Attached is a LCG showing two nice curves for V838 Mon, one visual, the other CCD. Both show good consistency in the overall trend (slightly downward over time) but offset by about a full magnitude!
I do realize that for objects such as ths one, extremely colored with emission lines, it can cause skewed results in different passbands.
Nevertheless, a full magnitude difference in two passbands which are supposed to be fairly close to each other by design is troubling.
Which should be considered more correct for this object, the visual or the V. Mainly in respect to reporting in general how bright this object is in this general passband ?
I am pretty new to the CCD photometry but have also done some visual readings in the past so I know something of the foiables of both. CCD is more precise, but it is also offers more opportunity for error and thus can easily be precisely wrong.
Here it is hard to imagine the visual readings being off by a magnitude, very easy for that to happen with the CCD.
If the photometry readings were mine, in light of the visual readings, I would go back and check carefully that I did not miss ID my comp star or enter the wrong bandwidth magnitude from the intended comp star - have been there already. One could imagine other issues. By comparison the visual process, as you know, is pretty bullet proof, if less capable of precison, once you get the hang of it.
I noticed some readings on a star last week where more than once, inside a week period, there were two photometric time series plainly overlaping in time but quite divergent in results. Having had to pull erronious results down once, I am now paranoid about doing it again. I am telling myself I will follow a ridgid QC check list going forward.
You raised this same question last March, when we still had the discussion group mail-list. I mentioned then that the reason for this discrepancy is due to the extreme red color of this old nova. Just to be safe, I looked at my recent (2013-02-19) data using the 61cm Mt. John telescope, where I get an untransformed V=14.41 for this star. The difference between my measure and the other CCD V-band measures is probably due to the inclusion by the other observer of one of the nearby faint companions, while I did psf-fitting.
A one-magnitude difference for this star between visual and CCD is just about right. The human eye does not replicate Johnson V for extreme objects.
On the other hand, the far more important issue is that V838 Mon is now nearly a magnitude brighter than at quiescence. It deserves more attention by the observers. If you want to transform your data, use B&V and not V&I.
Most of the recent CCD V data in the LCG is mine (DKS) since it's a regular target on my program - I try to get a measurement every few nights just in case something exciting happens again. I'm using a pretty mainstream setup: SBIG ST9 and Custom Scientific's newer interference-type V filter, and my transformation coeffs are quite small. However, since this object is so very red I purposefully do not report transformed values. I think it would be more misleading to apply a transform that's not really valid at this extreme B-V value, and that it would be better to report the raw v magnitude.
Arne: you're correct that I am doing aperture, not PSF photometry. I use a fairly small 16" diameter aperture that excludes all of the obvious nearby stars but on the DSS image there are a few very close faint stars that are probably contributing to my measurement. The difference between your recent measurement and mine could also be due in part to using different comp stars - I'm using the 11.956 (120) star as my single comp.
This light curve comprises 1 visual observer's work and 1 ccd observer's results. I would really like to see 2-3 more observers of each. This would make comments much more informed.
As Arne reminded me,yes we discussed this before. Using his V=14.4 and my average v=15.2, this gives from Stanton's formula B-V = (15.2-14.4)/0.21 = +3.8 which seems about right for an "extremely red old nova". So this may account for most of the difference. And DKS, whose CCD data is in the plot, admits some close companion contamination may account for a few tenths too.
I guess if it was a bit brighter it would looked colored like those ghostly carbon stars!
Like to be able to assist here, but the cloud cover here has been dreadful at best. I might get a peek at Sirius or Rigel through the thick low-lying cloud cover. If the situation resolves itself, I'll jump right on it for you all, and see what difference I can see from my vantage point.
This is very interesting object, both scientifically as well as observationally. Having multiple people observe it, especially over several months so that there is considerable overlap, could be a good example of how extreme objects are seen differently by different people. Everyone is "right"; just that no two people will get the same result. Don't try to force your estimates to correspond with others.
Likewise, even CCD observers will not get the same result, even if they transform their observations. The color is outside of the normal "fit" to get coefficients, and filter wing response variations will pick up different features that don't transform. Again, a good example of how to do everything right and still have no agreeement.
These kinds of targets are good to examine in the "mean", combining all results to get a kind of master estimate; or else to examine each observer independently. Getting spectra could be useful as they are less affected by the redness of the object.
This is very interesting object, both scientifically as well as observationally. Having multiple people observe it, especially over several months so that there is considerable overlap, could be a good example of how extreme objects are seen differently by different people.
Yes, as this attached LCG plot shows over the past 10 years, selecting all passbands, the object can appear about 11 magnitudes differently, from "naked eye" in H-band to 17th magnitude in B!
It finally cleared here at Maria Mitchell Obs, after nearly 3 weeks of clouds, rain, snow and 60 mph winds. Had a chance to observe V838 Mon. Got it between 14.19 and 14.23 in 16 integrations of 2 minutes each with th 24 inch. Error bars of .015 in each frame. Used the 123 star (B-V = 1.15) as a reference in Maxim DL.
Results posted to AID.
Hello Kiyota san
Good to meet you. Do I need an observer code to access this data base? I would be very interested in seeing the V838 Mon data.
No, you don't need any code/passwd to access these data. Just type: "V838 Mon" in the Object name field. I've check it yesterday.
Robert Fidrich (FRF)
Enough of the off-the-wall arm-waving about the 1 mag. discrepancy. Show me the variable's spectral irradiance distribution, or a derived color temperature, and tell me the telescope aperture used by LMK and any other current visual observers, and I will predict the magnitude difference to ~0.1 mag. Tentatively, it is m(vis)-V(CCD)=+1.0, disregarding possible contamination of the CCD value by close field stars.