resolution.uncertainty.pct: Compute the resolution uncertainty (last digit influence) -...

In claus-e-andersen/clanTools: R functions of general use for data analysis and presentation

Description

This function compute the relative standard uncertainty (k=1) for instruments with a given number of digits (say, 4) that is always used to their fullest extend (i.e. a reading of 1 would actually be 1.000 and a reading of 1.1E-9 would actually be 1.100 x 1E-9).

As the function gives the relative uncertainty, it cannot handle a reading which is zero (the function returns NA)..

We compute the standard uncertainty from resolution limitations as the last digit divided by the squareroot of 12 (i.e. u = 0.29 * step). This is consistent with GUM F.2.2.1 and the assumption of a uniform probability distribution. So, with four digits, a reading starting with 1 (like 1.002 or -1.192e-88) will lead to an estimated resolution uncertainty of about 0.029%. Note that we assume that the instrument always provides the given number of digits.

See also: resolution.uncertainty.abs

#' Example of a 0.029% resolution uncertainty:

resolution.uncertainty.pct(1004,digits=4,last.digit.step=1)

resolution.uncertainty.pct(1.1e-20,digits=4,last.digit.step=1)

resolution.uncertainty.pct(1.1e20,digits=4,last.digit.step=1)

Note that the function is vectorized, such that calls like the following is possible:

resolution.uncertainty.pct(c(1.2, 1.3, 0, 0, NA, NA, 2), digits=4,last.digit.step=1)

Usage

resolution.uncertainty.pct(1004,digits=4,last.digit.step=1)

Arguments

x	raw reading (can be vectorized).
digits	number of digits on the instrument display.
last.digit.step	resolution of last digit (e.g. 1 or 5).
min.value	do not report an uncertainty below this value (e.g. 0.001 %).
max.value	do not report an uncertainty above this value (e.g. 50 %).

Value

numeric value (pct uncertainty). May be vectorixed.

Author(s)

Claus E. Andersen

claus-e-andersen/clanTools documentation built on Oct. 23, 2020, 7:59 a.m.