Illustration of the grouping routine. Black points indicate centroid (μ) and FWHM (Θ) values of Gaussian components from the best fit solutions of unflagged neighbouring spectra. Blue shaded areas indicate the results of the first grouping, in which data points are only separated according to their μ values. Red shaded areas mark the results of the second grouping in which data points are additionally separated according to their Θ values. Blue squares and red stars indicate the initial guesses for the refitting with the first and second grouping approach, respectively.