Col-OSSOS: z-Band Photometry Reveals Three Distinct TNO Surface Types

Trans-Neptunian objects (TNOs) in the outer solar system exhibit a broad range of surface properties. The vast majority of TNOs are too faint for spectroscopic studies, so broad band surface reflectance is used to provide constraints on surface composition. TNOs, in general, have red optical colors in g-, r-, and i-bands; even the "neutral" TNO surfaces, sometimes referred to as "blue" in the literature, are slightly redder than solar. It is well accepted that the small dynamically excited TNOs and Centaurs exhibit a bimodal color distribution, with red and neutral classes (e.g., Tegler & Romanishin 1998; Peixinho et al. 2003, 2012, 2015; Tegler et al. 2003, 2016; Fraser & Brown 2012; Wong & Brown 2017).

Color surveys of TNOs in more than two bands have revealed additional complexities and correlations in the surface reflectance of these objects. Observations in i-band correlate strongly with g and r (Ofek 2012; Sheppard 2012), suggesting that across the g to i wavelength range, the same spectral feature is being probed. The dynamically excited TNOs also show correlations between the optical and near-infrared colors (Fraser & Brown 2012). The observed correlations in color have revealed different surface classes, though the exact number of classes is debated in the literature (Barucci et al. 2005; Fraser & Brown 2012; Dalle Ore et al. 2013).

TNOs are often subdivided based on their dynamical classifications into cold classical TNOs and dynamically excited objects, including scattering, detached, hot classical, and resonant TNOs (Brown 2001; Gladman et al. 2008). A range of criteria are used to identify cold classical TNOs in the literature; small eccentricities, semimajor axes between the 3:2 and 2:1 Neptune resonances, and an inclination cut at 4°–7° typically identify cold classical objects with minimal contamination from the hot population. Most cold classical TNOs exhibit similar red colors to the red excited TNOs, with the color distributions of both classes occupying similar ranges in B − R, g − r, and in the NIR (e.g., Tegler et al. 2003; Fraser & Brown 2012; M. Schwamb et al. 2017, in preparation). The red cold classicals, however, exhibit higher albedos (Brucker et al. 2009; Vilenius et al. 2014) than the dynamically excited red objects, implying that they occupy a different compositional class.

A number of different TNO taxonomies have been proposed. Barucci et al. (2005) use the colors derived from BVRIJ photometry to classify TNOs into four different classes using the G-mode statistical analysis method for classifying asteroids. A different technique that utilized a modified K-means clustering technique applied to multi-band optical photometry and optical albedos found 10 surface types (Dalle Ore et al. 2013). Fraser & Brown (2012) and Fraser et al. (2015) found that the small dynamically excited TNOs fall into only two surface classes, which exhibit a range of optical colors. The true number of TNO compositional classes remains an open question.

Extending photometric surveys of TNOs into the previously unexplored z-band provides new insight into TNO surface classifications. In part due to detector sensitivity, z-band photometry has not been utilized as a tool for probing TNO surface types. This wavelength range may be sensitive to the presence or absence of organics and silicates on minor planet surfaces. Here we present g-, r-, and z-band photometry of 26 TNOs. We find three distinct TNO surface types that result from classifying based on these surface colors.

The targets for this work are TNOs found by two large surveys. 23 targets are from the Outer Solar System Origins Survey (OSSOS, Bannister et al. 2016), a large TNO discovery survey executed on the Canada–France–Hawaii Telescope from 2013 to 2017. Three additional targets, the full known sample of 5:1 resonators (Pike et al. 2015), are included from the Canada–France Ecliptic Plane Survey (CFEPS, Petit et al. 2011, 2017). All 26 targets from these two surveys, listed in Table 1, are from a flux-limited sub-sample of these two surveys, with $r\lt 23.6$.

3.1. Observations

3.2. Sloan Digital Sky Survey (SDSS) Color Calibration

In order to compare the colors acquired using different bandpass filters on different facilities, it is necessary to properly characterize the flux measurement of each telescope and scale to a common system. The Col-OSSOS data are scaled to the Sloan Digital Sky Survey (SDSS) Release 13 (SDSS Collaboration et al. 2016) magnitudes from photometry of in-frame SDSS catalog stars. Those SDSS stars were then used to determine the color transform between the SDSS and Gemini filter sets. From the GMOS photometry, color terms between the GMOS (r_G, g_G, z_G) and SDSS (r_SDSS, g_SDSS, z_SDSS) systems were determined to be:

