Systematic uncertainty in the silicate strength

It is important to keep in mind that the main uncertainty in the silicate strength may not be the statistical uncertainty, which may be as low as 0.01 in silicate strength for a spectrum with very high S/N. Instead the dominant uncertainty may be systematic in nature, associated with the choice of interpolation method (spline or powerlaw continuum), or from the need to invoke deblendIRS.
To quantify the systematic uncertainty we have plotted the difference of the silicate strength solutions from both methods in the figure below. The difference is close to zero wherever the color is yellow. Where it tends to red the silicate strength from the spline method exceeds the one from the powerlaw method. Where the color tends to green, blue or purple the silicate strength from the powerlaw method is larger.
The power law method is prefered for spectra dominated by PAH emission, which are characterized by S_sil_powerlaw > -2 and EQW(PAH11)>0.1µm. See the box defined by the dashed lines.

To explore the systematic uncertainty in the silicate strength we compare the silicate strength solutions for both methods at the boundary of the two validity ranges.

Galaxies with power law based silicate strengths between -1.8 and -2.2 and EQW(PAH11)>0.1µm show a median systematic difference between spline and powerlaw based silicate strengths of -0.21. The 1-sigma dispersion of the measurements is 0.19. The large majority of measurements agree to within their uncertainty with the median silicate strength difference of -0.21.


Galaxies with EQW(PAH11) between 0.1/1.2µm and 0.1x1.2µm and with a power law silicate strength > -2 show a median systematic difference between the spline and power law based silicate strengths of zero, with a 1-sigma dispersion of 0.18.