The second RECTA / CAHA / DSA collaboration is a portrait of the Messier 57 nebula in Lyra.

These data were acquired mainly through the 1.23 meter Carl Zeiss telescope. It is a rather long exposure of the nebula taken during several nights from July to September:

- 18 hours of optical narrowband data in the H-alpha and OIII lines. This data was acquired entirely through the 1.23 meter telescope, and it gives an extreme deepness to the nebula’s halo. The H-alpha filter has a bandpass widht of 5 nm (it transmits part of the NII lines), while the OIII one has a bandwidht of 9 nm. Subexposures were all 20 minutes long.

- 3 hours of broadband data with standard photometric Johnson BVR filters. These data were acquired through the 3.5 meter telescope. It gives the required deepness to the background field, as well as the high resolution required to show all the detail in the halo and inner disk. Subexposures were 2 minutes in R, 3 minutes in V and 2 minutes in B. All the data were acquired in a white night.

- 45 minutes of infrared narrowband data, acquired through the 3.5 meter telescope and the Omega-2000 IR camera. The infrared image was taken through a filter centered in the molecular hidrogen emission at 2.12 microns. All subexposures were rather short (45 sec), as is usual in IR ground based observations.

The whole data set was calibrated manually with the ImageContainer and PixelMath tools.. The PixelMath formula is rather simple:

($T – bias)*Avg(flat)/flat

Where:

$T is the target image being calibrated,
bias is the master bias frame,
flat is the master, bias substracted, flat frame.

The image registration and integration was done automatically with the StarAlignment and
ImageIntegration tools, respectively.

The nebula, as seen through each filter, can be seen in the image below:

To combine the optical narrowband with the broadband one, we introduced the different emissions in the BVR images. The R filter transmits only the H-alpha line. But the B and V filter transmit in different proportions the OIII and H-beta emission lines. So we introduced the OIII and H-alpha images in both filters with varying weights.

Finally, the infrared image was processed separately and introduced in the processed optical image, as both images represent very different structures of the nebula.

This 6-channel mixing gave us the resulting emission line tones:

- The OIII emission has a teal tone.
- The combined H-alpha and H-beta gives a nearly violet emission. This gives us a good differentiation between optical hidrogen emission and the infrared one.
- Molecular hidrogen IR emission is represented as pure red.

All these operations were carried out with the PixelMath tool.

Color calibration was performed with the ColorCalibration tool, taking the whole light of the objects in the image as a white reference.

This image has two decisive steps: the deconvolution of the higher signal areas, and the dynamic range compression.

Deconvolution was done with the regularized Richardson-Lucy algorithm, and it was done only in the inner disk, where we have enough signal to noise ratio. The contrast between the ring and the halo of the nebula is very steep, so the mask to select the areas with enough signal is amlost a binary mask including only the ring and the central hole. This can be better understood if we make a 3D plot of the below linear H-alpha image:

The 3D plot (made with the 3DPlot javascript script for PI, ver 1.3, by Andrés Pozo, David Serrano y Juan Conejero) shows that even the hole at the center of the main disk of the nebula has a very steep illumination decrease respect to the ring:

To make a deconvolution to the outer halo (barely visible in the graphs due to its extreme faintness) would be impossible. Even if you deconvolve these areas with noise regularization, the wavelet-based algorithm would not detect any structure to deconvolve, as there isn’t enough signal strenght.

It’s interesting to compare our red image with the one taken by Hubble Space Telescope ten years ago:

Sorry, but the HST red image is heavily clipped. This side by side comparison shows the extremely good performance of this deconvolution algorithm, as well as the movement of the nebula structures during this 10 year period.

The inner nebula disk is extremely bright, compared to the outer halo. For that reason the dynamic range compression procedure is of crucial importance. Below we can see the stretched image prior to any dynamic range compression:

The dynamic range compression and processing was carried out with the HDR-Wavelet algorithm, as well as with the techniques described in the NGC 7331 processing notes.

A 3D plot of the above stretched image shows clearly the illusion we create when we stretch the data:

It seems that the center hole of the nebula has near the illumination level of the ring, and seems that there’s only a tiny part of the real contrast between the dist and the outer halo.

Please be sure to read the FAQs about this picture.