$$\begin{eqnarray}&&{r}_{{\rm{G}}}={r}_{\mathrm{SDSS}}-(0.060\pm 0.03)\times ({g}_{\mathrm{SDSS}}-{r}_{\mathrm{SDSS}})\end{eqnarray} \tag{ 1 }$$

$$\begin{eqnarray}&&{g}_{{\rm{G}}}={g}_{\mathrm{SDSS}}-(0.140\pm 0.03)\times ({g}_{\mathrm{SDSS}}-{r}_{\mathrm{SDSS}})\end{eqnarray} \tag{ 2 }$$

$$\begin{eqnarray}&&{z}_{{\rm{G}}}={z}_{\mathrm{SDSS}}-(0.027\pm 0.018)\times ({r}_{\mathrm{SDSS}}-{z}_{\mathrm{SDSS}}).\end{eqnarray} \tag{ 3 }$$

Similar techniques were utilized to measure SDSS-Subaru color terms (Rc_S, z_S, i_S), and calibrate the Subaru observations. The color terms were determined to be:

$$\begin{eqnarray}&&{{Rc}}_{{\rm{S}}}={{Rc}}_{\mathrm{SDSS}}-(0.044\pm 0.023)\times ({g}_{\mathrm{SDSS}}-{r}_{\mathrm{SDSS}})\end{eqnarray} \tag{ 4 }$$

$$\begin{eqnarray}&&{i}_{{\rm{S}}}={i}_{\mathrm{SDSS}}-(0.096\pm 0.019)\times ({g}_{\mathrm{SDSS}}-{r}_{\mathrm{SDSS}})\end{eqnarray} \tag{ 5 }$$

$$\begin{eqnarray}&&{z}_{{\rm{S}}}={z}_{\mathrm{SDSS}}-(0.077\pm 0.034)\times ({g}_{\mathrm{SDSS}}-{r}_{\mathrm{SDSS}}).\end{eqnarray} \tag{ 6 }$$

An additional color term was used to convert Rc_SDSS to r_SDSS based on the multi-band photometry of SDSS sources on frame (Jordi et al. 2006).

$$\begin{eqnarray}{{Rc}}_{\mathrm{SDSS}}-{r}_{\mathrm{SDSS}} & = & (-0.153\pm 0.003)\times ({r}_{\mathrm{SDSS}}-{i}_{\mathrm{SDSS}})\\ & & -(0.117\pm 0.003)\end{eqnarray} \tag{ 7 }$$

Photometry from Subaru and Gemini, converted to the SDSS system, are presented in Tables 2 and 3, respectively.

Table 3.  TNO Photometry Sequences of the 5:1 Resonators from Gemini

Survey	MPC	Filter	Magnitude	MJD	Exposure Time	Magnitude	Magnitude
ID	ID		(Gemini)		(s)	(Gemini)	(SDSS)
L3y02	2003 YQ179	r	22.91 ± 0.02	57308.63136	300
L3y02	2003 YQ179	r				22.91 ± 0.02	22.95 ± 0.02
L3y02	2003 YQ179	g	23.61 ± 0.04	57308.63533	300
L3y02	2003 YQ179	g				23.61 ± 0.04	23.72 ± 0.04
L3y02	2003 YQ179	z	22.50 ± 0.05	57308.63933	300
L3y02	2003 YQ179	z				22.50 ± 0.05	22.51 ± 0.05
HL7j4	2007 LF38	r	23.12 ± 0.02	57435.63929	200
HL7j4	2007 LF38	r	23.10 ± 0.02	57435.65007	200
HL7j4	2007 LF38	r				23.11 ± 0.01	23.14 ± 0.02
HL7j4	2007 LF38	g	23.59 ± 0.03	57435.64328	300
HL7j4	2007 LF38	g				23.59 ± 0.03	23.67 ± 0.03
HL7j4	2007 LF38	z	22.86 ± 0.04	57435.64728	300
HL7j4	2007 LF38	z				22.86 ± 0.04	22.87 ± 0.04
HL7c1	2007 FN51	r	23.69 ± 0.03	57432.52142	300
HL7c1	2007 FN51	r	23.68 ± 0.03	57432.55252	300
HL7c1	2007 FN51	r				23.68 ± 0.05	23.72 ± 0.05
HL7c1	2007 FN51	g	24.35 ± 0.03	57432.50984	300
HL7c1	2007 FN51	g	24.25 ± 0.03	57432.51365	300
HL7c1	2007 FN51	g	24.33 ± 0.03	57432.51746	300
HL7c1	2007 FN51	g	24.33 ± 0.03	57432.53317	300
HL7c1	2007 FN51	g	24.42 ± 0.03	57432.53697	300
HL7c1	2007 FN51	g	24.39 ± 0.04	57432.54078	300
HL7c1	2007 FN51	g	24.37 ± 0.04	57432.55649	300
HL7c1	2007 FN51	g	24.29 ± 0.04	57432.56030	300
HL7c1	2007 FN51	g	24.41 ± 0.05	57432.56410	300
HL7c1	2007 FN51	g				24.35 ± 0.05	24.45 ± 0.05
HL7c1	2007 FN51	z	23.06 ± 0.04	57432.52536	300
HL7c1	2007 FN51	z	23.11 ± 0.04	57432.52916	300
HL7c1	2007 FN51	z	23.25 ± 0.05	57432.54477	300
HL7c1	2007 FN51	z	23.41 ± 0.07	57432.54858	300
HL7c1	2007 FN51	z				23.18 ± 0.05	23.19 ± 0.05

