Infrared Excesses Around Bright White Dwarfs from Gaia and unWISE. II

We observed 183 targets with the InfraRed Array Camera (IRAC) in both warm channels with central wavelengths located at 3.6 (Ch1) and 4.5 microns (Ch2), respectively (Werner et al. 2004) under program number 14220. Each exposure was 30 seconds and 11 medium-sized dithers were used for each wavelength. Both point-response function (PRF) and aperture photometry were performed for every observation using the MOsaicker and Point source EXtractor (MOPEX) package. The measurement flux uncertainty was added in quadrature with an additional calibration uncertainty of 5% (Farihi et al. 2008). We report the PRF magnitudes unless the target did not appear properly subtracted in the residual image or the centroid position of the target between the Ch1 and Ch2 frames was significantly discrepant. In those cases, we reported magnitudes measured from aperture photometry. Observed targets have a typical measurement signal-to-noise ratio (SNR) of 17.5 in Ch1 and 18.2 in Ch2 after applying the systematic calibration uncertainty.

As they were selected for having unWISE excess, many of our targets are found in crowded fields with a risk of contamination from nearby sources even in the higher-spatial-resolution Spitzer images. The PRF-fitted photometry can mitigate this risk, but is not immune to cases of overlapping sources or nearby extended sources. The reliability of the Spitzer photometry was determined by examination of the PRF residual images, which we searched for evidence of over- or under-subtraction of the target source. We have flagged 9 of 183 targets for which the PRF was not cleanly subtracted and this Spitzer flag is indicated by the letter "s" in Table B2. Photometry of targets with the Spitzer flag is considered unreliable and any indication of excess is not likely to be real. We exclude flagged targets from excess statistics but report them in Table B2 for completeness. The remaining 174 targets with reliable Spitzer photometry are examined for infrared excess in Section 3.2.

We obtained new near-infrared photometry for a total of 116 white dwarfs using the Near-InfraRed Imager (NIRI; Hodapp et al. 2000) at Gemini North and 126 targets from Flamingos-2 (F2; Eikenberry et al. 2006) at Gemini South. In total, 235 white dwarfs were observed, accounting for some overlap between NIRI and F2. Observations were conducted under a variety of weather conditions. More information on the Gemini observations and data processing can be found in Appendix A. A small sample of the infrared photometry of targets observed by both Gemini and Spitzer is available in Table 2 with the full table containing all of the new photometry available in digital form.

Gemini-N/NIRI: We conducted observations of 116 white dwarfs using NIRI at Gemini North as part of programs GN-2018B-FT-208, GN-2018B-Q-406, GN-2019A-Q-303, GN-2019A-Q-403, GN-2019B-FT-111, GN-2019B-FT-216, GN-2019B-Q-237, GN-2019B-Q-408, and GN-2020A-Q-405. Each target was observed in the Maunakea Observatory (MKO) standard J, H, and K filters within a 120'' × 120'' field of view. Exposures were 10 seconds and a random dither pattern was employed for a total of approximately 20 exposures in each filter. Data reduction and frame stacking were handled by version 2.1.0 of Gemini's publicly available Data Reduction for Astronomy from Gemini North and South (DRAGONS) software (Labrie et al. 2019). Aperture photometry was performed with astropy's photutils (Astropy Collaboration et al. 2013) using its point-spread function (PSF) fitting capabilities. The reference stars were modeled independently to determine an appropriate aperture radius for each frame based on three times the median of the full-width half-maximum (FWHM) determined by the PSF fitting. Bright reference stars in the Two Micron All-Sky Survey (2MASS; Skrutskie et al. 2006), where the 2MASS K_s ≤ 15.5 mag, located within the field of view, were used to calibrate the zero points for all J, H, and K bandpasses. For J and H images, we used the same bright reference stars from K band unless the total number of reference stars was less than five, in which case we used all of the 2MASS stars contained within the field of view. The 2MASS photometry of reference stars was converted into the MKO photometric system in order to calibrate a static flux zero point for each image (Hodgkin et al. 2009). The typical SNR for our targets were 170, 170, and 140 for J, H, and K bands, respectively.

