A unified approach for reporting ARM measurement uncertainty

Poster PDF

Description

Uncertainty is often conceptualized to express one of two distinct parameters: either the estimate of the error of a measurement (accuracy) or the dispersion of the values that could reasonably be attributed to the measured variable (precision). The first uncertainty concept (accuracy) focuses on unknowable quantities, total error and true value, because knowing these quantities is impossible in an operational context. Therefore, the second uncertainty concept (precision) is becoming the current practice in metrology. The ultimate goal is to express all measurement uncertainties in accordance with the Guide for Uncertainty Measurements (GUM, Joint Committee for Guides in Metrology 2008), since it provides an internationally recommended protocol.

We tested these concepts by building an comprehensive inventory with current ARM uncertainty estimates, using inputs from each ARM instrument mentor, for each ARM instrument, and for the main ARM measured variables. Then we classified each uncertainty estimate either as instrument resolution, precision, accuracy, or non-standard estimate. For those cases where the uncertainty was reported as accuracy (i.e., calibration procedures), we further classified the calibration reference as a traceable standard, consensus procedure, or expert judgment. These methods provided a unified approach for reporting measurements uncertainties for all ARM instruments.

The results indicate the existence of measurement families that have common attributes of uncertainty type. Almost all of the ARM measurements have reported accuracy uncertainty. However, many of the measurements to date have not yet been characterized in terms of precision uncertainty. This will need to be done in order to match the higher metrology standards of the expression of uncertainty, which are consistent with GUM.