Evaluating the efficacy of a data cube treatment procedure for kinematic analyses: application to NGC 3115 and NGC 4699

TL;DR

The paper tackles how instrumental effects and high-frequency noise in data cubes can bias kinematic analyses. It applies a previously proposed treatment pipeline—differential atmospheric refraction correction, Butterworth spatial filtering, fingerprint removal, and Richardson–Lucy deconvolution—to GMOS/IFU data of NGC 3115 and NGC 4699, comparing three cube variants. Using pPXF to extract $V_*$ and $ obreak \sigma_*$ and Jeans Anisotropic Modelling with MGE mass profiles, it demonstrates that all cube variants yield compatible $M_{BH}$, $M/L_I$, and $eta$ within $1σ$, while the uncertainties are significantly reduced by the processing, with total improvements up to ~41–53% depending on the parameter and galaxy. The results validate the pipeline as a robust means to improve kinematic inferences from data cubes and provide insights into SMBH demographics, including a potential SMBH binary in NGC 3115.

Abstract

Data cubes have been increasingly used in astronomy. These data sets, however, are usually affected by instrumental effects and high-frequency noise. In this work, we evaluate the efficacy of a data cube treatment methodology, previously proposed by our research group, for analyses focused on the stellar and gas kinematics. To do that, we used data cubes of the central regions of the galaxies NGC 3115 and NGC 4699, obtained with the Integral Field Unit of the Gemini Multi-Object Spectrograph. For each galaxy, we analysed three data cubes: non-treated, filtered (with the Butterworth spatial filtering) and filtered and deconvolved (with the Richardson-Lucy deconvolution). For each data cube, we performed a dynamical modelling, using Jeans Anisotropic Models, to obtain, among other parameters, the masses of the central supermassive black holes. Both for NGC 3115 and NGC 4699, the values of the parameters provided by the dynamical modelling from the non-treated, filtered and filtered and deconvolved data cubes were compatible, at the 1-$σ$ level. However, the use of the Butterworth spatial filtering decreased the uncertainty of the parameters. The additional use of the Richardson-Lucy deconvolution decreased even more the uncertainty of the parameters. The complete data treatment procedure resulted in decreases of 41% and 45% in the uncertainties of the supermassive black hole masses in NGC 3115 and NGC 4699, respectively. These results indicate that our treatment procedure not only does not compromise analyses of data cubes focused on the stellar or gas kinematics, but actually improves the quality of the results.

Figures (10)

• Figure 1: Left: HST image, in the F814W filter, of each galaxy analysed in this work, with the GMOS FOV marked as a rectangle. Centre: image of the treated data cube of each galaxy, collapsed along the spectral axis. Right: average spectrum of the treated data cube of each galaxy.
• Figure 2: $V_{rms}$ maps (left) of the non-treated, filtered and deconvolved data cubes of NGC 3115, obtained with the pPXF technique, together with the simulated $V_{rms}$ maps (centre) corresponding to the best obtained models and the curves (right) extracted along the kinematic axis of the maps (shown as a black line). The position of the central SMBH (coincident with the stellar nucleus) is marked with a square. The elliptical black curves correspond to the isocontours of the data cubes, integrated along the spectral axis. All $V_{rms}$ values are in km s$^{-1}$.
• Figure 3: The same as in Fig. \ref{['fig2']}, but for the $V_*$ maps of NGC 3115.
• Figure 4: The same as in Fig. \ref{['fig2']}, but for the $V_{rms}$ maps of NGC 4699.
• Figure 5: The same as in Fig. \ref{['fig2']}, but for the $V_*$ maps of NGC 4699.