The characteristics of the experimental site are determined by the experimental objectives and the type of data to be obtained. Generally, plots are used to study physical phenomena affecting soil detachment and transport whereas complex hill slopes or small watersheds are gauged for examining deposition processes and therefore sediment yield.
In an experimental station for soil erosion studies, bounded runoff plots of known width, slope length, slope steepness, and soil type are monitored. Plot size depends on the physical process under investigation. Early basic data used for the development of the Universal Soil Loss Equation (USLE)  were often obtained from plots 40.5 m2 in size, having a standard length of 22.1 m .
Studies on sheet erosion due to overland flow need small plot width (2-3 m) and length (<10 m). Studies on rill erosion require plot lengths greater than 6-13 m . Sites should be selected to represent the range of uniform slopes prevailing in the farming area under consideration. Slopes ranging from convex to concave are used to evaluate the influence of the slope shape on soil loss. Because the plot area is known, soil loss can be expressed as kg/m2 per unit time, which assumes a uniform soil loss over the plot.
Figure 4.20. Evolution of a gully in a desert system. Source: .
Figure 4.20. Evolution of a gully in a desert system. Source: .
The equipment needs for a runoff plot are boundaries around the plot to define the measurement area, collecting equipment to catch plot runoff, a conveyance system (H-flume or pipe) to carry total runoff to a sampling unit, and a storage system (Fig. 4.21). The sampling device may consist of a Coshocton wheel  (Fig. 4.22) which has a water wheel, slightly inclined from the vertical, and a sampling head with a narrow opening (slot). With each revolution of the wheel, the slot cuts across the jet from the flume and collects a given portion of the runoff, which is transported to a storage tank.
A recently developed sampler is the Fagna-type hydrological unit  (Fig. 4.23). Runoff, cleaned of the coarser materials by passing it through a sedimentation tank, falls on a revolving pot supported by two U-shaped forks. When the pot is full, it turns completely upside down. A few cubic centimeters of the outgoing jet are intercepted by a sampling hole and conveyed to a small tank, below the hole level. When the pot is empty,
two coaxial pivots enable the pot mouth to return in the up position. Both the hydrograph and the runoff volume are measured by the number of times that the pot is emptied. The weight of suspended material is determined by the mean sediment concentration in the small tank and the measured runoff volume.
Both sampling devices (Coshocton wheel and Fagna-type unit) need field tests to calibrate the measurement system.
A simple method for measuring the sediment concentration is to store all runoff or to divide it using a sequence of tanks. In each tank the mean concentration Cm (g/L) is estimated from the concentration profile obtained by collecting samples of given volume at different depths. If Cm is calculated by integrating the concentration profile measured along the axial vertical of a tank wall, Bagarello and Ferro [6Q] illustrated that the actual concentration C (g/L) is linked to Cm according to the following relationship:
where the coefficient b depends on the eroded soil type, the sampled suspension volume from each location, the water depth h, the number of sampling locations, and the mixing time before sampling. Figure 4.24 shows, for three soil types (clay C, loam L, sand S), the relationship between the coefficient b and water depth.
Investigations of sediment yield are carried out at hill-slope or small-watershed scale. To measure discharge and suspended sediment concentration, a channel of known cross
section and slope usually is constructed. For channels of low slope, discharge measure-mentcanbe carried out using a Venturi meter channel  or by a free overfall . The Venturi meter channel is based on the principle of critical control section in which the relationship between the critical depth hc (m) and the discharge Q (m3/s) is definitive, independent of the channel roughness and of other uncontrollable circumstances. The
10 20 30 40 50 60
10 20 30 40 50 60
Figure 4.24. Relationship between coefficient b ofEq. (4.7) and water depth. Source: .
Venturi flowmeter is useful for water with suspended particles because it does not induce soil-particle sedimentation. A free overfall establishes in the mild channel a subcritical flow having a decreasing downstream water depth and a water depth at the overfall equal to hc (backwater curve type M2) .
For a steep-slope channel, the discharge of the uniform flow is calculated by the Chezy's law, measuring the flow depth. In this case, the channel length has to be calculated to avoid the roll-wave formation . The flow-depth measurement can be obtained by a nonintrusive, ultrasonic-level probe. The probe sends an ultrasonic pulse that, after partial reflection, is detected by the same sensor and converted back into an electrical signal. The time between transmission and reception of the pulse is directly proportional to the distance between the probe and the water surface.
The suspended sediment concentration can be measured by an infrared turbidity sensor. The excitation radiation is pulsely emitted into the flow at a defined angle by infrared transmitters. The suspended particles generate a scattered light that is received by scattered light receivers. The measured signals are processed to produce an output signal that is proportional to the suspended solid concentration. By coupling this local measurement to the discharge, the sedimentograph is obtained.
A global measurement of suspended sediment yield can be obtained by a water flow sample collector. Figure 4.25 shows the operating scheme of a typical portable sample collector. At the start of every sampling process, the built-in diaphragm pump pneumatically shuts off the dosing device (Fig. 4.25a) and via the dosing glass, blows the sample intake hose free. The sample is drawn in until the conductivity probe located at the top of the dosing glass responds (Fig. 4.25b). Then, the preprogrammed sample volume Vp is withheld and the surplus is returned to the source (Fig. 4.25c). The hose valve opens (Fig. 4.25d) and the sample under low pressure is drained off into the sampling bottle. The sampling procedure is controlled by a microprocessor that allows a sampling event proportional to time or to quantity.
Problems of representativeness arise when plot data are extrapolated to the hill slope and watershed scales. In fact, measurements from a bounded area provide little information on the local variability of erosion rates and the redistribution of soil within a field.
