astroinformatics.de

If you believe a single bright star is too simple to take a photograph, you probably failed. 

Introduction 

P Cygni is an LBV (luminous blue variable) star, a giant star located in the large asscociation Cygnus OB 1 (Van Schewick, H., 1968). P Cygni is bright enough to be visible for the naked eye in a dark and clear night. However, it is not easy to obtain pretty pictures of the famous star and its surrounding interstellar matter. If one wishes to obtain a picture of an extremely faint H II region, then the star formation Cygnus and the region around P Cygni probably is a good choice for a challenge and test of the equipment. The interstellar matter consists mainly of hydrogen gas of low density. The enormous luminosity of P Cygni hardens pretty pictures of the interstellar matter, and the star itself at the same time. Few professional publication records containing informations about the P Cygni region sounded like an adventure (Meaburn, J. et al., 1999). Some details have been evaluated after expensive image post-processing.

P Cygni currently is suspected to show variability in some spectroscopic and photometric details. While it is interesting to find out the reasons for the variability, an international observing campaign started with the organizations AAVSO, ASP and BAV (Pollmann & Bauer, 2011)

Observations and comparison to the POSS plates 

P Cygni is a giant star with a gas shell. It will be interesting to see, whether and how this shell can be made visible and observed over a certain period. I started my first attempt to get an idea about reasonable exposure time for astrophotography of the P Cygni region.

Figure 1: Image of P Cygni (center) and the surrounding H II Regions. Details about exposure are presented  at the end of this article. Click to enlarge. 

The image of P Cygni is practically saturated if an exposure time of more than a fraction of a second is used. Therefore, seems impossible to image both, the hydrogen gas and the star with broadband filters. With a narrow-band H-alpha filter the light of the star is reduced and the faint interstellar matter may be amplified with large exposure times. Again, the star would be saturated with too long exposure. Therefore, exposure time was limited to 30 seconds for each frame. To obtain a good limiting magnitude in the bright moon light, many exposures have been taken and stored. Finally the exposures were added together using a modified shift-and-add procedure. The final image now consists of 50 images.

Figure 2: Same region as shown in figure 1, but taken from a low resolution scan of a Kodak 103a-E plate of the Palomar Observatory Sky Survey (DSS2). Image obtained with the software Aladin of the Centre de Données astronomique des Strasbourg. While the plates show more details with contrast enhancement, this is the default view as shown by Aladin.

Noise and Limiting Magnitude

The multiplication of 50x30 seconds yields a total exposure time of 25 minutes. However, the effective exposure time regarding the limiting magnitude is less. This is based on a computation of the lowest signal in the image, which is dominated by the camera readout noise. A careful analysis of the resulting image yielded a standard deviation of 18 DU (digital units) for the total noise in the star-less areas of the resulting image. Assuming the quantum conversion rate for the Canon, 1 DU is equivalent to 0.84 e-at ISO 400 (C. Buil, unkown date), or in a first order approximation one photo electron per digital unit. You might think this Canon EOS produces lots of detector noise. However, this is not true. While the camera itself has a dynamic range of 14 bit, the resulting image is computed at double floating-point precision and streched to 16 bit (intensities multiplied by four). So 18 DU mean an average noise of 4.5 DU with the original 14 bit dynamic range of the camera. Of course, this is less than the readout and dark current noise of one single raw image and a result of an average process of 50 frames. The single raw frames have a standard deviation which is 7 times larger, which is close to the detector noise level evaluated earlier (Bauer, 2008). It comes out, that the faint illumination of the H II region is less than the amplitude of noise with the collected amount of images. The mean value of the sky background has a subtle offset of 4 DU (after dark frame subtraction) in the red image plane and the maximum illumination of the faint hydrogen is estimated to a value of 10 DU. Therefore, signal-to-noise ratio of these portions of the image is less than 1! The average number of photo electrons detected within every raw frame seems to have a value of 2.5 e-/pix in the brightest areas detected, which is the average photo flux detected flux over 30 seconds. This is not the common case of a  limiting magnitude, it is more likely the case of marginal detection. Because the intensitiy distribution is extended over a large area, the faint illumination is visible in both the Palomar plates and my own image. The expected amount of frames to obtain a signal-to-noise ratio of at least 1 is estimated to be larger than 200 frames. Exposure time therefore must be set to a value of more than 100 minutes (or 6000 s).

From earlier studies I detected a limiting magnitude of Vmag=20 with exposures at around one to two hours. This is given by a signal-to-noise ratio of the peak intensity of point light sources against the mean value of the noise (values around 50 DU to 300 DU of standard deviation). In this case the limiting magnitude is contrasted against the bright sky illumination of a suburban observatory site. In the case of my P Cygni picture, and with H-alpha filtering, most of the light pollution is cut by the filter. The photon noise (Poisson distribution) is much lower. Therefore, the intensity of the illumination may be estimated to a value of less than Vmag=20. However,  limiting magnitudes in a narrow-band filter are difficult to compare. In this case, the peak intensity of the H II region is set in contrast to observations of star clusters in distant galaxies taken with broadband filters. This limiting magnitude is not related to the stars detectable in the image, because their apparent magnitude is cut and reduced by the narrow-band H-alpha filter. Taking a limiting magnitude in this case is neither a photographic, nor a visual magnitude. And the faint glow of any H-alpha emission remains invisible to the naked eye. The limiting magnitude estimated depends mainly on consideratiopn of the signal-to-noise ratio of intensities and is compared to a different signal found in a different photograph with another filter and standard deviation of the noise.

