Thermal Emission from the Uranian Ring System
Edward M. Molter, Imke de Pater, Michael T. Roman, Leigh N. Fletcher
DDraft version May 30, 2019
Typeset using L A TEX preprint style in AASTeX61
THERMAL EMISSION FROM THE URANIAN RING SYSTEM
Edward M. Molter, Imke de Pater, Michael T. Roman, and Leigh N. Fletcher Astronomy Department, University of California, Berkeley; Berkeley CA, 94720, USA Department of Physics & Astronomy, University of Leicester, University Road, Leicester, LE1 7RH, UK
ABSTRACTThe narrow main rings of Uranus are composed of almost exclusively centimeter- to meter-sizedparticles, with a very small or nonexistent dust component; however, the filling factor, composition,thickness, mass, and detailed particle size distribution of these rings remain poorly constrained.Using millimeter (1.3 - 3.1 mm) imaging from the Atacama Large (sub-)Millimeter Array and mid-infrared (18.7 µ m) imaging from the Very Large Telescope VISIR instrument, we observed the thermalcomponent of the Uranian ring system for the first time. The (cid:15) ring is detected strongly and canbe seen by eye in the images; the other main rings are visible in a radial (azimuthally-averaged)profile at millimeter wavelengths. A simple thermal model similar to the NEATM model of near-Earth asteroids is applied to the (cid:15) ring to determine a ring particle temperature of 77 . ± . (cid:15) ring displays a factor of 2-3 brightness difference betweenperiapsis and apoapsis, with 49 . ± .
2% of sightlines through the ring striking a particle. Theseobservations are consistent with optical and near-infrared reflected light observations, confirming thehypothesis that micron-sized dust is not present in the ring system.
Corresponding author: Edward [email protected] a r X i v : . [ a s t r o - ph . E P ] M a y INTRODUCTIONTo date, observational data for the Uranian ring system were obtained from a combination of Earth-based stellar occultation measurements (French et al. 1991, and references within), visible-light andradio occultation data from Voyager 2 (e.g., Smith et al. 1986; Tyler et al. 1986; Gresh et al. 1989),visible-light HST (e.g., Karkoschka 2001a), and ground-based near-IR observations with adaptive op-tics (e.g., de Pater et al. 2006, 2007, 2013; de Kleer et al. 2013). Together, these observations revealeda complex system of ten narrow rings, three broad dusty rings, and at least thirteen associated smallsatellites (Smith et al. 1986; Karkoschka 2001b; Showalter & Lissauer 2006). The broad spectralcoverage of these observations permitted inferences about the particle size distribution of the rings,showing that the narrow rings (except the λ ring) are composed primarily of centimeter- to meter-sized particles, with a very small or nonexistent dust component (Gresh et al. 1989; Karkoschka2001c). The (cid:15) ring, the brightest and most massive of the narrow rings, was shown to maintainan appreciable eccentricity ( e = 0 . ∼ µ m) onthe Very Large Telescope (VLT). In contrast to images at visible/near-infrared wavelengths, whichshow the rings in reflected sunlight, our ALMA and VISIR images detect thermal emission from therings, a component that has never before been imaged. Any contribution from scattered light fromUranus or the Sun is < (cid:15) ring in Section 4. Finally, we present the measured azimuthal structure of the (cid:15) ring’s thermalcomponent in 5 before summarizing our results in Section 6. OBSERVATIONS AND DATA REDUCTIONWe observed the Uranian ring system at millimeter wavelengths with the Atacama Large (sub-)Millimeter Array (ALMA) and at mid-infrared wavelengths with the VISIR instrument on the VLTbetween December 2017 and September 2018. An observing log is shown in Table 1, and descriptionsof our data processing procedures are given in Sections 2.1 and 2.2 for the ALMA and VLT data,respectively. Images of the ring system at all four observed wavelengths are shown in Figure 1; eachclearly detects the massive (cid:15) ring. 2.1.
ALMA Data
The data in each of the three bands were flagged and calibrated by the North American ALMAScience Center using the standard data reduction procedures contained in the NRAO’s CASA soft-ware version 5.1.1. Standard flux and phase calibration procedures were carried out by applyingthe pipeline using the quasars listed in Table 1 as calibrator sources. The CASA pipeline retrievedflux calibration errors of 5.0%, 5.4%, and 5.3% at 3.1 mm, 2.1 mm, and 1.3 mm, respectively. Theflux calibration was sanity-checked by comparing the measured total flux of Uranus to previous
Figure 1.
