[Aavso-photometry] Re: [AAVSO-DIS] Errors and error reporting for
CCD observations
Wolfgang Renz
w_renz at onlinehome.de
Sat Jul 31 15:43:06 EDT 2004
Hello Lew
If you mean with "interstellar reddening" the first few miles, then you are right as the (B-V) is not corrected for the actual _interstellar_ reddening. There exists an other symbol "E(B-V)" thats used to represent the color shift due to interstellar reddening. The interstellar reddening has just to be obyed when you want to convert spectral types + luminosity class to (B-V).
The only effect of interstellar reddening on the determination of the instrumental coefficent might be when there are just very red stars available. But you can ckeck that by looking at the graphs and the standard deviations / correlation of the determination of your instrumental correction coefficents.
> CCD mag = V + Tc*(B-V) + offset
The Tc is usually defined as an instrumental correction coefficent for photometric nights at high celestrial altitudes (>> 30°). Its no more and no less.
> I have begun a preliminary study if my G16 camera and found
> that different fields give different transformation
> coefficients, Tc, for the same camera and telescope
> combination.
The different fields were probably not imaged at the same altitude and especially under the same atmospheric conditions. So it might be no wonder that you'll get also different instrumental correction coefficent. Using too few stars for the evaluation might also be a possible cause.
As the atmospheric extinction is dependent on the celestrial altitude and also depends on the color of the stars, we will end up with something like:
Mstd = Mccd + M0 + k * (B-V) + k' * X + k'' * X * (B-V)
Mstd: Brightness in standard band
Mccd: Brightness measured by CCD
M0: Instrumental+correction brightness offset
(B-V): Color index
X: Air mass
k: Instrumental correction coefficent
k': First order extinction coefficent
k'': Second order extinction coefficent
for a "full" transformation. Setting the two extinction coefficents to zero will of course have its effects on the instrumental correction coefficent and the offset.
An other source of error with no or single filter observations might be the fact that the (B-V) of the variable itself must be taken from literature or guessed by statistics. As the (B-V) varies for most variables (all pulsating varaibles; erruptive & symbiotic variables and CVs; but also for higher amplitude EBs) that might introduce an additional inaccuracy when the air mass / color index and the coefficients are large.
Example:
The (B-V) of CVs are usually listed as late spectral type at (B-V) ~ 1 representing the cool giant (that looses the mass) when the system is not in outburst. But when a CV is in outburst the color changes to an early spectral type with (B-V) ~ 0 as the disk and/or the surface of the (now brighter) small hot star is dominating the brightness of the system.
But what (B-V) has to be used when the system is in its brightening or fading state ?
If you don't measure it by your self, you won't know and you have to make a guess that might not be very accurate.
So no observation that was not done using at least two filters and was not fully transformed, might claim to be in the standard sytem.
In my opinion the current AAVSO system to classify CCD observations:
- CCD = Charge-coupled device (unfiltered)
- CR = Charge-coupled device (unfiltered) and reduced using R magnitude sequence
CV = Charge-coupled device (unfiltered) and reduced using V magnitude sequence
- CCDB = Charge-coupled device with (B) filter
CCDV = Charge-coupled device with (V) filter
CCDR = Charge-coupled device with (R) filter
CCDI = Charge-coupled device with (I) filter
is not sufficient to describe:
- how the obs was made:
- unfiltered (marked currently as CCD, CV or CR)
- single filter (marked currently as CCDB, CCDV, CCDR or CCDI)
- multiple filters (marked currently as CCDB, CCDV, CCDR or CCDI)
AND
- how it was reduced:
- using mags of a "unspecified" band
(marked currently as CCD)
Pure instrumental mags ? But how was the zero point set ? Mags of low accuracy ?
- using mags of the "nearest" standard band
(marked currently as CV or CR)
- using instrumental coefficient
(marked currently as CV, CR or sometimes even as CCDB, CCDV, CCDR or CCDI)
- using first order extinction coefficients
(marked currently as CCDB, CCDV, CCDR or CCDI)
- using first & second order extinction coefficients
(marked currently as CCDB, CCDV, CCDR or CCDI)
Looking at the most commonly used and/or wise combinations:
- unfiltered using the mags of a "unspecified" band and "unspecified" reduction
(marked currently as CCD)
- unfiltered using the mags of the "nearest" standard band
(marked currently as CV or CR)
- unfiltered using instrumental coefficient (see Arne's M67 project) and mags of a standard band
guessing the color index of the var
(marked currently as CV, CR or sometimes maybe even as CCDB, CCDV, CCDR or CCDI)
- single filter using mags of a standard band
(marked currently as CCDB, CCDV, CCDR or CCDI)
- single filter using first order extinction coefficient and mags of a standard band guessing
the color index of the var
(marked currently as CCDB, CCDV, CCDR or CCDI)
- multiple filters using first order extinction coefficient and mags of two standard bands measuring
the color index of the var
(marked currently as CCDB, CCDV, CCDR or CCDI)
- multiple filters using first & second order extinction coefficient and mags of two standard bands
measuring the color index of the var
(marked currently as CCDB, CCDV, CCDR or CCDI)
makes it obvious that there is a large range of accuracy within one AAVSO CCD observation classification.
