The B-ring’s surface mass density from hidden density waves: Less than meets the eye? M. M. Hedmana, P.D. Nicholsonb a Department of Physics, University of Idaho, Moscow ID 83844-0903 b Department of Astronomy, Cornell University, Ithaca NY 14853 Saturn’s B ring is the most opaque ring in our solar system, but many of its fundamental parameters, including its total mass, are not well constrained. Spiral density waves generated by 6 mean-motion resonances with Saturn’s moons provide some of the best constraints on the rings’ 1 0 mass density, but detecting and quantifying such waves in the B ring has been challenging because 2 of this ring’s high opacity and abundant fine-scale structure. Using a wavelet-based analyses of 17 b occultationsofthestarγ CrucisobservedbytheVisualandInfraredMappingSpectrometer(VIMS) e onboard the Cassini spacecraft, we are able to examine five density waves in the B ring. Two of F these waves are generated by the Janus 2:1 and Mimas 5:2 Inner Lindblad Resonances at 96,427 7 km and 101,311 km from Saturn’s center, respectively. Both of these waves can be detected in 1 individual occultation profiles, but the multi-profile wavelet analysis reveals unexpected variations ] in the pattern speed of the Janus 2:1 wave that might arise from the periodic changes in Janus’ P E orbit. The other three wave signatures are associated with the Janus 3:2, Enceladus 3:1 and . Pandora 3:2 Inner Lindblad Resonances at 115,959 km, 115,207 km and 108,546 km. These waves h p are not visible in individual profiles, but structures with the correct pattern speeds can be detected - in appropriately phase-corrected average wavelets. Estimates of the ring’s surface mass density o r derived from these five waves fall between 40 and 140 g/cm2, even though the ring’s optical depth t s in these regions ranges from ∼ 1.5 to almost 5. This suggests that the total mass of the B ring is a most likely between one-third and two-thirds the mass of Saturn’s moon Mimas. [ 2 v 5 1. Introduction 5 9 7 The B ring is the brightest, most opaque and probably the most massive of Saturn’s rings. 0 It is also among the least well understood, with even basic parameters like its maximum optical . 1 depth and its total mass being poorly constrained. Indeed, estimates of the B-ring’s mass range 0 from around 1×1019 kg (Cooper et al. 1985) to over 7×1019 kg (Robbins et al. 2010). The large 6 1 uncertainties in the B-ring’s mass and its typical surface mass density not only hamper efforts to : understand the structure and dynamics of this ring, but also complicate efforts to ascertain the age v i and history of Saturn’s ring system (Charnoz et al. 2009). X r One reason why the B-ring’s mass is so poorly constrained is that very few spiral density or a bending waves have been identified in this ring. These spiral patterns provide the most reliable estimates of a ring’s mass density, thanks to a very mature theoretical model that describes their formation and propagation (Shu 1984). Indeed, analyses of various density waves have yielded numerous mass density estimates of the A ring (Tiscareno et al. 2007, 2013), Cassini Division (Colwell et al. 2009) and C Ring (Bailli´e et al. 2011; Hedman and Nicholson 2014). However, thus far there have only been two waves that have yielded sensible mass density estimates for the B ring. The most prominent wave in the B ring is the one generated by the Janus 2:1 resonance, which lies in the least opaque, innermost part of the B ring. Analyses of Voyager occultation data yielded mass density estimates of around 70 g/cm2 for this region (Holberg et al. 1982; Esposito et al. 1983). On the opposite end of the B ring, Lissauer (1985) identified a bending wave due to the – 2 – Fig. 1.