Soil Surveys and Precision Agriculture - 2

So after Friday, your question must be how to get accurate boundary lines if we want to use soils as a basis of management zones.  First let me tell how not to.  Drawing circles around grid points won't cut it at all unless your grids are around 75 feet.  That is cost prohibitive. 

How do I do it?  First I look at the field.  I ride the boundary and then transect it as needed on my ATV just to see where the lines should be.  I have found it useful to spot where I think the boundary should be from a distance and then ride to the point and put a flag.  I use that method on subtle landscapes.  On "flat" ground, I try to separate highs from lows.  On sloping ground, I separate ridges from sideslopes.  Sometimes I use a probe to see if soils in a tentative management zone are similar enough to group them.  I separate by color and texture mostly when I am probing.  After I define the major zones in my mind, then I use my ATV mounted GPS to map them.  On areas bigger than 10 or 11 acres, a straight line is used to subdivide so that most areas are 10 acres are less in size.  My smallest zones are around an acre, but most are closer to 10 acres.  That is just a practical thing.  If there is  a way to use remote sensing to subdivide and maybe group smaller zones, that is ok. 

The next test is actually sampling.  Each core is examined to make sure it is similar to other cores in the sample.  I toss anomalies.  If opportunity for another subdivision shows up then I add it.  Sometimes when I am seeing some difference, but it is subtle, I will use the soil survey boundary if it looks ok.

Kelly Robertson gives excellent details on zone sampling in his blog. 

Soil zones should also be compared to yield zones to determine their usefulness.  When using yield maps, multiyear data is most useful.  Single year data may or may not be significant.  I am sending you to Kelly again to see what goes wrong with zones and grids.   His comments about what goes wrong with yield maps is excellent.  Calibration is the key.  You don't want field level accuracy with yield maps.