Variograms to guide spatial sampling for kriging

Detailed information on soil to manage polluted or agricultural sites is often prohibitively expensive to obtain. To sample adequately, the approximate scale of spatial variation needs to be known. If soil data are available, variograms can be computed and used to determine the kriging errors for several grid intervals and an interval selected to meet a specific tolerable error. In the absence of prior knowledge, if soil properties appear related to ancillary data such as aerial photographs, elevation or apparent electrical conductivity (ECa), individual or multivariate variograms of such data may indicate the scale of variation in the soil. If the scale of variation indicates too few data to compute a reliable variogram conventionally, it can be stimated by residual maximum likelihood or a standardized variogram from ancillary data can be used.