Sunlight refraction in the mesosphere of Venus during the transit on June 8th, 2004
P. Tanga, Th. Widemann, B. Sicardy, J.M. Pasachoff, J. Arnaud, L. Comolli, A. Rondi, S. Rondi, P. Suetterlin
SSunlight refraction in the mesosphere of Venusduring the transit on June 8 th , 2004 P. Tanga a, ∗ , T. Widemann b , B. Sicardy b , J. M. Pasachoff c , J. Arnaud d, ∗∗ , L.Comolli e , A. Rondi f , S. Rondi f , P. S¨utterlin g a Laboratoire Cassiop´ee UMR6202, Universit´e de Nice Sophia-Antipolis, CNRS,Observatoire de la Cˆote d’Azur, BP 4229, 06304 Nice Cedex 4, France b LESIA-Observatoire de Paris, CNRS, UPMC Universit´e Paris 6, Universit´eParis-Diderot, 5, place Jules Janssen, 92195 Meudon Cedex, France c Williams College Hopkins Observatory, Williamstown, MA 01267-2565, USA d Laboratoire Fizeau UMR6525, Universit´e de Nice Sophia-Antipolis, CNRS, Observatoirede la Cˆote d’Azur, BP 4229, 06304 Nice Cedex 4, France e Gruppo Astronomico Tradatese, Via Mameli 13, 21049 Tradate, Italy f Soci´et´e Astronomique de France, 3 rue Beethoven, 75016 Paris, France g Sterrekundig Instituut, Utrecht University, Postbus 80 000, 3508 TA Utrecht, TheNetherlands. Now: Institute for Solar Physics, The Royal Swedish Academy of Sciences,Alba Nova University Center, 106 91 Stockholm, Sweden
Abstract
Many observers in the past gave detailed descriptions of the telescopic as-pect of Venus during its extremely rare transits across the Solar disk. Inparticular, at the ingress and egress, the portion of the planet’s disk outsidethe Solar photosphere has been repeatedly perceived as outlined by a thin,bright arc (”aureole”). Those historical visual observations allowed inferringthe existence of Venus’ atmosphere, the bright arc being correctly ascribed tothe refraction of light by the outer layers of a dense atmosphere. On June 8 th ,2004, fast photometry based on electronic imaging devices allowed the firstquantitative analysis of the phenomenon. Several observers used a varietyof acquisition systems to image the event – ranging from amateur-sized toprofessional telescopes and cameras – thus collecting for the first time a largeamount of quantitative information on this atmospheric phenomenon. In thispaper, after reviewing some elements brought by the historical records, we ∗ Corresponding author ∗∗ Deceased on Sept. 11, 2010.
Email address:
[email protected] (P. Tanga)
Preprint submitted to Icarus October 30, 2018 a r X i v : . [ a s t r o - ph . E P ] D ec ive a detailed report of the ground based observations of the 2004 transit.Besides confirming the historical descriptions, we perform the first photomet-ric analysis of the aureole using various acquisition systems. The spatiallyresolved data provide measurements of the aureole flux as a function of theplanetocentric latitude along the limb. A new differential refraction modelof solar disk through the upper atmosphere allows us to relate the variablephotometry to the latitudinal dependency of scale-height with temperaturein the South polar region, as well as the latitudinal variation of the cloud-toplayer altitude. We compare our measurements to recent analysis of the VenusExpress VIRTIS-M, VMC and SPICAV/SOIR thermal field and aerosol dis-tribution. Our results can be used a starting point for new, more optimizedexperiments during the 2012 transit event. Keywords:
Venus, Venus atmosphere, Planet transits
1. Introduction
Since the mid 18 th Century, observers have reported unusual features ofthe telescopic image of Venus near the inferior conjunction, promptly at-tributed to its atmosphere. Some of them are at the reach of modest in-struments (although at small angular distance from the Sun), such as thecusp extension first described by Schroeter (1791), which tends to transformthe thin crescent of Venus into a ring of light (e.g. Russell, 1899; Dollfus &Maurice, 1965).One of the most relevant features pertaining to ground-based studies ofthe Venus atmospheric structure has been observed only during transits,close to the phases of ingress (between 1st and 2nd contact) ot egress (be-tween 3rd and 4th contact) as a bright arc outlining – in part or entirely –the portion of Venus’ disk projected outside the solar photosphere. Tradi-tionally the first account of this phenomenon was attributed to Mikhail V.Lomonosov (1711–1765) who reported his observations of the transit at St.Petersburg Observatory on May 26, 1761 (Marov, 2005). Actually the poorperformance of his small refractor hints that most probably other observers(such as Chappe d’Auteroche, Bergman, and Wargentin) were the first gen-uine witness on the same date (Link, 1969; Pasachoff and Sheehan, 2012).However, Lomonosov correctly attributed the putative phenomenon to thepresence of an atmosphere around the planet, refracting the sunlight in theobserver’s direction. 