Excess water that falls outside the soil is recycled via catch trays manufactured from galvanized aluminum plates that are arranged to collect excess water from either side of the individual oscillating nozzle (Figure 4).

Figure 4.

Excess water collecting trays.

During operation, the nozzles oscillate through “108°. Of this trek, the middle ≈ 68° applies raindrops onto the flume area below, and 20° is used at either end of the travel for overlapping, then the excess water gathered at the catch trays will go back to the water tank. Changing the frequency of nozzle oscillation using the stepper motor controller board, we can control and adjust the rainfall intensity coming by changing the waiting time and the consecutive sweeps (Figure 4). The stepper motor’s control system is composed of a DVP-14SS211T2 Programmable Logic Controller circuit board, EM806 Stepper Driver, a 4-button digital switch, and transistor switching circuits. The microcontroller runs in single chip mode using only internal random-access memory (RAM) and electrically erasable, programmable read-only memory (EEPROM) for data and program storage. The 4-button digital switch set the waiting time between consecutive sweeps of the spray manifold in 0.1-s increments. The transistor switching circuit is required to provide the correct voltage and current levels to the stepper motor driver module.

4. Rainfall simulators: Performance and calibration

The characteristics of simulated rainfall are reliant on the type of nozzle employed, the water pressure exerted, and the arrangement and movement pattern of the nozzles through the rotating bar (nozzle boom) [24]. Therefore, the performance of any rainfall simulator must be evaluated for the ability to produce different rainfall intensities (rainfall intensity calibration), the spatial uniformity of the simulated rainfall over the flume area, the rain Drop Size Distribution (DSD), the terminal velocity, and kinetic energy of simulated rainfall events.

4.1 Rainfall intensity calibration

To be effective in soil erosion experiments, a reliable rainfall simulator should have the capability to generate simulated rainfall events with a broad range of rainfall intensities. Typically, the rainfall intensity utilized in soil erosion experiments falls between 50 and 120 mm/hour, although it may be increased to 150 mm/hour for specific experiments or reduced to 30 mm/hour for less severe erosion studies [33]. The majority of present-day portable rainfall simulators, such as the QDPI rainfall simulator, manage the intensity of produced rainfall by manipulating two factors. First, the water flow/pressure from the primary water pump is regulated using a solenoid valve and pressure gauge. Second, the movement pattern of the nozzles is altered by employing a stepper motor controller and driver to modify the sweep and waiting time pattern. To verify the success of the apparatus (the rainfall simulator) in producing rainfall events with the desired intensities for the experiment or intended application, it must be calibrated using one of the following methods:

• Calibration using a flowmeter: This method involves measuring the flow rate of water through the nozzle that produces the raindrops. The flow rate can be measured using a flowmeter, and the rainfall intensity can be calculated by dividing the flow rate by the area of the nozzle.

• Calibration using a tipping bucket rain gauge: This method involves collecting the water that falls from the rainfall simulator using a tipping bucket rain gauge. The rainfall intensity can be calculated by dividing the volume of water collected by the time it took to collect it.

• Calibration using a high-speed camera: This method involves recording the artificial raindrops using a high-speed camera and analyzing the footage to determine the size and velocity of the raindrops. The rainfall intensity can be calculated using the size and velocity of the raindrops and the nozzle area.

• Calibration using a rain gauge array: This method involves setting up a rain gauge array around the rainfall simulator to measure the rainfall intensity at different distances from the simulator. The rainfall intensity can be calculated by analyzing the data collected by the rain gauge array.

It is important to note that the calibration method used will depend on the specific rainfall simulator being used and the accuracy required for the experiment or application. The rainfall intensities generated by the “Port-RFS” were evaluated and adjusted by taking into account the average volume of water collected in the flume area using the pan method. Additionally, the water discharge from each Port-RFS nozzle was measured by enclosing polyvinyl chloride (PVC) tubes around each nozzle individually and collecting the outflow for a duration of 5 min. To convert the collected data into flow rate in millimeters per hour, the recorded value was multiplied by 12.

Figure 5 shows the calibration of the Port-RFS where rainfall intensity can be controlled at rates from 60 to 150 mm/h using combinations of waiting and sweep periods. It is evident that the Port-RFS has the capability to produce simulated rainfall storms ranging from a minimum intensity of 60 mm/h to a maximum intensity of 150 mm/h. The control of rainfall intensity in the simulated rainstorm was found to be straightforward and efficient using the digital control panel.