Gemini-S/F2: For targets in the southern sky, we conducted observations of 126 white dwarfs at Gemini South using F2 as part of programs GS-2018B-FT-204, GS-2018B-Q-404, GS-2019A-Q-301, GS-2019A-Q-404, GS-2019B-Q-237, GS-2019B-Q-408, and GS-2020A-Q-409. Seven of the observed targets were also observed with NIRI. Each target was observed in the MKO J and H bandpasses, as well as the K_s bandpass, with a 6' circular field of view. As with NIRI, we used a random dither pattern around the target, but the exposure times were often between 10 and 30 s for J band, 6–10 seconds for H band, and 10–20 s for K_s band. Upwards of 20 exposures per target in each filter were taken. We used DRAGONS for data reduction and performed aperture photometry in the same way as with targets observed by NIRI using bright 2MASS reference stars within the field of view. Since the F2 K_s filter is similar to that of 2MASS (Leggett et al. 2015), no transformation was made for the K_s magnitude to convert it into the MKO system preceding the zero-point calibration. Typical SNR across all F2 targets and filters were 180, 140, and 110 for J, H, and K_s bands, respectively.

Some of the white dwarfs observed with Gemini have existing near-infrared photometry. The Gemini photometry was compared against the UKIRT Infrared Deep Sky Survey (UKIDSS) for NIRI and the VISTA Hemisphere Survey (VHS) for F2. The comparisons showed systematic linear offsets in the J band for NIRI and in the J and K_s bands for F2. We measured and corrected for this offset. Additional details, including the magnitude of the offsets, are shown in Appendix A.1.

The Gemini aperture photometry can be unreliable for reasons including blending with background objects or poor zero-point calibration. We found the uncertainty of our photometric results to increase up to 30% when the number of quality reference stars within the field of view is small, thus we have flagged all photometry where four or fewer reference stars were used. In cases where the Gemini photometry is flagged, we consider public photometry from UKIDSS, the UKIRT Hemisphere Survey (UHS), and VISTA instead for calculations involving near-infrared excess if they exist. The VISTA photometry is transformed into the MKO system. If no public near-infrared photometry is available, we use the flagged Gemini photometry as a last resort. In total, 81 targets were flagged out of 235 targets observed with Gemini Observatory. Flagged targets should be treated with additional caution.

SEDs were constructed using photometry from the Panoramic Survey Telescope and Rapid Response System (PanSTARRS; Chambers et al. 2019) DR1, Sloan Digital Sky Survey (SDSS; Ahn et al. 2014) DR12, VHS (Cross et al. 2012) DR6, UKIDSS (Lawrence et al. 2007; Hambly et al. 2008) DR11, UHS (Dye et al. 2017) DR1, 2MASS (Skrutskie et al. 2006), ALLWISE (Cutri et al. 2013), and unWISE (Schlafly et al. 2019). For our white dwarf photometric and spectroscopic models, we assume the effective temperature, T_eff, and surface gravity of the DA model fits reported in Gentile Fusillo (Gentile Fusillo et al. 2019). For each target, synthetic photometry of a DA white dwarf (Holberg & Bergeron 2006) ⁷ was scaled to fit the PanSTARRS, SDSS, or Gaia optical photometry using chi-square minimization methods. Hereafter, we refer to the DA white dwarf synthetic photometry as the "Bergeron model." For visual purposes, a blackbody model was adjusted to fit the J flux density of the photometric Bergeron model in the SED figures. The model flux and its corresponding uncertainty described in the following sections refer to the photometric Bergeron model.

All photometric magnitudes were converted into flux densities using associated zero points for each bandpass. Gemini photometry was converted into flux density using photometric zero points of the MKO system (Hodgkin et al. 2009). Filter transformation is described in detail in Appendix A.1.

