There have been discussions in the broader community about procedures for deriving photometry for saturated stars with various instruments. These procedures usually involve ignoring or masking out saturated pixels, and just using the unsaturated pixels, typically located in the wings of the PSF. Most of these discussions that I am aware of have taken place in the context of the Kepler/K2 mission (see here, here, and here (PDF).
It would be extremely valuable to apply such techniques to LSST. Is anyone working on this? It would be quite dependent on the structure and treatment of blooming effects and bleed trails, but it could potentially deliver useful photometry for targets at any brightness, effectively removing the LSST saturation limit. @mjuric, any thoughts about this?
We are not currently planning to do this, although it’s an interesting idea. We do need to estimate the wings of bright stars for a number of reasons, and this could be extended to measure fluxes of bright stars. If is also possible to use the flux in bleed columns (providing the gain is set low enough to set the bleed level below the 18-bit ADC saturation) — this works until the bleeds reach the edge of the chip/bleed stop.
There is also a proposal by @CStubbs to do a twilight survey with short exposure times to measure brighter stars than the regular limit (the camera team now thinks that it may be able to move the two shutter blades simultaneously, allowing a short effective exposure time).
You can be sure the A/D full scale and bleed-over full well will be set to never saturate the A/D, otherwise crosstalk correction would be doomed. This might work for the cases where the bleed trail doesn’t make it to the serial register, but we also plan to control bleed charge by running the serials during exposures so any charge that gets that far will be lost.
Thanks for the information. Those issues are certainly relevant if the effort is to truly conserve flux in bleed trails. But it should be possible to still extract fairly precise photometry from just the nonsaturated wings of the PSF, which would not be compromised by off-chip bleeding. Difference imaging would probably be required, especially in crowded regions, but the potential benefit would be large and broad across many areas of stellar science.
I am aware of the limitations of handling bleed trails, but it works well and is relatively easy.
I don’t quite see how difference imaging helps as the outer part of the PSF is usually pretty stable (and we’ll have high-latitude data as well as low-latitude data on almost all nights). If you’re interested in differential photometry I agree that there is some hope here.
In the past (e.g. SDSS) efforts to do this at a pipeline level failed because I didn’t have good centres, and because the outer parts of i-band images are strongly chromatic due to the properties of our CCDs. I think it could be done for LSST (see my reply above), but it’s not trivial and not currently part of the baseline – although it is only a relatively small extension of work that I think we will need to do to control blending by bright stars.
I’d really like to learn more both about very bright-star recovery from LSST data, and about the twilight mini-survey.
We’ve proposed a session for the #sci:lsst2017 in Tucson next month to determine what OpSim strategies to compute for the mini-surveys (the discussion will also include the deep drilling fields). Some (many?) of the mini-survey goals will include sampling stars that are quite bright, and the set of possible strategies that allow exploration of the bright end of the CMD (particularly for variability) I think is still relatively unexplored.
I think we’re going to be having the planning discussion on slack once the block schedule is released. In the meantime, the Community thread for that session is called “Candidate strategies for mini-surveys” within #sci:lsst2017 (I’m not sure how to link directly to the thread). Comments and contributions are welcome!
cc @jgizis , who would co-chair the mini-surveys and DDF session at LSST2017
Hi @pepper - we (Bulge DECam survey team) have been experimenting with photometry from bleed recovery in very crowded fields with DECam, which might be of interest here. I don’t know if this will be in a presentable state by the time #sci:lsst2017 happens in Tucson next month, but initial results seem promising.
Are you (Josh) planning to attend #sci:lsst2017? Perhaps we can interact more about this offline, or at that meeting.
Hi Will. Unfortunately, I will not be at LSST2017. But I will try to attend any sessions that are held remotely. I am still in the initial phase of working with a student on our current analysis. Part of what we are exploring is how reliable the PhoSim tools are at the bright end.
I think we would be interested in experimenting with the DecCam images and comparing them to some of the PhoSim results. I will email you separately to see about that.