Figure 5.

The Griffith University portable rainfall simulator (port-RFS) calibration curve for the rainfall intensities, relation between rainfall intensity, sweep and waiting times.

4.2 Spatial uniformity over the flume area

Obtaining a uniform distribution of rainfall across each section of the flume is crucial. Failure to achieve this can result in areas that receive more rainfall being more prone to erosion, compromising the accuracy of calculations based on the entire flume area. To ensure uniform rainfall distribution, the grid method is usually used to assess the spatial uniformity of simulated rainfall. It involves superimposing a grid with equidistant points onto the flume area and measuring the amount of rainfall that falls on each point using a rain gauge/graduated beaker. The gathered data for each point are then utilized to compute Christiansen’s uniformity coefficient (CUC), as shown in Eq. (1) [57].

$$\mathit{CUC}=100\left[1-\left(\frac{\sum \left(|D_i-D_m|\right)}{n\ast D_m}\right)\right]$$E1

where CUC is the coefficient of uniformity (%); Di is the depth of water in the graduated beakers (cm); Dm is the mean depth of water in rain gages/graduated beakers (cm); and n is the number of rain gages/graduated beakers. When the rainfall pattern is more uniform, the CUC value approaches 100%. According to Sousa, Mendes [58], several researchers consider that any CU values above 80.0% are acceptable for the uniformity of the rainfall distribution. However, some other studies have accepted a CUC value of 70% for large plot areas, as demonstrated by Luk, Abrahams [59]. The uniformity coefficients of the Port-RFS at various rainstorm intensities are presented in Table 1. All the coefficients exhibit high values, ranging from 86.55% to 91.8%. These values indicate a high level of uniformity across the measured experimental area. Figure 6 illustrates the distribution of rain intensities generated by the Port-RFS system for an average rain intensity of 100 mm/h across a flume area measuring 3 m × 1 m. The figure demonstrates that the incident intensities vary between 90 and 110 mm/h, with an average uniformity coefficient of 89.8%.

Rainfall intensity (mm/h)	130	120	110	100	90	80	70	60
Uniformity coefficient (%)	89.3	91.8	89.6	90.1	89.6	89.8	89.4	86.55

Table 1.

The calculated uniformity coefficient % for portable rainfall simulator (port-RFS) under different rainfall intensities.

Figure 6.

The spatial distribution of the simulated rainfall over the flume/plot area for the Griffith University portable rainfall simulator (port-RFS) (100 mm/h rainfall intensity).

4.3 The drop size distribution (DSD) and kinetic energy (KE)

The ability of any rainfall simulator to generate raindrops that approximate the volumetric size distribution of the droplets that occur during rainstorms in nature is highly influential in our judgment of the efficiency and quality of the rainfall simulator design, as the distribution of grain size over the different classes of drop sizes (volume in mm3) affects the total kinetic energy generated from the simulated rainstorm, whereas the kinetic energy of a single drop is a function of a grain’s mass, which is related to its size (volume) as well as its terminal velocity when it hits the ground [49, 60]. In general, the sizes of raindrops in nature range from 0.5 mm in diameter to the large drops associated with heavy rainfall and reaching up to 6 mm in diameter, with median droplet diameter varying depending upon the storm intensity but usually ranging from 2 mm to 3 mm [9, 46].

There are numerous methods and instruments for measuring raindrops, which can be divided into two main groups: manual and automated techniques. These approaches are used to determine the raindrop size distributions and the average size of raindrops for simulated rainfall events.

Manual rain drop measurement techniques include the stain method that involves using dyed absorbent paper to measure the stains left by raindrops [61], the flour pellet method that uses finely sieved flour to create dough pellets from raindrops [62, 63], and the oil immersion method that measures raindrops in a vessel containing oil [64]. While these methods are simple and inexpensive, they are time consuming, the accuracy of the results obtained from it depends on the skills and experience of the researcher, and do not provide immediate data readings.

On the other hand, there are various automated techniques available for measuring raindrops, including but not limited to: the displacement disdrometers [65], the photographic method [66, 67], acoustic disdrometers [68], the radar technique [69, 70], and the optical spectra pluviometers [71].