2n the following, adopting a denomination widely used in the historicalaccounts, we will often call this arc “aureole”. Since both the aureole andthe cusp extension occur close to the planet terminator, they are also col-lectively known as ”twilight phenomena”. For detailed historical reviews theinterested reader can refer to Link (1969) and Edson (1963). Of course, inthis context we neglect the initial motivation for the observation of transits:the determination of the Astronomical Unit by the solar parallax (proposedby E. Halley in 1716), whose interest is purely historical today.Venus transits are rare, as they occur in pairs 8 years apart, each pairseparated by 121.5 or 105.5 years, alternating between descending node(June pairs: 1761/1769, 2004/2012), and ascending node (December pairs:1631/1639, 1874/1882, 2117/2125). As a consequence, data concerning theaureole are correspondingly sparse and, up to the last event, they have beenobtained by simple visual inspection, mainly by using refracting telescopes ofmodest aperture (typically up to 15-20 cm). This limitation was mainly dueto the constraint of organizing complex expeditions including the delicatetransportation of all the instruments.The 2004 event represents a giant leap in the observation of Venus tran-sits, as the modern imaging technologies available allow for the first time aquantitative analysis of the atmospheric phenomena associated to the transitof Venus.We performed measurements of the aureole on the original images ob-tained through several different instruments, and compared them both witha simple refraction model and with observations obtained in the past. Thiswork summarizes the aspect of Venus close to the Sun’s limb during theJune 8, 2004 transit as observed by ground–based instruments. A recent pa-per (Pasachoff et al., 2011) deals with imaging using NASA’s then operatingTransition Region and Coronal Explorer solar observatory (TRACE).Since the ground-based observations were not specifically organized be-forehand, data at our disposal are rather heterogenous. In order to bridge thegap between past visual observations through small telescopes and today’stechnologies, we decided to consider accounts obtained both with profes-sional instruments and low-cost amateur telescopes, either by CCD imagingor by direct image inspection by experienced observers. In fact, as shownin recent studies of distant solar system objects based on stellar occultationcampaigns (e.g. Widemann et al., 2009) while CCD imaging offers today themost valuable quantitative measurements, small telescopes and visual obser-vations allow a significant increase of constrains on the phenomena - and in3he case of Venus, the most direct comparison to past reports. The resultsobtained from the analysis of the most significant image sets, representing acertain variety of instrument size and quality, are illustrated in this paper.The paper is organized as follows. First, we describe the conditions ofthe 2004 event, and the reconstruction of limb geometry (section 2). Wethen provide an extensive review of the measured spatial and temporal vari-ations for the brightness of the aureole in wide or narrow-band photometry(section 3 and 4). We then address the basic physical principles of the at-mospheric differential refraction model producing the aureole (section 5) andwe use it for the interpretation of the observations (6). In Section 7, themodeling of the aureole is compared to recent analysis of the Venus Expressobservations (VIRTIS-M, VeRa and SPICAV/SOIR) regarding the thermalfield and cloud-deck altitude and haze distribution pertaining to this study,as well as ground-based mid-infrared spectroscopy of non-LTE CO emission.The comparison is discussed.
