NIRI instrument monitoring

The instrument monitoring zeropoint pipeline for NIRI, a set of Python scripts, is running each day and looks for observations of standard stars from a set of standard fields used by Gemini. If files are found the data is reduced using the NIRI specific Pyraf tasks (nprepare, niflat, nresidual, nisky, nireduce) provided by Gemini's IRAF package as well as two standalone scripts (nirlin.py and cleanir.py) to linearise the data as well as clean noise features sometimes superimposed by NIRI's GNAAC detector controller. The flats are taken each day after an observation was executed. They are a set of open and closed shutter images of a calibration lamp in GCAL.

The photometry is done on each individual, reduced image. In a first run IRAF's apphot phot task is used to identify the stars based on their catalogue position and the WCS of the images. To account for uncertainties in the WCS as well as high proper motion standard stars (e.g. FS29) a relatively large centering box is used. So mis-identifications of random peaks do happen, but are most often filtered by the pipeline in subsequent steps.

The next step is to filter out saturated sources using imexam 'm'. Then another imexam 'a' is run on all the unsaturated stars to get an estimate of the FWHM, which is later used to determine the aperture size as four times the median FWHM. This imexam run also refines the center coordinates of the stars which are sometimes slightly off due to the large centering box in the initial phot run.

This is followed by a round of excluding mis-identifications: For each successful detection the peak of the fitted imexam profile needs to be 90 ADUs above the sky, and the allowed recentering in imexam should have been below 2 arcsec compared to the original center estimated by the phot task. The supposed star is also rejected if the FWHM is outside reasonable ranges which are set differently for guided and unguided images as well as for AO and non-AO ones. For the remaining stars the sky background is determined as a sky per pixel using IRAF's apphot fitsky task. The sky annulus is set to the aperture plus a safety distance of 20 pixels and the sky dannulus is set to 10 pixels.

The final aperture photometry run on the stars is done using the AstroPy photutils aperture_photometry task. Using the X and Y centers obtained in the last imexam run aperture_photometry determines the total sum within a circular aperture determined by the FWHM. To avoid getting skewed fluxes by being too close to the detector edges stars which are within 0.5 times the aperture of the edges are excluded.

The final flux is calculated as the total sum minus the sky contribution. The photometric error of the flux is calculated in the standard form as a Poisson noise based on the flux itself, the stddev of the sky value, the area, and the number of sky pixels. However, it turns out that this error is significantly smaller compared to the variations we see between zeropoints from the typically 5 different images taken per star, so they are not used in the subsequent analysis. The error bars on the plots are the standard deviation of the ZPs for each image per night. The results for the individual stars as well as nightly averages are given in the two different kind of datafiles on this webpage.