While the disdrometer method has been particularly successful over the past decade due to its ability to generate a large number of measurements [72, 73] and its efficiency in measuring the impact of raindrop splash on soil detachment [74] and erosion caused by changes in soil cover [75], the old flour method [62] is a widely accepted, standardized test method [46, 49, 63]. Using the flour method, the mean diameter of the raindrops that came out of each examined rain event was measured and could be calculated using Eq. (2).

$$D_r=\sqrt[3]{\left(\frac{6}{\pi }\right)W{m}_R}$$E2

Where D_r is the mean raindrops’ diameter (mm) and W is the mean weight of the raindrops (mg) and m_R is the ratio of the mass of the raindrop to the mass of the pellet, which is obtained using the flour-calibration line [76].

Using Eq. (3), the kinetic energy of individual raindrops could be determined after the median size distribution (D50) of the rainfall simulator was easily estimated from the previous step:

$$\text{KE=}\mathrm{½}{\text{mv}}^2$$E3

where KE is the kinetic energy (Joule); m is the mass (kg) of the raindrop (calculated from the relation between the volume, density, and mass); and v is the terminal velocity (m/s) at which the drop hits the soil surface where the values for examined rainfall intensities by examined rainfall simulator could be obtained from the American Society for Testing and Materials (ASTM) [77] chart that correlates the fall velocity, fall height, and raindrop diameter.

The threshold kinetic energies needed to initiate soil detachment (erosion) by raindrop impact were listed and discussed by Salles, Poesen [78]; they stated that the threshold kinetic energy required to initiate the detachment of soil particles by raindrop impact declines with increasing median grain-sized diameters, starting from 0.001 mm until D50 reaches values near 0.1–0.2 mm. Once D50 becomes larger, the variation in the threshold kinetic energy changes and increases with median grain diameter of the soil.

By utilizing the flour method, the drop size distribution of the Port-RFS was examined, resulting in a median size distribution of 2.15 mm; Figure 7 presents the drop size distribution pattern observed with the Port-RFS. In Figure 8, the relationship between the generated rainfall intensities and the corresponding kinetic energy (KE) per second per flume area is depicted. The measured KE values for rainfall intensities of 80, 90, and 100 mm/h were found to be 1.96, 2.2, and 2.45 Joule/Sec. flume area, respectively. Based on these calibration data, the kinetic energy and drop sizes generated by the Port-RFS were deemed satisfactory for initiating soil erosion in the range of 0.001 mm < D50 < 2.5 mm, which is considered suitable for most soil samples.

Figure 7.

Raindrop size distribution for the Griffith University portable rainfall simulator (port-RFS) under different simulated rainfall intensities.

Figure 8.

The relation between rainfall intensity and the kinetic energy (KE) of the simulated rainfall events.

5. Rainfall simulator: the experimental design and data analysis

The experimental design of soil erosion experiments using a rainfall simulator is crucial in obtaining reliable results. The rainfall simulator is designed to simulate natural rainfall events and generate runoff, which can be collected and analyzed for sediment concentration, steady-state runoff rate, and other parameters. The duration and intensity of the rainfall events can be controlled, which allows researchers to investigate the effects of different factors, such as soil type, slope gradient and length, vegetation cover, and land use, on soil erosion.

The experimental design of soil erosion experiments using rainfall simulators should include a detailed description of the soil characteristics, such as texture, organic matter content, and aggregate stability. The location and orientation of the experimental plots, as well as the size and shape of the rainfall simulator nozzle, should also be specified. The experiments should be repeated several times to ensure the reproducibility of the results. The collected runoff should be analyzed for sediment concentration, particle size distribution, and other relevant parameters. Statistical analysis of the data should be performed to determine the significance of the observed differences and to identify the most important factors affecting soil erosion.

Data analysis of soil erosion experiments using rainfall simulators can be challenging due to the complexity of the processes involved. The measured variables are often interdependent, and the relationships between them can be nonlinear. Various statistical methods, such as regression analysis, analysis of variance (ANOVA), and principal component analysis (PCA), can be used to analyze the data. The results should be presented in a clear and concise manner, with appropriate graphs and tables. The conclusions should be based on sound statistical analysis and should be supported by experimental data. The implications of the results for soil management and conservation should also be discussed.

As a practical example of what can be done with soil erosion experiments using rain simulators, Kibet, Saporito [79] presented a protocol along with a video showing how to conduct the experiment; MINErosion 3.4 software [33] also contains a descriptive file that explains the steps for conducting the rainfall simulator experiment and how to calculate rill and interrill soil erodibilities using it.