Images of the Uranian ring system at 3.1 mm (ALMA Band 3; 97.5 GHz), 2.1 mm (ALMA Band4; 144 GHz), 1.3 mm (ALMA Band 6; 233 GHz), and 18.7 µ m (VLT VISIR; 100 THz). The synthesizedbeams of the ALMA images are shown as grey ellipses in the bottom left corner of each image, and an imageof a point source is shown in the bottom left corner of the VLT 18.7 µ m image. The planet itself is maskedsince it is very bright compared to the rings. UT Date & On-Source Minimum Maximum Flux & Gain PhaseWavelength Start Time Time (min) Baseline (m) Baseline (m) Calibrator Calibrator3.1 mm 2017-12-03 23:52 42 41.4 5200 J0238+1636 J0121+11493.1 mm 2017-12-06 23:42 42 41.4 3600 J0238+1636 J0121+11492.1 mm 2017-12-27 22:56 22 15.1 2500 J0238+1636 J0121+11491.3 mm 2018-09-13 07:02 24 15.1 1200 J0237+2848 J0211+105118.72 µ m 2018-09-04 08:14 42.3 — — HD009692 & HD013596 —18.72 µ m 2018-10-03 05:18 42.3 — — HD008498 & HD010380 —18.72 µ m 2018-10-13 05:28 42.3 — — HD011353 & HD040808 — Table 1.
Table of observations. millimeter-wavelength observations; the result of this exercise is shown in Figure 2. To improve thequality of the map, iterative phase-only self-calibration was performed using a procedure similar tothat outlined in Brogan et al. (2018) using solution intervals of 20, 10, 5, and 1 minutes in thatorder. To reduce ringing in the image plane from the presence of a bright planet with sharp edges, auniform limb-darkened disk model of Uranus was subtracted from the data in the UV plane, as donein e.g. de Pater et al. (2014, 2016). The disk-subtracted data were inverted into the image plane anddeconvolved using CASA’s tclean function. The ring geometry at the time of observation in eachband, taken from JPL Horizons and the Planetary Ring Node, are given in Table 2.2.2. VLT Data
The 18.72 µ m (Q2 filter) VISIR image combined observations from three nights in September andOctober 2018; details of these observations are listed in Table 1. Data from each night were reducedwith standard infrared chopping and nodding techniques using the European Southern Observatory https://ssd.jpl.nasa.gov/horizons.cgi https://pds-rings.seti.org/ Figure 2.
ALMA-derived (red; this paper) and literature (black; Gulkis & de Pater 1984; de Pater & Gulkis1988) measurements of the average millimeter- and radio-wavelength brightness temperature of Uranus’sdisk.Wavelength d (AU) B obs (degrees) Ω (degrees) w (degrees)3.1 mm 19.2 36.8 192 337 - 3412.1 mm 19.6 36.4 192 101.3 mm 19.1 43.7 190 318.72 µ m 18.9-19.2 42.7 - 43.9 190 351 - 44 Table 2.
Observing geometry of the Uranian ring system at the time of our observations. Ω denotes thelongitude of the ring plane ascending node, and w denotes the argument of periapsis of the (cid:15) ring. Rangesof values are shown for the 3.1 mm and 18.7 µ m observations because they were taken over the course ofseveral nights. VISIR pipeline, and resulting images were flux calibrated via comparison to observed standard starsusing custom-written
IDL routines. The three calibrated images were then combined and weightedby the inverse of their errors squared to yield the final, absolutely calibrated mean image. Randomerrors were estimated from the standard deviation of the background sky, and we estimate a 20%systematic error in radiance due to the uncertainty in the stellar flux and sky subtraction (Dobrzycka& Vanzi 2008). RADIAL PROFILES AND TOTAL FLUX MEASUREMENTSWe measured the radial profile of the rings in each ALMA image by integrating the rings in azimuthas follows. First, the images were divided into elliptical annuli using the astropy -affiliated photutils
Python package (Bradley et al. 2019). Each annulus was centered at the center of Uranus and givenan eccentricity and angle of rotation such that it mimicked a circular annulus projected into Uranus’sring plane. Each annulus was given a ring-plane width of 500 km, and the ring-plane radius of theannulus inner edge was varied in steps of 500 km from 35000 km to 64500 km. The planet Uranuswas masked out to a distance equal to its mean radius plus three times the ALMA full beam widthat half power in each observing band. The photutils package was used to compute the geometricfraction of each pixel contained within a given annulus, weight the flux within that pixel accordingto that fraction, then sum up the weighted flux of all unmasked pixels. This provided a total fluxmeasurement within the exact elliptical region specified by each annulus (ignoring any masked pixels).The result was a radial flux profile of the ring system in which each unmasked pixel was countedexactly once total, splitting its flux between two annuli if it fell on an annulus edge. The error σ n oneach flux measurement was taken to be the RMS noise in the image (computed far from the planetand rings) times the area of the annulus (not including any masked pixels) in units of the synthesizedbeam area. The observed profile is shown in Figure 3.The radial flux profile was modeled as the sum of flux contributions from the 6/5/4, α / β , η / γ / δ ,and (cid:15) ring groups. Each of the narrow main rings (6, 5, 4, α , β , η , γ , δ , (cid:15) ) was projected sepa-rately (i.e. one 2-D model per ring, ignoring all other rings) onto a very high resolution (0.01 (cid:48)(cid:48) /pixel)2-D grid with proper semimajor axis, eccentricity, inclination, and observing geometry using the PyAstronomy package in