Sources of statistical error in the model include uncertainty in the temperature, parallax, and surface gravity. The model error was computed by fitting a Poissonian probability distribution to a set of apparent magnitudes generated from a Monte Carlo simulation. For each magnitude, the temperature was randomly sampled from a Gaussian distribution of its mean and standard deviation. A resulting magnitude was obtained from each set of white dwarf parameters by referring to the pure-hydrogen model white dwarf atmosphere grid from the Bergeron model. We applied an uncertainty floor of 5% in the model flux to represent the systematic uncertainty of fitting every target with a DA white dwarf model while lacking information on each individual target's spectral type. If the white dwarf parameters are off or it is a different spectral type than the assumed DA, it can lead to an erroneous stellar temperature affecting the predicted infrared flux in the bands of interest. As a secondary check, we have also performed the infrared stellar model flux calculations for all targets using DB parameters from Gentile Fusillo et al. and DB models from Bergeron, finding a median flux difference of 4% although other white dwarf models may result in a larger discrepancy. A sample of representative SEDs are shown in Figure 1 and the rest are submitted as digital content.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

View figure set (320 panels)

Tarball of figure set images

With the new infrared data presented in Section 2 and the Bergeron stellar model fluxes (Holberg & Bergeron 2006), we are able to re-assess the infrared excess for each target. Following the methods outlined in Paper i , we use both the flux excess metric and mid-infrared color excess metric to search for infrared excesses. We define the flux excess metric, χ_i, as

$$\begin{eqnarray}&&{\chi }_{{\rm{i}}}=\displaystyle \frac{{F}_{\mathrm{obs},{\rm{i}}}-{F}_{\mathrm{mod},{\rm{i}}}}{\sqrt{{\sigma }_{\mathrm{obs},{\rm{i}}}^{2}+{\sigma }_{\mathrm{mod},{\rm{i}}}^{2}}},\end{eqnarray} \tag{ 1 }$$

where the indices, i, indicate any single bandpass. F_obs,i and ${F}_{\mathrm{mod},{\rm{i}}}$ are the observed flux and model flux respectively, while σ indicates the flux error. An alternative version of the flux excess metric used in Paper i measures magnitude excess with the observed and model magnitudes as well as their errors. In our sample, there was no significant difference between the flux and magnitude excess. Many of the observed targets, including all white dwarfs in Sample E, were identified in Paper i with both unWISE magnitude excess metrics exceeding the thresholds χ_W1 > 5 and χ_W2 > 5. We calculated Spitzer χ_ch1 and χ_ch2 for 56 infrared excess candidates in Sample E and 118 white dwarfs in Samples A–E. We also measured Gemini χ_J, χ_H, and χ_K values for 70 infrared excess candidates in Sample E and 165 white dwarfs in Samples A–E with significant overlap between targets observed by Spitzer and Gemini. We show the excess Spitzer χ_ch1 and χ_ch2 values for all observed candidates in Tables B1 and B2.

In Figure 2, we show the distribution of Spitzer Ch1 and Ch2 chi values for all of the observed targets. There is a locus of points clustered around the chi values of 〈χ〉_ch1 = 0.54 ± 0.91 and 〈χ〉_ch2 = 1.05 ± 1.25, as indicated by the solid green lines. This offset from zero is not likely due to real excess but rather originating from systematics of the photometry or source confusion. As discussed in Paper i, many of the unWISE candidates are likely false positives, with the unWISE excess being the result of nearby unresolved objects. Though the Spitzer data have much higher quality, we still expect some contamination from nearby sources in our PRF fluxes, which could be contributing to the offsets. To identify targets with true flux excess, we assume the negative dispersion around the locus is due to statistical fluctuation. A Gaussian profile is fitted to a synthetic distribution created by mirroring the negative dispersion across the most populated bin. The mirrored distribution is shown by the red dashed histogram outline in Figure 2. We consider infrared excess to be statistically significant as observed by Spitzer if ${\chi }_{{\rm{i}}}\gt \left(3\sigma +\mathrm{offset}\right)$ in both IRAC Ch1 and Ch2. Therefore, the Spitzer flux excess thresholds are χ_ch1 > 3.27 and χ_ch2 > 4.80 and both conditions must be satisfied for a target to have confirmed flux excess. We found 43 targets with flux excess in Sample E and 17 targets with flux excess in Samples A–E.

