Error is the difference between the measured value and the ‘true
value’ (NPL, 1999). Errors can come from the measuring device
itself, including bias, changes due to wear, instrument drift, electrical noise
and device resolution. Other errors can
be introduced by difficulties in performing the measurement and by operator
skill. To avoid sampling error, sufficiently
dense measurements in space and time should take place to make sure that full
variability is captured e.g. diurnal cycles, variations across a site.
Errors can be random or systematic (NPL, 1999). With random errors, each measurement gives a
different result, so the more measurements (of the same thing) the better the
estimate and the more certain the measurement becomes. Systematic errors arise from a bias, e.g., a
stretched tape measure, and more measurements do not produce a better estimate
of the ‘true value’.