— Overview of the B ring. The profile shows the normal optical depth of Saturn’s B ring as a function of ring radius derived from an occultation of the star γ Crucis on Cassini Orbit (“Rev”) 089 observed by the VIMS instrument. The background shading indicates the five distinct zones in theBring(Colwellet al.2009), andthedashedlinesmarkthelocationsofmeanmotionresonances discussed in the text. 4:2 vertical resonance with Mimas in the Voyager imaging data. Bending waves can be analyzed much like density waves, allowing Lissauer (1985) to estimate a local surface mass density of 54±10 g/cm2. More recently, analyses of normal modes on the B-ring’s outer edge have found generally comparable mass densities in this region (Spitale and Porco 2010; Nicholson et al. 2014). The few existing measurements provide very limited information about the B-ring’s surface mass density because they only sample two locations in a very complex ring. As shown in Figure 1, the B-ring can be divided into five broad regions with very different optical depths (Colwell et al. 2009). The Janus 2:1 wave occupies the less opaque BI region, while the Mimas 4:2 wave is found in the BV region, where the ring’s structure is highly variable due to its proximity to the highly non-circular outer edge at 117,500 km. To date, no-one has reported mass density estimates for the central BII, BIII and BIV regions, which include the most opaque parts of Saturn’s rings. However, there are several strong satellite resonances in these regions that could generate relatively intense density waves. One is the Mimas 5:2 resonance, which is found in the BII region, where the optical depth appears to switch between values around 2 and near opaque. This wave has been identified in occultation profiles (Bratcher and Colwell 2013), but a mass density has not yet been published based on this feature. In the BIV region, there should be strong resonances with Janus (3:2) and Enceladus (3:1), but neither of these has been identified in the observational data because this part of the B ring contains intense fine-scale stochastic optical depth variations that obscure any organized structure from a density wave. Finally, the Prometheus 3:2 and Pandora 3:2 resonances lie in BIII, the nearly-opaque core of the ring, and so high-resolution occultations, which measure – 3 – the amount of light transmitted through the rings, have very low signal-to-noise. HereweinvestigateallsixoftheseLindbladresonancesintheBringusinganewwavelet-based technique that combines data from multiple occultations in order to identify weak coherent signals due to waves that are not obvious in single measurements. These methods reveal potential wave signatures from the Janus 3:2, Enceladus 3:1 and Pandora 3:2 resonances (but not the Prometheus 3:2 resonance). Together with the Janus 2:1 and Mimas 5:2 density waves and the Mimas 4:2 bending wave, these wave features provide mass density estimates spanning a wide range of optical depths, and thus provide a much clearer picture of the B-ring’s total mass. Our measurements indicate that regions with optical depths ranging between 1 and 5 all have mass densities less than 150 g/cm2. This is consistent with recent investigations of other parts of Saturn’s rings, which demonstrate that large variations in the ring’s optical depth are often not associated with comparable variations in its surface mass density (Colwell et al. 2009; Bailli´e et al. 2011; Tiscareno et al. 2013; Hedman and Nicholson 2014). At the same time, these B-ring mass densities are well below the values that have been considered in some recent studies of the rings’ opacity, history and spectral properties (Robbins et al. 2010; Charnoz et al. 2011; Hedman et al. 2013) but may be consistent with estimates based on the rings’ charged particle emissions and thermal properties (Cooper et al. 1985; Reffet et al. 2015). Section 2 briefly reviews the relevant aspects of density wave theory, while Section 3 pro- vides information about the occultation data used for this investigation. Section 4 describes the wavelet-basedtechniqueswehavedevelopedtoisolateandquantifythewavesignalsassociatedwith particular resonances. Section 5 discusses the potential wave signatures associated with each of the B-ring Lindblad resonances, and the surface mass densities implied by these features. Section 6 discusses the implications of these new estimates. 