(for recording) Jennifer will work on after she improved conversion formula using the 2023 April/May run.

I'm working on the colour-term from GAIA to the AG cameras, and have the sample code from Arnaud for the calculation. 

Have any standard stars been systematically observed with the AG?  If so, what are the stars/visit IDs. Otherwise, I can run through the observed stars looking for appropriate candidates and observing conditions. 

If we do not care too much about the absolute zero-point of the AG, then
1 - select visits with clear skies
2 - identify Gaia stars in those visits
3 - retrieve G, Bp, and Rp photometry of the stars from the gaiaDB
4 - plot AG_mag - G against Bp - Rp, where AG_mag = 2.5log(AG_flux) + zero-point  (you might want to worry about the vignetting in AG_flux)
5 - fit a curve
6 - set the AG zero-point so that AG_mag - G =0 at Bp - Rp = 0
You end up with AG_mag = G + f(Bp - Rp) and I think this is what we want.

Thanks, that matches what I was planning if there weren't systematic calibration observations. 

Filtering the data by seeing (using the spot size), flux (eliminating faint sources), and sources near the edge of the detector (via the flags), gives sets of reasonably well behaved fluxes/colours. Fitting each set of sources using a linear fit with outlier rejection to derive the colour term and a zero point  produces generally consistent values, with some scatter. Looking at the fits, in some fields, particularly those with a small number of sources, the outlier rejection obviously needs to be fine-tuned. 

Could you share a few plots? I think we want a higher order function to fit the data. Probably 2nd or 3rd order polynomials. As Jim mentioned the other day, we can estimate the color term using the response functions and a stellar library. I can do that and you can compare it with yours, if that is useful.

For the plots below

• Data filtered by magnitude (GAIA g < 18.5)
• Seeing (via median FWHM in frame, excluding camera 2)
• sources near edge of image removed
• linear fit with outlier rejection
• each exposure was fit separately, as there's enough variation in transparency over the observations to make the fitting 
• the derived parameters of the fits plotted in a histogram

for the plot of AGmag - Gaia G vs GAIA rp - GAIA bp the best slope and intercept  are -0.30 and -30.45 (with a bin size of 0.01).

The first two figures show the histograms; the results are consistent within the bin size for small changes in limiting magnitude;  increasing the limiting magnitude, lowering the seeing threshold or combining images produces a larger scatter. 

The next two pairs of plots show example colour and magnitude plots for two fields, one with a small number of sources, one with a large. 

Higher order fits will, I think, need to be left for systematically taken measurements of standard stars in photometric conditions. 

In your mag_AG - G vs. Bp - Rp plots, I see stars have Bp-Rp ~ -2. This is too blue. See, e.g.,

 https://sci.esa.int/web/gaia/-/60198-gaia-hertzsprung-russell-diagram

agCam.pdf

The PDF attached above has notes about the AG calibration and validation of measured results. 

I'm assuming that this ticket is where the analysis for INSTRM-2073 is being done.

1. Why are you fitting for the slope of the fit to Gaia? The only reason for that would be concerns about non-linearity. Are we worried about non-linearity in the AG chips? If so, what data do we have? And should we take some more, e.g. photometry as a function of sky background. I don't think this is likely to be a first order problem (a percent or so at worst?) so for now please set the slope to be 1.0 and only fit the offsets. Is this the cause of the non-linearity you report below c. 19.5; in particular, is this onset independent of the sky brightness? In a similar way, we should be fitting the slope of the colour correction once and for all for GAIA v. PFS, unless you have evidence that it's different for the different AGs (possible, but unlikely).
2. What is causing the outliers in the plots v. GAIA? When you look at them, does something jump out? What is the RMS to the fit after clipping? Is it consistent with the expected errors in the photometry (both GAIA and PFS)?
3. If you don't have enough GAIA stars, have you looked at the match to SDSS or PanSTARRS? If you haven't done this, and don't have the tooling setup to do it, ignore this for now. If you do have more catalogue matches available, though, it'll be easier to understand our photometry.
4. Please remake all the mag v. mag plots as plots with the residual on the y-axis (e.g. AGMag - G v. G). Please make these after as well as before the colour correction derived from GAIA BP - RP.
5. Is your photometry independent of seeing? If not, what aperture would you have to use to make it seeing independent (for bright stars, of course)?
6. How well can you tie the photometry from the different AGs together? E.g. what's the mean scatter, both raw and after you've solved for the side-to-side and chip-to-chip offsets (possibly as a function of seeing)
7. How consistent are the zero points from standard catalogues derived from bright and faint stars (after a colour cut to restrict the fits to reasonable coloured objects, e.g. the ones that you think will have g-r of c. 0.0? How about the other cut into red and blue samples? You'll probably want to do this once you've established that each AG chip is on the same system.
8. How much do the fluxes of individual stars vary during an exposure? How do these correlate between stars and between AG chips? If you take out the average variability, how consistent are the fluxes? Can you find nights when there doesn't seem to be any transparency variation?
9. Once you've taken out transparency variations, is the scatter in the flux of each star consistent with your quoted errors?

(This is for recording) Jennifer will work on 1. and 4. first. Some items may be addressed in the new tickets.

An updated report is attached. 

agCamNew.pdf

(Recapping activities in the last a few months)

• For photometry itself, Jennifer
  • applied bias and flat-fielding, but the resultant scattering didn't change
  • changed parameter for sep, but  the resultant scattering didn't change
• For fitting of mag(GP) vs mag(AG),
  • She filtered objects with bad seeing, as well as objects flagged as non-MS in gaia catalogue.
• Applying the fitting to each half region of each camera separately, variation gets small (0.05mag), and residual deviates from linear at fainter sources
• She will upload the latest version of the slides that she explained at ICS/PFI telecon in November/December 2023 by the end of this week, in order to see if the current formula is good enough or not
• I also requested to use the following visits where the same stars were observed in different AGs, to check her fitting
  • visit = 98353 – 98360 (ngc6633, PA=90)

• • visit = 98371 – 98378 (ngc6633, PA=-90)

Report is attached. report.pdf

Earlier this week, we talked about how/who to do what again. Considering the role of agccActor, Jennifer will focus on fixing INSTRM-2060 and measure centroid to guiding (and MCS). Calculating photometry to measure transparency taking account into a color term will be taken over to someone outside ASIAA.