Affiliation
None
Mon, 12/06/2021 - 06:14
Every plot of exoplanets I have seen shows some scatter (residuals) and error bars in it. Assume a perfectly guided configuration where the target star remains centered on the same set of pixels throughout the entire transit, and everything else in the configuration is optimized and correctly set. Given that, what are the main contributors to residuals and error bars on the final plot? Is it the Read Noise in the CCD? Dark current? Shot noise? Or something else?
Ed
Hi Ed,
Residuals are the difference between the transit model data and the observed data. They are therefore a measure of how close your observed data matches an ideal exoplanet model fit. Thus, they are a function of both astrophysical causes, as well as systematics, such as transparency and seeing conditions, local interference (e.g., observing over a roof radiating heat), CCD/CMOS camera effects, etc. Error bars are a measure of uncertainty for a given data point and are a function of such things as camera noise (read-out noise, dark current) and source and sky noise.
Dennis
Hi Ed,
Most everyone uses the uncertainties given by their analysis program. Most of those programs either use Poisson error or else calculate the error based on the (comp-check) uncertainty or the average measurement uncertainty from within an ensemble. The empirical real-life measurements from using the comp star(s) or check star are usually preferred over the theoretical Poisson noise value.
The major contributors are: Poisson error (remember, both target and comp have to have signal/noise greater than 1000) and scintillation (which contributes 1mmag at 100sec exposures for a 30cm telescope observing at airmass 1.4). However, for any given target, other contributors may become important. Even flat fielding errors when the stars don't fall on the same pixel frame after frame can be a large contributor, especially if the flats don't have signal/noise > 1000. Achieving routine high precision takes practice. The best ground-based exoplanet transit light curves that I've seen have come from large professional telescopes like the MMT, with the stars profile spread out with an engineered diffuser. For more information on possible noise sources, I suggest getting a copy of Steve Howell's Handbook of CCD Astronomy.
Read noise is rarely an issue with these bright targets. It is a far more important contributor with spectroscopy or narrow-band imaging.
Arne
Thanks Dennis and Arne, The scintillations are one aspect of accuracy that we cannot avoid (at this time). I am at the finishing stages of building my observatory and my goal is to get the finest, most accurate exoplanet data possible with my equipment. I am glad to hear that read noise is not a major contributor to residuals since I just bought a CCD camera with a read noise of 9 e- at 500 kHz. This is about the slowest I would want to run it. I could have greatly reduced the read noise if I had gone with a CMOS, but I am familiar with the CCD format so (for better or worse) I stayed with it. The camera has minimal dark current as it can be cooled to 50 degrees below ambient.
One thing I have picked up recently that you might be interested in is that the camera USB connection to the computer should be on a completely separate cable from the connection for the guidescope because throwing all that CCD's traffic down the same connection as the guide scope could momentarily block the guide function. Each USB root usually has two USB ports on it, so you would want to have the camera on its own USB root. Thay way the guide scope and mount can continue their tracking uninterrupted.
Ed