2. Theoretical Background Spiral density wave patterns are generated in the rings near Lindblad resonances with periodic gravitational perturbations from either Saturn’s various moons or asymmetries in the planet’s internal structure. At these locations, the periodic perturbations induce organized radial epicyclic motions in the ring-particles’ orbital motions, which in turn generate a spiral wave pattern in the ring’sopticaldepththatpropagatesthroughtherings. Agoodreviewofthedetailedtheorybehind these structures is provided by Shu (1984), and we summarize some of the key aspects of these calculations here for the sake of clarity and to introduce the notation used in this paper. A generic (m + (cid:96)):(m − 1) inner Lindblad resonance with a satellite occurs when the ring- particles’ radial epicyclic frequency κ is an integer multiple of the difference between the frequency of some periodic perturbing force Ω and the particles’ mean motion n: p m(n−Ω ) = κ (1) p In a differentially-rotating disk with finite mass density, these perturbations organize the motions oftheringparticles, generatingapatternconsistingofmspiralarmsthatrotatesaroundtheplanet at the rate Ω . So long as the fractional optical depth variations δτ/τ are sufficiently small, they p should be described by the following function of ring radius r, inertial longitude λ and time t: δτ = A(r)cos[φ (r)+φ (λ,t)], (2) r λt τ – 4 – where A(r) is a radius-dependent amplitude of the density variations, while φ and φ are phase r λt parameters. Notethatφ dependsonlyontheringradius, andφ dependsonacombinationofthe r λt observed longitude and observation time. Note that φ governs the overall trends in the pattern’s r wavelength that are common to all the occultation profiles, while φ affects the exact positions of λt the wavecrests within each occultation. Sincethepatterncreatedbytheresonanceconsistsofmspiralarmsrotatingaroundtheplanet at the rate Ω , the longitude/time-dependent part of the pattern’s phase φ can be written as: p λt φ (λ,t) = |m|[λ−Ω (t−t )] (3) λt p 0 where λ and t are the observed inertial longitude and time, and t is a reference epoch time, which 0 for this analysis corresponds to 2008-200T00:00:00 UTC (a time near the middle of the interval spanned by the occultations used in this analysis). For a first-order ((cid:96) = 0) Lindblad resonance, Ω equals the mean motion of the relevant satellite, but for higher-order resonances the pattern p speed is a more complex function of the satellite’s orbital parameters. For all Lindblad Resonances, Equation 1 can be used to express the pattern speed in terms of the orbital properties of the ring material: 1 (m−1) 1 Ω = n(r )− κ(r ) = n(r )+ (cid:36)˙ (r ) (4) p L L L L m m m where n(r ), κ(r ) and (cid:36)˙ (r ) are the orbital mean motion, radial epicyclic frequency and apsidal L L L precession rate at the radial location of the exact resonance r . L Meanwhile, the radius-dependent part of the phase φ can be derived from the density-wave’s r dispersion equation and the resulting radius-dependent radial wavenumber of the pattern k(r). For sufficiently weak waves at sufficiently large distances from the resonance, the perturbations from the density wave should cause the ring’s surface mass density to oscillate quasi-sinusoidally as a function of radius with a wavenumber k(r) given by the following expression: dφ 3(m−1)M (r−r ) r P L k(r) = = , (5) dr 2πσ r4 0 L where r is the radial location of the exact resonance, M is the mass of the central planet (Saturn L P inthiscase)andσ istheundisturbedsurfacemassdensityoftherings. Theexplicitdependenceon 0 σ iswhymeasurementsofthewave’sradialwavenumberataspecifieddistancefromtheresonance 0 provideestimatesoftherings’localsurfacemassdensity. Integratingthisexpression(andassuming a constant mass density), we find that the radius-dependent part of the phase should be given by the following asymptotic expression: 3(m−1)M (r−r )2 P L φ (r) = +φ , (6) r 4πσ r4 0 0 L where φ is a constant phase offset. 