Download table as:  ASCIITypeset image

3.3. Determining Colors from TNO Photometry

In Figure 1, we present the measured colors of the TNOs in g − r and r − z. The z-band measurements increase the distinction between different surface types suggested by the g- and r-band measurements for the TNOs. The r − z colors of the objects span a large range of values, ∼0.7 mag. In Figure 1, the r − z versus g − r structure indicating different surface reflectivity becomes apparent. The commonly reported g − r bi-modality dividing surface color types at $g-r\sim 0.75$ noted in previous studies (e.g., Peixinho et al. 2015) is indicated in the histogram. Given the size of our sample, unsurprisingly, we do not find statistically significant evidence of bi-modality. As this topic is discussed quite thoroughly in other works with samples more appropriate to test the existence of bi-modality, we adopt the accepted conclusion that the excited populations possess two compositional classes as evidenced by their bimodal optical colors (Tegler & Romanishin 1998; Peixinho et al. 2003, 2012, 2015; Tegler et al. 2003, 2016; Doressoundiram et al. 2007; Fraser & Brown 2012; Wong & Brown 2017). However, Figure 1 indicates that dividing these objects requires more than a simple optical color cut; we use a surface model based on Fraser & Brown (2012) to describe the color range occupied by two dynamically excited TNOs compositional classes. Because of their albedos, cold classical objects are expected to have different surface properties (Brucker et al. 2009; Lacerda et al. 2014; Vilenius et al. 2014). The cold classical objects form a third surface type identifiable in g − r and r − z.

Zoom In Zoom Out Reset image size

The dynamically excited TNOs have both a neutral and a red surface group in g − r. The "neutral" objects in g − r (g − r ≲ 0.75) show a roughly linear trend of increasing r − z colors with increasing g − r colors. The Spearman rank test (Spearman 1904) finds that these colors are correlated (ρ = 0.82) with 3σ significance. The red dynamically excited TNOs have a different trend of increasing r − z colors with increasing g − r colors at 2σ significance, but with a shallower slope (ρ = 0.72). With g − r colors alone, it is unclear if the dynamically excited TNOs with $0.6\lesssim g-r\lesssim 0.8$ belong to the red or neutral surface groups. Due to the clear correlation of g − r and r − z, however, the difference in surface colors of these two classes becomes obvious; those objects with red r − z are neutral class members.

A model of the two types of dynamically excited TNO surfaces is presented in Figure 1. We modeled the approximate range of color occupied by a TNO surface class after the approach of Fraser & Brown (2012): a simple Hapke surface model (Hapke 2002) defines the overall reflectivity of a mix of two materials with different surface reflectivity. The Fraser & Brown (2012) geometric mixing model uses different reflectivity of two materials in each of the filters; these materials combine in varying amounts to reproduce the possible range of color occupied by a TNO surface class. Only three materials, one common between the two populations, are all that are necessary to reproduce the colors of the two surface types. The precise surface reflectivity used in the models is not informative because only the relative reflectivity in the different bands affects the material's color; as a result, we describe only the comparative reflectivity as in Fraser & Brown (2012). The Hubble Space Telescope Wide Field Camera 3 filters used in the modeling by Fraser & Brown (2012) are sufficiently different from the filter selection here that this new data set provides independent confirmation of the published models.