6. Using the rainfall simulators’ results for landform design and assessing the erosional stability of mining rehabilitation sites

Open-cut mining involves a larger disturbance of surface area compared to underground mining. In Australia, explosives are used to blast the deep solid overburden above the mineral or coal seam, which is then mechanically removed using trucks and shovels or draglines. Draglines are the most commonly used methods in Central Queensland open-cut coal mines, which operate at high speeds and result in landscapes consisting of long parallel spoil-piles that are highly saline, dispersive, and erodible. These spoil-piles can be over 50–60 m high and have slopes at an angle of repose of around 75% or 37°. It is legally required for mining organizations to rehabilitate the land by law, and therefore, disturbed open-cut post-mining landscapes must be rehabilitated to an approved post-mining land use. The principal rehabilitation process is shown in Figure 9.

Figure 9.

Pictorial representation of the process of rehabilitation of post-mining landscapes on open-cut coal mines is Central Queensland, Australia.

The most expensive part of rehabilitating a mining site is creating a suitable landscape for vegetation growth by reshaping and preparing overburden dumps, which requires costly earthworks. The new landscape must be able to withstand geotechnical failure and surface erosion caused by rainfall and runoff. Steep slopes can lead to severe erosion, causing rehabilitation failures, gully erosion, acid mine drainage (AMD), and salt discharges. Mine sites must evaluate the potential annual erosion rates from rehabilitated areas and report them to regulators, such as the Department of Environment and Science in Queensland, which monitor and enforce rehabilitation compliance. To minimize costs and prevent rehabilitation failures, it is essential to predict the soil erosion rates for the suggested landscape design before its construction. Since the soil and overburden materials used to build the engineered landform have varying degrees of erodibility, accurately measuring their erodibility values is necessary to predict potential erosion rates.

The mining rehabilitation industry has traditionally used predictive equations that rely on soil properties like texture to estimate erodibility values for soil/overburden materials used in constructing engineered landforms. While this approach was easy to use, its unreliability in predicting soil erosion rates led regulators to require actual measurements of soil erodibility values using rainfall simulators in either a field or a laboratory setting. Once these erodibility values have been measured, they are used in combination with more sophisticated erosion/deposition models such as WEPP, MINErosion, and SIBERIA to predict and assess the erosional stability of the proposed landform design.

As an example of using rainfall simulators to estimate and measure soil erodibility values, Sheridan, So [32] conducted a study in which they measured the rill and interrill erodibilities and slope adjustment factors of 34 soil/overburden materials using a portable rainfall simulator tilting flume (measuring 3 m long × 0.8 m wide and with a slope adjustable from 0 to 50%) at the University of Queensland Erosion Processes Laboratory. The materials were exposed to a 100 mm/h rainstorm for 30 min (equivalent to a 1-in-20-year event in Central Queensland) at a 20% slope, followed by slopes of 5, 10, 15, and 30 for 15 min each. At these simulated rainfall intensities, a steady state was quickly reached. Data from rainfall simulation at 10% slope were used to determine interrill erodibility, and data from the overland flow experiments at 20% slope were used to calculate rill erodibility coefficients. Sheridan, So [32] found that soils were generally more erodible than overburdens, as many of the overburdens either contained considerable amounts of rock or had a strong sealing ability. The strongly aggregated high clay soils tended to be the most erodible, followed by the lighter textured sandy loams and loamy sands. Soils or overburdens with 20–30% silt tended to form strong, raindrop impact seals under rainfall and consequently had very low erodibilities. Khalifa [80] expanded the dataset obtained from this research by including information from 93 additional samples (46 soil samples and 47 spoil samples) gathered from six coal mines in Central Queensland. This was done to capture the diversity of spatially distributed samples across each of the selected mining sites. The data collected by Khalifa [80] were used to feed MINErosion 3 & 4 erosion/deposition models for post-mining rehabilitation [81], resulting in the creation of an embedded database file containing the rill and erodibility values for 34 open-cut mine sites in Central Queensland. Additionally, the model is capable of analyzing rainfall simulation data to calculate erodibilities and predict erosion rates on an annual or event-based basis.

Currently, an increasing number of environmental consulting companies operating in the mining sector have constructed and utilized rainfall simulators in a number of environmental rehabilitation projects. Their primary purposes include assessing the erodibility factors of materials intended for constructing the designated land cover and estimating final landform erosion stability. These rainfall simulators have been utilized in several Australian mines, such as the North Parkes Mine in New South Wales, the Ranger Uranium Mine in the Northern Territory, the Carmichael Coal Mine and Mt. Rawdon gold mine in Queensland, and the Yallourn Coal mine in Victoria.