2. Geometry of the transit in 2004
For simplicity and following Link (1969) hereinafter we call “phase” ( f )of the event the fraction of Venus diameter external to the solar photosphere.A value f =0 corresponds to the planet entirely projected on the Sun, tangentto its limb. When f =0.5, the planet centre will be exactly on the solar limb,and so on.In Fig. 1 the orientation of the disk of Venus relatively to the solar limbis given, both for ingress and egress phases. In both cases the temperatelatitudes are tangent to the Sun for f = 0 and f = 1. At first contact theSouth pole of Venus remains projected longer on the sky, while the Northpole is the first one to enter on the solar disk . The sequence is invertedbetween third and fourth contact, such that it is always the South pole tobe observed externally to the Solar disk for a longer time.The total limb crossing for the disk of Venus lasted 18.9 minutes and theapparent radius of the planet was 28.9 arcsec. In the following we will always use the IAU convention i.e. the North pole is the onelying on the northern side of the ecliptic. Figure 1: Sketch of Venus’ disk orientation at the ingress (left panel) or egress (right) of thetransit. The greyed area corresponds to the Sun’s photosphere at the epoch when the Venusdisk is externally tangent to the Sun (first and fourth contact: t and t respectively).The Solar limb is also indicated at the second and third contact (labeled t and t ) andwhen projected on the center of Venus. The solid arrows indicate the direction of theapparent motion of the planet relative to the Sun. The vertical dashed line correspondsto the sky North-South direction, while the dash-dotted one represents the sky-projectedrotation axis of Venus. . Observations and measurements The European observers providing the data sets described further on hadparticularly favourable conditions around the end of the event, while theingress of the planet on the solar disk was observed at much higher airmass,i.e. at low elevation above the horizon. However, visual observers undergood sky conditions and employing a magnification higher than ∼ × hadno particular difficulty in identifying the bright aureole outlining the Venusdisk between 1st and 2nd contact, while it was crossing the Solar limb (i.e.for f < Figure 2: Drawings of Venus at the beginning (three leftmost panels) and end of the event(remaining two panels), as seen visually by an expert amateur astromer (Mario Frassati,Italy) through a 20 cm Schmidt-Cassegrain telescope. Courtesy of the archive of SezionePianeti, Unione Astrofili Italiani.
For a quantitative analysis of this phenomenon we rely on images ob-tained by CCD cameras through different telescopes. Given the casual na-ture of most images of the aureole that have been produced during the eventthrough instruments of all sizes, not all of them are suited for a compre-hensive analysis. We thus selected the representative sample of observationspresented in the following, with appropriate image quality and informationcontent. Table 1 summarizes the actual contributions that were collectedand analyzed, as presented in the following.6 able 1: Measured observations. The availability of ingress (”In” column) or egress (”Ex”)images is given. The number of measured images follows. These are final images, as aresult of an averaging process when needed. The final column identifies the filter nameand/or their central wavelength/width. See case by case detailed explanations in the text.
Observer Instrument Site In Ex N. images λ /FWHM (nm) J. Arnaud Themis Tenerife (Spain) - X 50 VL. Comolli 20 cm Schmidt-Cass. Tradate, Italy X X 50 VP. Suetterlin DOT La Palma (Spain) X 21 430.5/1.0, GX 21 431.9/0.6, Blue cont.X 21 655.0/0.6, Red cont.X 21 396.8/0.1, Ca II HS. Rondi 50 cm Tourelle refr. Pic du Midi (France) X 32 NaD1/10
Images have undergone a standard calibration process (dark current andflat–field correction). The brightness of the aureole, when present, was mea-sured. Fig. 3 shows an example of CCD image sequence between 3rd and4th contact on June 8, 2004. The bright arc or aureole is clearly visible inthe three frames obtained at an increasing distance of the Venus disk centerfrom the Sun. For illustration purposes a contrast enhancement is applied.The inhomogeneity of the aureole is clearly visible on the motionward limbof Venus.