Python . These super-resolution model rings were convolved with the Gaus-sian synthesized beam of ALMA, then summed into a radial profile in the same way as the data (seeprevious paragraph). The result of this exercise was a 1-D pseudo-Gaussian profile of each ring oneach observing date that accounted for its eccentricity. These ring profiles were weighted accordingto their equivalent widths at visible wavelengths as measured by stellar occultations (French et al.1986), added together, and then scaled by eye to match the data to produce an a-priori model radialprofile of the ring system. A suite of models was then produced in a Markov Chain Monte Carlo(MCMC) framework implemented using the Python emcee package (Foreman-Mackey et al. 2013),wherein the fluxes F , F αβ , F ηγδ , and F (cid:15) of the four ring groups were allowed to vary freely whilethe relative contribution from each ring within a group was set according to its visible-wavelengthequivalent width (French et al. 1986). The flux of the 1-D model profile F m at all radii r was thusfully specified by the fluxes of the four ring groups. Letting θ = [ F , F αβ , F ηγδ , F (cid:15) ], the likelihoodfunction ln p was given byln p ( F | r, σ, θ ) = −
12 Σ n (cid:104) ( F n − F m ( θ )) σ − n + ln( σ − n ) (cid:105) (1)where σ n was the variance of the measured fluxes at each radius F n . The MCMC simulation producedposterior probability distributions of F , F αβ , F ηγδ , and F (cid:15) , which are given in Appendix B FiguresA1, A2, and A3 for images at 3.1 mm, 2.1 mm, and 1.3 mm wavelengths, respectively. The probabilitydistributions show that mild degeneracy between the fluxes of the ring groups is present, and wasmost apparent at 1.3 mm where the spatial resolution was poorest. The mean flux values of eachgroup are given in Table 3. The uncertainties given in that table are a quadrature sum of the standarddeviation of the MCMC-derived probability distribution and the flux calibration error (Section 2).Those mean values were used to generate the best-fitting radial profiles shown in orange in Figure3, and random sets of θ values drawn from the MCMC samples were used to produce the light graymodel profiles shown in the same figure. Ephemeris data were taken from JPL Horizons and the Planetary Ring Node. https://github.com/sczesla/PyAstronomy http://dfm.io/emcee/current/ Figure 3.
Radial flux profiles of the Uranian ring system at 3.1 mm, 2.1 mm, and 1.3 mm. The data,binned into 500 km intervals, are shown as blue dots. The best-fitting model is shown as an orange line, anddraws from the distribution of allowed models are shown as thin grey lines. Locations of the main rings arelabeled.
Wavelength B obs Total Flux, F ν (mJy)(mm) UT Date (degrees) 6/5/4 α / β η / γ / δ (cid:15) . ± .
15 1 . ± .
19 2 . ± .
22 13 . ± . . ± .
67 6 . ± .
80 8 . ± .
89 36 . ± . . ± . . ± . . ± . . ± . Table 3.
Total millimeter flux measurements of each ring group. B obs denotes ring opening angle withrespect to the observer. Note that two observations were taken at 3.1 mm, on 2017-12-03 and 2017-12-06,and co-added. The lower resolution and SNR of the VLT Q2 image precluded retrieval of the (cid:15) ring flux froma radial profile. We instead measured the total flux within an elliptical annulus of 3 beam-widthsthickness ( ∼
33 pixels) centered on the (cid:15) ring. To minimize contamination from the bright planet,we masked the planet such that the pixels within two beam widths of the planet’s outer edge wereomitted from the calculated ring flux. The choice of the mask size was informed by line scans throughthe images along the ansae, where the contributions from the rings and planet were most cleanlyseparated. Due to the viewing geometry, the PSFs from the ring system and the planet partlyoverlapped away from the ansae; the masking consequently removed more pixels from these portionsof the contributing ellipse and potentially underestimated the flux of the rings. To evaluate theaccuracy of our approach, we applied the same method to synthetic images with azimuthally-uniformrings of known brightness, blurred by the observed PSF of our images and corrupted by differentmanifestations of synthetic noise. These tests showed that neglecting any attempt at correcting forthe planetary contribution overestimated the ring flux by nearly 20%, while the simple mask tendedto underestimate the ring flux by less than 10%. We therefore conservatively estimate our measuredring system total flux is accurate to within 20%. Finally, we assumed the contribution of the (cid:15) ringto the total flux of the ring system was the same at 18.7 µ m as at 3 mm to get an approximate (cid:15) ring flux.We find the best fit to the 3.1 mm radial profile with total fluxes relative to the (cid:15) ring of 9 . ± . . ± . . ± .
