Four Lessons From
A Wetting Front Detector
The Irrigation Association Of Australia
2002 National Conference And Exhibition

Richard Stirzaker
CSIRO Land and Water, PO Box 1666, Canberra ACT 2601

Joyce Wilkie
Allsun Farm, Gundaroo NSW 2620

ABSTRACT

Wetting front detectors were installed on-farm in a drip irrigated pumpkin crop and a sprinkler irrigated garlic crop. The wetting front detector is a funnel-shaped instrument that is buried in the soil. The funnel concentrates the downward movement of water particles so that saturation occurs at the base of the funnel. The free (liquid) water produced from the unsaturated soil activates an electronic or mechanical float, alerting the farmer that water has penetrated to the desired depth. The detectors retain a sample of soil water that is used for nutrient monitoring. Four principles emerged that challenged the farmers' perceptions of how they were irrigating. First, the wetting patterns under drip penetrated deeper into the soil than they had imagined. Second, the wetting fronts from rain or sprinkler irrigation did not penetrate as deeply as they expected. Third, high concentrations of nitrate were measured during the first month after planting from the water samples retained in the detectors. Fourth, it was easy to misjudge the onset of exponential growth and its impact on water use. In each case the farmers found it easy to take remedial action. Irrigation intervals were shortened for drip and the duration of irrigation was lengthened for sprinklers. Extra effort was made to limit water applications in the early stages so that nitrate was not moved below the root zone. Lastly the farmers were alert to the rapidly escalating demand for water at the onset of exponential growth and the importance of avoiding water deficits during the period when yield is most affected. The experience showed that the basics of irrigation scheduling could be captured using a simple tool and simple information in a relatively short period of time.

INTRODUCTION

Against the background of poor adoption of irrigation scheduling tools by farmers, the FullStop wetting front detector was developed in answer to the question "what is the simplest information that would help an irrigator make a better decision?" (Stirzaker et al. 2000). In a range of experimental trials, the wetting front detector performed well in comparison to other methods of scheduling (Hutchinson and Stirzaker 2000, Stirzaker 2002). This paper evaluates how useful the detectors were in the hands of irrigators.

The evaluation took place on a small market garden near the town of Gundaroo in the Southern Tablelands of NSW. A range of high quality organic vegetables is direct marketed to subscription clients and restaurants. The owners had not used irrigation scheduling tools before, but were highly motivated to save water both because of limited supply and their commitment to environmental stewardship. They were keen to use the wetting front detector because of its simplicity and low cost.

In previous work the wetting front detectors had been used in "control" mode. Electronic detectors were connected to solenoid valves and automatically shut off irrigation when the water reached the required depth. The "control" method worked well, but its success depended on choosing the right combination of detector depth and irrigation frequency. In this study wetting front detectors were used as a learning tool; that is the farmers started with their own experience, and then modified their practice according to feedback from the detectors.

MATERIALS AND METHODS

The soil was a red chromosol with a sandy loam topsoil 300 mm deep overlying a light clay. The pumpkin crop Cucurbita pepo var delicata was planted on 30 December 2000 on raised beds spaced 1 m centre to centre. Each bed had a row of drip tape with 2 l/h emitters spaced 0.5 m apart, with seeds planted adjacent to each emitter. Compost was added before planting at a rate of approximately 60 m3/ha. This was incorporated in the top 200 mm of soil.

The pumpkin crop was harvested on 20 March 2001, and the crop residues removed. The beds were reformed, compost added at the same rate as above. The drip irrigation was removed and sprinklers set up with an application rate of between 10-15 mm/h. Garlic Allium sativum was planted in 4 rows per bed with 100 mm between the bulbs on 25 April.

Ten electronic wetting front detectors and five mechanical detectors were installed in the pumpkin crop. All detectors were placed with the rim of the funnel 200 mm below the soil surface directly below an emitter. Earlier work showed that the detectors record the wetting front when it is approximately 100 mm below the rim of the funnel, hence the depth of measurement for this crop was 300 mm. The electronic detectors were connected to a Campbell Scientific CR10X logger that recorded the time the float was up (water in the detector) and time the detector reset (water withdrawn from the detector by capillary action). The time and duration of irrigation was logged by a pressure transducer and rainfall logged using an automatic rain gauge. One emitter was connected to a short length of 4 mm tubing and placed directly into the rain gauge to monitor variations in irrigation rate.

Ten electronic and ten mechanical detectors were set up in pairs for the sprinkler irrigated garlic crop. The upper detector of each pair monitored wetting fronts at a depth of approximately 200 mm and the deeper detector at a depth of 300 mm. Electronic detectors, rainfall and irrigation were logged as above.

The farmers remained in complete control of the irrigation timing and duration. The mechanical detectors send up a float to give a visible indication that water has reached them. This information was immediately available to the farmers and influenced subsequent irrigations. The logged record was viewed several times during each crop, which further influenced their irrigation decisions.

