[Aavso-photometry] When to submit "Fainter Than" versus actualnumbers.

[Jeff Hopkins]

Hello Michael,

I have multiple photometry professional papers, books and manuals 
published, both as author and coauthor and many professional papers 
do indeed claim that accuracy.  But again, the data in the papers you 
mention were probably taken with a CCD.  See

http://www.hposoft.com/Astro/Publications.html

While determination of the sky condition could warrant determination 
of standard deviations of comparison stars, for normal work, I do not 
think that necessary. Standard deviation of magnitudes is usually 
determined from multiple program star magnitude determinations during 
a short interval (less than and hour for long term variables). Aside 
as noted above, doing standard deviations on the comparison stars is 
rather meaningless as the whole point is determining the program star 
magnitude. That is why we bracket the program star measurements with 
comparison star measurements and do this a minimum of three times. 
The comparison star can vary slightly over time due to sky 
conditions, but the bracketing and averaging tend to smooth that out. 
Each determination of the program star then adjusts for small changes 
in the comparison star (sky condition). That's the beauty of 
differential photometry. I know most CCD work uses a different 
technique than single channel photometry and I originally thought the 
CCD approach should be more accurate. I may have been wrong or 
perhaps the observers doing CCD photometry could fine tune their 
technique some.

BTW, for bright star work my counts can approach a million or more 
per second (Algol V was around 1.5 million counts per second) with 
dark counts (noise) under 100 which can be reduced to zero if the PMT 
is cooled. The dark counts are not important for bright star work as 
they get subtracted out when the sky reading is subtracted. There is 
still sky scintillation noise which is much greater, however. The 
basic rule of thumb to increase accuracy is if you have done 
everything else correct is to get more photons. You do this by either 
observing brighter stars, using a large objective, increasing 
integration (gate) time (only helps up to a point, usually 10 
seconds) or a combination of those.

Remember too, the brighter the star for a given aperture, the lower 
the signal to noise and the potential increase in accuracy.

I'm truly surprised there are not more people doing single channel 
photometry, particularly photon counting. It is certainly much 
easier, much cheaper and appears to be more accurate. I'd be happy to 
help anyone interested in pursuing that.

Jeff

At 12:43 -0700 1/25/05, Michael Koppelman wrote:
>I won't take issue with what you claim. I've read many, many 
>professional papers would would not claim < 0.001 accuracy in their 
>measurements. Perhaps you are just that good, for which I 
>congratulate you.
>
>However, as has been discussed many times on this list, 1/SNR type 
>of errors on a single frame do not take into account many effects 
>which increase the uncertainty such as changing sky conditions, 
>movement within the aperture due to tracking or scintillation, color 
>terms and atmospheric extinction. Most people I have discussed this 
>with agree that the standard deviation of multiple measurements of 
>non-variable stars in the field are the best indication of the 
>uncertainty and take all these factors into account. It is certainly 
>not valid to throw out the measurements that increase the scatter of 
>these constant stars. My rule is I only throw out images *before* I 
>do the photometry, based on bad tracking, clouds, planes or 
>whatever. For non-time-series observations I still take at least 8 
>frames so I have some valid statistical basis to determine my 
>uncertainty.
>
>While some, I know, tire of these discussions, I personally love them.
>
>Cheers,
>Michael Koppelman
>http://www.lolife.com/astronomy/
>
>On Jan 25, 2005, at 11:36 AM, Jeff Hopkins wrote:
>
>>  As for 0.002 m errors being a vast understatement of the error, I 
>>take case with that, however, it maybe true for CCD photometry. 
>>Doing single channel photon counting I routinely produce SDs of 
>>three sets of measurements in the area of 0.001 and sometimes even 
>>less.

-- 
Jeff Hopkins
HPO SOFT
*********************************************************
  Small minds speak about people * Average minds speak of events
************   Great minds speak of ideas!    ****************
*********************************************************
Phoenix, Arizona                           Atlanta, Georgia
7812 West Clayton Drive              12170 Boxwood Circle
Phoenix, AZ 85033-2439 U.S.A.    Alpharetta, GA 30005 U.S.A.
(623) 849-5889 (Fax)                  (770) 619-3322 (Phone/Fax)
www.hposoft.com                         dataman.home.mindspring.com
4th Dimension Developer/ FileMaker Pro Expert
BASIC/C/C++ Programmer/Web Site Developer