6% for the 6/5/4, α / β , and η / γ / δ groups. These values are close to therelative equivalent widths at visible wavelengths, which French et al. (1986) give as 7%, 16%, and15%, respectively. However, at shorter wavelengths we find the 6/5/4 and η / γ / δ groups to be muchbrighter than expected based on the equivalent width measurements, by factors of 2.1 and 1.5 at2.1 mm and 4.4 and 2.7 at 1.3 mm. This discrepancy may be due to unmodeled contributions fromdiffuse millimeter-sized dust between the inner rings, a larger fraction of millimeter-sized grains inthose ring groups than in the (cid:15) ring, or systematic errors due to imperfect subtraction of the PSFaround the bright planet. The latter effect may be especially consequential at 1.3 mm due to thepoorer resolution of our data at that frequency.The ring is bathed by the cosmic microwave background (CMB) from every angle, except fromthe angles where it is blocked by Uranus. The final reported brightness temperature of Uranus wascorrected to account for the blocked CMB photons according to the prescription in de Pater et al.(2014), which gives correction factors of 1.02 K, 0.59 K, and 0.12 K at 3.1 mm, 2.1 mm, and 1.3 mm,respectively. If that CMB correction is undone (that is, Uranus is assumed to be fainter than it reallyis), then the CMB appears to reach the rings fully isotropically and thus be scattered isotropically.The interferometer is insensitive to scattered CMB photons from the rings since it, too, is bathedisotropically by the CMB; thus, the measured flux from the rings does not need to be corrected forthe CMB (Dunn et al. 2005; Zhang et al. 2019). THERMAL MODELING OF THE (cid:15)
RINGWe converted the (cid:15) ring total flux measurements derived in Section 3 into spectral radiance (here-after, “brightness”) units to produce a coarse spectrum of the ring. To do so, we simply divided thetotal flux measurements by the projected geometric area of the (cid:15) ring, which was computed as follows.Let the measured width of the ring at periapsis and apoapsis be denoted w p and w a (19.7 km and96.4 km for the (cid:15) ring Karkoschka 2001c). The inner and outer edge of the ring are defined by twoellipses sharing one focus (the center of mass of Uranus) but with semimajor axes a and eccentricities e such that w p and w a take their measured values. The inner ellipse has a inner = a − ( w p + w a ) / a outer = a + ( w p + w a ) /
4. To keep the focus at the same location requiresthe equation a outer − c outer = a inner − c inner + w p to be satisfied, where c = ae is the focal distance ofthe ellipse. The equation is satisfied when c outer − c inner = ( w a − w p ) /
2. Thus the area is given by A r = πa outer (cid:112) − e outer − πa inner (cid:113) − e inner (2)Projecting into the observing geometry and scaling to the observer-Uranus distance yields an angulararea of Ω r,obs = A r sin B/d = 1 . × − steradians.We compare these four brightness measurements to a simple thermal model modified from theNEATM model for near-earth asteroids (Harris 1998) to constrain the properties of the (cid:15) ring. Themodel gives the measured temperature T of a ring particle as T = L (cid:12) (1 − A B )4 π ¯ ησ(cid:15)d (3)where L (cid:12) is the solar luminosity, d is the heliocentric distance to the asteroid, (cid:15) is the bolometricemissivity of the asteroid, A B is the Bond albedo, σ is the Boltzmann constant, and ¯ η is the “beamingfactor”, a catch-all correction factor for low-phase-angle beaming, rotation, thermal inertia, andgeometry (see Section 4.1 for a discussion of how ¯ η differs from the standard NEATM model’s η ).To translate this temperature into a brightness S ν we use the equation S ν = f ( B ) B ( ν, T ) (4)where B ( ν, T ) is the Planck function at frequency ν and temperature T , and f ( B ) is the ring’s“fractional visible area”, the fraction of paths from the observer through the projected geometricring area that strike a ring particle (as opposed to passing through to background space).We performed a Markov Chain Monte Carlo (MCMC) sampling of our model’s parameter spaceusing the emcee Python package (Foreman-Mackey et al. 2013) to explore the probability distributionsof the free parameters in the thermal model. We allowed four parameters to vary: η , (cid:15) , and A B inEquation 3, and f ( B ) in Equation 4. We treated η , (cid:15) , and f ( B ) as completely unknown, but provideda Gaussian prior on A B with mean value 0 .
061 and standard deviation 0 .
