The 3.6 µm and 4.5 µm calibration star normalized flux densities were monitored and periodically checked to make sure that the aperture photometry of the calibration stars remained nominal throughout the mission (see also Section 8.3.7) for more details. In Figure 4.22, the primary calibrators are shown in blue, red, black, and green colors; the secondary calibrators are shown with a multitude of repeating colors as they cycle through the year. The primary calibrator data points are mostly not visible behind the more frequently observed secondary calibrator data points, but any deviations can be observed.
Figure 4.22: The 3.6 µm (left) and 4.5 µm (right) calibration star normalized flux densities as a function of time from 2013 to 2020. All flux densities are measured by aperture photometry and include 21 primary and secondary calibrators denoted with different colors. The line (and gap) in December 2015 denotes an anomaly in which IRAC was off and no observations were taken. There is no difference seen before or after this anomaly. See the text for the meaning of the various colors, and more information about the plots.
The flux densities have been measured using aperture photometry corrected for both the location of the center of the star with respect to the pixel (pixel phase effect) and the location on the array (array location dependence). The flux densities for each star measurement were normalized to the median value for that star. Error bars are calculated as the sigma clipped standard deviation in the bins, divided by the square root of the number of data points in each bin. The black line at normalized flux density = 1.0 is shown to aid the eye in determining what a flat line would look like. All the data have been processed with pipeline version S19.2. These 21 stars include full array and subarray observations, as well as many different exposure times and well depths.
Over the course of the entire mission, we do see a decrease in sensitivity of IRAC photometry of order 0.1% per year in channel 1 and 0.05% per year in channel 2. The suspected cause is radiation damage to the optics, which can be expected in space environment.
4.13.2 Skydark Bias Offsets
To monitor for any changes in the bias, the median value of the skydarks was checked after they had been pipeline processed. Only one frame time (12 seconds) was consistently used for monitoring, as similar trends in the background value as a function of time over the mission were seen in skydarks taken with all frame times. The 12 second frame time was chosen for its balance between read noise and background noise, so that changes in the bias level can be seen most clearly.
Figure 4.23: Median value of the 12 second skydarks plotted over time since the beginning of the warm mission. The vertical dotted lines denote two anomalies in which the array was reset. The median value was observed to settle to a new nominal level after the anomalies. The red lines give the shape of the predicted seasonal zodiacal variation normalized to the median value of the skydarks in January 2012, to help with the search for any trend changes in skydark values. See the text for more information.
For every combined 12 second skydark frame, a median value of the array is calculated and plotted with error bars over the course of the warm mission (Figure 4.23). Error bars were calculated from the Poisson electron noise and read noise error added in quadrature, as determined in the pipeline processing. There does appear to be a slight trend (< 0.1% per year) of decreasing bias in both channel 1 and 2. This could be due to a decrease in the zodiacal background as the distance of the spacecraft has increased from Earth, or it could be due in part to a decrease in the sensitivity of the optics. Any change in the bias in the skydark frames would also appear in the science frames, but because the darks were subtracted from the science frames, this does not cause any change in the overall calibration of the instrument.
The two vertical lines in the plot denote two anomalies, which occurred on 2014 August 2 and 2015 November 28. The first one caused the arrays to reset with less than the usual applied reverse voltage across the detectors. After normal operation was restored, the skydark bias levels settled back to a level 3.2% higher in both channels. The second event caused IRAC to go to a safe mode. After recovery, the 3.6 µm array quickly settled back to the same higher bias, however, the temperature of the 4.5 µm array was below nominal for over a week while off. Once its normal temperature was restored, the skydark bias quickly settled to a level slightly lower than it was between the anomalies.
4.13.3 Radiation Hits
The number of radiation hit affected pixels (“bad pixels”) was tracked using the calibration observations taken with IRAC once a week. During the mission, the statistical baseline of the average number of affected pixels per second was established as approximately 4 - 6 saturated pixels per second (see Figure 4.24). The trend observed in the IRAC data follows the inverse of the solar cycle, as expected, with less cosmic rays during the peak of the solar activity.
Figure 4.24: Average number of pixels per second affected by cosmic rays in the 100 second skydarks since the beginning of the warm mission. The spike in 2012 is due to a solar flare that occurred during the time the skydark calibration data were being taken.
The number of cosmic rays affecting an IRAC image was determined from the 100 second skydark measurements taken once a week. The dark frames consisted of 27 individual images taken in a 3×3 mapping pattern in the “Best NEP dark” region chosen at the beginning of the mission. This was within the continuous viewing zone and contained no bright sources that would saturate in 100 second images.
The individual images were registered and stacked to determine the affected pixels per second, using the number of pixels above a radiation cutoff level as a criterion (after masking bright stars in the image). These images were also used to count the number of bad, noisy, and dead pixels.
Hot and noisy pixels changed with time, as some pixels recovered and others became hot or noisy. Pixels were deemed to be noisy if the standard deviation was greater than 50 DN, and hot if the flux retained was > 5σ of the flux of the surrounding pixels. Dead pixels, or non-responsive pixels, overall increased by 10 - 20 pixels during the mission (see Figure 4.25). These were trended since the beginning of the mission, with the result that more than 98% of pixels remained usable until the end of the mission.
Figure 4.25: Number of pixels determined to be hot, noisy, or dead during each observing campaign in the 100 second skydarks since the beginning of the mission. Hot and noisy pixels changed over time as some pixels recovered and others became hot or noisy. Dead pixels only increased by 10 - 20 pixels over the entire mission.