Rainout Shelters: Some Basic Principles of Design and Operation

By A. Blum



Rainout shelters are designed to protect a certain area of land against receiving precipitations so that an experimentally controlled drought stress can be imposed on that area. Many types of rainout shelters were designed and used, with better or lesser results. This is not a comprehensive review of this system but a concise discussion of the most important issues in designing, constructing and operating rainout shelters.


There are two main designs: (1) static and (2) movable. Within the moveable design there are automatic/motorized and manual versions. The automatic version is signaled to move over the protected plot by a rain sensor and an electric drive system. The manual version is moved either by manually switching the drive on ("manually driven") or by manually pushing it ("manually pushed") over the protected plot. The 'manually pushed' must by lightweight and hence it is cheaper and can cover a limited land area. The automatic version is becoming less popular because of reliability problems and cost. Hence, many automatic types become 'manually driven' after the first failure of the automatic system. The earlier automatic shelters were designed without any consideration for light transmission because they were not expected to remain over the protected plot for a long time.

The manual version is moved from its parking spot onto the protected plot whenever a rain is expected and not when the rain begins. It is moved into the parking space whenever rain is expected to cease completely. Good weather forecasting service is therefore important. If forecasting is unreliable, better have the shelter over the protected plot more time than expected. Therefore, the shelter construction must allow sufficient light inside as well as some ventilation.

The direction of the protected plot and the parking place of the shelter should be designed so that the parked shelter will not shade the plot in the morning or the evening and that the direction of the wind would not allow rain to blow under the shelter. The shelter parking place cannot be used for growing experimental plots.


Automatic rainout shelter at UGA Griffin Station turfgrass research, used to protect experiments performed in containers. However, the shelter is very dim.


The area around the shelter should be managed in consideration of surface runoff and the drainage and diversion of rainfall flowing off the shelter especially under storm conditions. Hence ditches and other means of protection should be properly placed around the installation, with respect to topography. Consideration should be given also to the fact that horizontal flow of water in a saturated soil can be appreciable.

Often experiments under a rainout shelter involve control (non-stressed) plots. If the protected plot is not covered by the shelter for an extensive period of time, than the control plots may be placed outside the shelter and exposed to rainfall (or irrigation). However, if a perfectly controlled experiment is planned, then the control plots should be placed also under the shelter and irrigated. Drip or manual irrigation is a good way to assure proper separation between control and stress plot treatments. The two treatments should constitute separate blocks under the shelter. Experiments were also performed by using containers (pots or large vessels such as PVC tubes for root studies) placed under a rainout shelter. Rainout shelters provide an environment which is much closer to the ambient than a standard greenhouse.




This type of shelter is actually a well ventilated greenhouse that can be quickly rain-proofed or ventilated, on and off, at will (Fig.1A). The principle is that when it is not closed for rain exclusion in must be well ventilated to equilibrate with the external ambient conditions. Since it is permanently covered by a (transparent) roof, the water regime inside must be regulated by irrigation. Control (non-stress) plots are grown inside the shelter and are fully irrigated. Both the walls and part of the roof can be widely opened for ventilation when there is no rain. Irrigation can be provided by overhead sprinklers or by drip irrigation. Consider the possibility that drip irrigation might result in a different root system as compared with that under other irrigation methods which supply larger quantities of water in shorter time periods, similar to rain. Some drip irrigation systems can be programmed to deliver larger than normal quantities of water per unit time.




Fig.1A Inside view of a static rainout shelter at the Shanghai Agrobiological Gene Center, China. The shelter is in an open (ventilated) position. (B) Manual rolling of sidewalls. (C) An electric motorized system for rolling part of the roof cover made of a plastic sheet. Rice growing inside displays a systematic designated gradient of stress towards the walls where a drainage


These shelters are usually constructed with strong frames and agricultural grade, light transmiting plastic walls as well as transparent strong plastic roof panels. Strong winds and the prevalence of hail should be considered in the construction of the roof. The plastic walls can be rolled up and down easily, manually or by electrical motor (Fig.1B). The opening of the roof can be motorized if part of the roof is made with the same material as the walls (Fig.1C). Various methods are available for opening the roof, which depend on cost and materials.

Both the roof and the ground surrounding the structure should provide full protection against vertical or horizontal leakage of water into the shelter. Consider also that leakage can also occur by saturated soil water flow under the foundation of the shelter. If possible, periodic measurements of temperature, humidity and light should be logged as well as soil moisture data – depending on the type of work performed under the shelter.


Polyethylene covered "tunnels" are being used as rainout shelters at the Broom's Barn Research Center (Suffolk, UK). Fig.2 demonstrates their use over experimental sugar beet plots. Each single structure is 150 m long, 3.5 m high and 30 m wide. The frame parts are available commercially (Google search 'poly tunnel' or contact the Station). The lower sides can be opened for ventilation (Fig.2-middle) and a trench is dug along both sides to collect rainwater. Drip irrigation can be installed inside if certain plots require irrigation (e.g. non-stress controls).

The tunnels have a significant effect on crop microclimate, but do not create combinations of weather factors that are unlikely to occur in practice. The polyethylene is 100 mm thick and transmits more than 90% of the incident PAR when new. By the end of the season this decreases to approximately 80%; therefore fresh polythene is installed each year. Air temperature at crop level rarely exceeds outside temperatures by more than 1 ºC. Even with open sides and ends, the largest effect of the tunnels is reduced wind-speed. In 1999, for example, wind-run was decreased under the tunnels by 50-60% so that, during the period the crop was covered, estimated cumulative evapotranspiration was 202 mm; this was 24% less than that of the crop outside the tunnels.