006 (Karkoschka 1997). Ineach MCMC iteration, a set of these parameters was drawn and plugged into the NEATM model,yielding a Planck function. The value of that Planck function at the ALMA- and VLT-observed
Figure 4. (a)
Brightness and (b) brightness temperature comparison between retrieved NEATM model(red line: maximum-likelihood model, translucent black lines: random draws from probability distribution)and (cid:15) ring brightness measurements (blue points). c) Probability distribution function of (cid:15) ring temperature. (d)
Probability distribution function of fractional visible area of (cid:15) ring. The mean value and quartiles areshown as dashed and dotted black lines, respectively. frequencies was compared with the measured ring group brightness using a likelihood function similarin form to Equation 1. The simulation constrains the fractional visible area to f ( B ) = 0 . ± . T = 77 . ± . (cid:15) ring total flux in Figure 4.That figure also shows posterior probability distributions for f ( B ) and T , and a full “corner plot”displaying the one- and two-dimensional projections of the posterior probability distribution is shownin Appendix Figure A4. The two figures show that ¯ η and the emissivity are completely degeneratewith one another, and the retrieved albedo values are constrained only by the Karkoschka (1997)prior; however, the fractional visible area f ( B ) and temperature T (derived by combining ¯ η , (cid:15) , and A B ) are strongly constrained and minimally degenerate with one another.Our derived fractional visible area is consistent with the Karkoschka (2001c) model, which gives f ( B ) ≈ . B = 40 ◦ ( B ranges from 37 ◦ to 44 ◦ forour observations). The emissivity of the ring cannot be constrained since it is degenerate with ¯ η and both are unknown a-priori. However, the product of the two quantities is determined to be¯ η(cid:15) = 1 . ± .
17. If we assume the ring particles are perfect graybody emitters such that (cid:15) = 1 − A B ,0then ¯ η takes a best-fit value of 1.72. This ¯ η value is larger, and the observed temperature is lower,than expected from the Standard Thermal Model for near-Earth asteroids (STM; Lebofsky et al.1986), which assumes very low thermal inertia and/or slow rotation such that all flux is emitted onthe sun-facing side of the particle, leading to a predicted temperature T ST M = 79 . η issmaller and T higher than expected for a particle with thermal timescales long enough and rotationfast enough that the particle’s temperature is independent of longitude as in the Fast-Rotating Model(FRM) (Lebofsky & Spencer 1989), which yields T F RM = 71 . ≈ ◦ ; see Section4.1 for a description of our ring particle thermal model). This finding may indicate that the ringparticles are closer to STM-like than FRM-like; that is, their thermal inertia is low enough, and theirrotation rate is slow enough, that their dayside and nightside temperatures are different at a givenlatitude. However, beaming due to surface roughness tends to increase the amount of flux an observersees, leading to lower ¯ η values and higher observed temperatures (Spencer et al. 1989; Lagerros 1998,and references within). Surface roughness is not included in our model, so we cannot disentangle theeffects of beaming from the effects of thermal inertia.4.1. Ring Particle Model
The NEATM-like model we employ defines ¯ η as a property of the ring particles as a whole accordingto Equation 3. This differs slightly from the original NEATM model (Harris 1998), which defines η asa function of the temperature at the sub-solar point T such that T ∝ η and T ( µ ) = T µ / , where µ is the cosine of the emission angle. That is, ¯ η represents a geometric average while η does not. Todetermine the physical meaning of our observed value of ¯ η for the (cid:15) ring, we produced a toy modelof a ring particle. The model assumes a spherical particle with no surface roughness and no lateralheat conduction within the particle. The mean diurnal insolation ¯ Q for such a particle is given by¯ Q = L (cid:12) πd π (cid:16) h sin φ sin δ + cos φ cos δ sin h (cid:17) (5)(Pierrehumbert 2010), where δ is the sub-solar latitude of the particle, φ is the latitude at a givenlocation on the particle, and h is the hour angle at sunrise and sunset in radians at a given location(cos h = tan φ tan δ ), which is equal to π times the fraction of the day that location spends insunlight (note that h is set to π at the summer pole and zero at the winter pole). The temperatureat each point is given by T ( φ ) = ¯ Q / , and the total flux from the particle at the observed frequenciesis found by applying a Planck function at each point on the model particle, then taking the mean.We retrieve a blackbody temperature T of the particle as a whole by fitting a single Planck functionto these model flux values. Finally, ¯ η is found by comparing T to the subsolar temperature of theSTM at the same distance: 1 / ¯ η = ( T /T ,ST M ) . Taking T ,ST M = 88 . T = 79 . η = 1 . T = 64 . η = 3 . (cid:15) ring means its particles are likely toundergo frequent collisions that lead to a large spread in rotation axis orientations. However, thetotal angular momentum of the ring system should dictate that the average rotation axis is alignedwith that of Uranus, so we assume this average rotation axis (sub-solar latitude ≈ ◦ ) for all ringparticles. Spatial variations in h /π and ¯ Q are mapped in Figure 5 for this geometry. In this caseour model yields T = 71 . η = 2 .
