Thursday, December 16, 2010

What we learned from the Zimpeto couve crop

This is the final post on the Zimpeto couve crop. So what did we learn?

It is possible, with the measurement techniques we had available, to grow a big couve crop with very little water, and in our case, no added nutrients.

We could do this by watching the watermark sensors to check the soil did not get too dry, and at the same time ensuring that the 500 mm detectors were only activated occasionally, to check that the soil did not get too wet.

Our irrigation strategy was very influenced by the nitrate readings. The very high readings around day 13 meant that we needed to be careful not to leach the nitrate out of the profile.

For example, we measured above 500 mg nitrate/L at 300 mm depth at the start of the experiment. At a soil water content of 25%, 500 mg nitrate/L adds up to around 85 kg N per ha in the top 300 mm of soil. We add another 10 kg in the irrigation water. This would be a large proportion of the total nitrogen fertiliser the crop requires.

How well could we have done if we just had the detectors and not the nitrate, conductivity and watermark sensors?

We can draw some lessons from this experience which suit this soil type and irrigation system as follows:
When we do not activate the detectors at 300 mm depth, the plants might be getting just enough water, or more likely not quite enough. When we activate the 500 mm detectors, the soil is very wet.

We could come up with the following guidelines:

1) When the plants are small, irrigate frequently but do NOT activate a 300 mm detector. This will ensure the nutrients are not leached.

2) As the plants start to grow quickly, activate the 300 mm detector once or twice per week

3) When the crop is at a yield-sensitive growth stage (for vegetables this is usually flowering time, or ‘hearting’ time for leafy crops) activate the 500 mm detectors a couple of times.
There is one remaining lesson. We must minimise the leaching of nutrients, but we must also manage the salt. By the end of the season the wet side had an extra 624 kg Salt per ha in the roozone. The dry side had more (660 kg of salt / ha), applied in the irrigation water.
A few larer irrigation events towards the end of the season, when the soil nutirent content tends to be low, will ensure salt levels do not build up too much in this soil. By deep I mean activating the 500 mm detectors. The detectors placed at 700 mm were too deep to be useful.

Salt and nitrate leaching?

Here we are using a mini disk permeameter to get some idea of how fast the water moves into the soil, and how wide the wetting patterns become. This all helps us to manage salt and nutrients.

We pumped water from the aquifer at an electrical conductivity of 0.8 dS/m. This equates roughly to 0.5 g of salt for every litre of water applied. So when we apply 159 mm of irrigation water we also add 865 kg of salt per ha. We did not measure which salts were in the water, but most of it is likely to sodium chloride. Plants don’t like sodium chloride. We can let it build up a bit in the soil, but eventually we have to apply extra water – more than the plants requirements, to leach the salt out.

The groundwater also contained 30 mg/L of nitrate. So when we are irrigating we are also applying some fertiliser.

On the 'wet side' we applied 159 mm of water, which has 865 kg of salt dissolved in it and this included 11 kg of nitrogen (in the nitrate form).

On the 'dry side' we applied 134 mm of water, which has 686 kg of salt dissolved in it and this included 9 kg of nitrogen (in the nitrate form).

That takes care of the input side of the equation. By collecting water from the wheelie bins - the water that the couve did not use - we can complete the output side of the picture (on a per hectare basis).

On the wet side we collected the equivalent of 18.6 mm of drainage water which had 241 kg of salt dissolved in it which included 11 kg of nitrogen (in the nitrate form).

On the dry side we collected the equivalent of 2.7 mm of drainage water which had 26 kg of salt dissolved in it which included 0.1 kg of nitrogen (in the nitrate form).

What did we catch in the bins?

One of the buried wheelie bins is just visible in the left of the picture. Up to day 40, the 300 mm depth detectors collected water 3 times. Then on day 41, after a few larger irrigation events, the 500 mm detectors collected their first samples. The detectors at 700 mm did not collect water, and there was no water in the bins.

A very large irrigation was applied on day 66 (13 hours or 8.45 littres per dripper or 28 mm averaged over the whole area). This time the 300 and 500 mm detectors collected, and water also started to collect in the bins. We measured drainage of 2.7 mm from the drier side and 18.6 mm from the wetter side.

We did NOT collect ANY water in the detectors at 700 mm depth.

Water must have gone past this depth because we measured water in the 900 mm deep wheelie bins. However none of this infiltrating water could be collected in a FullStop funnel at 700 mm depth. This tells us something about the limitation of the technique. They do not work when placed too deep.

A New Trick

The sandy soil was easy to dig and had no structure at all, so we could try a new trick. We went to a hardware store and bought a couple of the largest “Wheelie” bins we could find. These were dug into the soil and gravel placed in the base. Now we needed a means of showing how much water collected at the bottom of the bins. We used the 'base piece' and 'float housing' from the wetting front detectors to make a ‘riser’ to the surface. The filter screen in the base piece allowed the water into the riser. The 4 mm off-take on the base piece and tubing was used to suck water out. Then we placed a thin round stick into the float housing (where the floats normally go) so we could measure the depth of water inside – much the same as we use a ‘dipstick’ to measure the amount of oil in the car engine.

The bins were refilled with soil in the order it was removed from the hole and the drip tape went back over the top. Two couve plants were then planted in each bin. Now any water or nutrient going past the couve roots would collect at the bottom of the bin (900 mm) and could be measured accurately.

You can see the setup going in on youtube at

Monday, December 13, 2010

How good were we really?

This was a good crop, grown with very little water. The first detector samples at 30 cm depth showed that we had lots of nitrate in the soil, left over from the chicken manure applied to previous crops. So we managed with just the nutrients already in this sandy soil. But how good were we really?

For those who follow this blog you will know that I think across five strands of enquiry to answer this question:

1. LOCAL KNOWLEDGE: what do the people I’m working with already know? What do they do and why, and how did they arrive at these practices?
2. THERMODYNAMICS: Measure the water going on - by flow meter, irrigation run time, collecting water in cups (or whatever) and compare with the theoretical amount that can be evaporated - from a weather station, a model, an evaporation pan (or whatever). It does not matter if the method is rough – as long as it is consistent.
3. SOIL WATER STATUS: I go for the simplest measurement protocol and, once I have some local experience, target only a few depths that can give me most of what I need to know (usually Watermark sensors at two depths)
4. WETTING DEPTH: Wetting front detectors that give a visual indication to the irrigator of how deep the irrigation water has penetrated into the root zone
5. SOIL SOLUTION: Measure the electrical conductivity and nitrate concentration from the water samples collected from the wetting front detectors

The reason for five strands of enquiry is that they are independent, but complementary lines of evidence. Frequently, if you look at one strand alone, you think you are doing OK (say soil water status: strand 3). Then you look at nitrate leaching or salt accumulation (strand 5), and it gives a rather different picture. So then you need additional information, like the ratio between water applied and potential evapotranspiration so you can troubleshoot (strand 2).

Wetting front detectors are useful at showing how deep the water is going, and the salt and nutrients that are being carried with the water. But they are not perfect. For example, infiltrating water is captured by the funnel, but when the soil in the funnel is saturated, and the soil outside the funnel is slightly drier, then water can be ‘wicked’ out of the funnel as fast as it arrives. This happens especially in fine sandy soils with no structure, and also when the wetting front detectors are placed quite deep. The water is passing the detector quite slowly, and although some water is accumulating inside, it is wicked out again before the float can pop up.

This was a perfect site to test the ‘problem’ of deep detectors, and to see how much water and nutrient could go past.