Figure 3: Three images concerning the final phases of the transit obtained by the DOTtelescope in the ”G” band. Each image has been altered in contrast (gamma = 0.4) inorder to better show the thin and faint arc of light outlining (partially or completely) theVenus limb. The orientation is the same as in Fig. 1. The bright saturated area is thesolar photosphere. The progressive reduction in extension and brightness of the aureolewith time is clearly visible.
The arc’s photometric profile was obtained by the integration of the signalin a ring centered on the planet, outlined in Fig. 4. The full ring is dividedinto sectors of identical angular extension (as seen from the center of theVenus disk), each corresponding to a flux integration area. Among them,7nly those sectors projected on the sky are then considered. The background N Figure 4: Scheme of the procedure adopted for the measurements and implemented ina custom software. Here black is used to represent the sky background and the diskof Venus. The four white circles define three concentric rings, with the aureole entirelycontained in the central one. The nearby external and internal rings are used for evaluatingthe background. Each ring is divided in sectors of equal angular size. One of themis represented for the aureole (solid rectangle) and the corresponding background areas(dotted). The dashed angle represents the position angle (relative to the celestial North)used for representing the brightness profiles in following plots. The dashed-dotted linerepresents the orientation of the rotation axis of Venus. is evaluated on two rings, one on the inside and the other on the outside ofthe main ring. Their outer and inner edges (respectively) are also plottedin Fig. 4. Each background ring is divided in sectors as the measurementannulus. This way, each patch on the main annulus is accompanied by twobackground patches at the same position angle (relative to the Venus diskcenter) but on opposite sides. Their average flux, weighted by their surfaces,yields the final background value to be used for the given measurement patch.In the images we selected, the solar disk is not saturated, so a normal-8zation relative to a selected photospheric region is possible. We decided touse the value of brightness measured at 1 Venus radius from the solar limb.On the images, the geometric limb is estimated by the analysis of the radialbrightness profile of the solar disk. The photometric normalization value isprovided by the average of the flux in a square window of the same surface asthe sectors used for measuring the aureole brightness. A final normalizationis done for deriving the plotted values, corresponding to the flux coming froman unresolved, thin segment of aureole 1 arcsec in length, normalized to thebrightness of a square patch of photosphere having a surface of 1 arcsec .In the following, we present technical details of each data set, with thecorresponding aureole luminosity derived. The error bars given in the plotsare computed by considering the Poissonian statistics of the photometricnoise. We indicate by N s and N b the number of photoelectrons comingfrom the aureole and the background, respectively, in each arc bin. Theuncertainty that we consider is σ = ( N s + N b ) . . In our case, N b (cid:29) N s , thusthe only relevant contribution to noise comes from the background, and is afunction of the distance of the arc element from the solar limb. The cameraread-out-noise is cleary smaller than the background contribution and can bediscarded as well. The same applies to noise associated to the dark current,since exposure times are extremely short.From a practical point of view, we have verified that σ appears to be fairlyconstant over the entire arc, growing of about 50% only for measurementpoints very close to the solar limb. For this reason, we choose to show only oneerror bar, common to all curves, except in the case of observations obtainedat the DOT telescope which are used for modelling the aureole flux. In therelated plots, two error bars separately for low and high f values, provide anadditional information on the amount of noise variation in time. Only the ingress has been imaged by this station, since clouds have pre-vented a clear view of the final phases. Given the image scale, the radius ofVenus is 142.5 pixels. For a set of selected images in which the aureole iswell visible at the visual inspection, its integrated flux was determined us-ing different widths of the measuring ring, from 2 to 14 pixel. The resulting”growth curve” (Howell, 1989) shows the convergence toward the inclusion ofthe