[Aavso-photometry] Flat frame quality and its effectonphotometricaccuracy

Sun Sep 7 23:58:13 EDT 2008

Mike,

Thanks for the precisions. If I understand you correctly, the key assumption
in your procedure is that the large scale patterns in your flats are
additive (due to scattered light) rather than multiplicative (vignetting),
and therefore you used the polynomial fit to remove the large scale
variations from your master flat. In my flats, the difference between the
brightest (centre) and darkest parts is about 5%, similar to what you
reported for yours. I use a Celestron C11 with a ST8XME camera (similar to
your system), but I also use a 0.6 X focal reducer. When I take flats at
dusk and at dawn, the telescope is pointing to different regions of the sky,
and I take several flat frames each time. My observation is that all my
individual flat frames look almost exactly the same (visually I can't tell
the difference between them)for a given filter. If a significant part of the
large scale variations in my flats was due to scattered light, wouldn't one
expect that there be variations between individual flat images taken at
different times and different telescope orientations? Do you see such
variations in your individual flats? The fact that there aren't for my
system makes me suspect that the majority of the large scale variations in
my images are not due to light scattering. Of course, my interpretation
could be wrong, so please don't hesitate to correct me.    

Lionel CTE

-----Original Message-----
From: aavso-photometry-bounces at mira.aavso.org
[mailto:aavso-photometry-bounces at mira.aavso.org] On Behalf Of Mike Potter
Sent: Sunday, September 07, 2008 10:44 PM
To: Aavso-photometry at mira.aavso.org
Subject: Re: [Aavso-photometry] Flat frame quality and its
effectonphotometricaccuracy

Mike, Lionel, et.al:

     I did not explain very well what I did.  Basically I made the
assumption that there was no vignetting occurring - so any large-scale
variations in the flat fields must be due to scattered light.  So I fit a
6th order (nothing special about that - it was chosen because it was about
the highest power curve that would not respond quickly enough to "flatten"
out the largest dust motes) surface to my master flat for the evening and
SUBTRACTED it so that the flat was roughly 1.0 everywhere.  That "flattened"
master flat was then used to calibrate the data.  Now, since I do have lots
of scattered light in the actual "science" images, the resulting calibrated
images don't look as nice; there is scattered light that causes a bright
spot toward the middle of the frame, very much mimicking what you might
expect to see if the edges of the chip were mildly vignetted (intensity down
by about 5% in the outer corners).  But I had drawn up a very rough model of
the C14 OTA and I don't think any of the chip - or at most just the outer
corners - is at all vignetted.  The extra scattered light in the calibrated
images gets subtracted out by measuring the sky around the stellar image so
it's not a problem measurement-wise - it just won't win any astrophotography
contests.  All dust motes are completely gone.

     Previously I had nice, almost ruler-flat backgrounds in all of my
calibrated images, but had noticed that light curves from comparative
photometry of pairs of supposedly non-varying stars which were in widely
separated positions on the chip frequently displayed jumps of up to 0.05
magnitudes when I would do the "meridian flip".  This simple procedure seems
to work; though I've only applied it to two nights' data it worked perfectly
both times, completely removing the "jump".   I'm guessing that as long as
the ratio of "actual" sky to scattered light remains roughly the same, and
as long as there are no very bright objects in the fov that it should work.
And, of course, there can't be any actual vignetting.  The shape of the
fitted surface should also be useful to subtract out the scattered light
from individual images, though it would have to be individually scaled and
subtracted for each image.  Any changes in the optical path would
necessitate making another fit - but it's so easy it might be something I do
every time I make a flat.  
     Also - I should have mentioned - I create the flats by observing the
twilight sky - typically 25 or so images, all with 20,000-29,000 adu using
an SBIG ST8.  I keep the exposures longer than 4 seconds to minimize
variations that might be caused by the shutter (which ARE evident in
1-second integrations).  The individual flats are bias and dark subtracted,
then scaled and median combined.

Mike

-----Original Message-----
From: Michael Newberry [mailto:mnewberry at mirametrics.com]
Sent: Sunday, September 07, 2008 8:11 PM
To: lcatalan at lakeheadu.ca; mike at orionsound.com; Gary Walker
Cc: Aavso-photometry at mira.aavso.org
Subject: Re: [Aavso-photometry] Flat frame quality and its effect
onphotometricaccuracy

Gary,

Do you mean Mike Potter? He's the one who first mentioned the 6th order
polynomial. Assuming yes, then I would answer that the coefficient values
don't have to reproduce from night to night in order for the correction to
be justified or valid. My understanding is that Mike is not trusting a
specific set of values that he applies every night but, rather, he is
computing the fit from the image of interest and then applying that to the
same image. I am not addressing whether the coefficient values themselves
would be of interest or value, but I believe his goal was simply to flatten
the image using whatever coefficients resulted. I don't mean to second guess

Mike here. I am just giving the general principle behind the process he
used.