3. This temperature is lower, and ¯ η is higher, than retrievedfrom the data.1 Figure 5. Left:
Fraction h /π of one rotation period that a model ring particle spends in sunlight. Right:
Diurnal mean insolation ¯ Q of a model ring particle, normalized to the insolation received by a non-rotating(STM-like) model particle at the sub-solar point.5. AZIMUTHAL STRUCTUREWe searched for azimuthal structure in the (cid:15) ring by breaking the ring into 30 bins, each subtending12 ◦ of azimuth and having a width equal to two times the FWHM of the ALMA point-spread function.The measured brightness in these bins as a function of angle from periapsis is shown in ExtendedData Figure 6. We find the (cid:15) ring to be a factor of 2-3 brighter at apoapsis than at periapsis at3.1 and 2.1 mm, in good agreement with stellar occultation measurements (French et al. 1986),visible/infrared reflected light observations (Karkoschka 1997; de Pater et al. 2002, 2013) and radiooccultation measurements (Gresh et al. 1989). No clear periapsis-apoapsis asymmetry is seen at 1.3mm or 18.7 µ m; however, the signal-to-noise ratio is lower in those data, and due to the poorerresolution of those data the measurements are contaminated by the presence of the inner rings to agreater extent. Note that we exclude the portions of the ring nearest (in projection) to Uranus dueto contamination by imaging artifacts arising from the very millimeter-bright planet. SUMMARYUsing ALMA and the VLT VISIR instrument, we have observed the thermal emission componentof the Uranian ring system for the first time. Our observations of the (cid:15) ring agree very well with theKarkoschka (2001c) model, supporting its derivation of the fractional visible area as a function ofring opening angle, and the 3.1 mm brightnesses of the inner main ring groups take the same ratio asobserved at visible wavelengths. The consensus between our millimeter and mid-infrared observationsand literature visible-wavelength observations shows that the properties of the main rings remainthe same at any observed wavelength despite the fact that our observations are not sensitive tomicron-sized dust. This finding confirms the hypothesis, proposed based on radio occultation results(Gresh et al. 1989), that the main rings are composed of centimeter-sized or larger particles. Asimple thermal model similar to the NEATM model for asteroids was applied to determine thatthe (cid:15) ring particles display roughly black-body behavior at millimeter/mid-infrared wavelengths at atemperature of 77 ± Figure 6. (cid:15) ring brightness as a function of azimuth. The grayed-out regions are contaminated by artifactsfrom the bright planet. A clear periapsis-apoapsis asymmetry is present at 3.1 mm and 2.1 mm wavelengths. rate slow enough, to induce longitudinal temperature differences between their dayside and nightside.Observations at higher spatial resolution, achievable using ALMA, would resolve the inner main ringsseparately and quantify the millimeter-sized dust component if present.
Software:
PyAstronomy, photutils, emcee, IDL
Acknowledgements:
This paper makes use of the following ALMA data: ADS/JAO.ALMA A. SCATTERED LIGHT CONTRIBUTION
Here we adopt a simple analytical radiative transfer calculation (after Chandrasekhar 1960) todetermine the contribution of scattered millimeter-wavelength light from Uranus to the millimeterbrightness of the (cid:15) ring. The radiative transfer equation reads: dI ν dτ ν = S ν − I ν (A1)where S ν = j ν /α ν is the source function. For scattered light from direction ( θ (cid:48) , φ (cid:48) ) into direction( θ, φ ), the source function is given by S ν ( θ, φ ) = (cid:36) ν (cid:90) π p (Θ)4 π I ν ( θ (cid:48) , φ (cid:48) ) d Ω (cid:48) (A2)where p (Θ) is the scattering phase function, Θ is the scattering angle, (cid:36) ν is the single-scatteringalbedo, and (cid:82) π I ν ( θ (cid:48) , φ (cid:48) ) d Ω (cid:48) is the flux density incident on the rings from the entire 4 π steradian sky.Since light from Uranus is the only contributor to the incident radiation at the (cid:15) ring, (cid:90) π I ν ( θ (cid:48) , φ (cid:48) ) d Ω (cid:48) = I ν,u Ω u,r (A3)where I ν,u is the specific intensity of Uranus at the observed frequency and Ω u,r is the solid angle ofUranus as seen from the ring. Then the source function of the rings reads S ν = (cid:36) ν p (Θ)4 π I ν,u Ω u,r (A4)To convert this into a flux density as seen from ALMA, we must multiply by the solid angle of therings with respect to the observer Ω r,obs : F ν,r,obs = (cid:36) ν p (Θ)4 π I ν,u Ω u,r Ω r,obs (A5)The specific intensity is the same for any observer, so the same I ν,u measured by ALMA would bemeasured by an observer at the (cid:15) ring, and thus I ν,u = F ν,u,r Ω u,r = F ν,u,obs Ω u,obs (A6)4where F ν,u,r and F ν,u,obs denote the spectral flux density of Uranus as seen from the rings and fromEarth, respectively. Substituting Equation A6 