How to run a catch-can test

What you need

You need a set of identical catch devices, a way to read them, and a timer. Purpose-made catch cans with a printed scale read directly in inches or millimetres. Straight-sided containers such as tuna tins work too if you measure depth with a ruler. Graduated cylinders paired with any container work as well, as long as you know the container's throat area, since depth equals collected volume divided by throat area. The arithmetic, and the throat areas for the common can sizes, are in converting mL catch readings.

Identical devices matter more than fancy devices. Mixing container shapes inside one test changes the effective collection area from can to can and corrupts the uniformity statistics.

• 8 or more identical catch devices per zone; 16 to 24 gives a steadier low-quarter figure
• A timer for the zone runtime
• A notebook or the Irrigaudit workspace to record each reading in order
• For mL readings: a graduated cylinder and the throat diameter of your containers

Placing the cans

Set cans in a regular grid across the whole irrigated area, not just where coverage looks good. Common practice places cans roughly a half spacing from heads and from edges, then fills in the middle of the zone at even intervals. Include the areas that prompted the audit: dry corners, slopes, and spots shaded by shrubs.

Keep cans level and clear of grass blades that would deflect spray into or away from the opening. On slopes, push a small terrace flat for each can rather than letting it lean.

Running the test

Run the zone long enough for the average can to collect a comfortably measurable depth. For fixed sprays that is commonly 10 to 15 minutes; for rotors and rotary nozzles, which apply water more slowly, 20 to 30 minutes or one or more full rotation cycles. Record the exact runtime; the precipitation rate divides by it directly.

Read every can, including the empty ones. A zero is data: it says part of the zone received nothing during the test, and dropping it silently inflates both uniformity figures.

A worked example: 16 cans, 12 minutes, one fixed-spray zone

Here is a complete test with nothing left out. A fixed-spray zone, 16 identical catch cans on a regular grid, the zone run for exactly 12 minutes. The depths in inches, in the order the cans were set out: 0.42, 0.38, 0.31, 0.45, 0.29, 0.40, 0.36, 0.22, 0.47, 0.34, 0.41, 0.18, 0.39, 0.44, 0.33, 0.27.

Sorted, the four lowest are 0.18, 0.22, 0.27 and 0.29. Those four are the low quarter, because the low quarter is the lowest 16 divided by 4 readings rather than a fixed count, and they are the only readings the DU figure is built from. The table below is exactly what Irrigaudit returns for this list, with the derivation of each line.

Note how far apart the two uniformity figures sit. CU is 81 percent and DU is 68 percent from the same 16 numbers, because CU averages the deviation of every can while DU looks only at the driest quarter. Quoting the higher of the two as the zone's uniformity is the single most common way an audit flatters a zone that has a dry corner.

Turning readings into results

The precipitation rate is the average catch depth divided by the runtime, scaled to inches per hour. The low-quarter DU is the average of the lowest quarter of readings divided by the overall average. What DU and CU each measure is worth reading before you act on either. Christiansen CU is based on the average absolute deviation from the mean. The Irrigaudit workspace computes all three as you type readings in, flags the low-quarter cans, and grades the DU against the DU target bands for the head type.

With fewer than four cans there is no meaningful low quarter, so treat any DU from a tiny test as an anecdote rather than a measurement. The workspace says so explicitly when the sample is that small.

From DU to a runtime, which is the point of the whole exercise

A uniformity figure on its own changes nothing. What it buys you is a defensible runtime, because a zone that waters unevenly has to run longer than its precipitation rate alone suggests if the driest part of it is to get what it needs.

Take a net weekly requirement of 1.25 inches on the zone above. At a measured 1.77 in/hr the bare arithmetic is 1.25 / 1.77, or 42 minutes a week. Nobody waters to the average, though, because the average is not where the grass dies.

Irrigaudit stretches that baseline by a scheduling multiplier, and it offers the two that are actually in use. The Irrigation Association auditor multiplier, 1 / (0.4 + 0.6 x DU), accepts some deficit in the driest quarter and gives 1.24 at a DU of 0.678, so 42 minutes becomes 53. The stricter reading, 1 / DU, waters the low quarter to full need and gives 1.47, so the same zone asks for 63 minutes. Ten minutes a week separates those two interpretations of the same measurement, which is why the workspace makes you choose rather than picking for you.

Put an area against it and the cost of the uneven coverage becomes visible. At 0.623 gallons per square foot per inch of water, 53 minutes a week on an 1,800 square foot zone at 1.77 in/hr is about 1,737 gallons. Raise the DU and that number falls without anyone watering less often; the extra minutes exist only to compensate for the spread in the readings.

The four mistakes that make a test meaningless

Every one of these produces a number that looks like a measurement. The workspace flags the first two and cannot detect the others, so they are worth knowing by heart.

• Too few cans. The low quarter of three readings is one reading, so DU becomes the ratio of the single driest can to the mean. Irrigaudit marks any test below four readings as a small sample for exactly this reason, and the honest reading of such a result is anecdote rather than audit.
• Dropping the zeros. An empty can is the most informative reading in the set, and deleting it raises both the mean and the low-quarter mean. A zone with one dead head can post a respectable DU once its evidence has been thrown away.
• Mixing container shapes. Depth is volume over throat area, so two different openings in the same test are two different scales. The uniformity statistics have no way to know and will quietly report the difference between your containers as a difference in the sprinklers.
• A runtime too short to read. If the average can holds a tenth of an inch, a reading error of a thirty-second of an inch is a third of the signal. Run long enough that the readings sit well up the scale, and record the runtime exactly, because the precipitation rate divides by it directly.

Keep going