Furthermore, the utilization of rainfall simulators extends to the assessment of erosion control techniques. By providing a controlled environment, rainfall simulators allow for the testing and evaluation of various measures, such as terracing, vegetation cover, and mulching, to determine their efficacy in reducing soil erosion and improving landform stability [33, 82]. These experiments are valuable in assisting mining operators in selecting the most appropriate and cost-effective erosion control practices tailored to their specific site conditions.

Additionally, rainfall simulators are instrumental in evaluating and monitoring the success of rehabilitation efforts for a specific mine site. After mining operations cease, the restoration of ecosystems and the mitigation of long-term environmental impacts are of utmost importance. Rainfall simulators could be used to facilitate the assessment of rehabilitation effectiveness by comparing the erosional stability of reclaimed landforms with undisturbed reference areas. This evaluation provides invaluable feedback on rehabilitation techniques and offers guidance for future restoration practices.

7. Conclusion

The use of rainfall simulators in assessing the erosional stability of mine site landforms has emerged as a valuable tool for the mining industry. These simulators enable mining operators, researchers, and regulators to measure the material erodibilities, predict erosion rates, estimate the erosional stability, evaluate erosion control measures, and assess the success of land rehabilitation efforts. By providing valuable information on potential erosion hotspots and effective erosion control practices, rainfall simulators contribute to the adoption of responsible mining practices, mitigating environmental impacts, and advancing sustainable land management in the mining sector. When combined with field monitoring and validation, rainfall simulators assume a crucial role in supporting the long-term environmental sustainability of mining operations.

Acknowledgments

The authors express their gratitude to ACARP (the Australian Coal Association Research Project) for their prior support in researching the use of rainfall simulators for rehabilitating mining sites, and to the Australian Rivers Institute-Griffith University for their academic support. Additionally, the corresponding author wishes to acknowledge RGS Environmental for their support and encouragement during his work on this chapter.

A. Appendix

The calculation of Christiansen’s uniformity coefficient (CUC) for three different rainfall events simulated by our portable rainfall simulator.

	30 mm/hr. Designed event			60 mm/hr. Designed event			90 mm/hr. Designed event			100 mm/hr. Designed event
Pan no.	Measured rainfall intensity (mm/hr)	Di	|Di-Dm|	Measured rainfall intensity (mm/hr)	Di	|Di-Dm|	Measured rainfall intensity (mm/hr)	Di	|Di-Dm|	Measured rainfall intensity (mm/hr)	Di	|Di-Dm|
1	28	13.83	1.61	56	28.00	2.29	84	42.17	3.88	91	45.55	6.34
2	37	18.56	3.12	62	31.04	0.75	86	43.18	2.87	88	43.86	8.03
3	34	17.21	1.77	57	28.34	1.95	81	40.49	5.57	88	43.86	8.03
4	38	19.23	3.79	62	31.04	0.75	92	45.88	0.17	101	50.61	1.28
5	31	15.59	0.15	65	32.35	2.06	98	48.82	2.77	106	52.94	1.05
6	35	17.54	2.10	61	30.36	0.07	86	43.18	2.87	101	50.61	1.28
7	31	15.29	0.15	61	30.59	0.29	92	45.88	0.17	112	55.88	3.99
8	38	18.82	3.38	66	32.94	2.65	98	48.82	2.77	118	58.82	6.93
9	29	14.71	0.73	61	30.29	0.00	96	48.24	2.18	118	58.82	6.93
10	27	13.53	1.91	62	31.18	0.88	95	47.65	1.59	118	58.82	6.93
11	29	14.41	1.03	67	33.53	3.24	93	46.47	0.41	118	58.82	6.93
12	21	10.59	4.85	58	28.82	1.47	94	47.06	1.00	106	52.94	1.05
13	22	11.23	4.21	64	31.91	1.62	99	49.65	3.59	106	53.19	1.30
14	35	17.71	2.27	61	30.70	0.41	101	50.61	4.55	103	51.28	0.61
15	35	17.54	2.10	59	29.35	0.94	88	43.86	2.20	88	43.86	8.03
16	26	13.16	2.28	55	27.33	2.97	88	43.86	2.20	101	50.61	1.28
17	28	14.12	1.32	60	30.00	0.29	100	50.00	3.94	106	52.94	1.05
18	30	14.84	0.60	55	27.50	2.80	86	43.18	2.87	101	50.61	1.28
		Dm	15.44		Dm	30.29		Dm	46.06		Dm	51.89
		n	18		n	18		n	18		n	18
		Sum |Di-Dm|	37.37		Sum|Di-Dm|	25.42		Sum |Di-Dm|	45.61		Sum |Di-Dm|	72.35
		CU	86.55		CU	95.34		CU	94.50		CU	92.25

Calculation of Kinetic energy for different rainfall events simulated by portable Rainfall simulator.