into Equation A5, we have F ν,r,obs = (cid:36) ν p (Θ)4 π Ω u,r Ω r,obs Ω u,obs F ν,u,obs (A7)The solid angle is defined as Ω = A/d , where A is the projected area of an object and d is thedistance from that object to the observer, soΩ u,r Ω u,obs = A u d obs A u r r = d obs r r (A8)where A u is the cross-sectional area of Uranus and r r is the distance from Uranus to the (cid:15) ring.Combining this with the fact that Ω r,obs = A r,obs /d obs , where A r,obs is the cross sectional area of the (cid:15) ring with respect to the observer, we end up with F ν,r,obs = (cid:36) ν p (Θ)4 π A r,obs r r F ν,u,obs (A9)This derivation assumed that the entire area of the ring is illuminated by light from Uranus as seenby the observer. However, due to shadowing and gaps in the ring, only a fraction of the ring f ( B u , B )is illuminated, where B u denotes the ring opening angle for an observer on Uranus and B denotes thering opening angle for an observer on Earth. In reality the flux comes from a distribution of angles:the diameter of Uranus is roughly 52 . ◦ at the distance to the (cid:15) ring, giving ring opening anglesbetween − . ◦ and 26 . ◦ . To evaluate f ( B u , B ) precisely is a difficult task, since one must integrateover many Uranus-ring-observer angles. To simplify the problem we shall make two assumptions.First, we assume B u is well represented by the geometric average opening angle B u = 18 . ◦ (theorbital inclination of the ring with respect to Uranus is i = 0 . ◦ , which is negligible). Second, weassume that half of the ring area illuminated by light from Uranus can be seen from Earth, to accountfor surfaces that point away from the observer. Under these assumptions, f ( B u , B ) = f ( B u ) /
2, andwe can again use the Karkoschka (2001c) prescription to calculate f ( B u ). Since the photons originatefrom Uranus itself the phase angle is 0 ◦ and the model gives us f ( B u , B ) = f (18 . / .
18. Thephysical interpretation of such a low value is that at small incident angles particles cast long shadowson each other, blocking much of the flux. Applying this correction to Equation A9 gives us F ν,r,obs = f ( B u , B ) (cid:36) p (Θ)4 π A r,obs r r F ν,u,obs (A10)Since the particles in the (cid:15) ring are assumed to be (cid:38)
10 cm in diameter (Gresh et al. 1989; Karkoschka2001c), we assume a phase function p (Θ) = 1 appropriate for geometric scattering.Our ALMA observations give the total 3 mm flux density of Uranus to be F ν,u,obs = 8 .
80 Jy. Weassume the ring particles have albedo (cid:36) mm = 1 − (cid:15) mm = 0 .
05. For simplicity we take r r = 51148km, the average of the (cid:15) ring semimajor and semiminor axes. At ring opening angle B obs = 36 . ◦ ,the projected ring area is A sin B obs = 1 . × km . Plugging these numbers into Equation A10yields F mm,r,obs = 28 µ Jy. This value is a factor of ∼
500 smaller than the measured ring flux at 3.1mm. Since both the planet and rings are at least relatively near the same temperature, the ratiobetween the reflected thermal emission from the planet and the direct thermal emission from the rings5is nearly wavelength-independent, and similarly small values are found in the mid-infrared. Whilemany simplifying assumptions were made in this derivation, such a small scattered light contributionmeans that scattered light from Uranus can be ignored.Equation A10 can be applied to quantify the contribution of scattered radio flux from the Sun withthe following modifications. The fraction f ( B (cid:12) , B ) can once again be taken from the Karkoschka(2001c) model, where B (cid:12) is the ring opening angle as viewed from the Sun (in fact, this geometry isthe scenario for which the model was developed). Instead of the Uranus-ring distance r r , the distancefrom the Sun to the ring d (cid:12) ,r should be used. Finally, the Sun’s radio flux at Earth F ν, (cid:12) ,obs replaces F ν,u,obs . Thus we have F ν,r,obs = f ( B (cid:12) , B ) (cid:36) p (Θ)4 π A r,obs d (cid:12) ,r F ν, (cid:12) ,obs (A11)At the time of our 3.1 mm observations, the phase angle was very nearly 2 ◦ , so the Karkoschka(2001c) model gives f ( B (cid:12) , B ) = 0 .
5. Alissandrakis et al. (2017) used ALMA observations to find asolar brightness temperature of 7250 ±
170 K at 100 GHz, as well as a solar radius of 964 . ± . (cid:48)(cid:48) at 1 AU at 100 GHz. Assuming the same temperature and radius at 97.5 GHz (3.1 mm), we get F ν, (cid:12) ,obs = 1 . × Jy. Finally, taking d (cid:12) ,r at the time of observation from JPL Horizons, wefind F ν,r,obs = 0 . µ Jy, which is ∼ µ m wavelengths, we find F ν,r,obs = 7 . ∼
2% of the measured flux; this is much smaller than the 20% VLT fluxcalibration error and so is ignored.6 B. EXTENDED DATA
Figures A1, A2, and A3 display “corner plots” for the MCMC-derived total flux measurementsdiscussed in Section 3. Figure A4 displays a corner plot for the thermal model discussed in Section4.