Zoom In Zoom Out Reset image size

The mid-infrared single-color excess, Σ_ch1−ch2, hereafter shortened to "color excess", is defined for the two warm bandpasses Spitzer Ch1 and Ch2 as

$$\begin{eqnarray}&&{{\rm{\Sigma }}}_{\mathrm{ch}1-\mathrm{ch}2}=\displaystyle \frac{{m}_{\mathrm{obs},\mathrm{ch}1}-{m}_{\mathrm{obs},\mathrm{ch}2}-{m}_{\mathrm{mod},\mathrm{ch}1-\mathrm{ch}2}}{\sqrt{{\sigma }_{\mathrm{obs},\mathrm{ch}1}^{2}+{\sigma }_{\mathrm{obs},\mathrm{ch}2}^{2}+{\sigma }_{\mathrm{mod},\mathrm{ch}1-\mathrm{ch}2}^{2}}},\end{eqnarray} \tag{ 2 }$$

where all quantities and uncertainties are in units of magnitude. The numerator measures the difference between the observed and model color, which is normalized against their uncertainties added in quadrature. In Paper i , we considered a target to have color excess if Σ_W1−W2 > 3. For our Spitzer-observed sample, we found the mean color-excess metric 〈Σ〉_ch1−ch2 = 0.31 ± 0.59. We minimized the effect of the infrared-excess bias in our Spitzer-observed sample by removing targets where flux excess is observed in either Ch1 or Ch2 from the mean color-excess metric estimate. Using the same Σ > 3 threshold consistent with Paper i, we find 23 candidates in Sample E and 5 targets in Samples A–E with color excess, all but 2 of which also host a flux excess. As seen in Figure 2, roughly 40% of observed white dwarfs in our sample exhibiting flux excess also have color excess. We show the full table of all white dwarfs with Spitzer infrared excess and a table of all Spitzer-observed white dwarfs without infrared excess in Appendix Tables B1 and B2.

In this paper, white dwarfs are considered to have an IR excess when either the flux excess or the color excess metric exceeds the threshold. In this study, we have identified a total of 62 systems with Spitzer-confirmed infrared excess across all Samples ABCDE, 44 of which are from the unWISE-selected candidates in Sample E originally from Paper i .

We have observed 56 unWISE-selected infrared excess candidates from the final sample of 188 candidates which makes up Sample E in Paper i . With the excess metrics outlined in previous sections and the new Spitzer photometry, we find 44 targets from Sample E with either flux or color excess, of which 22 satisfy both excess thresholds. Previous studies found that over 90% of the known sample of white dwarfs with circumstellar dust disks exhibit both statistically significant flux and color excess (Wilson et al. 2019).

In Sample E, we find 21 stars with only flux excess and one star with color-only excess. We show the number of stars with infrared excess in Table 3. The confirmation rate for Spitzer excess in the 56 observed Sample E targets is 79%. We also find 7 targets with significant J-band excess, indicating that the white dwarf is likely to have a low-mass companion.

Out of the observed Sample E white dwarfs, we found four white dwarfs with flux excess in only Spitzer Ch2 bandpass, including one star with only color excess. This scenario could indicate the presence of cooler dust, similar to the case of HS 2132+0941 (Bergfors et al. 2014). For the 12 Spitzer-observed Sample E candidates without a Spitzer flag or confirmation of excess, 2 of the white dwarfs exhibited only Ch2 excess and did not satisfy either of the flux or color excess thresholds. All 12 candidates exhibited unWISE excess which is ruled out by the higher-quality Spitzer photometry. The Spitzer residuals show one or more distinct nearby sources that were captured and blended within the large beam size of unWISE and the unWISE excess was the result of source confusion. These limitations were resolved using the new Spitzer photometry. We show an example of such a case in the top left panel of Figure 1.