$D_r=\sqrt[3]{\left(\frac{6}{\pi }\right)W{m}_R}$				(Eq. 1)	$V=\frac{4}{3}\pi {\left(\frac{D_r}{2}\right)}^3$					(Eq. 4)
Rainfall Intensity	Number of pellets	weights of dry pellets (g)	Average weight of dry pallets (mg)	mR	Dr (mm)	Volume mm^3	Density of water 1 mg/mm^3	mass of an individual raindrop mg	terminal velocity (from the curve below) m/s	Kinetic Energy, KE (μJ)
100	350	2.98	8.520	1.1	3.3588	19.84127	1	19.84127	5.5	300.1
90	387	2.53	6.537	1.1	3.0168	14.37598	1	14.37598	5.5	217.4
60	435	2.29	5.257	1.1	2.7736	11.17245	1	11.17245	5.5	169.0
45	521	1.98	3.797	1.1	2.4670	7.86145	1	7.86145	5.5	118.9
30	567	1.79	3.155	1.1	2.3196	6.534952	1	6.534952	5.5	98.8

Using the flour method to measure the drop size distribution of the Port-RFS.

Measuring the water flow rate for different rainfall events simulated by portable RFS.

Time interval	Rainfall intensity	Water pressure (kPa)	Flow rate L/min
2	30	382	0.1
5	40	384	0.3
7	50	386	0.6
10	60	388	0.8
15	70	391	1
20	80	393	1.2
25	90	394	1.5
30	100	395	1.6

Database of Soil/Spoil Properties (Measured by Khalifa [80]) and Rill/Interrill Erodibilities Assessed with a Portable Rainfall Simulator at 16 Mine Sites in Central Queensland [32, 33, 80].

