I'm trying to improve the way I carry out measurements of magnitude. What are the main causes of scatter around a measurement? Suppose I take a series of images of a non-variable star one per minute over an hour. I've taken two reference stars from an AAVSO chart. I use one of the reference stars as my object and the other as the actual reference star. I should get 60 magnitudes all the same but I won't because there will be some scatter around the true value. What are the mains causes of the scatter and what would be typical values?
There must be a dozen or so noise sources involved with our measurements.
May I suggest Berry and Burnett's "The Handbook of Astronomical Image Processing"?
Also Henden's excellent CCD videos class really helps overcome operator carelessness.
Also, cameras have ADCs that have dynamic ranges that are usually 13 of the 16 bits. The amplifiers that feed the ADCs have noise figures. If the amps happen to be on the cooled chip, you get a better noise figure. The sensor arrays themselves are not perfect from pixel to pixel, or in the way they dump their electrons to the amplifiers.
Not much you can do about all that hardware nonsense except run as cold as you can to reduce thermal noise. I find the biggest sources of error is individual observing techniques, calibration care, and sky conditions. Calibration care means, how good are your flats, darks and bias frames, how recent are your transform coefficients, etc.
Other things can add 10s of mmags. Like tracking, focusing, meridian flips, signal to noise (S/N, or, how bright is the star?) and comps that are a little bit variable. Then there are stated errors on the magnitude of the comps. There are a wealth of systems of filter band-passes, and bug-a-boos like U filters that leak IR.
If you do a time series of something like a 10 or 20 mmag exoplanet, you learn a lot about your own skill levels and about your particular equipment. You will see the meridian flip shift from a not perfect flat. You will see high cirrus that you only suspected might be there. You will see imperfect tracking that allows the target stars to land on different pixels. You will see the dancing scintillation from the atmosphere when you zoom in on a star and run a movie of 300 successive 40 second images. You will find all those unknown small-amplitude delta-scuti that you inadvertently chose for comps. You might see interference from cosmic rays, meteors, the moonlight, satellites. And, once you get some big sources of error resolved, you will see more noise for dimmer stars.
Hope some of this noise helps.
Where can I find "Henden's excellent CCD videos class" that you mentioned when replying to Steve regarding Noise of Camera measurements?
I searched the AAVSO YouTube channel but I am not confident I found the video(s) you meant. If you may, please indicate the link(s) for Arne Henden's video(s).
The Henden course is not free. However, in a practical sense, the course was worth five times it's cost to me. As a bonus, I made a small donation to AAVSO just by taking the course. It was a double win for me; I improved my data quality, and I supported an organization that would archive my improved data for future generations.
No one will look at my gravestone in a hundred years, but my Henden-improved, AAVSO-archived data will likely be looked at and used by some bright young post-doc, and possibly provide some new understanding of astrophysical processes.
Also, the better the data, the more likely it can be used for something unconventional. I was fortunate to spend a decade plying experimental cosmology. I can't rule out that AAVSO data could be used to answer the gravitational or emergent space-time mysteries. For example; stars, though small in the universe, are themselves huge laboratories compared to any gravitational experiment we could conduct on our planet.
Arne's CCD course also provides excellent technical references pertaining to sources of noise.
Yes, the link you posted took me to the "AAVSO CCD School video purchase - Members" page.
The video set is very complete, with the titels listed below. Thank you.
COMPLETE SET - $200 (save $35 over the purchase of all individual videos)
Introduction - $15 (77 min)
The Good and Bad of Science - $15 (88 min)
Astrophysics - $15 (84 min)
CCD Detectors - $15 (154 min)
Filter Systems - $10 (72 min)
Bias & Dark Current - $10 (53 min)
Flat Fielding - $15 (150 min)
CCD Camera Selection & Calibration - $15 (185 min)
Atmosphere - $10 (62 min)
Its All About Time - $10 (52 min)
Statistics - $15 (80 min)
Image Processing - $10 (32 min)
Photometry Concepts & Suggested Reading - $15 (56 min)
Transformation - $15 (95 min)
Advanced Procedures - $15 (148 min)
Project Review - $10 (39 min)
Curious Cases - $10 (39 min)
Point Spread Function Fitting - $15 (70 min)
Thanks Ray, I think what I'm particularly interested in is Precision. Suppose I take 1 hours worth of 60 second exposures. My field of view is about 1 arc minute and I pick a reference star which is near the object star and it's about the same magnitude. I also adjust the exposure such that the maximum pixel value is about 2/3 maximum. The sensor on my camera is controlled at -20c. So I think I should measure the same magnitude for my object star on each image because each image was taken under the same conditions and the reference star and the object star are close together in the sky. The magnitude might be wrong which is an issue of accuracy but I think the precision should be good. So I was trying to think what kind of thing affects accuracy.
Thanks everyone for your replies. I have started looking at exoplanet transits and it's made me aware of some problems with my observations and I'm trying to address those problems.