In Paper i , it was estimated that the false-positive rate for the final 188 infrared excess candidates in Sample E could be as high as 60% based on the expected frequency of white dwarfs with dust disks or brown dwarf companions. At first glance, our confirmation rate of 79% for the Spitzer-observed Sample E candidates is almost double the expected rate of 40%. However, recalling that the targets for the Spitzer program were chosen before the Sample criteria in Paper i was established, we can understand this difference as a selection effect. As shown in Table 4, we break down the Spitzer confirmation rate in Sample E by the unWISE excess designation (color or magnitude), which was a selection factor for our Spitzer targets. The unWISE magnitude excess in Paper i is analogous to flux excess in this study. As most known white dwarfs with dust disks exhibit both color and flux excess, we have chosen to observe the highest proportion of targets in this category. Candidates with color-only excess as indicated by the unWISE W1 and W2 bandpasses are under-observed by comparison.

Table 4 also shows that only a quarter of the targets in the color-only excess unWISE category are confirmed by Spitzer. Though only 8 color-only unWISE excess candidates were observed, these results indicate that color-only excess candidates from Paper i are much more likely to be false positives than candidates with both a color and magnitude excess. Interestingly, there is only 1 target observed in Sample E with Spitzer-confirmed color-only excess, making it considerably rarer than the proportion of candidates indicated by Paper i 's unWISE-based study. This indicates that the color of contaminants is typically redder than the white dwarf and the confused flux is responsible for the color-only excess determination, consistent with Barber et al. (2014).

Based on the new Spitzer data, the 95% Spitzer infrared excess confirmation rate in the Colour+Mag unWISE excess category indicates that the remaining 9 unobserved targets which have not been previously confirmed are the best candidates for future study. Within the same Colour+Mag unWISE category, there are 18 known white dwarfs with infrared excess which have all been shown to host circumstellar dust disks. Details of the 9 remaining unobserved targets are shown in Table 5.

As the targets for follow-up with Spitzer were chosen before the criteria for Paper i were finalized, we have also observed 118 white dwarfs with Spitzer outside of Sample E. Some of the targets in Samples A–E chosen for follow-up observation with Spitzer show evidence of excess in unWISE photometry, but were filtered from the final sample due to poor crossmatching, signs of contamination, or other reasons. In Samples A–E, 18 of the 118 white dwarfs were confirmed with Spitzer infrared excess. The confirmation rate of 15% is significantly lower than the 79% of Sample E candidates. All of the Spitzer photometry for observed targets in Samples A–E can be found in Appendix Tables B1 and B2. Here, we summarize the white dwarfs with confirmed Spitzer infrared excess in Samples A–E.

Four of the targets show both flux and color excess. One of them, GaiaJ2223-2510, is a heavily polluted DB white dwarf (Jeffery et al. 2020) previously mis-identified as a hot subdwarf (Geier et al. 2017). Another target, GaiaJ0147+2329, is a known infrared variable with a dusty debris disk also known as Gaia 0145+234 (Wang et al. 2019). Its SED is shown in the bottom right panel of Figure 1. GaiaJ1814-7355 and GaiaJ2015+5531 are two new targets with Spitzer-confirmed flux and color excess. Only one target from Samples A–E, GaiaJ1343-0453, exhibits color-only excess.

In Samples A–E, 13 targets exhibit flux-only excess and have existing near-infrared data, many of which were observed with Gemini. Three of these targets, GaiaJ0433+2827, GaiaJ1731-1002, and GaiaJ2026+5925, likely host a stellar or sub-stellar companion based on the strong near-infrared excess in the J, H, and K bands. For the remaining 10, there is no clear evidence of excess in the J, H, K bandpasses, so we are unable to conclude that their excess originates from a binary companion. As with the confirmed infrared excesses from Sample E candidates, these targets are also worthwhile for further investigation into the nature and characteristics of their apparent infrared excess.

Our Spitzer observations confirm a total of 62 white dwarfs with flux or color infrared excess, 10 of which have excess likely to be attributed to low-mass companions rather than debris disks based on the observed J-band excess. The remaining 52 bright white dwarfs with Spitzer-confirmed infrared excess have the potential to double the known sample of white dwarfs with dusty debris disks. Additional spectroscopic studies will allow for further investigations into metal contamination, atmospheric typing of the white dwarf star, and constraining stellar parameters. Among other things, low-resolution follow-up optical spectroscopy can be used to find gas emission lines from dust disks and measure white dwarf atmospheric parameters. High-resolution follow-up can measure atmospheric pollution and infrared spectroscopy will be able to distinguish spectral features of brown dwarf companions from circumstellar dust disks. As Spitzer is now decommissioned, this is one of the final large samples of white dwarfs with infrared excess confirmed by the Spitzer Space Telescope.

