This discussion of rainfall simulators will address only field applications, although it can be argued that it is in laboratory research that simulators are most useful.
The main advantages are:
� The ability to take many measurements quickly without having to wait for natural rain.
� Tbe able to work with constant controlled rain, thereby eliminating the erratic and unpredictable variability of natural rain.
� It is usually quicker and simpler to set up a simulator over existing cropping treatments than to establish the treatments on runoff plots.
The disadvantages are all related to scale:
� It is cheap and simple to use a small simulator which rains onto a test plot of only a few square metres, but simulators to cover field plots of say 100 m� are large, expensive and cumbersome.
� Measurements of runoff and erosion from simulator tests on small plots cannot be extrapolated to field conditions. They are best restricted to comparisons, such as which of three cropping treatments suffers least erosion under the specific conditions of the simulator test, or the comparison of relative values of erodibility of different soil types.
� Simulators are likely to be affected by wind, but having to erect windshields undermines the advantage of simplicity.
Any consideration of using rainfall simulators must start by defining exactly what information is required. Simulators can be a useful tool for some purposes but quite unsuitable for others, and the objectives will affect most factors when choosing the most appropriate type of simulator. For example, in studies of infiltration and runoff it is not necessary for the simulated rainfall to have precisely the same characteristics as natural rain. In other studies it may be important that the erosion processes are not distorted by the simulated rain being different from natural rain. The required size of the test plots may dictate the best type of simulator: for example small plots may be suitable for studies of relative erodibility, but larger plots would be required for measuring rill erosion.
Some examples of experiments for which simulators are appropriate are:
� the relative protection afforded by different plant densities;
� the relative protection afforded at different times during the growing season;
� studies of relative erodibility;
� studies of soil infiltration characteristics;
� erosion and runoff from up and down slope row crops.
Some examples of studies for which simulators are not suitable are:
� crops grown on a contour, because the plot borders interfere with the normal water flow;
� any comparison of treatments which have only minor differences because under field conditions, experiments with rainfall simulators will suffer from considerable uncontrollable experimental variation;
� studies of physical processes which require accurate variation of rainfall characteristics such as changes in kinetic energy or intensity.
There are few commercial suppliers of rainfall simulators so it usual for research workers to build their own. However, there is a large amount of literature reporting the building and testing of rainfall simulators so it is usually practical to copy a previous design rather than to start from the beginning. In particular, a huge variety of commercially-available spraying nozzles have been tested. Also anyone contemplating building and using a simulator would be well advised to seek advice from other researchers.
Meyer (1988) offers two useful pieces of advice:
"Research with rainfall simulators involves many problems and pitfalls, and most researchers are glad to help others avoid problems they have encountered."
"Researchers should avoid becoming so involved in developing and improving simulators that little time is left for their use. The goal of rainfall simulator research should be the collection of accurate, useful data, not a perfect rainfall simulator."
The comments in Chapter 3 on the statistical design of field plot experiments are equally applicable to experiments with rainfall simulators. It is essential to have replications and to calculate the variance between plots in order to compare that with the variance between treatments. Also there must be randomization of the location of plots to avoid bias resulting from soil variation.
There is a huge amount of literature on the design, construction and operation of rainfall simulators, and some reviews are listed in Further Reading. Large simulators using test plots of 100 m� or more are valuable for the study of cropping treatments under something aproaching field conditions, and examples from the USA, Australia, and Israel are illustrated in Plates 6, 7 and 36. These machines are expensive and need teams of trained operators, so are outside the scope of this Bulletin, which will look at some simple and inexpensive simulators.
It is desirable that all the physical characteristics of natural rain should be reproduced as accurately as possible, but some latitude may be acceptable in the interests of simplicity and cost. The main characteristics are:
� Drop size. Raindrops vary from the minute droplets in mist up to a maximum of 6 or 7 mm diameter. This is a physical upper limit to drop size and above this any drops which form from the coallescence of more than one drop are unstable and will break up into smaller drops. The median drop diameter by volume lies between 2 and 3 mm and varies with intensity, as shown in Figure 50.
� The distribution of drops of different sizes varies. Cyclonic rain in temperate climates is mainly composed of small and average size drops, but high-intensity tropical thunderstorms have a greater proportion of large drops.
� Fall velocity. Falling raindrops reach a maximum (or terminal) velocity when the force of gravitational acceleration is equalled by the resistence of the drop falling through the air. The terminal velocity is a function of drop size and increases up to about 9 m/s for the largest drops, as shown in Figure 51.
