INTERPOLATION ERROR ON DIGITALLY RECORDED SPECTRA

Author Institution:Interpolation is often used to obtain the wavelengths or wavenumbers of spectral lines. Estimation of error due to interpolation is important in the operation and improvement of the sampling process. When the spectra are recorded sequentially by changing an experimental parameter (i.e. angle or voltage), the increment of this experimental parameter always has some uncertainty. In a simple case when two reference lines are located at data points 0 and n and the error of each increment is random and Gaussian with dispersion $\sigma^{2}$, the average dispersion for the interpolation error is calculated to be $\frac{n\sigma^{2}}{6}$. Derivation and application of this formula will be presented