The age-old question of SNR in Astrometrica


Enrico Prosperi
 

Hello All,
Below is the result obtained with Astrometrica on the same 20 images of 10s each of (99942) Apophis of last April 2, summarized in the file PhotReport.txt

OBSERVER: E. Prosperi, S. Prosperi
OBS CODE: 160
TELESCOPE: 0.36m f/10.2 0.36-m f/10.2 Schmidt-Cassegrain + CCD
EXPOSURE JD: Mid-exposure, not corrected for light time
---------------------------------------------------------------
JD mag Flt SNR ZeroPt Cat Design.
---------------------------------------------------------------
2459307.29294 16.673 R C 17.08 23.164 Gaia2 (99942) Apophis
2459307.29294 16.725 R C 46.85 23.164 Gaia2 (99942) Apophis
2459307.29294 16.728 R C 16.86 23.164 Gaia2 (99942) Apophis
----- end -----

The 3 lines relate to the measurement on the same 20 images using the 3 different STACK options, in detail, Average, Add and Median, in the order.
Note how, while in the case of mean and median sum the results are practically identical, in the case of the sum (Add) the SNR is greater than almost 3 times!
Well, I believe that there is no reason why the SNR in the case of the sum of images is different from that which results in the case of the average of images.
This unless an improper calculation of the SNR by Astrometrica.
And this is precisely the case, sometimes confessed by Herbert himself following questions in the this electronic group.
How to behave then? I use the practice of evaluating the SNR estimation (SNR*) so: SNR* = SNR * 3 when I stack my images with the Average or with the Median sum in order to evaluate the goodness of my measurements,
even though I am aware that it is not a correct way to proceed.
Any comments on this?
Thanks for every contribution to this question.
Sincerely,
Enrico Prosperi
#160 Castelmartini


Richard Miles
 


Enrico,
 
As a general rule, I find that the SNR as displayed in the Photometry file is very close to twice as large compared with values I obtain with other software that I know is accurate.
 
As to the SNR that comes up when you click on an object, that tends generally to be an understimate of the true value.
 
So the 'true' SNR of your Apophis measurement listed below may very well be close to 9.
 
Richard Miles
BAA

----- Original Message -----
Sent: Monday, April 12, 2021 12:19 PM
Subject: [astrometrica] The age-old question of SNR in Astrometrica

Hello All,
Below is the result obtained with Astrometrica on the same 20 images of 10s each of (99942) Apophis of last April 2, summarized in the file PhotReport.txt

OBSERVER: E. Prosperi, S. Prosperi
OBS CODE: 160
TELESCOPE: 0.36m f/10.2 0.36-m f/10.2 Schmidt-Cassegrain + CCD
EXPOSURE JD: Mid-exposure, not corrected for light time
---------------------------------------------------------------
JD mag Flt SNR ZeroPt Cat Design.
---------------------------------------------------------------
2459307.29294 16.673 R C 17.08 23.164 Gaia2 (99942) Apophis
2459307.29294 16.725 R C 46.85 23.164 Gaia2 (99942) Apophis
2459307.29294 16.728 R C 16.86 23.164 Gaia2 (99942) Apophis
----- end -----

The 3 lines relate to the measurement on the same 20 images using the 3 different STACK options, in detail, Average, Add and Median, in the order.
Note how, while in the case of mean and median sum the results are practically identical, in the case of the sum (Add) the SNR is greater than almost 3 times!
Well, I believe that there is no reason why the SNR in the case of the sum of images is different from that which results in the case of the average of images.
This unless an improper calculation of the SNR by Astrometrica.
And this is precisely the case, sometimes confessed by Herbert himself following questions in the this electronic group.
How to behave then? I use the practice of evaluating the SNR estimation (SNR*) so: SNR* = SNR * 3 when I stack my images with the Average or with the Median sum in order to evaluate the goodness of my measurements,
even though I am aware that it is not a correct way to proceed.
Any comments on this?
Thanks for every contribution to this question.
Sincerely,
Enrico Prosperi
#160 Castelmartini