soil name	interillK	rillK	Infiltration	Coverfactor	EC	pH	CEC	ESP	Wiltingpoint	Fieldcapacity	Organic C	Clay%	Silt%	Sand%	Rock%	BD
Curragh sandstone overburden	2,133,601	4.43	22.61	0.96	0.69	8.89	26.9	26.1	0.12	0.29	0	22.56	25.24	52.2	1.56	1.59
Curragh medium heavy clay soil	6,012,859	11.94	23	0.98	0.75	7.67	60.2	6.5	0.17	0.37	1.6	57.13	17.26	25.62	0.9	1.36
German creek rocky overburden	870,685	5.82	21.98	0.67	1.43	7.57	24.6	10.1	0.08	0.17	0	4.63	28.49	66.88	16.16	1.73
German creek clay loam soil	4,039,506	8.12	21.98	0.9	0.71	7.57	13.2	13.9	0.06	0.17	0.79	22.58	7.92	69.5	4.25	1.59
Goonyella permian overburden	3,048,501	5	15.64	0.97	1.37	8.82	36.5	39.8	0.16	0.31	0	26.85	22.17	50.98	1.31	1.48
Goonyella light clay soil	2,573,589	7.24	19.36	0.96	0.2	6.3	10.1	7.9	0.11	0.18	1.44	44.23	7.23	48.54	1.58	1.58
Gregory overburden	1,629,811	8.06	18.21	0.85	1.05	7.8	16.2	6.5	0.07	0.15	0	9.76	18.16	72.08	6.66	1.79
Gregory clay loam soil	3,969,008	12.72	20.06	0.93	0.07	6.33	13.6	0.5	0.09	0.21	0.78	30.33	5.51	64.16	2.8	1.56
Norwich park rocky overburden	85,664	0.13	70.97	0.44	0.71	8.34	14.6	27.7	0.07	0.12	0	17.22	25.19	57.59	33.14	1.77
Norwich park tertiary overburden	2,874,002	0.29	14.09	0.89	0.48	7.29	15.4	35.4	0.11	0.2	0	31.91	13.04	55.05	4.47	1.61
Norwich park sandy clay loam soil	3,582,186	41.22	12.03	0.97	0.14	6.56	3.5	29.2	0.04	0.11	0.56	15.72	2.71	81.57	1.22	1.73
Oaky Ck shale overburden	796,145	8.93	21.03	0.83	1.6	7.6	30	6.9	0.06	0.16	0	0.72	24.13	75.1	7.34	1.67
Oaky Ck clay loam soil	2,937,891	9.01	10.06	0.98	0.25	7.1	10.2	9.8	0.07	0.17	1.39	22.88	8.47	68.65	0.97	1.61
Wandoan shale overburden	3,123,865	3.09	8.15	0.86	0.85	8.17	25.9	31.3	0.13	0.3	0	26.57	36.62	36.81	6.05	1.41
Wandoan light clay soil	1,861,215	0.83	8.38	0.98	0.11	8.07	23.9	3	0.11	0.23	2.43	33.54	17.94	48.52	0.72	1.48
Callide overburden	1,043,065	5.05	8.94	0.73	0.22	6.68	12.7	5.5	0.08	0.15	0	18.48	11.91	69.61	12.77	1.59
Callide sandy clay loam soil	1,682,373	22.11	8.94	0.91	0.04	5.19	1.4	22.8	0.05	0.13	2.3	19.02	4.31	76.68	3.59	1.59
Blackwater rocky overburden	535,766	17.78	8.86	0.81	0.47	8.62	23	15.9	0.1	0.23	0	20.65	29.26	50.08	8.28	1.65
Blackwater heavy clay soil	5,783,402	21.9	7.58	0.94	0.3	8.69	40.2	10.6	0.21	0.44	2.35	57.65	14.2	28.15	2.45	1.4
Blair Athol overburden	1,377,148	8.68	4.5	0.81	0.28	7.66	7.6	11.8	0.07	0.16	0	17.72	10.48	71.8	8.55	1.62
Blair Athol sandy loam soil	2,314,952	6.89	2.72	0.86	0.1	7.06	8.6	4	0.06	0.15	1.33	23.69	15.31	61	6.04	1.36
Tarong lithosol soil	1,047,697	6.82	7.06	0.66	0.35	4.99	4.9	14.4	0.09	0.19	3.49	30.06	12.13	57.82	16.77	1.47
Tarong krasnozem soil	760,859	8.06	8.51	0.8	0.09	4.75	2.7	3.3	0.13	0.22	2.63	37.84	6.92	55.24	8.86	1.21
Saraji overburden	4,199,042	2.29	12.74	0.98	0.44	9.44	26.5	28.5	0.13	0.3	0	34	9.08	56.92	0.98	1.5
Saraji sandy loam soil	2,250,000	12.42	5.84	0.97	0.14	8.67	5.8	5.5	0.04	0.13	1.04	14.01	9	76.99	1.26	1.71
Newlands rocky overburden	878,384	7.67	13.01	0.78	0.3	8.16	14.8	3.1	0.07	0.14	0	13.82	18.3	67.88	9.93	1.53
Newlands sandy clay loam soil	3,228,555	2.27	4.66	0.96	0.57	8.93	24.7	36.8	0.1	0.22	0.68	28.58	10.4	61.02	1.49	1.52
Peak downs overburden	1,216,710	1.01	6.22	0.84	1.55	8.55	25.1	52.7	0.12	0.27	0	27.45	42.91	29.64	6.96	1.29
Peak downs loamy sand soil	1,221,103	32.83	6.87	0.94	0.22	7.53	9.5	17.1	0.05	0.14	0.8	20.17	7.03	72.8	2.63	1.44
Moura Medium Clay Soil	2,281,317	7.43	8.03	0.9	0.26	7.81	34.6	9	0.2	0.47	1.67	48.85	15.73	35.42	4.37	1.44
Moura rocky overburden	980,397	1.73	5.46	0.5	0.43	7.88	25.1	19.2	0.14	0.29	0	24.13	25.38	50.49	28.01	1.48
Moura permian overburden	3,297,501	4.81	0.14	0.88	0.75	8.57	32.6	31.3	0.12	0.26	0	31.06	16.66	52.28	4.93	1.18
Tarong sandstone overburden	2,500,000	6	15	0.98			2.2	3.26	0.12	0.26	0	12	16	72	8
Kidston	260,394	0.61	5	0.5	0.06	8.5	17.4	8.04