The accuracy of the above "no or fixed coefficient reduction" entries might be increased significantly by using more comps with averaging instead of just a single comp + check star. Thats especially true and necessary when using low accuracy mags (e.g. visual AAVSO chart mags or faint USNO mags).
Determining coefficients dynamically (e.g. for first & second order extinction) is usually done by averaging and/or linear regression of many comps and will lead therefore also to more accurate results.
Using many comp stars has at least the effect of nearly eliminating their statistical noise on the final result. So you might still double your accuracy when all other factors are already optimized.
In my opinion there should be introduced at least one more highest-quality class for multi color observations with using one or both extinction coefficient for reduction. That might be as B, V, R, I or by adding an additional R for advanced reduction to the currently used abbreviations.
Using an additional final R might be also applied to CV/CR to motivate the observers to apply at least the instrumental corrections to their unfiltered observations.
Cear skies
Wolfgang
--
Wolfgang Renz, Karlsruhe, Germany
Rz.BAV = WRe.vsnet = RWG.AAVSO
----- Original Message -----
From: "Lew Cook" <lcoo at yahoo.com>
To: <aavso-discussion at mira.aavso.org>
Sent: Friday, July 30, 2004 7:47 PM
Subject: [AAVSO-DIS] Errors and error reporting for CCD observations
> There are many ways to skin a cat. Reported CCD
> observations contain numerous sources of errors. For
> unfiltered work, and to a lesser extent filtered work, lack
> of good transformation coefficients from V magnitude to CCD
> magnitude is a prime source of systematic error.
>
> Transformation coefficients themselves are a source of
> error. CCD magnitude is transformed to an instrumental
> magnitude from V magnitudes using the formula
>
> CCD mag = V + Tc*(B-V) + offset
>
> I have begun a preliminary study if my G16 camera and found
> that different fields give different transformation
> coefficients, Tc, for the same camera and telescope
> combination.
>
> I suspect the amount of interstellar reddening affects the
> coefficient more than any other factor. For example, NGC
> 7790 gave me a coefficient of 0.46 while the field of CG
> Cep gave 0.67. The stars at CG Cep are all quite red,
> suggesting a lot of absorption. If the field I am working
> in has high absorption and the stars are reddened, then I
> will be systematically off when I use the transformation
> coefficient from NGC 7790.
>
> So, as Arne has repeatedly drummed into our heads,
> unfiltered data has problems that filtered data does not.
> Then again you get more light, and noise is reduced. Yes,
> filtered data is best, but that loses a magnitude at least.
> You can compensate with longer exposures, but there you
> lose time resolution for rapid variations. It is like
> asking a lawyer for advice: "On one hand ... but one the
> other ... "
>
> CCD Cameras and telescopes will have different
> transformation coefficients. You cannot directly compare
> one observer+camera+telescope with another using unfiltered
> data even when transformed. There WILL be differences,
> systematic, between them. Some observers do not know the
> transformation coefficients and ignore the correction. You
> have a mixed bag of data sets and an intelligent analysis
> will apply corrections to the data sets as a set to correct
> systematic errors.
>
> The best practical way to get a good light curve is having
> 2 observers observe the variable at the same time. Find
> the difference empirically and apply it as a correction to
> get a consistent light curve.
>
> Another source of errors is statistical noise. Those who
> report only it as their error are not reporting the whole
> story, but it may be the best data they have, or the
> easiest to get ahold of.
>
> Yet another source of error is imperfect flats. I have
> trouble getting flats that are truly flat better than 0.01
> magnitude from corner to corner on my frames. I flat one
> flat against another and look at the differences. Even if
> you have perfect guiding, that could introduce a systematic
> error of 0.005 magnitude from the center of the field to an
> edge. My drive is imperfect and the field wanders from the
> east edge to the west edge of the chip (if I let is), so
> there is some systematic error there.
>
> I report the standard deviation of the check star
> magnitudes throughout the night where the check star is
> about the same magnitude as the variable. If the variable
> is a bit brighter than the check star, I still report the
> SD of the check star. I try not to claim higher percision
> than I have, however when the variable is varying wildly,
> that number will be high when the variable is bright and
> low when dim. Of course, I could just leave that field
> blank, but I'd rather report something.
>
> Yes, there are differences between observers in
> (especially) unfiltered CCD data sets, but anyone analyzing
> the data cannot take the magnitudes as gospel as reported.
> You have got to use your head, just as with visual observations.
>
> =====
> Regards, Lew - from CBA Concord -
> http://www.geocities.com/lcoo/cbacal.htm
> My other observing site is CBA Pahala
> http://www.geocities.com/lcoo/pahala.htm
> Get Donna's Book "GHOLSON ROAD: Revolutionaries and Texas Rangers"
> at
> http://www.authorhouse.com/BookStore/ItemDetail.aspx?bookid=18043
> _______________________________________________
> Aavso-discussion mailing list
> Aavso-discussion at mira.aavso.org
> http://mira.aavso.org/mailman/listinfo/aavso-discussion
More information about the Aavso-photometry
mailing list