The coefficient values and the order of the fit needed to correct or reduce
differential vignetting (multiplicative) and scattered light (additive) vary

because the sources of non-flatness come from all sorts of external
variables. Here are some sources of non-flatness (I'm mixing the two types
here): sagging of the primary mirror, repeatability or constancy of position

of an aperture stop like the primary baffle or filter cell, the amount and
direction of light pollution, imperfect dark correction if the dark current
is not uniform over the CCD or if the cold finger does not uniformly spread
the temperature across the chip, the moon's phase and direction relative to
the FOV, and which particular bright stars are in or near the FOV. It is a
complex situation! By "differential vignetting", I mean that the basic
vignetting is corrected by the flat field correction and whatever is left
over, is differential vignetting.

Michael Newberry

----- Original Message -----
From: "Gary Walker" <bailyhill at aol.com>
To: <lcatalan at lakeheadu.ca>; <mike at orionsound.com>
Cc: <Aavso-photometry at mira.aavso.org>
Sent: Sunday, September 07, 2008 3:18 PM
Subject: Re: [Aavso-photometry] Flat frame quality and its effect
onphotometricaccuracy

>
> Hello All
>
> Mike Simonsen, I have a question for you.? Do you know the 
> coefficients of

> the 6th order polynomial?? If so, it would be interesting to try this 
> on several nights and see if they are stable, or change considerably 
> each night.? If they do change, then I would not trust them.? However, 
> when you

> do get a stable polynomial, after making changes to the scope, flat 
> routine or whatever, you have good evidence that they might be 
> appropriate.
>
>
>
>
>
> Thanks
>
> Clear Skies
> Gary Walker
> Maria Mitchell Observatory
> 4 Vestal Street
> Nantucket, Mass  02554
>
>
>
>
>
>
> -----Original Message-----
> From: Lionel Catalan <lcatalan at lakeheadu.ca>
> To: 'Mike Potter' <mike at orionsound.com>
> Cc: Aavso-photometry at mira.aavso.org
> Sent: Sun, 7 Sep 2008 5:41 pm
> Subject: Re: [Aavso-photometry] Flat frame quality and its effect 
> onphotometricaccuracy
>
>
>
>
>
>
>
>
>
>
>
> This is very interesting if it consistently works to eliminate these 
> jumps in magnitude after meridian flips. Do you know why "flattening" 
> a flat with a 6-th order (polynomial?) would result in a better flat? 
> This sounds like magic to me at this point, but there may be a 
> perfectly reasonnable reason of which I am not aware. Since this 
> thread started a few weeks ago, I noticed that there were differences 
> between flats taken at dusk and dawn.
> Michael Newberry suggested that these differences may be due to slight 
> changes in the position of the primary mirror in the OTA as the mount 
> does