In Figure 4, we show a comparison of effective temperature and surface gravity between the known sample of white dwarfs with debris disks and all of the Spitzer-confirmed excess white dwarfs. The effective temperature and surface gravity are DA white dwarf model fits to the Gaia DR2 photometric data reported in Gentile Fusillo et al. (2019). Two-sided Kolmogorov–Smirnov (KS) tests show that the overall distributions between the two samples are not significantly different, as demonstrated by the high p values of 0.11 and 0.06 for the surface gravity and effective temperature respectively. Though the distributions are similar, there are candidates outside of the current parameter space occupied by known white dwarf debris disks to the increased sample size. If these Spitzer excess white dwarfs are confirmed to host dusty debris disks, they will probe a much wider range in surface gravity and effective temperatures, enabling new discoveries and increasing the range of environments where these disks can exist.

Zoom In Zoom Out Reset image size

The 62 Spitzer-confirmed infrared excess targets presented here are prime targets for further follow-up. For example, Dennihy et al. (2020b) and Melis et al. (2020) independently discovered emission from gaseous debris disks around a total of 9 white dwarfs, 8 of which were candidates considered in this study. Wang et al. (2019) identified GaiaJ0147+2329, shown in Figure 1, as a dusty disk with high infrared variability, possibly due to a tidal disruption event in progress. This shows the potential for this list of Spitzer-confirmed infrared excess white dwarfs to result in new discoveries.

In this work, all magnitudes are reported in the Maunakea Observatory (MKO) system and the appropriate filter transformations were applied. As not all fields have MKO-system photometry available, we have used 2MASS as the universal calibrator. For our standard stars, we converted the 2MASS J, H, K_s magnitudes into the MKO system using the transformations which were measured empirically for regions of low reddening, specifically Equations (6), (7), and (8) from Hodgkin et al. (2009).

For NIRI, all three of its near-infrared filters are similar to UKIDSS, which is already in the MKO system. We transformed the 2MASS photometry of reference stars according to the above color equations before measuring the target magnitudes. For F2, although J and H filters are in the MKO system, the K_s filter profile is very similar to the 2MASS K_s filter (Leggett et al. 2015). Thus, the transformation of 2MASS K_s to the MKO system is applied after measuring the target magnitude.

Comparisons between 2MASS and UKIDSS photometry show a systematic deviation in the K filter wavelength regime when the targets are fainter than roughly 15.5 magnitude due to a sensitivity limit in 2MASS (Skrutskie et al. 2006). For this reason, we choose reference stars with K_s magnitude brighter than 15.5 mag unless there are fewer than two reference stars within frame. The same reference stars used to perform photometry in the K bandpass are reused for both the J and H bandpasses whenever possible.

Further comparisons between our 2MASS-calibrated Gemini photometry against existing near-infrared photometry from UKIRT's WFCAM and VISTA show some systematic linear offset. Correcting for this offset was found to improve the reliability of the Gemini photometry in fitting with white dwarf photospheric models. We show this offset by comparing the uncorrected Gemini photometry with a sample of white dwarfs that have existing survey photometry (UKIRT/VISTA) in Figure A1. The linear offsets for NIRI J and F2 J were found to be −0.12 ± 0.04 mag and −0.05 ± 0.01 mag respectively when compared against WFCAM J. The linear offset for K_s was 0.15 ± 0.03 mag when compared against VISTA K_s. All offsets were calculated with a weighted average where higher weight was applied for brighter stars with lower uncertainty. The VISTA magnitudes were converted into the MKO system according to the literature, specifically Equations (16), 17, and 18 from González-Fernández et al. (2018). For the remaining bandpasses, NIRI H, NIRI K, and F2 H, there was not enough existing photometry to determine robust statistical offsets and what little data there was did not show significant discrepancy between the Gemini and WFCAM or VISTA photometry. Therefore, no linear offsets were applied to NIRI H, NIRI K, and F2 H.