Figure A1. “Corner plot” showing the one- and two-dimensional projections of the posterior probabilitydistributions of the MCMC-retrieved total fluxes of each ring group in Band 3 (3.1 mm), derived fromcomparing our model to the observed the radial profile. Figure A2. “Corner plot” showing the one- and two-dimensional projections of the posterior probabilitydistributions of the MCMC-retrieved total fluxes of each ring group in Band 4 (2.1 mm), derived fromcomparing our model to the observed the radial profile. Figure A3. “Corner plot” showing the one- and two-dimensional projections of the posterior probabilitydistributions of the MCMC-retrieved total fluxes of each ring group in Band 6 (1.3 mm), derived fromcomparing our model to the observed the radial profile. Figure A4. “Corner plot” showing the one- and two-dimensional projections of the posterior probabilitydistributions of the MCMC-retrieved thermal model parameters.
Alissandrakis, C. E., Patsourakos, S., Nindos, A.,& Bastian, T. S. 2017, A&A, 605, A78,doi:
Bradley, L., Sipocz, B., Robitaille, T., et al. 2019,astropy/photutils: v0.6,doi: . https://doi.org/10.5281/zenodo.2533376 Brogan, C. L., Hunter, T. R., & Fomalont, E. B.2018, arXiv e-prints. https://arxiv.org/abs/1805.05266
Chandrasekhar, S. 1960, Radiative transferde Kleer, K., de Pater, I., ´Ad´amkovics, M., &Hammel, H. 2013, Icarus, 226, 1038,doi: de Pater, I., Gibbard, S. G., & Hammel, H. B.2006, Icarus, 180, 186,doi: de Pater, I., Gibbard, S. G., Macintosh, B. A.,et al. 2002, Icarus, 160, 359,doi: de Pater, I., & Gulkis, S. 1988, Icarus, 75, 306,doi: de Pater, I., Hammel, H. B., Showalter, M. R., &van Dam, M. A. 2007, Science, 317, 1888,doi: de Pater, I., Sault, R. J., Butler, B., DeBoer, D.,& Wong, M. H. 2016, Science, 352, 1198,doi: de Pater, I., Dunn, D. E., Stam, D. M., et al.2013, Icarus, 226, 1399,doi: de Pater, I., Fletcher, L. N., Luszcz-Cook, S.,et al. 2014, Icarus, 237, 211,doi:
Dobrzycka, D., & Vanzi, L. 2008, in The 2007ESO Instrument Calibration Workshop,Springer, 139–144Dunn, D. E., de Pater, I., Wright, M.,Hogerheijde, M. R., & Molnar, L. A. 2005, AJ,129, 1109, doi:
Foreman-Mackey, D., Hogg, D. W., Lang, D., &Goodman, J. 2013, PASP, 125, 306,doi:
French, R. G., Elliot, J. L., & Levine, S. E. 1986,Icarus, 67, 134,doi:
French, R. G., Nicholson, P. D., Porco, C. C., &Marouf, E. A. 1991, Dynamics and structure ofthe Uranian rings, ed. J. T. Bergstralh, E. D.Miner, & M. S. Matthews, 327–409French, R. G., Elliot, J. L., French, L. M., et al.1988, Icarus, 73, 349,doi:
Gresh, D. L., Marouf, E. A., Tyler, G. L., Rosen,P. A., & Simpson, R. A. 1989, Icarus, 78, 131,doi:
Gulkis, S., & de Pater, I. 1984, in NASAConference Publication, Vol. 2330, NASAConference Publication, ed. J. T. BergstralhHarris, A. W. 1998, Icarus, 131, 291,doi:
Karkoschka, E. 1997, Icarus, 125, 348,doi: —. 2001a, Icarus, 151, 51,doi: —. 2001b, Icarus, 151, 69,doi: —. 2001c, Icarus, 151, 78,doi:
Lagage, P., Pel, J., Authier, M., et al. 2004, TheMessenger, 117, 12Lagerros, J. S. V. 1998, A&A, 332, 1123Lebofsky, L. A., & Spencer, J. R. 1989, inAsteroids II, ed. R. P. Binzel, T. Gehrels, &M. S. Matthews, 128–147Lebofsky, L. A., Sykes, M. V., Tedesco, E. F.,et al. 1986, Icarus, 68, 239,doi:
Nicholson, P. D., De Pater, I., French, R. G., &Showalter, M. R. 2018, The Rings of Uranus,ed. M. S. Tiscareno & C. D. Murray, 93–111Pierrehumbert, R. T. 2010, Principles ofPlanetary ClimateShowalter, M. R., & Lissauer, J. J. 2006, Science,311, 973, doi:
Smith, B. A., Soderblom, L. A., Beebe, R., et al.1986, Science, 233, 43,doi:
Spencer, J. R., Lebofsky, L. A., & Sykes, M. V.1989, Icarus, 78, 337,doi: