Checking and calibrating the neutron probe

The CPN 503 neutron probe can be easily checked by taking a count in a water drum - it is simple preventatative maintenance.

Taking a standard count in a 200 l drum of water (242784 bytes)

Taking a standard count in a 200 l drum of water.

A more permanent solution would have the tube sealed with silicon in a hole cut in the top of the drum, and a block of wood or brick should be put in the drum to separate the bottom of the aluminium tube from the steel drum.

All counts are taken with the detector at the centre of the water.

There are two reasons for doing a water drum count...

  1. A quick count (8*16 second) as a routine check on the stability of the probe.
  2. A standard count (32*8 second) if new equations need calculating.

The quick count should be done regularly, preferably every time the probe is used. If the data is logged and posted to the software (e.g. as Site 1) the counts will be automatically recorded, and any drift in the instrument can easily be identified later -see here for details. The average of the 8*16 second counts should vary by no more than about plus/minus 200 counts.

The standard count should be done at the beginning of the season, and after the instrument has been serviced or repaired. The test is done automatically by pressing <STD> and <START> on the 503DR.

The Chi-Sqvalue, a measure of the stability of the instrument, will be calculated

At the end of the test...

MS 24400new standard count
<STEP>PS 24500current standard count
<STEP>MCHI 1.06Chi-Sq of the STD test
<STEP>to cycle through values
<ENTER>to save the new standard count

Immediately after a service the Chi-Sq value should be in the range 0.9-1.1, and after normal field use the acceptable range is range is 0.75-1.25. The average of the 32 counts should not change by more than about plus/minus 100[*].

The standard count is saved in the neutron probe head and uploaded with the readings in the hader line, but is not normally used.

The standard count can be used in count ratio form - see below and water drum method.

Calibration Equations

The neutron probe counts neutrons that are returned after they have 'bounced' off hydrogen (water) particles in the soil. To calculate the soil moisture content the counts need to be multiplied by a calibration equation. A linear equation is of the form...

Soil Moisture = Count * Slope + Intercept

The slope and intercept will vary with different soil types and variations between probes.

The variation between probes can be corrected if each probe has a set of counts taken in a water drum, with the assumption that if a probe has a 10% higher count in water than another probe, then it will also read 10% higher in soil. A large mass of water is used as the standard rather than the wax shield because it is a constant across the world, and absorbs all neutrons.

To allow equations derived with different probes to be compared they should be in water drum count ratio form...

Soil Moisture = (CountInSoil)/CountInWater) * Slope + Intercept

Probe for Windows can accept equations in either form - see calibrations

Changing Calibration Equations

If the standad count has changed significantly, and the 503DR has been checked, the new equations can be calculated from...

Correction = OldWaterDrumCount / NewWaterDrumCount

NewSlope = OldSlope * Correction

For example, say the standard count has changed fom 19,450 to 19,960, and the original equation was...

VSW = 0.00343 * Count + 1.42


Correction = 19,450 / 19960 = 0.974

NewSlope = 0.00343 * 0.974 = 0.00334

VSW = 0.00334 * Count + 1.42

If the equations are in water drum count ratio form they do not need changing if the water drum count changes.

Deriving Calibration Equations

Calibration equations can be derived by taking soil samples from around a neutron probe tube. The volumetric soil moisture content of each of the samples is then calculated by measuring the water content (by weight) and the bulk density of the sample. At the same time the counts from the neutron probe are recorded.

At least 10 samples will be required at each depth or soil type when the site is dry, and a further 10 samples when the site is wet. Ideally further samples should be taken at intermediate moisture contents. The soil tests must be done carefully, and in particular make sure that there are adequate tests at the dry end.

The moisture content sample results are then plotted against the corresponding neutron probe counts, and the best straight line plotted through the points.  This line is the calibration equation, relating neutron probe counts to volumetric soil moisture.
In fact the relationship is non-linear, but it is generally assumed that a straight line approximation gives acceptable results over the range of soil moisture encountered in agricultural soils, and noting the wide scatter in the measured soil samples.

The equation can be used with other neutron probes if the water drum count ratio method is used, with Count/WaterDrumCount being plotted instead of just Count.

A warning - unless the tests are done extremely carefully you may have an unrepresentative equation. Check what the calculated soil moisture will be with a low count measured in a very dry soil.

Some references on calibrating neutron probes, and some equations for different soil types, are here.

Converting CPN factory calibration to Probe for Windows form

The CPN factory calibration is in units of ins/ft per CountRatio, with the standard count taken in the wax shield and not in a water drum.

The wax shield count will be less than a water drum count, and is also less consistent.

The Probe for Windows calibration is in units of % per Count.

The conversion needs to allow for '%' instead of 'ins/ft' as well as the standard count.

The CPN factory calibration has 2 values - A (slope) and B (intercept) - and the Standard Count saved in the probe is also required.

Press <STD> on the probe head; or it is item 4 in the header line of the post file.

The A and B values are in lines 2-17 of the post file.

Slope = (A / StandardCount) * (100/12)

Intercept = B * (100/12)

As an example, if the CPN Slope (A) = 1.953; CPN Intercept (B) = -0.084; standard count (shield) = 64841

Slope = (A / StandardCount) * (100/12) = (1.953 / 64841) * (100/12) = 0.0025

Intercept = B * (100/12) = -0.084 * (100/12) = -0.7

Acceptable variation in water drum count

From the CPN 503DR manual...

The difference should be less than 0.707 * Sqrt(AverageCount) for 95% of standard count tests.

For an average water drum count of 25,000, the difference should be less than 0.707 * 160 = 110, although up to 5% of tests can be outside this limit.

Recording water drum counts in the software

To record the water drum counts in the software, create a new site with site number say '1', and set the 'depths' to be 10,20,30... 60,80, and RootZone1 to be 0-100.

The number in the root zone 1 row will be in mm - divide by 10 to get the average VSW%.

e.g. If 638mm is displayed, the average VSW = 63.8%.

This number should be reasonably constant, and could be plotted as a time graph.

If the equation constant is subtracted from the calculated VSW (e.g. 63.8+1.0 = 64.8), the result is the slope of the calibration equation expressed using the water drum count ratio method - see here for some examples.