Tone Ag Consulting Ltd.

Except in laboratory conditions (and maybe not even then), soil tests are not 100% accurate and are subject to a margin of error. The target accuracy is about +/- 15 % with a precision level of about 80%. For example, if you had a field with a nitrogen (N) level of 20 and you sampled it 10 different times, you might end up with results such as 12, 17, 18, 21, 21, 21, 22, 23, 23, 26. 8 of the 10 are within 3 (15%) of 20 so this is a good soil test. (See Sampling for Soil Testing by Swenson from NDSU for a light mathematical treatment of soil sampling.)

Take away lesson: Don’t focus on the exact number given by the soil test – what matters is if your field is low, medium or high for a given nutrient.

As a reminder, accuracy is how closely the numbers reflect the true value. Precision is how likely further tests are to give similar values. More details on accuracy and precision.

Factors Affecting Soil Test Accuracy

There are a number of factors that can affect the accuracy of a soil test.

• Sampling representative areas of a field. This is something that often gives us trouble with new clients. They get results that are orders of magnitude greater than expected. Upon reviewing the sample points with them, we find out that the northwest corner of the field used to have a dairy barn and manure was dumped there regularly. An inexperienced soil sampler may also miss things such as saline spots which will throw the results off.
• Sample depth. If the indicated sample depth does not reflect the true sample depth, the results will be incorrect.
• Number of samples – I will discuss this further below
• Laboratory variability. Different labs use different tests and have different people evaluating the results. This is out of the soil sampler’s hands, but if you get a result that seems odd it’s worth asking the lab to rerun the sample.

Number of Samples

Here in Manitoba, getting an accurate level for N is pretty easy. Even for manured soils, 34 samples can get you within 5 ppm accuracy with 90% precision. If you’re willing to go down to 80% accuracy, even fewer samples are needed.

The issue we’re dealing with right now is the number of samples needed to accurately determine (P) Phosphorus levels for a field. In a project carried out by DGH, they got from 188 to 582 samples for 5 ppm accuracy with 90% precision, with more samples needed on clay soils and fewer on till soils. A study by Scott Dyck and Cliff Loewen (currently of Agra-Gold) came up with similar results for P variability in manured fields. Taking 500 samples is not economical as it would take about 3 hours to sample one field. There are practical considerations as well – this would be over 50 gallons of soil and would weigh over 400 pounds.

We are working on finding a reasonable compromise between accuracy and economics to provide the best soil sampling service to our clients!