Zoom In Zoom Out Reset image size

There are also no significant trends in the zero-point difference with calibrator brightness or color, suggesting the photometric transformation and linearity correction is acceptable. However, the systematic offset suggests there are unrecognised errors at the ≲10% level in the 2MASS to MKO photometric transformation and/or in the NIRI/F2 linearity correction (the 2MASS stars are brighter). The 2MASS photometry is more uncertain than the survey photometry, and the linearity correction and system transformation is larger between 2MASS and raw data, therefore we apply this offset to our measurements to put them onto the MKO system.

This section discusses a variety of factors which affect photometric data quality and evaluates how each factor affects the result. We will discuss observing conditions, image quality, and misalignment of coordinates in the output image of our observations with both NIRI and F2.

Of the 16 programs observed using Gemini's North and South near-infrared imagers, eight of the largest programs were poor weather programs, characterized by observing condition constraints of Cloud Cover (CC) in the 70th, 80th, or "Any" percentiles. Each percentile corresponds to a percentage of time with a certain transparency based on long term data for Maunakea. Higher percentiles indicate the potential for more cloud coverage, and thus a greater loss of signal. Poor weather proposals also do not place any restriction on the desired Image Quality (IQ) and Water Vapour (WV) content. For the 215 targets observed by either Gemini's NIRI or F2 instrument, 150 were observed in poor weather conditions, split between 69% (80/116) of NIRI observations and 88% (111/126) of F2 observations. The majority of targets were observed in photometric conditions of IQ85 or better, indicating a FWHM of less than 0.85 arcseconds in the J bandpass.

In Figure A2, all of the measured aperture photometry obtained from our algorithm is compared against the 2MASS photometry of the reference stars. The figure does not show an appreciable systematic deviation with brightness between our photometry with 2MASS in the J, H, K filters for either NIRI or F2 instruments. We find that although the 2MASS precision decreases for fainter stars, the primary contributing factor to the Gemini photometric precision is the number of reference stars used to calculate the Gemini photometry. Furthermore, even though the majority of the targets were observed under poor weather conditions, we find that the standard deviation in the difference between 2MASS and our photometry to be ≲ 0.1 mag. Since the field of view of F2 is 6' × 6' circular field compared against the 120'' × 120'' square field of NIRI, there are often more reference stars in F2 which improves both the accuracy and precision of the measured magnitudes. We conclude that the measured Gemini near-infrared photometry is reliable even if targets are observed under poor weather conditions as long as there are a good number of bright 2MASS reference stars within the field of view.

Zoom In Zoom Out Reset image size

The image quality (IQ) at Gemini Observatory is a percentile defined for a target at zenith with a profile FWHM below a wavelength-dependent maximum threshold. The percentile following the IQ represents the percentage of time when the FWHM is below a defined threshold and is linked to astronomical seeing. ⁸ Point sources of the reference stars and our target were modeled using astropy's photutils, where the aperture radius was designated as three times the median of the FWHM among all stars detected in the frame determined by the PSF fitting. Figure A3 shows that the median aperture radius in each IQ category increases with the deteriorating quality, independently recovering the desired result without directly referencing the observed IQ of each frame.

Zoom In Zoom Out Reset image size

The World Coordinate System (WCS) of the output images for both NIRI and F2 can be offset from the true WCS when compared against other public surveys. In all of the observations performed in this study, a unique linear correction applied to each individual image was sufficient to correct for the WCS offset compared against 2MASS. Some images show non-linear warping, but the effects are negligible compared to the linear offset. We find that the magnitude of the median offset for NIRI is typically 11, split into −106 in the X direction and −033 in the Y direction, while the median F2 offset is 89, split into 547 in the X direction and 707 in the Y direction.

We also assess how the photometric zero point determined from 2MASS reference stars is dependent on the observation conditions. Table A1 shows the median zero points in magnitude for each filter in the two Gemini instruments. For the set of observations performed in this study, the measured zero points are roughly comparable between CC50 and CC70, but decrease sharply with CC80, consistent with a greater loss of signal under those observing conditions.