We calculated model reflectivities for the three surface components that well represent the g − r and r − z surface colors of our TNO sample. Similar to the Fraser & Brown (2012) model, we find that the red and neutral surface types can be approximately accounted for by two different red-end components, but with a common neutral-end component. Our data imply that this neutral component is a roughly neutral reflector through grz, though it could be slightly less reflective in z. The z-band reflectivity was not explored in Fraser & Brown (2012), but this lower z-band reflectance is consistent with their speculation that the neutral material is silicates. Our two surface models have different red components, which combine with the neutral/blue component in different ratios to produce the range of colors of each of the two dynamically excited spectral classes. The models account for the range and distribution of grz colors of the dynamically excited objects quite effectively.

The cold classical TNOs occupy a different range of g − r and r − z space than do the dynamically excited objects; they are red in g − r and less reflective in z-band. The objects identified in Figure 1 as cold classical TNOs were selected from the classical sub-population based on pericenters >38 au, semimajor axes 39.4–48.0 au, and inclinations <6°. Photometry in the g-, r-, and z-bands demonstrates a clear difference between the red cold classical surfaces and dynamically excited TNO surfaces. The Spearman rank test finds that the r − z and g − r colors of red cold classical objects do not show a statistically significant correlation; this may be due to the small sample size or the underlying color distribution. The cold classicals also include 2013 UL₁₅, which shows clear variation in r − z between the two observing epochs; though the color variation is large, both measurements fall within the surface colors of cold classicals. (Although TNOs typically do not display variability in g − r, variability has been identified beyond ∼0.9 μm (Fraser et al. 2015).) As it is unclear if the cold classicals have correlated colors or if they clump in grz space, we do not model these surfaces using a geometric mixing model. In our sample of cold classicals, the only exceptions to the unique red cold classical surfaces are those two objects identified as belonging to the recently identified class of blue binary objects on cold classical orbits (Fraser et al. 2017). These blue binary TNOs have colors consistent with the dynamically excited population in all bands studied here: g-, r-, and z-bands, which supports the theory that these objects formed inward of their current location and were pushed outward and hence share a primordial origin with the dynamically excited TNOs (Fraser et al. 2017).

Previous work by Ofek (2012) provides further evidence that the red cold classical objects occupy a distinct region of g − r versus r − z color space. The photometry of TNOs from the SDSS presented by Ofek (2012) in which g, r, and z photometry are available is presented in Figure 2. Due to the depth of the SDSS, this sample is entirely dynamically excited TNOs. The larger uncertainties in the photometry and the non-simultaneous color measurements obscure the distinction between the two dynamically excited surface classes, but the range of colors is representative of the dynamically excited objects as a whole. The surface reflectance of the dynamically excited TNOs measured in Ofek (2012) do not extend into the region we have identified as cold classical TNO surfaces, confirming that these surface colors are unique to cold classical objects.

Zoom In Zoom Out Reset image size

We verify the unique surface properties of cold classical objects by using the Kolmogorov–Smirnov (KS) test (Peacock 1983). We compared the sample of dynamically excited objects to the dynamically cold objects using both the one and two-dimensional KS test. To ensure a fair comparison, we only consider objects with $g-r\gt 0.85$, the color range occupied by the dynamically selected sample of cold classical objects. Bootstrapping random simulated samples was used to calibrate both the 1D and 2D KS test results. From the 2D KS test applied to the g − r and r − z colors, we find that the probability that the red dynamically hot and cold TNOs share the same parent distribution is only 2%. Similarly, the 1D KS test on the r − z distribution reports only a 1% probability that the two samples share the same parent color distribution. It appears that the cold classical objects possess a different grz color distribution than do the dynamically excited objects. Our findings support the assertion drawn from the high albedos of cold classicals (Brucker et al. 2009) and the analysis of non-simultaneous colors of TNOs in the Hyper Suprime-Cam Subaru Strategic Program, which revealed indications of distinct cold classical surfaces (Terai et al. 2017); the cold classical objects possess a unique surface type.

In the wavelengths studied in Fraser & Brown (2012), the red cold classical colors overlapped significantly with the red dynamically excited surfaces, as is the case in g − r and g − i in other previous studies. The grz surface reflectance of red cold classicals and red excited objects provides a new diagnostic for identifying interlopers in the cold classical region and red cold classical surfaces elsewhere in the Kuiper Belt. In grz, the red cold classical objects clearly exhibit a different compositional class than the excited TNOs, in agreement with the high albedos exhibited by these objects compared to the duller excited TNOs (Brucker et al. 2009; Vilenius et al. 2014).

