Looking for image test set(s) of variables

[han59]

At the moment the nights are short and bright and it is difficult for me to image variable stars. I'm working on a photometry extension of my freeware program and would be interested in existing set of images to test the program photometric capabilities . If anybody has an set an existing set of images which he/she could share that would be nice.

The program has currently the following functionality:

- Read/use FITS, DSLR raws and calibrate them using darks, flats & flat-darks.

- Astrometric (plate) solving the images.

- Detect the stars on each image and compares them with a Gaia star database in Johnson-V to determine a mean flux/magnitude relationship. This I call a photometric calibration.

- Measure one star in the series and report the estimated magnitude against JD and HJD.

- Export to a spreadsheet by select, copy & paste.

The aim with the image set(s) is to test the accuracy of the measured magnitude, check the HJD calculation against existing measurements and to detect possible bugs or ways to improve. The program doesn't use BAA or AAVSO reference stars, just Gaia.

Han

 

 

[Helen]

Hi Han,

You could use the LCO archive https://archive.lco.global/?q=a&RLEVEL=&PROPID=&INSTRUME=&OBJECT=&SITEID=&TELID=&FILTER=&OBSTYPE=&EXPTIME=&BLKUID=&REQNUM=&basename=&start=2018-06-01 00%3A00&end=2018-11-30 23%3A59&id=&public=true .  Grae me @coatesg has been using the SGL account with Faulkes Telescopes to gather variable star data (that we at FT are going to use for education resources).  That will all be available publicly via the LCO link above 😎

Helen

[han59]

Thanks Helen. Looks good. Can unpack them using Funpack.exe. Can solve them (0.4 meter) telescope.  Now I have to find images relevant to variable stars in the 385000 images archive....

Edited June 8, 2019 by han59

[Helen]

 Search for this one to start V0735 Lyr 

Helen

[Helen]

Or CS Cas 

Or ASASSN-V J183347.59+162557.2

Helen

[han59]

Hello Helen,

I have V0735 Lyr series.  Images are solved except one which is only noise. The guiding and focus was not always super. In some images the stars are pretty oval and my software doesn't always sync on the star  V0735 Lyr.  The star reaches an ADU level of about 8000 max But that is a good test case. I will work on it and report.

Thanks again for the suggestion.

Han

[han59]

Bij relaxing for oval stars, I get useful data for V735 Lyr out of the images:

Edited June 8, 2019 by han59

[coatesg]

Nice one Yes - the data on LCOG isn't always pretty, but tracking (and indeed focus) isn't critical for variable work as long as the relevant size circles are used in the analysis. 

There's a second variable in the 0735lyr set (it's 0734lyr). You might want to look at OV Lup - that's a Mira type sequence that I'm running and I haven't plotted yet and there's 3 bands available.

I have some sets of data from my own scope that happy to share as well - mostly HADS stars, but have a nice series showing two eclipses of an EB type eclipsing variable. Drop me a message if you'd like them.

[han59]

These test sets help to improve the program and remove some bugs. The best approach is first to binx2 all images. The oversampling and poor guiding (if any) is confusing the program. I also integrated funpack.

Helen, I have not idea what this red star  ASASSN-V J183347.59+162557.2  is, but below the data the program produces. 

@coatesg I will have a look to OV Lup. The program need to recognize the object as a star. This is a different setup then AstroimageJ? If its too oval, the program will assume it is a double star and ignore it. I'm interested in your data as well.

Han

 

Edited June 9, 2019 by han59

[han59]

@coatesg , Here a plot of OV Lup. This time I used the raws rather then the reduced/calibrated versions. The reduced versions contained partly negative values and my software could not work well with that. Furthermore some images are exposed with 20 second rather then 10 seconds. I could not process the R and B versions since my star database contains only V.

Some FITS keywords in LCOGT files are non- standard like ccd temperature and if flat or dark are applied.

Han

Edited June 9, 2019 by han59

[coatesg]

The catchy named star above, as well as CS Cas, is a W Virginis (Cepheid II with period>8 days) star - curve looks good - CS Cas is a bit atypical with the hump after maximum.

OV Lup is a Mira and has a period of about 2.5 months from memory. The exposures in each band should be consistent, unless I've mucked up the request... I'll take a look tomorrow...

AstroimageJ places centroids using Dec/RA coords - if you choose the right parameters, then having trailed stars doesn't matter too much, as long as the plate solve works!

Edited June 9, 2019 by coatesg

[han59]

Using the three image set above, I'm happy about the software performance as long the stars are reasonable round.  Some images are really challenging.  The main remaining question for me is how good are the magnitude values.  The program uses every Gaia star it can recognise in the image as reference and calculates the mean flux/magnitude value of typical 50 stars. Flux/magnitude outlier values are ignored using simple statistics.  The only problem is that the database has a 0.1 magnitude resolution (earlier designed for compactness) but that's compensated by the number of reference stars used. Using an other public image test set it was able to detect a 20 mmag dip of Wasp-12b. 

You can see the OV-Lup exposure times in one of the columns. Just noted that the table is not sorted correctly on time, but that doesn't influence the graph.

Han

[coatesg]

Darn! I set the wrong filter for the 20s exposures - should have been R. Have cancelled and re-setup those... grr. I'll try and sort out a couple of datasets for you.

 

[coatesg]

One thing to note here, is that the BAA/AAVSO sequence stars are chosen so that they are not variable (or at least shouldn't be!). if you're comparing against *all* field stars from Gaia, I think you need some kind of way of eliminating variables. Having lots of them helps in this regard, but some kind of check to ensure the quoted mag, and the ensemble check of comparison magnitude would be useful. A good check here i the V735Lyr set - it has V734Lyr in it - that varies substantially over the course of that dataset. 

The BAA spreadsheet has something similar (but more by way of a quality check) by doing a calculation to ensure the magnitude from the data you're using comes out correct for the comparison stars in view. 

 

[han59]

Graeme,

Many thanks for the sets. The image quality is good, much better then LCOGT.  Especially the CC Com set was helpful to eliminate a bug in the statistics.  Since the image field is small (0.35 degrees) it finds average about 8 Gaia stars up to magnitude 17 which are not saturated and have enough signal to noise ratio.

With respect to variable stars in the field,  my first idea was to eliminate them using statistics, but that doesn't work for long periods. The best solution would be to create a new Gaia star database without any star having variable flag. That requires some work I like to prospone to the future. First the program should work easy and reliable. Below the graph of CC Com using your data. Below a second graph of one  weak star  in the image demonstrating the stability of the measurement.

There is not much data available about V734 Lyr. In the Gaia archive, see below it doesn't have a variable flag. So any new variable free star database has to wait till Gaia release dr3 or later.

Han

And this KP-Lyn using your data.

 

 

choose files... Click to choose files