I am taking soil samples on an agricultural plot and will be using the results as input to create a semivariogram for kriging in ArcGIS. This plot features raised planting beds in which the test crop is planted. Ideally I am looking to take soil samples using a stratified random approach; with an equal number of samples collected from different management zones within the plot.
I am wondering whether taking samples from the planting beds only will result in valid output from the kriging process. Clearly, using this method, the sample points will be biased in their distribution; linearly, along the planting beds.
Will taking samples from the planting beds only result in a valid interpolation surface?