The g-, r-, and z-band photometry show that these TNOs have three surface types. The TNO colors are consistent with the known bi-modality in g − r (e.g., Tegler & Romanishin 1998; Peixinho et al. 2003, 2012, 2015; Tegler et al. 2003, 2016; Fraser & Brown 2012; Wong & Brown 2017). The addition of the z-band measurement makes it possible to separate TNOs where the g − r surface groups overlap and determine which surface group the TNOs belong to: neutral excited, red excited, or red cold classical. The red cold classical surfaces occupy a distinct region of the grz color space, and the two neutral cold classicals (blue binaries) are consistent with the dynamically excited objects. The neutral and red excited TNOs show two different correlated slopes between the g − r and r − z colors. This reddening is well represented by the two-component geometric composition model (Fraser & Brown 2012).

The variation in z-band color is indicative of TNO surface properties, and several materials are speculated to be present on TNO surfaces. "Tholin" is an organic compound that has been reddened through irradiation (Roush & Dalton 2004); a material of this type is typically thought to be responsible for the red spectral slopes of TNOs in g-, r-, and i-band, which should extend with the same slope through the z-band as well. However, if TNO surfaces include contributions from an iron-rich material, such as olivine or pyroxene (Clark et al. 2007), these materials are less reflective in the z-band. The inclusion of a silicate material in the mixing model, such olivine or pyroxene material, would result in a range of z-band reflectivity. A silicate component for TNO composition was a good match for the neutral component of the TNO surface models from Fraser & Brown (2012), and this is better demonstrated in the models in Figure 1 where the neutral component in g and r has a reduced reflectance in the z-band. We speculate that cold classicals have surfaces richer in silicates than the red excited objects or perhaps a different surface silicate material. A larger sample of precise multi-band photometry or spectroscopy searching for silicate absorption could further constrain the TNO surface components.

The three 5:1 resonators in our sample have a range of surface colors in g − r and r − z. Two of the objects are consistent with the surface model of red excited objects, and the third is consistent with a neutral excited TNO surface; none of the 5:1 resonators resemble the cold classical object surfaces. The known 5:1 resonators have surfaces consistent with the dynamically excited population, which implies the dynamically excited and distant resonant objects share the same source population. Dynamically excited populations display a range of g − r surface colors, seen here and in previous work (e.g., Tegler & Romanishin 1998). This range of surface colors may have resulted from formation in different locations closer to the Sun (Brown et al. 2012), followed by scattering into the outer solar system. Pike et al. (2015) speculates that the 5:1 objects are captured from the scattering objects, and the range of 5:1 resonator surface colors is consistent with capture from the dynamically excited scattering object colors.

We find that z-band photometry provides a powerful tool to more precisely discriminate between three different surface groups and clearly identifies the red cold classical TNO surfaces as unique in the Kuiper Belt. These data show that when TNO colors overlap in g − r, the z-band can be used to effectively divide the TNOs into three surface classifications: red cold classical TNOs, dynamically excited red TNOs, and dynamically excited neutral TNOs. TNOs are sufficiently bright in the z-band for this measurement to be a reasonable addition to a TNO color survey. Expanding the use of z-band photometry would provide a useful tool for tracing the dynamical history of the region, as it enables the identification of cold classical surfaces outside of the classical belt as well as the identification of hot classical object interlopers on cold classical orbits.

This research was based on observations obtained at the Gemini Observatory, acquired through the Gemini Observatory Archive, and processed using the Gemini IRAF package, which is operated by the Association of Universities for Research in Astronomy, Inc., under a cooperative agreement with the NSF on behalf of the Gemini partnership: the National Science Foundation (United States), the National Research Council (Canada), CONICYT (Chile), Ministerio de Ciencia, Tecnología e Innovación Productiva (Argentina), and Ministério da Ciência, Tecnologia e Inovação (Brazil). This research was also based on data collected at Subaru Telescope, which is operated by the National Astronomical Observatory of Japan. The authors recognize and acknowledge the very significant cultural role and reverence that the summit of Maunakea has always had within the indigenous Hawaiian community. We are fortunate to have the opportunity to conduct observations from this mountain. M.B. acknowledges support from UK STFC grant ST/L000709/1. M.E.S. was supported by Gemini Observatory.

Software: Astropy v-1.3 TRIPPy v-0.5.1 SciPy v-0.17.1 Matplotlib v-1.5.1.