[Aavso-photometry] Uncertainty limits

[Richard Huziak]

Keith,

Besides, if someone else by chance or by plan also measures the star (+ 
large error) and submits, then those obs can be taken with yours and the 
error effectively reduced. :-)  You may want to plot the last 100 days 
of S Cyg and take a look.  PVA and I have been looking at this field 
every clear night to study suspected (now confirmed) variables and the S 
Cyg data is a byproduct of our study.  We are exposing for the much 
brighter (10 - 13m) vars, and the SNR for S Cyg is horrible - around 2. 
 But if you do a statistical fit, our data will make a good curve.  At 
the same time, all visual obs are less thans, so even CCD obs with 
errors in this case of +/- 0.5 mag can be seen as good. :-)  (BTW - the 
'bar' of data at 2453668 is a time-series that was used to check another 
star in the field.  My data starts and Vance's end the time series, and 
if you expand the data, you can see we are in pretty good step even if 
the SNR is very low and the data is somewhat scattered).

I'd also like to suggest that when you shoot any field, you shoot it 
more than once a session - say 3 - 5 times, in immediate succession. 
 Single hit measurements can have unknown errors - scintillation, 
passing cirrus or cosmic rays or whatever, and you can never tell. 
 Imaging a few, then either submitting all 3 or 5 frames of data, or 
averaging and submitting 1 refined estimate should give more confident 
data.  I see a lot of single hit CCD points as seen on the LCG not being 
significantly better than visual for accuracy at times, likely for these 
random error-creating reasons. :-)

rick (HUZ)



Dirk Terrell wrote:

>On Thu, 10 Nov 2005 10:49:54 -0600, Keith Graham wrote:
>
>  
>
>>So, I was wondering if it is best to simply scrap observations with such a high uncertainty? My thinking is that even with such a high uncertainty, the information is still better than  a < observation.
>>    
>>
>
>If the error estimates are correct, then the data can be properly used
>in an analysis. Keep the data plus error estimate rather than giving a
>< observation.
>
>Dirk
>
>---
>Dr. Dirk Terrell
>Computer and Mission Systems Section Manager/Astronomer
>Southwest Research Institute, Boulder, CO
>303-546-9670
>http://www.boulder.swri.edu/~terrell/
>
>
>_______________________________________________
>
>Aavso-photometry mailing list
>Aavso-photometry at mira.aavso.org
>http://www.aavso.org/mailman/listinfo/aavso-photometry
>
>Support the AAVSO by clicking on the Amazon.com link at www.aavso.org before making your purchase. The AAVSO will get ~5% of the purchase price as a donation with no increased cost to you. 
>
>  
>

-- 

* * * * * * * * * * * * *
Richard Huziak
Manufacturing Engineering
SED Systems
Saskatoon, SK, Canada
tel. (306) 933-1676
<huziak at SEDSystems.ca>
* * * * * * * * * * * * *