Aerial view of 4 structures.

Inside the structure. Note the open lower sides for ventilation and the rainwater collecting trench.


A stressed sugar beet plot. Note the border rows adjacent to the trench on the left.

Fig.2. Broom's Barn Research Station model of "polytunnels" rainout shelter. Station web site: http://www.rothamsted.bbsrc.ac.uk/broom/sbrindex.html. Information courtesy of Dr. Eric Ober.





Many versions of motorized moveable shelters were constructed. The "classical" design used in many of these shelters consist of the same principle. A roof /walls structure is mounted on wheels on a track. The structure can be driven by electric motors. The drive can be switched on and off manually or via an electronic signal from a rain sensor. The structure is relatively heavy and often not allowing sufficient light inside.

A very large and elaborate moveable rainout shelter has been constructed at the National Key Laboratory of Crop Genetic Improvement Huazhong Agricultural University, Wuhan, China (Fig.3). Besides covering a very large area this installation also includes an elaborate soil drainage system and control which allows growing paddy rice and draining the paddy water at will in order to impose soil moisture stress. The installation is used in large scale screening of genetic and breeding material. The large structure is composed of two parallel units. Each unit is constructed of four relatively transparent roof sections (Fig.2.-inset) which can be driven (two at a time one under the other) into the parking spaces on the two sides of the protected plot.

Large area rainout shelters can adopt a multi-cover system which acts like a bellows and expands over the protected area (Fig. 3-1). This system has the advantage of using small parking area for the moveable roof while avoiding very large and heavy roof construction.



Fig.3. A large motorized moveable rainout shelter at the National Key Laboratory of Crop Genetic Improvement, Huazhong Agricultural University, Wuhan, China. Rice experimental materials grow in the protected area while the shelter is parked on the edge of the area. Inset: the installation set at full cover.


Fig. 3-1. Multi-cover installation combined of 4 roof sections covering the protected plot. Each section travels on its own tracks and sections are parked one under the other when they are off the protected plot. Notice the water drains on each section (The Luoyang Station of the Luoyang Academy of Science, Henan China).


Less elaborate and also less costly light, manually-driven rainout shelters allow cost-effective control over smaller land area. These are becoming popular after the initial design by the ICRISAT (Patancheru, India) and the discussions held under the Rockefeller Foundation sponsored workshop entitles “Field Screening for Drought Resistance in Crop Plants with Emphasis on Rice”, held there in 2000. Several units of the ICRISAT model in one field are depicted in Fig.4. In this example rice experiments are performed in the protected area while the parking area is used for some seed increase. The original design is available in the Appendix, courtesy of Mr. Ravindranath, Senior Manager, ICRISAT. This unit consists of a wheel-mounted light structure wrapped with polyethylene sheet. The unit travels on iron tracks. The lower part of the walls in this design is too open and should be extended to the ground level as much as possible. Front and back sides are open here, but this may not be allowed under conditions where the direction of rain and wind might cause rain to penetrate the shelter. Rolled up polyethylene front and back walls (e.g. Fig.1A) or any other arrangement that will allow lifting these walls during the installation travel might offer a solution. Always consider that rain can be driven by a strong wind.



Fig.4. Lightweight manually movable rainout shelters at ICRISAT Center, Patancheru India. The construction plans are presented in the Appendix.


Fig.5. Lightweight manually movable rainout shelter at Narendra Deva University of Agriculture and Technology, Faizabad, India. Inset: the wheel in its U-shaped track.



This design by ICRISAT was in principle repeated in constructing the low cost rainout shelter at NDUAT Faizabad (Fig.5). Here, the u-shaped tracks are more robust and hold the iron wheels well. Wheels should contain well greased ball bearings.

Because of their light weight these rainout shelters can be very unstable under windy conditions. Strong winds can break the construction or lift the unit altogether off its tracks. More solid construction might be needed where strong winds are expected. Simple anchoring arrangements are essential in both the parking space and the protected plot to secure the structure in case of expected windy conditions. The simplest arrangement is that used for anchoring light aircrafts in airports, using cables that can be fast-connect to a weight or a stake lodged very firmly into the ground. These are needed in the four corners of the shelter. Another option is an arrangement to secure the structure onto its rails, as long as rails are well anchored to the ground. Do not underestimate the power of wind.


Another simple version of a lightweight rainout shelter was seen at the experiment station in Merredin, Western Australia (Fig.6). Here the structure is static but the cover is provided by a moveable plastic “curtain”. The structure is similar to a large tent. A portable frame is anchored to the ground with steel cables. Steel cables run along the over the protected plot. A strong sheet of plastic (not shown) is connected to the cables so that it can be manually stretched over the plot or moved off horizontally like a curtain. The “curtain” is attached to the horizontal cables by special clips that can smoothly move on the cable (Fig.6-inset). This installation can be disassembled readily and moved to another field.



Fig.6. “Frame and curtain” version of a lightweight rainout shelter seen at the experiment station in Merredin, Western Australia. Inset: the clip used to hold the movable curtain.



Contributed material to this article will be considered.





Advanced rainout shelter design:


Automated Rainout Shelter’s Design for Well-Defined Water Stress Field Phenotyping of Crop Plants






 Simple rainout shelter design: