We reached out to Frank Valdes, NOIRLab Pipeline and Observatory Scientist, about the processing of DECam images. He writes:
"The DECam image data products, such as SkySub, typically have a magnitude characterization. I want to emphasize scientific photometry requires more calibration by the user than the Community Pipeline can do. For instance, the CP only processes data independently by filter and therefore there is no color information which is required for absolute magnitudes of specific objects. What the CP provides. is a magnitude zeropoint that is an average over all source instrumental fluxes that match sources in some photometric catalog. Generally the photometric catalog is Pan-STARRS1 for dec>-30 and Gaia EDR3 for dec <-30. The matched objects (primarily stars) use the best or nearest bandpass to the observations. DECam data may not be in the same filters but, for instance, if the DECam filter is g and dec>-30 then the Pan-STARRS g-band calibration is used. For dec<-30 the Gaia G magnitudes are matched regardless of observation filter.
The characterization provides two magnitude zero points in the header. One, MAGZERO, is directly in the units of the pixels and the estimated magnitude is
mag = MAGZERO - 2.5 * log10 (instrumental flux)
THe other is MAGZPT which has the exposure time taken out (i.e. for instrumental values per second) so that it can trace transparency variations independent of exposure time. To use it the change to above is straightforward:
mag = MAGZPT - 2.5 * log10 (instrumental flux / exposure time)
Note that the gain does not enter because the data has been flat fielded to try and make the gain uniform over the field of view.
Again, this is just a start on scientific photometry and any science you want to do will require actual photometric calibrations. Also calibration depends on additional factors such as how the integrated flux is measured (aperture photometry vs PSF photometery vs …) The CP uses isophotal photometry with some gaussian assumptions. The magnitudes you get, in the bandpass of the reference catalog, is roughly correct to 0.1 mag.
I hope this gets you started on the challenges of scientific photometry.
Frank Valdes, NOIRLab Pipeline and Observatory Scientist"
Further, our own Mike Fitzpatrick notes regarding your comparison with SDSS:
SDSS magnitudes are "asinh magnitudes" computed as
m = -2.5/ln(10) * [asinh((f/f0)/(2b)) + ln(b)]
and not with the standard flux conversion which may explain the difference."