Handheld Geographic Information Systems

.
the leader in
Handheld Geographic Information Systems^tm
....

GPS Mapping

Soil Sampling

Instrument Recording

Download Software

Training Centers

Contact Us

Home Page

Smart Soil Sampling

By using modern soil sampling techniques; we can improve the quality and reduce the cost of soil sampling. HGIS can create Grids of Squares, Hexagons, or Points on any orientation.

Proper Grid Sampling
Proper grid sampling is an art as well as a science. Look at any smoothed soil sample map. If you can recognize the sample points - the data has a problem. You should be able to recognize the indigenous features in the map. We generally do not record enough samples to develop a mathematical solution. This is for a number of reasons:

the number of samples required to prove the data is high,
high variability within a locality (T-tests can fail), and
the time and cost.
Indigenous knowledge

Pay careful attention to indigenous factors - such as soil type, topology, and especially history. Use this indigenous knowledge to determine areas of more and less interest. Pay more attention where it is most interesting or important – a hill that is low yielding - to see what's wrong. A large flat area may often be treated as a single feature where there is minimal variation. Determine how many features (or management zones ) need to be specified. A mathematical solution of the entire field is not necessary.

Proper Sample Technique
To take a proper soil sample, the sample needs to be representative of all of the soil in the grid, not just the location at the center of the grid cell. Each sample should be made up of sub-samples.

First, use the GPS to locate the approximate center of the target cell. Second, go a few meters in several directions from the center to obtain partial samples (for example, off the four corners of the truck). Mix the samples together in one bag to represent the overall sample for that grid cell. Therefore, one grid sample may be made up of 4 or more actual samples. This adds some time to the collection process, but does not increase the lab analysis cost. And it greatly increases the accuracy of the data.

Smart Sampling
If you treat soil sampling as simply a mathematical problem, you will spend too much money. Smaller grid cells are appropriate for small features, large cells for larger features. For small features, each grid cell should be several times smaller than the feature so that we get at a minimum of three samples per feature. Medium features may be covered by 5-10 grid cells. More cells increase the confidence. Large features never need more than 30 cells. It may be better to skip grid cells that fall in a transition area; excellent results can be obtained by analysis software such as Vertical Mapper for MapInfo or Surfer which interpolate across empty spaces.

Analysis
Don't blindly believe all the lab sample results. Overlay the sample results on top of the feature map. See how they fit. Perhaps some samples need to be discarded. Occasionally the sample was taken in a small area which actually contains soil from a nearby feature. Perhaps other points should be copied or moved on the map to better fit with the feature map.

Some believe that the average/median of samples in a region (or management zonet) should be assigned to the whole region. It is this author's belief that when faulty samples are relocated or eliminated, excellent results can be obtained by a smoothing application. This is because there may also be errors in the zones's exact boundary. In either case, results are improved when non-representative samples are removed.

Summary
By using the Partial Sampling, Indigenous Knowledge, and proper analysis; we can improve the quality of soil sampling. We can also accomplish sampling at a reasonable cost. Using samples within the same feature to confirm each other eliminates bad samples.

StarPal HGIS tm
StarPal HGIS can create a target over any shape region at any target size. Create a point at each sample location (map below). Auto_ID will generate a unique bag ID. Attribute fields may also be filled in with field observations. The result can be exported to compatible applications for analysis.