� Kinetic energy is the energy of a moving body, and the kinetic energy of rainfall is the sum of the kinetic energy of the individual drops. Kinetic energy is a function of the size and fall velocity and is often used as a desirable parameter for a simulator because it is known that kinetic energy is closely related to the ability of rainfall to cause erosion. The kinetic energy of rainfall varies with intensity as shown in Figure 52, with an upper limit at about 75 mm/h. This upper limit is a consequence of the upper limit of the size of raindrops in that the highest intensities have more drops but not of an ever-increasing size, so the energy per volume of rain does not increase above intensities of 75 mm/h. The energy per second does, of course, increase with intensity at all levels of intensity. The intensity of rainfall is not related to mean annual rainfall - arid or semi-arid rainfall can reach intensities as high as in the humid tropics, although less frequently.
� Rainfall intensity or rate of rainfall can vary rapidly in natural rainfall, but it is usually not practical or necessary to build into rainfall simulators the ability to change intensity during a test. It is usual to choose and design for a single value of intensity, for example 25 mm/h to simulate temperate rainfall, or 75 mm/h for tropical or semi-arid rainfall.
� Uniformity of distribution of rainfall over the test plots is desirable.
Non-pressure droppers
Many simple simulators have used the principle of drops forming and dropping from the tip of tubes connected to a water supply. The size of drop is related to the size of the tubing. Metal, glass or plastic tubing has been used or hypodermic needles which are manufactured to a high degree of accuracy. An array of tubes of different sizes may be used to produce rain of different size drops (Plate 37).
The advantages of this method are that the size of the drops and their fall velocity are constant, the distribution of rainfall across the test plot is uniform and can be achieved with low water pressures.
The disadvantages are that unless the device is raised up very high, the drops strike the test plot at a velocity much lower than the terminal velocity of falling rain, and therefore the values of kinetic energy are also low. A large drop of 5 mm diameter needs a height of fall of about 12 metres to reach terminal velocity and this is difficult to achieve in field conditions. To some extent this can be compensated by using larger drops than in natural rainfall. Another disadvantage is that the size of the test plot is limited by the practicalities of constructing a very large drop forming tank. A simulator using this approach and mounted on a small trailer has been successfully used for many years in Venezuela (Plates 38 and 39).
The simplest possible form of spray, but which may be perfectly suitable for some simple applications, is a spray from a watering can, or the rose connected to a pressurized hosepipe as in Figure 53. Most commercial roses are drilled with all the holes of the same size, but it is easy to achieve a mixed drop distribution by drilling holes of different sizes. A basic problem with sprinklers of this type is that, like non-pressure drop formers, they only achieve a low impact velocity unless falling from a considerable height. With pressure sprays the impact velocity can be increased by pointing the spray downwards so that it leaves the nozzle with a velocity dependent on the pressure and then accelerates as it falls.
Another very simple simulator using a reciprocating garden spray is shown in Figure 54. The oscillation is controlled by a simple water turbine whose rotary action is converted into simple harmonic motion. This means that the distribution is not uniform as there is a dwell at each extreme, so a test plot using this principle should be located in the central part of the spray pattern.
Many types of spraying nozzle are commercially available, some designed for other purposes and some designed especially for rainfall simulators. A major difficulty is that if the spray is to include drops of the largest size which occur in natural rain, then the nozzle opening has to be large - about 3 mm diameter. But even with low water pressures the intensity produced from nozzles of this size is higher than natural rain (Elwell and Makwanya 1980). It is therefore necessary to have some kind of interruption of the spray to reduce the intensity to that of natural rain. In Meyer's 'Rainulator' two methods were used (Figure 55). The spray nozzles were mounted on an overhead carriage which traversed backwards and forwards across the plot, and also the flow of water to the nozzles was switched on and off by solenoid valves. This simulator and its derivatives are very efficient, but because they were designed for operation on large plots they are complicated and expensive. Most subsequent developments have therefore been concerned with designing simpler or smaller machines. One such variation was designed by Dunne, Dietrich and Brunengo (1980) for field use in Kenya, shown in Figure 56. A trolley carrying the spray nozzle is pulled backwards and forwards along an overhead track by two operators pulling on ropes.
Another approach is a machine based on a commercial rotating-boom irrigation machine shown in Figure 55 and Plate 7. Each boom carries the water supply to a number of nozzles on each boom which rotate slowly, powered by a water turbine. The machine is set up between two test plots so that rain can be applied simulataneously to both plots. Plot lengths up to 15 m can be rained on by one machine, or for longer plots two machines can be used (Swanson 1965; Hinkle 1990).