0 3. Observational Data This analysis uses stellar occultation data obtained by the Visual and Infrared Mapping Spec- trometer (VIMS) instrument onboard the Cassini spacecraft (Brown et al. 2004). During these observations the instrument measures the brightness of a selected star repeatedly as it passes be- hind the rings. While VIMS measures the brightness of the star at multiple wavelengths, for this – 5 – Table 1: Longitudes, Ephemeris Time (seconds past J2000 in TDB) and wave phases φ for the λt occultationsbyγ Crucisusedinthisanalysis(computedusingtheindicatedm-numbersandpattern speeds). Rev Ja2:1 Mi5:2 Pd3:2 En3:1 Ja3:2 m=2 m=3 m=3 m=2 m=3 Ωp=518.24◦/day Ωp=635.99◦/day Ωp=572.79◦/day Ωp=393.89◦/day Ωp=518.24◦/day 184.8◦ 185.3◦ 185.9◦ 186.4◦ 186.4◦ 071 266191216. 266190496. 266189456. 266188496. 266188384. 142.7◦ 156.4◦ 55.2◦ 303.2◦ 269.9◦ — — 185.5◦ 186.0◦ 186.0◦ 072 — — 266806000. 266805040. 266804928. — — 31.7◦ 80.7◦ 334.2◦ 184.0◦ 184.5◦ 185.1◦ 185.6◦ 185.7◦ 073 267423904. 267423184. 267422144. 267421184. 267421088. 113.4◦ 292.5◦ 16.4◦ 222.1◦ 45.9◦ 183.0◦ 183.6◦ 184.2◦ — — 077 269856064. 269855328. 269854304. — — 94.8◦ 220.4◦ 241.6◦ — — 182.8◦ 183.4◦ 184.0◦ 184.6◦ 184.6◦ 078 270464544. 270463808. 270462784. 270461856. 270461760. 354.9◦ 102.7◦ 19.2◦ 215.8◦ 47.9◦ 181.5◦ 182.1◦ 182.8◦ — — 079 271043200. 271042432. 271041344. — — 250.2◦ 280.9◦ 28.9◦ — — 180.7◦ 181.3◦ 182.1◦ 182.8◦ 182.9◦ 081 272318080. 272317312. 272316224. 272315232. 272315104. 75.1◦ 205.9◦ 231.6◦ 233.5◦ 172.5◦ 180.3◦ 181.0◦ 181.8◦ 182.5◦ 182.5◦ 082 272953856. 272953088. 272952000. 272951008. 272950880. 7.4◦ 204.9◦ 185.9◦ 196.0◦ 251.2◦ 179.4◦ 180.1◦ 181.0◦ 181.7◦ 181.7◦ 086 275501376. 275500640. 275499520. 275498528. 275498432. 44.8◦ 105.2◦ 276.9◦ 6.4◦ 127.5◦ 179.2◦ 179.9◦ 180.7◦ 181.4◦ 181.5◦ 089 277406464. 277405696. 277404608. 277403616. 277403488. 230.7◦ 154.9◦ 187.2◦ 275.8◦ 46.2◦ 205.9◦ 205.2◦ 204.4◦ 203.8◦ 203.7◦ 093 280042592. 280041728. 280040480. 280039360. 280039232. 340.3◦ 339.6◦ 34.7◦ 48.4◦ 204.6◦ 191.8◦ 191.9◦ 191.9◦ 192.0◦ 192.0◦ 094 280679008. 280678208. 280677056. 280676000. 280675904. 237.6◦ 284.0◦ 296.6◦ 339.8◦ 232.9◦ 186.3◦ 186.6◦ 187.0◦ 187.4◦ 187.4◦ 096 282012064. 282011328. 282010272. 282009312. 282009216. 75.0◦ 349.0◦ 46.2◦ 53.7◦ 346.9◦ — 186.5◦ 186.9◦ 187.3◦ 187.3◦ 097 — 282700000. 282698944. 282697984. 282697888. — 260.7◦ 29.2◦ 254.2◦ 194.3◦ 218.7◦ 217.3◦ 215.5◦ 214.1◦ 213.9◦ 100 285031424. 285030592. 285029376. 285028320. 285028192. 278.7◦ 6.8◦ 206.0◦ 300.5◦ 101.9◦ 218.7◦ 217.3◦ 215.5◦ 214.1◦ 213.9◦ 101 285858560. 285857728. 285856544. 285855456. 285855328. 76.0◦ 100.7◦ 315.2◦ 318.6◦ 337.7◦ 218.4◦ 217.0◦ 215.3◦ 213.8◦ 213.7◦ 102 286683744. 286682912. 286681728. 286680640. 286680544. 256.3◦ 237.4◦ 102.7◦ 354.2◦ 248.2◦ – 6 – analysis we use only data obtained at wavelengths around 3 microns, where the ring is especially dark. Each of these brightness measurements is tagged with a precise time stamp, which (together with the relevant spacecraft trajectory information stored in the NAIF SPICE kernels; Acton 1996) allows us to compute the radius and inertial longitude where the starlight passed through the rings. Based on the positions of sharp edges elsewhere in the rings, we can confirm that these calculations are accurate to within one kilometer. Global fits to the ring geometry provide small corrections to the spacecraft’s trajectory that improve this accuracy to within a few hundred meters (French et al. 2011). This particular investigation examines 17 occultations of the star γ Crucis obtained in 2008, corresponding to “Revs” (Cassini orbits) 71-102. These occultations are especially useful for study- ing the B ring because γ Crucis is a very bright star that lies in Saturn’s far southern skies. The line of sight to the star therefore passes through the rings at a very steep angle (62.35◦), reducing the light’s pathlength through the rings and increasing the