Enrico Prosperi
 

Thanks for your reply, Richard, but my question was also about the difference between the two SNR evaluations of the stacks of the same set of images as summed or averaged: the only difference it would be a constant, i.e. the number of the images!
Cheers,
Enrico


>----Messaggio originale----
>Da: rmiles.btee@...
>Data: 12-apr-2021 16.01
>A:
>Ogg: Re: [astrometrica] The age-old question of SNR in Astrometrica
>
>Enrico,
>
>As a general rule, I find that the SNR as displayed in the Photometry file is very close to twice as large compared with values I obtain with other software that I know is accurate.
>
>As to the SNR that comes up when you click on an object, that tends generally to be an understimate of the true value.
>
>So the 'true' SNR of your Apophis measurement listed below may very well be close to 9.
>
>Richard Miles
>BAA

> ----- Original Message -----
> From: Enrico Prosperi
> To: astrometrica@groups.io
> Sent: Monday, April 12, 2021 12:19 PM
> Subject: [astrometrica] The age-old question of SNR in Astrometrica
>
>
> Hello All,
> Below is the result obtained with Astrometrica on the same 20 images of 10s each of (99942) Apophis of last April 2, summarized in the file PhotReport.txt
>
> OBSERVER: E. Prosperi, S. Prosperi
> OBS CODE: 160
> TELESCOPE: 0.36m f/10.2 0.36-m f/10.2 Schmidt-Cassegrain + CCD
> EXPOSURE JD: Mid-exposure, not corrected for light time
> ---------------------------------------------------------------
> JD mag Flt SNR ZeroPt Cat Design.
> ---------------------------------------------------------------
> 2459307.29294 16.673 R C 17.08 23.164 Gaia2 (99942) Apophis
> 2459307.29294 16.725 R C 46.85 23.164 Gaia2 (99942) Apophis
> 2459307.29294 16.728 R C 16.86 23.164 Gaia2 (99942) Apophis
> ----- end -----
>
> The 3 lines relate to the measurement on the same 20 images using the 3 different STACK options, in detail, Average, Add and Median, in the order.
> Note how, while in the case of mean and median sum the results are practically identical, in the case of the sum (Add) the SNR is greater than almost 3 times!
> Well, I believe that there is no reason why the SNR in the case of the sum of images is different from that which results in the case of the average of images.
> This unless an improper calculation of the SNR by Astrometrica.
> And this is precisely the case, sometimes confessed by Herbert himself following questions in the this electronic group.
> How to behave then? I use the practice of evaluating the SNR estimation (SNR*) so: SNR* = SNR * 3 when I stack my images with the Average or with the Median sum in order to evaluate the goodness of my measurements,
> even though I am aware that it is not a correct way to proceed.
> Any comments on this?
> Thanks for every contribution to this question.
> Sincerely,
> Enrico Prosperi
> #160 Castelmartini
>
>
>
>
>
>
>


Herbert Raab
 


Hello Enrico,

thank you for bringing ma attention to this. This is clearly not what is expected.
I will investigate this when I find some time, but current constrains from my job,
time is limited and it will take a while...

 - Herbert


Enrico Prosperi
 

Don't worry Herbert.
We will continue to use Astrometrica intensively, as we do in the last 25 years, confident that, as soon as you can, you will be able to clarify, with the usual disposability, the reasons for this dilemma.
Have a nice week-end.
Clear skies,
Enrico

>----Messaggio originale----
>Da: hraab2006@...
>Data: 15-apr-2021 19.11
>A:
>Ogg: Re: [astrometrica] The age-old question of SNR in Astrometrica
>
>Hello Enrico,
>
>thank you for bringing ma attention to this. This is clearly not what is expected.
>I will investigate this when I find some time, but current constrains from my job,
>time is limited and it will take a while...
>
>- Herbert
>
>
>
>
>
>