Another very popular device which has been copied and developed in many countries is the rotating disc originally designed by Morin, Goldberg and Seginer (1967) and illustrated in Figure 55 and Plate 40. A fixed nozzle sprays continuously, but the soil is intermittently shielded from the spray. The nozzle is directed vertically downwards, and just below it is a metal disc which rotates in the horizontal plane. A radial slot is cut in the disc, and each time this passes under the nozzle a short burst of rain passes through to the plot below. The proportion of the spray which passes is determined by the angle of the slot. This design allows the use of large nozzles which give the right drop size distribution and kinetic energy but which, when spraying continuously, produce excessive intensities.
For simple, home-made, do-it-yourself experiments with simulators one of the best approaches is to use the reaction of the jet to rotate an upside down irrigation sprinkler. An early device of this type shown in Figure 55 and Plate 5. This was designed to be light and portable so that it could be taken to remote areas with poor road access. After the machine is set up it can be positioned in turn over each of the six hexagonal plots which are arranged in circular pattern around the supporting mast. Six replications of each test can thus be made without moving the machine. A later model suitable for larger plots was developed at Silsoe College (Plate 41). A major advantage is that this type of machine can be assembled from off-the-shelf components, as was done in China (Plate 42).
Potential sources for the required pressure are gravity or by pumping. In hilly country it may be possible to generate sufficient pressure by piping water down from a higher reservoir, or from a storage tank on top of a tall building. The pressure generated is approximately 10 kN/m� for each metre of head.
Higher pressures are usually achieved by pumping. Peristaltic pumps which squeeze a flexible tube are very simple, but only appropriate for small flows. Centrifugal pumps are more appropriate for most simulators, and are readily available in all sizes directly coupled to electric motors or IC engines. Submersible electric pumps can be used for direct abstraction from streams, and hydraulic rams can be used to pump water from a steeply-falling stream to storage at a higher level.
Usually the researcher is looking for the lowest operating water pressure which will give the required drop size distribution and uniformity. An alternative, to date confined to laboratory use, is to inject air into the water supply, which gives higher discharge pressure for better uniformity without increasing rainfall intensity (Shelton, von Bernuth and Rejbhandari 1985; Hinkle 1990). Although mainly a laboratory technique, it should be possible to adapt to field use using compressed air cylinders.
The main factors are power sources, water supplies and access. Most simulators need a power source for motors and pumps, the only exceptions being those using gravity. Small reliable diesel- or petrol-powered generators are available but they are not cheap, and one more thing to be carried to the site. Some small simulators can run on electricity from batteries, but lead-acid car-type batteries are heavy and awkward to carry, and dry batteries, while suitable for electronic equipment, are expensive as a power source for motors or pumps.
Small simulators of the nozzle dropper type may need only small supplies of water because they can be targeted onto the test plot with little wastage outside the plot. Spraying systems need larger supplies, partly because they usually run at higher intensities, and also because the sprays usually cover a larger area than the test plot. It is important to calculate the amount of water which will be required, and how it is going to be delivered to the site. The spinning disc type and oscillating types can be fitted with a device to catch and recirculate the rain not going into the plot, but this has to be done without affecting the rain onto the plot. Large drops from leakages are a common problem.
Access is important. A site close to an all-weather road is so much easier to operate; indeed really large machines like the rainulator have to be able to take large trucks and trailers right to the site. But the sites to be investigated may not be easily accessible, so many simulators are designed to be carried by or operate from a four-wheel-drive vehicle.
Another practical consideration is reliability. Things never work as well in the field as when tested at the workshop. Components get dropped or bent in transit; pipes get clogged; pumps jam; motors burn out. The key is to make a field simulator as simple as possible, robust, easy to repair and with as few moving parts as possible.
Runoff plots used with rainfall simulators are the same as miniature runoff plots discussed in Chapter 3, and the same considerations apply to the plot boundaries, the collecting trough, piping the runoff and sediment to containers, and recording the volume of the runoff and weight of soil.
Much labour is required for setting up the large machines such as the rainulator. Several simulators are designed to reduce this. The Australian simulator in Plate 6 is made of lightweight materials, and can be picked up in one piece by a mobile crane and swung onto a new plot in minutes. The rotating boom in Plate 7 covers two plots from one position, and the rotating nozzle type in Plate 5 allows six replicated plots for each setting up.
Tests have shown that the results from simulator trials are considerably affected by the initial soil moisture of the test plot, and to reduce this undesirable variable it is usual to specify some standardized pre-wetting treatment. One method is saturating the soil 24 hours before the tests, so that the soil is approximately at field capacity. An alternative is to apply 25 mm of rain at 100 mm/h four hours before each test.
The amount of simulated rainfall during the test must be measured. One approach is to cover the whole plot with a collecting sheet and measure the rain caught in a fixed period. This is done before and after the test to make sure there is no change during the test. Alternatively, measurements can be made during the test either by small rain-gauges installed on the test plot, or by collecting channels across the plot, as in Plate 7.