signal transmitted through the rings. Furthermore, theseoccultationswereallobtainedfromverysimilarobservinggeometriesoverarel- atively short period of time, which facilitates the comparisons between the various opacity profiles described in Section 4 below. Table 1 provides a summary of this data set, giving the occultations that cover each of the relevant resonances, along with the inertial longitudes and times where the line of sight to the star crossed the resonant radius r . These numbers, along with the pattern L speeds appropriate for each resonance, are then inserted into Equation 3 in order to compute the expectedφ valuesforeachoccultationandeachofthewavesconsideredinthisstudy(seeTable1). λt TheresponseoftheVIMSinstrumentishighlylinear,sothemeasuredsignaliseasilytranslated intoestimatesofthetransmissionT throughtheringbyfirstsubtractingthemeansignalinaregion where the ring is nearly opaque (105,700-106,100 km from Saturn’s center) and then dividing the resultingbrightnessmeasurementsbythemeansignallevelinaregionunobstructedbyringmaterial (either just outside the B ring in the Huygens Gap at 117,700-117,750 km or, if these data are missing, outside the entire ring system beyond 145,000 km). This transmission can be converted into estimates of the normal optical depth τ = −ln(T)sin(B), where B = 62.35◦ is the elevation n angle of the star above the ring plane. Note that the signal from the unocculted star corresponds to about 250 counts per 100 meters of radius in most of these occultations (with some variability among the occultations due to how well the selected pixel captured the star). Since the read noise of the instrument is low (around 1 count), this means that these occultations have sufficient signal- to-noise to discern sub-percent variations in the transmission on sub-kilometer radius scales, and detect a finite signal through the rings even where the optical depth exceeds 4. During each occultation, VIMS recorded the average stellar signal every 20-40 ms, which corresponds to a radial range of 200-400 meters. This sampling scale is larger than both the projected stellar diameter (70-100 m) and the Fresnel zone (60-70 m), and so determines the effective resolution for these observations. In order to simplify comparisons between the profiles and facilitate the multi-profile wavelet analysis described below, the transmission, longitude and time parameters for every profile of each ring feature were re-sampled and interpolated onto a uniform radial grid sampled every 100 meters (well above the resolution of any given occultation). 4. Multi-profile wavelet analysis The Janus 2:1 and Mimas 5:2 waves are clearly visible in individual occultation profiles, but noneoftheotherdensitywavescanbeclearlyseenwithinasingleprofile. Identifyingthesewavesis – 7 – difficult not only because the signal levels are low, but also because the relevant parts of the B ring containintenseshort-wavelengthvariationsthatvarystochasticallyfromoccultationtooccultation. These variations obscure any coherent signal from the relevant density waves. Fortunately, we can use wavelet-based methods to combine data from multiple profiles and thereby isolate the density wave signals. The desired density wave patterns have wavelengths that should vary with radius across the ring (see Equation 5), so these waves are most easily identified using wavelets. Continuous wavelet transformations are analogous to localized Fourier transformations and have already proven to be extremely powerful tools for quantifying the properties of waves in planetary rings (Tiscareno et al. 2007; Colwell et al. 2009; Bailli´e et al. 2011; Tiscareno et al. 2013; Hedman and Nicholson 2013, 2014). For this investigation, we compute the wavelet transform for each resampled profile with the standard wavelet routine in the IDL language (Torrence and Compo 1998), using a Morlet mother wavelet with ω = 6. This yields a wavelet transform W (r,k) for each profile i as a function of 0 i radius r and wavenumber k. Note