There doesn’t seem to be an exoplanets category in the chat, so I’ll ask this question here. I’ve had a request to observe a planet transit where the transit depth is only 0.056 mmag, i.e. 0.00056 mag, known from K2 and TESS data. I tend to think that the minimum per-observation rms scatter in ground-based photometry is about 3 mmag. This seems to be limited mainly by jitter in atmospheric transparency on short timescalesobserving against multiple comp stars. But in principle, given enough such data, and if the timescale is useful for a transit event (say), one could beat down the mean-errors-of-the-mean well below a millimag. But I don’t have a good feeling for this, whereas the experts here might have the experience of doing so. Can anyone point to some published results with errors in this range, such as fitting to a lightcurve or transit?
I don’t have published results but the smallest transit depth I’ve seen reported in the SG1 group reports is 1.5 ppt. There was also one transit where there was a ‘hint’ of a transit at 1.0 ppt! Both of these were observed with a 1m scope in the Atacama?
BTW, I assume you meant 0.56 mmag = 0.00056 mag?
I doubt you will be able to detect the 0.5 mmag transit with ANY statistical confidence. Perhaps a ‘guess’!
Thanks, Ken, for the added advice. Yes, one too many zeros on the millimags... So I agree the proposed observation can't be done from the ground with the usual sorts of hardware + sky --- a couple millimags, okay, but down well below 1 mmag is unrealistic. Our postdoc making the request wants to know the ingress time, even more difficult to determine.
Some context for the curious: an example of what I think I can do is in:
(see the upper-left plot in Figure 2)
…whereas the upcoming event is for this star:
(see Figure 9, noting the ~10x smaller vertical scale-range compared to the above)
Another example of precision limits in data I've taken is:
…where Figure 1 here shows a lightcurve for this M-dwarf in a locked system with a (yet unseen) white dwarf. The smoothed-mean looks pretty nice given the 3 mmag per-point rms. But this is a case where you have many cycles to average over, not the unique event of a planet transit.
Brian, I’m sure you know this already but all one can do is try if you think your observing system and observing location has any hope of detecting signal changes that small. Of couse, smoothing high-rate-sampled photometry can beat down the high-frequency noise to help but the signal has to be there and not overwhelmed by other noise sources in your observing system. Which are you going to use?
We’ve given up on trying this upcoming transit of interest since the transit depth is simply too small. The idea would have been to use the Lowell 42-inch (1.1-m) telescope, but the comp stars available for HD 73344 and the target being rather bright (V ~7, we could use a narrowband filter) makes it a poor target.
I note in the current JAAVSO the report about HAT-P-54 shows an event using the Kuiper 61-inch telescope in the Catalinas north of Tucson. See the bottom panel of Figure 2 there. The trace shows rms scatter of 0.47% in the residuals, i.e. about 0.005 mag. They could have binned the data into 5-image averages (or even 5 clock-minutes) to cut down on the scatter somewhat, but it still would get them only to 2-3 mmag or so. So errors in this range seem to be pretty well enforced in ground-based data.
Hi folks! We published a paper back in 2020 that looked into the precision that a 6-inch (15.24-cm) scope could get - we tentatively got a limit of 0.52%/minute for this sized telescope (see top panel of Fig. 7), but more study is needed.
Somewhere I saw a proposal for a spacecraft with a “standard” artificial star to improve ground-based photometry. I have not seen a formal link error budget for the proposal but can’t imagine it will solve the standard propagation of light from astronomical source to ground added propagation noise problem. Not sure if a LEO or a GEO satellite was proposed. If one were lucky and could get the orbiting artificial star and your variable in the same FOV that might be interesting but what is the probability of that? Just thought I would mention it.
We happened to have a talk about this at Lowell a month or two ago by one of the project managers. The claim was the Landolt etc standards were good to only 2-3 percent, which is incorrect. The idea is an LEO spacecraft that shines some multi-wavelength 'NIST-traceable' lasers downward at short intervals. So you have to have filters for the specific wavelengths and you must have 'best' skies at the moment it passes over. The stellar magnitudes in the visible were quite faint (like mag 16 or something), so you'd need a big telescope to get any precision in the short time interval involved. It didn't look workable to me for any telescope I would have access to. So for a planet transit, you'd have to have that spacecraft sitting in your field effectively for a whole night, which isn't gonna happen.
The Landolt and Cousins E-region standards are good to ~2 mmag in the mean around the sky, with per-star scatter of about 7 or 8 mmag. So for external calibration for us mere mortals they are entirely sufficient. It would be nice to extend those to many other fields and also to somewhat brighter stars to expand the basis of comparison. I suppose the GAIA photometry does that, but the very broad passbands makes them problematic for many applications.