Turfgrass soil sampling, part 3 of 7
One research project suggests 20 subsamples be composited to get a representative sample
There’s an article by Donohue, Evaluation of soil nutrient variability for development of turfgrass soil test sampling methods, that describes results from sampling a 2,000 m2 (21,520 ft2) lawn at the Virginia Tech Turfgrass Center Laboratory.
 showing Mehlich 1 phosphorus (P) test results; the units are mg P/dm3 (mg/L).](/media/donohue_2002_fig3_hu779d098045f079840291eec307d8a490_140660_f99b85afac3db748192e0c63d7bc0151.webp)
221 possible sampling locations were identified, on a 3 m (9.8 foot) grid pattern across this area. I’ll now give some quotes directly from the article:
“To determine the optimum number of subsamples required for a representative composite sample for the turf area, the overall landscape was first evaluated to identify unusual spots that one should avoid when sampling. This evaluation was made after the grid was marked off, but before samples were actually collected.”
Donohue then excluded some areas near a road that may be affected by salt application, areas near a flower bed that may have had fertilizer overthrow, and steeply sloped areas and low areas.
“This left a total of 163 spots for the subsampling study.”
“Soil test results from the 163 locations within the uniform area were entered into the Virginia Tech mainframe computer for evaluation. For determining the optimum number of subsamples for a representative composite sample, 1, 5 10, 15 and 20 subsamples/composite sample were chosen for evaluation. For each subsample number, five separate composite samples were ‘collected’ by the computer and compared to determine variation in test results.”
In short, random samples were drawn from the 163 soil tests (for n = 1, 5, 10, 15, and 20), then these random samples were averaged as if they were a composite sample, and the means of those random samples and their variation were compared to the known mean of all 163 soil tests.
And here’s the final result.
“Within the uniform area selected in the study, the number of subsamples required for a representative composite sample varied from five subsamples for soil pH to 20 or more for P. Based on the results of this study, 20 subsamples per composite is suggested for a representative sample from a 1,000—2,000 m2 turfgrass area.”