> a
> meridian flip. But I suppose that changes in scattered light patterns 
> as the telescope changes orientation might also account for these 
> differences. In addition, there is the problem of slight lack of 
> repeatability of the filter wheel positioning. Nevertheless, it would 
> be great if the flat processing technique that you mention did the 
> trick. I just wonder what are the physical reasons for it to work.
>
> Lionel CTE
>
> -----Original Message-----
> From: aavso-photometry-bounces at mira.aavso.org
> [mailto:aavso-photometry-bounces at mira.aavso.org] On Behalf Of Mike 
> Potter
> Sent: Sunday, September 07, 2008 2:40 PM
> To: Aavso-photometry at mira.aavso.org
> Subject: Re: [Aavso-photometry] Flat frame quality and its effect 
> onphotometricaccuracy
>
>     I know it's been a while since this thread was active, but it's 
> been nagging at me ever since I read through it.
>
>     I've frequently noticed exactly what you described in my own data.
> After following the discussion here I decided to try a very simple 
> experiment.  A prime example of the problem you describe showed up in 
> some data I gathered early this week of the newly-minted WZ Sge star 
> currently in outburst in Andromeda (Var 08 And @ 020025.5+441019).  A 
> plot of the difference in brightness of two of my comparison stars 
> showed a jump of about 0.05 magnitudes after a meridian flip.  I'd 
> seen that happen before, but never 0.05 magnitudes - more typically 
> 0.02 mag at worst.  I use a C14 with a SBIG ST8xme.  The chip is 
> pretty small, and I use the AstroPhysics 2"
> visual back that attaches directly to the 3.25" tailpiece of the C14 OTA. 
> I
> did a very quick sketch of the optical path to the chip on the camera 
> and figured there must be very little or no vignetting across such a 
> small chip.
> Still, my flat fields show a decrease towards each corner of about 
> 4-5% in brightness.  So I essentially made the assumption that there 
> was no vignetting occurring, and that all of the low-frequency 
> variation in the flat had to be due to scattered light.  I used the
"Correct Background"
> task
> in Mira to "flatten" the flat field using a 6th-order curve for each 
> dimension (depending on the size of the largest "real" features (dust
> motes)
> in the flat you may want to use a lower-order curve).  Re-processing 
> the data using the "flattened" flat completely solved the problem.  It 
> also resolved a problem I had wherein differential magnitudes between 
> pairs of non-varying stars changed from night-to-night, typically at 
> roughly the same level (few %).
>
>     Basically, in my case, the scattered light appears to be coming 
> from internal reflections inside the OAG and, most significantly, from 
> the baffle tube inside the C14 OTA, producing a "bright" spot near the 
> center of the optical axis.  My flat-fielded images do look less 
> "pretty", but at first blush it appears I'm getting better photometry.
>
> Mike Potter
>
>
>
> -----Original Message-----
> From: aavso-photometry-bounces at mira.aavso.org
> [mailto:aavso-photometry-bounces at mira.aavso.org] On Behalf Of Lionel 
> Catalan
> Sent: Monday, August 11, 2008 4:25 PM
> To: aavso-photometry at mira.aavso.org
> Subject: [Aavso-photometry] Flat frame quality and its effect on 
> photometricaccuracy
>
> I've noticed that the main source of systematic error in my 
> photometric analysis is now due to the quality of calibration flat 
> frames. I take sky flats at dusk or dawn with an exposure time 
> adjusted to achieve a pixel value approximately 50% of the linearity 
> range of my camera (SBIG ST8XMEI).
> I usually combine 10 to 25 individual flat frames to make master flat 
> frames. I make master flats for each filter that I use. Because I use 
> a German equatorial mount, the stars change position in the CCD image 
> after a meridian flip. This change sometimes causes jumps or drops of 
> up to 0.015 magnitude in the target or check stars. I attribute these 
> jumps to less-than-perfect flats. I reason that an error of just 1.5% 
> in the master flat pixel values would cause a systematic error of 
> 0.015 magnitude, and this error only shows up during meridian flips or 
> when the stars slowly drift in the image (I try to avoid that drift by 
> using an autoguider to keep my target star centered in the image). 
> I've also noticed that if I try to calibrate a flat done on one night 
> with a master flat done a previous night (without changing the camera 
> orientation in between), then I don't get a perfectly flat image 
> having just random noise. Instead, I can see the edges of dust donuts, 
> and these patterns in the calibrated flat represent about 1% 
> variations above or below the image average pixel value. So, in 
> summary, I don't know how to further improve the quality of my flats 
> to avoid (or
> reduce) these errors.  I'd be grateful for any suggestions.
>
>
> Lionel Catalan
>
> _______________________________________________
>
> Aavso-photometry mailing list
> Aavso-photometry at mira.aavso.org
> To change options or unsubscribe, goto 
> http://www.aavso.org/mailman/listinfo/aavso-photometry
>
> _______________________________________________
>
> Aavso-photometry mailing list
> Aavso-photometry at mira.aavso.org
> To change options or unsubscribe, goto 
> http://www.aavso.org/mailman/listinfo/aavso-photometry
>
> _______________________________________________
>
> Aavso-photometry mailing list
> Aavso-photometry at mira.aavso.org
> To change options or unsubscribe, goto 
> http://www.aavso.org/mailman/listinfo/aavso-photometry
>
>
>
>
>
> _______________________________________________
>
> Aavso-photometry mailing list
> Aavso-photometry at mira.aavso.org
> To change options or unsubscribe, goto 
> http://www.aavso.org/mailman/listinfo/aavso-photometry
>
> 

_______________________________________________

Aavso-photometry mailing list
Aavso-photometry at mira.aavso.org
To change options or unsubscribe, goto
http://www.aavso.org/mailman/listinfo/aavso-photometry