that this is a complex function, and so in general can be written as Wi = AieiΦi, where Ai(r,k) and Φi(r,k) are the (real) wavelet amplitude and phase. Note that if we have a pure sinusoidal signal at a given wavevector k , then Φ (r,k ) is the phase of the wave 0 i 0 as a function of position. Also, one can define the wavelet power P (r,k) = A2 i i Previous work by Colwell et al. (2009) and Bailli´e et al. (2011) combined wavelet data from multipleoccultationsinamannerthatenabledthemtoidentifywavesthatweretooweaktodiscern in individual occultation profiles. This method amounts to co-adding the wavelet power from the various occultations (technically, they co-added the values of the WWZ transform, a version of the wavelet power that is better optimized for unevenly sampled data, see Foster 1996). With this approach, peaks in individual wavelet power maps that are due to stochastic opacity variations in individualprofilesareaveragedout, whilepersistentsignalsatparticularwavelengthsandlocations remain in the averaged power. This technique does improve the signal-to-noise for weak waves and is an especially good method for searching for waves with unknown pattern speeds. However, this method does not clearly identify the density wave signals in the central and outer B ring. Fortunately, the waves of interest here have known pattern speeds, enabling us to use a method that is even better at isolating the signatures of density waves. As mentioned in Section 2, spiral density waves are patterns consisting of m tightly wrapped spiral arms rotating around the planet at a predictable pattern speed Ω . In a given occultation p cut, this pattern produces a quasi-sinusoidal opacity variation described by Equation 2. However, depending on the observed longitude and time, each profile will have a different value for the phase parameter φ . The crests and troughs of the wave will therefore appear at different locations in λt different profiles, and the wavelet phases associated with these patterns will vary from occultation to occultation. Fortunately, the waves considered here are all generated by resonances with known satellites, and so have known values of m and Ω , and so we can use Equation 3 to compute the p expected φ for each occultation profile and use this information to isolate the desired signal from λt those specific waves. Let us denote the calculated value of φ (assuming a given m and Ω ) for the i-th occultation λt p profile as φ . If the track of the star behind the ring was in the radial direction and the star’s i apparent radial motion across the ring feature were fast compared to the perturbing moon’s orbital motion, then φ would have the a constant value for the entire profile. However, in reality both i the observation time and observed longitude vary slightly as the star passes behind the wave, so that φ is a function of radius, albeit a very weak one. In any case, we can use φ to compute the i i – 8 – phase-corrected wavelet for each profile: W (r,k) = W (r,k)e−iφi(r) = A (r,k)ei(Φi(r,k)−φi(r)) (7) φ,i i i Recall that Φ is the observed wavelet phase, while φ is the expected longitude/time-dependent i i part of the wavelet phase for a spiral density wave with the selected m-number and pattern speed. Hence for any signal in the wavelet due to the desired density wave, the corrected phase parameter Φ −φ will equal φ (r) (see Equation 2) and should have the same value for all occultations. Thus i i r any signal from such a pattern should persist in the average phase-corrected wavelet: N 1 (cid:88) (cid:104)W (r,k)(cid:105) = W (r,k), (8) φ φ,i N i=1 where N is the number of occultations. By contrast, any pattern that does not have the selected Ω should have different phases in the phase-corrected wavelet and thus should average to zero in p (cid:104)W (cid:105) in the limit of large N. The signal-to-noise for the selected waves should therefore be much φ better in the average phase-corrected wavelet than it is in wavelets from individual profiles. In fact, we