I am basically totally new to the topic, but am lucky to have a good instrument that I hope will add to the sum of knowledge in time.
i have just done a long first run on BL CAM using my scope hosted on the itelescope network and images will be sent to VPHOT (i hope). I just realised that i checked the dither images option when taking the images. is this a mistake? will it mess up my image analysis in vphot and render the run useless?
i have the option to download the 100+ images, align in ccdstack or in pixinsight and then upload them to vphot again - though that would take a while and be a waste of the automated upload to vphot. should i do so or is there alignment in vphot that can handle this.
as a general principle should i dither images or not...my thought is that if i am stacking them and if vphot can handle the alignment automatically then yes would be the answer -if i am not stacking then there is no point i suppose...
and one final "extension" question, conceptually i think i can get a better result from a stacked image than unstacked. is there a facility to have a "rolling stack" such that each image in a time series is stacked say with the previous 3 and following 3 images?
thanks for any help as i get started here.
HI Gavin,
VPHOT works from the plate-solved image, so dithering will not affect its ability to find the target and comparison stars and do the photometry. I do not recommend stacking them with other software packages unless you understand the photometric issues involved in sub-pixel alighment techniques.
That said, the other question is about how best to handle a long time series. If you want to average several points together and submit them as a single measure, there is a nice program called Boxster on our website that will perform this duty and give good error analysis. That will improve the signal/noise in a submitted measure at the expense of time resolution. I don't recommend "rolling or running averages" as each point is then not unique - it depends on measures prior and subsequent, which usually screws up statistical analysis.
Congratulations on having an iTelescope hosted system! I (and others) will be happy in getting you established in the photometric world!
Arne
Arne,
I'd like to learn more about the photometric issues in stacked images. Can you expand on this or point me to a source of information?
I've been using AIP to stack images of a faint target (AO Her now between 18 and 19 V) before uploading them to VPhot. Are you aware of any photometric issues with how AIP does stacking?
Phil Sullivan spp
i'm off on
thanks Arne
i'm off on business for a few days so may not get to do much this week, but look forward to communicating with you on it thereafter. thanks for taking the time to give me the advice...
arne
am going to try to work a bit on bl cam so that i can optimise my system and methods before moving to other objects.
one thing that i was looking for quick guidance on is exposure time. - my scope is a cdk 17 and my camera an fli 16803 with filters in lrgb ha sii oiii, V, B and CBB (exoplanet).
i attach the simple one image photometry report, and a report from a time series run last week. (if the file attachment works)
i was a bit disappointed with the error report on the time series report saying an error of 0.12 (which looked big to me compared to other reports i had seen folk file) on a mag of 13.1.
am i misunderstanding the whole thing? or - and here is my question - am i exposing too short at 60 seconds for that target as i am only getting 170 of snr...?
many thanks for any advice as ever,
gavin
Your image duration of 60s is fine. A SNR of 170 gives a poisson error of less than 0.01. Look at the single image photometry screen shot where error is 0.012. Expected since it includes other sources of error as well.
The error reported for the time series (0.12) just takes into account the variation of the target (0.4 mag) over the duration of your series. In other words, the target varies by a lot more than 0.01 during the duration so its "error" is "large". Look at the error of the check star (0.01) which is a constant target over the same duaration.
Your procedure is fine!
Ken
Hi Phil,
The main issue is a combination of your pixel sampling of the image profile, and the algorithm used for the alignment.
If you use integer pixel alignment, so that you align to the nearest pixel, then there are no issues. If you use sub-pixel alignment, so that your software apportions parts of a raw pixel into multiple other pixels, then the problem is how that operation is performed. Remember that you are taking a smooth, gaussian profile star that is usually circularly symmetric, and having that flux fall on square pixels. With typical image scale, you do not sample the image profile very well. This is fine for aperture photometry, but if you want to shift an image from one place to another, that poor sampling comes into play. Does the software model the image profile with a gaussian profile, and then divide the raw pixel accordingly? Does it just use a linear interpolation, or no interpolation at all?
So in general, if you are aligning images to make photometry easier, do it to the nearest pixel. Otherwise, the results will depend on the particular situation you have. Most of the time you won't see much difference, but I don't like to have another factor to worry about.
Arne
Thanks Arne. I'll see if I can learn how AIP does image alignment. I'll also try out Boxster.
Phil Sullivan