Water samples were removed from the detectors at weekly (summer) or fortnightly (winter) intervals. Nitrate test strips (Quantofix, Macherey-Nagel, Duren) were used to give an immediate approximate measure of the concentration of nitrate moving past the detectors.

Lesson #1: Drip - shorten the interval between irrigation events

The detector installation depth of 300 mm in the drip-irrigated pumpkin crop was chosen because it marked the transition between the topsoil and subsoil. Since fewer roots were observed in the subsoil, it was reasoned that there was little point in pushing wetting fronts below 300 mm if the water might subsequently be difficult for young plants to extract. The very first irrigation showed how difficult this goal could be. Just 14 minutes of irrigation, or 612 cm3 per emitter, was enough to activate five out of ten electronic detectors at 300 mm. On an area basis this equated to an irrigation depth of 1.2 mm (Figure 1).

The next irrigation on January 5 was 0.7 mm, and only one detector responded. Two days later an irrigation of 1.1 mm set off seven of the ten electronic detectors. Over the first three weeks it became clear that 1-1.5 mm (12-18 minutes) would set off 5 to 7 detectors; less than 1 mm would set of just one or two of the ten. Clearly very small changes in irrigation elicited a large response from the detectors.

Figure 1.

The relationship between the amount of drip irrigation (left axis) and number of detectors that responded to each irrigation (right axis). The open bars at the right represent rainfall, not irrigation.

Rain on January 25 demonstrated the difference between complete and partial wetting of the soil surface (open bars in Figure 1). Rainfall of 11.7 mm was not sufficient to set off any detectors, and a further 14.8 mm the following day still had no impact. It took 25.1 mm of rain a week later set off 3 detectors, before a large rainfall event of 31.9 mm set off nine of the ten detectors.

There are two reasons for the small amounts of drip irrigation required during the early stages. First, the diameters of the wetting patterns averaged 20 cm, representing 6 % of the soil surface. Second, the only loss of water was soil evaporation from the small wetted area and some transpiration from the seedling. Once a detector had tripped, the soil between 100 and 300 mm remained close to the upper drained limit. Wetting fronts move quickly through wet soil, hence the short irrigation required.

Lesson #2: Sprinkler irrigation - lengthen the duration of each event

Rain during the drip irrigated pumpkin crop had already alerted the farmer to the fact that more than 15 mm was required to get the wetting front down to 300 mm, unless the soil was very wet. The actual amount of water required is a function of initial water content of the soil. This is the principle behind the operation of the wetting front detector. For a given soil/irrigation rate combination, the speed of propagation of the front is proportional to the initial water content (Philip 1969). Dry soil would therefore require a long irrigation and wet soil a short irrigation.

Detectors were placed at depths of 200 and 300 mm for the sprinkler irrigated garlic crop. It is preferable that wetting fronts do not penetrate as deep under sprinkler as they do under drip irrigation. This is primarily because the entire soil area is wetted by sprinklers. A second reason relates the way soil water redistributes after irrigation has ceased. Under drip irrigation, water is pulled sideways by capillarity as well as downwards. Under sprinkler irrigation, all redistribution is downward.

The garlic crop was planted in late autumn, and no irrigation was required until late spring. Figure 2 gives an example of how the detectors responded to rainfall during the early stages. Over an 18 hour period there was 23.9 mm of rain falling at a fairly constant rate of 1.3 mm/h. The soil was moist prior to this, as rain had fallen four days earlier. All five of the electronic detectors at a depth of 200 mm responded after 9.1 to 11.2 mm of rain. The five electronic detectors at 300 mm responded after 12.8 - 23.3 mm.

After a break of 18 hours the rain started again with a further 4 mm. In this case all ten detector at 200 and 300 mm responded after just 2.1 ­ 3.5 mm. This illustrates the point concerning initial water content and amount of water needed to trip the detectors. It took 23.3 mm to trip all detectors when the soil was moist, and just 3.5 mm when the soil was very wet.

Figure 2.

The response of detectors to rainfall during the early stages of the garlic crop. The solid line shows the cumulative rainfall over a three-day period. The horizontal bands denote the period when the first and last detectors at depths of 200 and 300 mm responded. The horizontal band on 14 June shows the period in which all detectors responded after rainfall resumed.

Five sprinkler irrigations were applied in the spring/summer. Though the weather was now warm and the crop at maximum leaf area, the detector record shows that, in general, too much water was applied. For each irrigation except 22 November, all five detectors at 200 mm were activated. On 20 Oct, 6 Nov and 28 Nov three or more detectors at 300 mm were activated. This demonstrates that water was moving past 300 mm and into the clayey subsoil. From this small data set, it appears 20-30 mm per irrigation would be appropriate. The interval between irrigations could be lengthened if more detectors responded and shortened if fewer responded the previous time.

Table 1. The dates and amount of irrigation water applied to the garlic crop and the number of mechanical detectors that responded at 200 and 300 mm.