Erroneous measure of brightness

It shall be mentioned, that some authors calibrate the magnitude of an H II region or nebula to the brightness of a star found within the same photograph. However, this is a very bad idea. This procedure will not reflect the true apparent magnitude of the nebula, e.g. compared to the visible spectral range. If one wishes to evaluate distant H II regions in distant galaxies like Messier 51, these H II regions are sometimes given with a brightness of around 12 mag, while they are hardly visible in a photograph with a limiting magnitude of Vmag=20. So this procedure will not give expected results anyway. A difference of 8 magnitudes between the detection limit of a star and the H II region represents an error of a factor 500 of the true value! This happens, because the brightness of the emission nebula integrated over the spectral range of the filter remains the same in H-alpha, as in the wider spectral range of a broadband filter. Any contrast obtained and therefore the value of the limiting magnitude, however, is worse with a broadband filter due to a larger standard deviation of the sky background noise. On the other hand, use of a narrow-band interference filter, like an H-alpha filter, means every single star appears much darker, than it appears with a broadband V (visual) filter. Further reasons for any difference is given, when brightness is expressed as an absolute magnitude or a surface brightness. This is why some magnitudes will not match expectations of an astrophotographer. The method presented in the preceeding section will compare magnitudes based on the intensity found in a pixel, or in a small and comparable portion of the image (e.g. 2x2 pixel). Therefore, it will resemble the total flux of an assumed point light source related to certain signal-to-noise constraints.

Conclusion

The interstellar matter around P Cygni is recovered within an image taken with a DSLR. The very faint H II region is demonstrated and even dark clouds are recovered from the observed signal. Structures identified are found below the limit of marginal detection. Despite of marginal detection, it is an amazing result for a digital single reflex camera with a 20 cm class telescope and short exposure time. It seems impossible to obtain similar results from the raw frames without further expensive image processing. This reflects well the difficulties found even with larger exposure time of 7000s at a larger telescope and using CCD detectors reported in the literature (O'Connor et al., 1998). To achieve the result presented in this work, the author developed a new process chain for a DSLR camera and a new noise model resulting in a new image calibration process including a dedicated noise filter to recover the signal. The discovery of the faint H II region demonstrated a successful application of the new technique. The shell of P Cygni, however, is not demonstrated with this technique. The Giant Lobe close to P Cygni and discussed in the literature (O'Connor et al., 1998 and Meaburn, J. et al., 1999) is found below the detection limit of my own photograph. The shell around P Cygni is hard to detect due to the enormous brightness of the star itself. The gas shell is believed to lie within a small portion of the image located around the star and therefore needs high-resolution imaging techniques to be resolved.

Literature

Bauer, T., 2008. "The RAW Image File Format Problem - Applications of Digital SLR Cameras in Astronomy and Science", Proceedings of the 4th Annual Meeting on Information Technology & Computer Science at the BA-University of Cooperative Education, H. Weghorn (Ed.), Stuttgart, ISSN 1614-2519

Buil. Ch.: "Comparaison des Canon 40D, 50D, 5D et 5D Mark II", last visited on 1. February 2011, -> goto website 

Meaburn, J. et al., 1999. "The Kinematical Association of a Giant Lobe with the Luminous Blue Variable Star P Cygni". Astrophysical Journal, vol. 516, issue 1, pp. L29-L32

O’Connor, J. A. et al., 1998. The Manchester occulting mask imager (MOMI): first results on the
environment of P Cygni. Monthly Notices of the Royal Astronomical Society, Volume 300, Issue 2, pp. 411-416

Pollmann E., Bauer, T., 2011. International Observing Campaing: Photometry and Spectroscopy of P Cygni, The American Association of Variable Star Observers, AAVSO Newsletter January 2011, p.12 

Van Schewick, H., 1968. "Eigenbewegung und absolute Helligkeit des Hüllensterns P Cygni". Zeitschr. für Astrophysik, Vol. 68, p.229

"The Aladin Sky Atlas", Centre de Données astronomique des Strasbourg, last visited on 1. February 2011, -> goto website of Aladin 7

Observational data 

 

Telescope: Vixen VC200L, focal reducer f/6.4, Sphinx SXD
Camera: Canon EOS 40D-a, ISO 400, Astronomik H-alpha 12nm filter
Exposure: 50 x 30 s, manually selected
Calibration: Dark (50 images), sky flatfield (50 images)
Image Processing: Shift & add with correction of subpixel movement, improved noise reduction
Date of exposure: 10 October 2010, 19:50 h MEZ
Software: ArgusPro SE, some post-processing with Photoshop CS3
Remark:

The H II regions are hardly visible at the limit at Vmag<20.