I am fairly new to variable star photometry and very new to doing time series analysis. Last night I collected around 2.5 hours of data on one of the campaigns looking at flare stars (https://www.aavso.org/aavso-alert-notice-789).
This morning I prepared my images by calibrating and aligning them with ASTAP. I also did plate solves on them.
After uploading them I created a sequence for the analysis apertures and did a time serial analysis. I was surprised to see a fair amount of scatter as well as some real outliers. Looking into them, it appear that on the outlier images, the apertures are being placed slightly differently than in most cases. Stepping through the images I found them placed correctly most of the time but occasionally they would be off, such that not all the star was in the aperture (sometimes on half the star was, giving an anonymously low value).
I've captured some screen shots showing the issue. Note that I placed an on-screen ruler next to the target star to show it is the aperture moving, not the star.
http://www.otherwise.com/AAVSO/Image_5.png (NSV 23386 is outside the aperture)
http://www.otherwise.com/AAVSO/Image_6.png (Star 103 is outside)
http://www.otherwise.com/AAVSO/Image_7.png (NSV 23386 is outside the aperture again but in a different direction)
http://www.otherwise.com/AAVSO/Image_14.png (Star 113 is outside the aperture)
Then I will get long stretches of images where all three stars are well within the aperture. I will repeat that it looks like the aperture placement is incorrect. The star position is staying fixed.
What could be causing this? It seems impossible to do the time series analysis with this issue. All I can think of this that ASTAP is doing the registration or perhaps the WCS solving in some way that VPhot is not designed to handle.
I can make the images and sequence available is someone wants to look into it.
Bill
Just to follow up on this... I was thinking that perhaps the alignment (registration) that ASTAP was doing might have confused things. I had ASTAP do only calibration. I uploaded these without plate solving, to let VPhot solve them. I still have the same problem where the apertures are not placed correctly in some images.
I deleted those and uploaded a second set where ASTAP did the plate solving. Same issue.
I would like to get to the bottom of this since I can't submit time series results until this is resolved. Very mysterious (to me).
Please share the whole timeseries to SGEO
George
SGEO
Please also share the sequence that your using:
Admin / Sequence / select and Share to SGEO
Thx,
George
I will also point out another problem I see. This morning I uploaded a time series using the "B" filter on the same star (the first was "V"). If I:
Thanks for the help…
George,
Thanks for the help. I've shared the time series as well as the sequence used for analysis (NSV 23386). The time series files are slightly different from what I was using yesterday, but they still show the problem (although maybe not quite as much). These have been calibrated and plate solved by ASTAP but I made sure not to align them.
I was using an aperture size of 7 pixels. I found the automatic aperture of 1.5 * FWHM was too small in general. using something smaller makes the problem worse.
Look particularly at images 5, 6, and 7 (starting from earliest time). You can clearly see the aperture being placed in different positions, sometimes cursing off part of the star (particularly if you use a 5 pixel aperture).
Bill
Bill:
George shared half your images with me along with your sequence. When I run a time series, I don't really see your problem? The target was always within the measurement aperture. Light curve looked reasonable.
1. Can you send us a screenshot of your VPhot Tools>Settings> Centroid Determination settings?
2. You may consider not binning 2x2 to avoid being undersampled. Oversampled (a little) is safer.
3. Measurement Aperture of 7 pixels is TOO large. Look at the star profile and see where adu counts return to sky background. That is what you should look at to determine the size of your aperture. I used 4 pixels BUT could have used less! (See binning issue above) Don't let your eyeball make you increase the aperture because it looks like a little of the star is beyond the aperture. The gap will remove that issue.
3. How many images were you seeing a problem on? A few or a lot?
4. BTW, sometimes (not often) you cannot trust the location of an aperture when you use the blue arrows to scroll the images.
5. If a few images give a mag that looks clearly abnormal (all by its lonesome), delete them? I removed two obvious outliers.
Comments?
Ken
I wrote George several…
Ken,
I wrote George several days ago and told him not to spend time on this right now. I have been unable to reproduce it with other data sets. I also deleted the VPhot data that originally caused the problem. It *may* have been that I asked ASTAP to calibrate and align the time series rather than just calibrate them.
1. My centroid settings are here. I don't think I have changed them from the default: https://otherwise.com/AAVSO/CentroidSettings.png
2. I don't think I'm under sampling. My stars are spread over 7 or 8 pixels generally.
3. Yes, I've been playing around with different aperture sizes. I didn't know this: "Don't let your eyeball make you increase the aperture because it looks like a little of the star is beyond the aperture. The gap will remove that issue."
3b. There were 5 or 6 out of more than 100 that showed it. I noticed them because they were outliers in the days plot.
4. I don't think I did that. What was strange was that some apertures in an image were off and others were fine.
5. Yes, I suspect this might have been ones with the aperture problem.
If you haven't done so, look at my first post. It has images showing the issue. I am not making this up but as I say, it is not totally reproducible. It *may* have been related to "aligning" the images in ASTAP.
Bill
Bill:
The rule of thumb per the Nyquist Criterion is that you want about/at least 2 pixels across the Full Width Half Maximum (FWHM, seeing parameter), not across the full visible stellar disk.
Since it is difficult to get exactly 2 pixels across the FWHM due to the physical pixel size constraint (microns) of the camera, it is better to have 2-3 pixels across the FWHM rather than the 1-2 pixels. Again it is better to be oversampled (2-3+ pixels/FWHM) than undersampled (<2 pixels/FWHM) as you are (1.5 pixels/FWHM). Yes, you will get even more pixels inside the measurement aperture.
With respect to the aperture radius, you would want to select a radius that corresponds to the radius at which the star profile returns to sky background level, not a lot more. In that case, you would increase noise by including a lot of sky background.
Ken