Lesson #3 - Nitrate leaching when the crop is young

Each time a wetting front is detected, a sample of water is retained in the detector. This sample was used for rapid assessment of the nitrate status of the soil using nitrate test strips. At the start of the season the nitrate-N levels were high for both crops, even though no artificial fertilisers were used. In the case of the drip irrigated pumpkin crop nitrate-N dropped from 60 to 23 mg/l during the early crop stage when total irrigation was only 10 mm. Thereafter nitrate N remained fairly constant before falling sharply again during the period of exponential growth. It is important to note that the nutrient concentrations would be much higher in the 80% of the soil volume outside that wetted by the drip emitters. Thus the timing of rainfall and hence water and nutrient uptake would have an enormous impact on crop nutrition.

Figure 3.

The change in nitrate-N measured from samples stored in the detectors at 300 mm from the drip irrigated pumpkin crop.

Fewer water samples were available from the garlic crop. Since it was not irrigated during the early stages, samples could only be collected after rain. Nevertheless nitrate-N levels fell sharply after the rains in early June. The nitrate-N levels at 200 mm were quite low by early August, but still moderate at 300 mm, indicating that the topsoil had not been fully flushed. The nitrate-N level at 300 mm had fallen to low levels by mid October, the period when the crop was growing rapidly.

Figure 4.

The change in nitrate-N measured from samples stored in the detectors at 200 and 300 mm from the sprinkler irrigated garlic crop (left axis). The line without symbols shows the cumulative rain plus irrigation (right axis).

Lesson #4 - Misjudging the onset of exponential growth

The pumpkin crop was irrigated every second day during the first month. With one exception, the first nine irrigations activated 5 or more detectors. The subsequent nine irrigations activated 2 detectors or less. Even after the heavy rain on 4 and 5 February, when the soil profile was fully wetted, too little irrigation was given. It was not until 14 February, when the irrigation amount was increased to over 6 mm and the interval shortened to daily, that five or more detectors were consistently activated.

Figure 5.

The response of detectors (right axis) to irrigation amount (left axis) in the pumpkin crop from sowing to harvest. The total number of detectors was ten.

The rapid escalation in water use, from around 0.5 mm/day in mid January to 5 mm/day in mid February reflects the period of exponential vegetative growth. The crop was also growing into increasing temperatures. Flowering and fruit set occur during the latter part of this period, the time when the yield of many vegetable crops is most susceptible to water deficits (e.g. Rudich et al. 1977). Thus, if stress is going to occur at all, it is most likely to occur when the yield is most vulnerable, as deficits accumulate over the exponential growth period.

CONCLUSION

Irrigation scheduling is often portrayed by scientists as an exercise in accuracy - the idea that there is a defined refill point and upper drained limit and a precise amount of water can be added to satisfy the crop without wastage. Things look different on the farm. There are clear differences in the size of plants hence transpiration, especially during the early stages of growth. The drip emitters in this study were rated at 2 l/h but varied between 2.3 and 2.7 l/h. The sprinklers were less uniform.

Farmers are well aware of this variability. Moreover they often cannot irrigate exactly on cue, either because water is being used elsewhere on the farm, or some other cultural operation requires the irrigation to be withheld. Of greater importance, the farmer must optimise many tasks simultaneously, from soil preparation to marketing. The key question from the farmer's point of view is what is the value of information in reducing uncertainty, and what does it cost to get that information (Pannel and Glenn 2000).

In this study the wetting front detectors quickly honed in on the most important issues to be addressed by the farmer, as outlined in the four lessons above. They did not resolve the question of accuracy, but helped the farmer to move in the right direction. After all the soil is a buffer and each irrigation event need not be accurate. It is not important to be right every time ­ just important not to be consistently wrong.

In the words of the farmer involved in this trial, "the detectors provided a point of dialogue between the experience of the farmer and the language of the scientist". Essentially the detectors are a learning tool. They help the irrigator to evaluate their own practice and to modify this practice as their knowledge and confidence grows.

REFERENCES

Hutchinson, P.A., Stirzaker, R.J. (2000) A new method and device for scheduling irrigation. Irrigation Association of Australia, May 23-25, 2000 National Conference, p 584-592.

Pannell, D.J. and Glenn, N.A. (2000) A framework for the economic evaluation and selection of sustainability indicators in agriculture. Ecological Economics, 33, 135-149.

Philip, J.R. (1969) Theory of infiltration. Advances in Hydroscience, 5, 215-296.

Rudich, J., Kalmar, D., Geizenber G.C. and Harel, S. (1977) Low water tensions in defined growth stages of processing tomato plants and their effects on yield and quality. Journal of Horticultural Science, 52, 391-399.

Stirzaker, R.J., Hutchinson, P.A. and Mosena, M.L. (2000) A new way for small farm irrigators to save water. In: Proceedings of the 6th International Micro-irrigation congress, 23-26 October, 2000, Cape Town. South African National Association of Irrigation and Drainage, Cape Town, p. 4.3, 1-10.

Stirzaker, R.J. (2002) When to turn the water off; scheduling micro-irrigation with a wetting front detector. Irrigation Science (in press).