find that the average phase-corrected wavelet can yield a clearer detection of weak waves than even the averaged wavelet powers used by Colwell et al. (2009) and Bailli´e et al. (2011). In order the illustrate the utility of the average phase-corrected wavelet, it is useful to consider two distinct wavelet powers. First, we define the average wavelet power: N 1 (cid:88) P¯(r,k) = (cid:104)|W |2(cid:105) = |W |2. (9) φ φ,i N i=1 This is independent of the individual wavelet phases and so is equivalent to the average value of wavelet powers from the individual profiles P : i N 1 (cid:88) P¯(r,k) = (cid:104)|W |2(cid:105) = |W |2, (10) i i N i=1 and is therefore similar to the statistics used by Colwell et al. (2009) and Bailli´e et al. (2011). Second, define the power of the average phase-corrected wavelet as: (cid:12) (cid:12)2 N (cid:12) 1 (cid:88) (cid:12) P (r,k) = |(cid:104)W (cid:105)|2 = (cid:12) W (cid:12) . (11) φ φ (cid:12)N φ,i(cid:12) (cid:12) (cid:12) i=1 Note that in this case we perform the averaging prior to taking the absolute square, while the opposite is true for P¯. Recall that for any real variable x the difference (cid:104)x2(cid:105) − (cid:104)x(cid:105)2 is positive definite and equivalent to the variance of x. The difference between these two quantities P¯ −P φ is similarly a positive quantity determined by the variance in the real and imaginary components of the wavelet among the various occultations. Hence P must have a value between 0 and P¯. In φ fact, P can only equal P¯ when Φ −φ is the same for all occultations, as would be the case if the φ i i opacity variations are entirely due to a single wave with the specified pattern speed and m-number. For any other signal, P will be less than P¯ because of the finite scatter in Φ −φ . Indeed, as N φ i i approaches infinity, P should approach zero for these signals (provided we can sample all possible φ values of Φ −φ ). The ratio between these two powers: i i P (r,k) φ R(r,k) = , (12) P¯(r,k) – 9 – Fig. 2.— Sample analysis of the Prometheus 7:6 wave in the A ring. The top panel shows the transmission through the A ring as a function of radius from the Rev 89 occultation by γ Crucis. The two density waves clearly visible in this profile are due to the Pandora 6:5 and Prometheus 7:6 resonances. The second panel shows the average wavelet power P¯ for the γ Crucis occultations, with clear diagonal bands associated with both waves. The third panel shows the power of the average phase-corrected wavelet P , assuming m = 7 and a pattern speed appropriate for the φ Prometheus 7:6 resonance (the exact resonance location is marked by the vertical dotted line). Note that this highlights the right-hand wave. The fourth panel shows the ratio of the above powers R, and shows only the signal from that wave. Finally, the bottom panel shows the peak value of R as a function of radius and assumed pattern speed, parameterized as a displacement δr from the expected Prometheus 7:6 resonance location (marked with a horizontal dotted line). Note that the maps of P¯ and P use a common logarithmic stretch, while the maps of R use a linear φ stretch. – 10 – Fig. 3.— Extracting wavelength information from the Prometheus 7:6 wave in the A ring from the average phase-corrected wavelet. The top panel shows the mean normal optical depth τ of the n ring, along with the range of optical depths among the various profiles. The second panel shows the reconstructed fractional variations in T derived from the average phase-corrected wavelet data for wavelengths between 1 and 10 km. The third panel shows the wavenumber of the pattern as a function of radius. For the sake of clarity, only data where the peak power ratio was above 0.5 are shown. The fourth panel shows the estimated surface mass density σ derived from this wave, and the bottom panel shows the estimated opacity τ /σ. Note that in this case the opacity oscillates n because of the residual variations in the average normal optical depth associated with the wave.