The problem here is twofold. First,
the chemical composition of a canopy is variable
within a given volume. Using a sample to represent
the whole canopy requires the imaginative
use of three-dimensional spatial statistics, giving
the most weight to the uppermost layers in the
canopy, rather than simple random sampling. The
second problem is that the concentration of a
chemical measured in the laboratory may differ
significantly from its concentration in the field.
There are three reasons for this.