Tracer techniques afford an alternative to the use of plots and a means of overcoming the problems of measurement representativeness and spatial variability. The most widely used tracer, which possesses the greatest potential in soil erosion and sediment yield studies, is the radionuclide caesium-137, which has a half-life of 30 years. It is a fission reaction product of atmospheric thermonuclear weapons test and has been released in significant quantities into the stratosphere. The 137Cs technique (Fig. 4.26) is based on the following key assumptions: (1) local fallout distribution is uniform; (2) 137Cs fallout is adsorbed rapidly onto soil particles; (3) subsequent redistribution of 137Cs is due to sediment movement; and (4) estimates of rates of soil loss can be derived from measurements of soil137 Cs inventories.
As shown in Fig. 4.26, 137Cs fallout reaches the land surface by precipitation. It is strongly adsorbed onto clay and organic particles and is essentially nonexchangeable . Adsorption on soil particles is rapid with distribution in undisturbed soil profiles showing an exponential decrease with soil depth. More than 75% of the total inventory usually occurs in the top 15-20 cm, indicating that vertical translocation is minimal. The measurement of the 137 Cs content of the soil in each point of interest is easy to obtain because it involves collection of a single core sample and a laboratory measurement of 137Cs activity (mBq/cm2) using standard gamma-spectrometry equipment. If little or no soil loss has occurred, the total 137Cs loading will remain similar to that at the local reference site Csrif (an undisturbed site with the original distribution of fallout activity). The watershed reference site is a little area in which no erosion or deposition processes occur.
Where soil erosion has occurred, 137Cs also will have been lost, leading to a reduced loading. Conversely, where soil deposition takes place, an increase in 137Cs activity will be found. The geostatistical analysis of local 137Cs activity measurements allows establishment of the spatial distribution of the 137Cs percentage residuals Y = (137Cs — Csrif)/Csrif which is correlated with the mean annual sediment yield spatial distribution [4, 65].
In many laboratory and plot studies, a rainfall simulator is used to reproduce a storm of known kinetic energy and drop-size distribution. Rainfall simulators are ideal tools for distinguishing the effects of single factors on soil erosion and for carrying out a large number of experimental runs in a short time. Desired characteristics of rainfall simulators [66, 67] include a wide range of intensities, drop sizes, fall velocities, and impact energy characteristics similar to those of natural rainstorms; uniformity of spray pattern and of drop-size distribution on the plot area; and continuous application of rain or minimum time between simulated raindrop applications if intermittent.
In rainfall simulators, developed after 1930, the use of nozzles is practically the only method available to produce a drop distribution that includes a large range of drop sizes. Drop size and velocity at impact, equal to the terminal velocity, similar to those of natural rain can be obtained; however, a high simulated rainfall intensity also is obtained because water is delivered under pressure. To reduce rainfall intensity without modifying drop sizes and fall velocities of the simulated rain, intermittent spray is used in many nozzle rainfall simulators. The main disadvantage of this last solution is that the average intensity is made up of high-intensity periods and zero-intensity periods.
Recently, Leone and Pica [68-70] demonstrated the need to reproduce a simulated rainfall with both the same momentum per unit time and surface area ps [N/m2] and the same kinetic energy per unit time and surface area Ps [W/m2] as those of natural rainfall. According to those authors, the rainfall momentum is responsible for soil aggregate detachment due to the drop impact on the soil surface, whereas rainfall kinetic energy is responsible for rainfall detachability and transportability.
By reanalyzing the experimental data of Laws  and Gunn and Kinzer , Leone and Pica obtained the following relationship between the terminal velocity Vd (m/s) and the raindrop diameter d (cm):
By using Eq. (4.8) and describing the raindrop size distribution with a Gamma probabilty distribution (as suggested by meteorological data collected by Ulbrich ), Leone and Pica  deduced the following relationships for estimating natural rainfall momentum pn (N/m2) and kinetic energy per unit time and surface area Pn (W/m2):
where I is the natural rainfall intensity (mm/h).
Leone and Pica suggested that the same momentum and kinetic energy of the natural storm event can be reproduced using test duration and rain intensity in rainfall simulation . The portable rainfall simulator proposed by these authors, which is a modified version of the Guelph II simulator , uses a Fulljet nozzle (Spraying System Co.) for which the following relationships for estimating momentum ps (N/m2) and kinetic energy per unit time and surface area Ps (W/m2) are available:
in which Is (mm/h) is the intensity of the simulated rainfall.
By equating natural and simulated rainfall momentum [Eqs. (4.9) and (4.11)], the following equation for calculating the simulated rainfall intensity is obtained:
The ratio between the duration of the simulated rainfall ts (h) and the natural rainfall tn (h) is obtained by equating the kinetic energy per unit surface:
ts Pn 3.16 x 10-31117
Ps 0.448 x 10-311-42
For example, a natural rainfall intensity equal to 100 mm/h has to be simulated with respect to momentum per unit time and area, using a simulated rainfall intensity Is equal to 136 mm/h. With respect to kinetic energy per unit time and area, the same storm has to be simulated using a rainfall duration 1.44 times the natural rainfall duration.
Different rainfall simulators have been proposed. For example, Meyer and Harmon  developed an intermittent simulator for small fields or laboratory plots (1 x 1 m2). Moore et al.  proposed the Kentucky rainfall simulator for use in field studies, which also may be used for relatively large areas. Recently, a small rainfall simulator, which does not allow reproduction of natural rainfall properties, was proposed by Kamphorst  for soil erodibility studies.
Was this article helpful?