complete flux coming from the arc. The flux was considered to be com-pletely inside the ring at a width of 9 pixel, then used for all measurements.The arc was sampled at angular steps of 9 degrees.9lso, it was found that due to the faintness of the aureole, a useful im-provement of the SNR was obtained by averaging images by groups of 3.The final image set obtained this way was much easier to measure than theoriginal single frames. In the following discussion of Pic du Midi data we call”images” the members of this final set.The arc is present on 16 images, and a photometric profile was derivedfor each. Given the poor signal-to-noise ratio (SNR) the curves were lateraverage by four, obtaining the result presented in Fig. 5. The profiles havebeen trimmed in order to exclude the region in which the measurement ar-eas (main or background patch) are contaminated by the photosphere. Inother words, only the portion corresponding to the arc projected on the skyis plotted. In the case of the Tourelle telescope, the background stronglycontaminates the signal when the planet is about halfway into the ingress.We thus had to be very conservative concerning the fraction of arc to beconsidered.Despite the averaging, the result is still affected by a significant noise,as indicated by the size of the vertical bars in Fig. 5, representing the 1-sigma level. For this reason, small scale fluctuations in the brightness profileprobably do not have a strong statistic significance if considered separately.However, some general trends are present and repeat from one curve to theother. For example, a general slope of the curve bottom is present. Also,at large values of f (lowest curves) a maximum in intensity is detectableclose to the South pole of the planet. Interestingly, the maximum is notexactly centered on the pole, a feature that repeats in the other data setsdescribed further on. The intensity minimum coincides rather well withthe position angle of the equator of Venus (bar on the left). Even at latertimes, represented by the curves at the highest values, the trend of growingbrightness toward the pole is present. In all cases, if we neglect the extremetips of the arc very close to the photosphere, the aureole flux is less thanabout 2% the photospheric value, and is detectable down to about 0.2%. L. Comolli observed the event by a digital video-camera, with exposuretimes variable between 1/2000 and 1/50 sec. Only the shortest exposureshave been used, since they do not present saturation on the solar photosphere,thus allowing a normalization equivalent to those presented in the other cases.Each analyzed image is the result of the sum of the best 200-300 frames over ∼ igure 5: Photometric profile of the aureole during ingress, obtained at the Tourelletelescope, Pic du Midi. The plotted value represents the flux coming from a segment of1 arcsec of the aureole, normalized to the flux coming from a 1 arcsec surface of thesolar photosphere at 1 Venus radii from the solar limb (measured on the same set ofimages). The horizontal axis shows the position angle along the Venus limb, with theusual conventions in the equatorial reference: origin of angles at the celestial N, increasingtoward E. The two vertical lines represent the position angle of the equator (left, ”E” label)and the South pole (right, ”S”). The flux from the aureole increases with time. Uppercurves, with the planet about halfway across the solar limb, cover a short angular rangesince their extremes are strongly contaminated by a high background. Labels indicate thevalues of f associated to each curve. igure 6: Photometric profile of the aureole during ingress, obtained by L. Comolli, nor-malized as the other image sets. Vertical bars indicate the entity of the 1-sigma uncertaintyfor each data point. The left bar represents the position angle of the equator, while theright bar refers to the South pole. Labels indicate the values of f associated to each curve. ∼ Themis data are complementary to the Pic du Midi observations, sincefrom this site it was the exit phase to be imaged, both through a narrow-bandH-alfa filter and a filter centered at 430 nm. The image sampling is lower(Venus radius ∼
70 pixels) but a large image rate is available (1 image each1.06 seconds). The sequence was thus divided into different sets of images,containing 20 to 60 frames each. The most populated sets correspond to thelargest values of f , containing an extremely faint aureole slowly fading intothe background.Inside each set, each image was re-sampled with a re-alignment of the diskof Venus relatively to an image chosen as reference. An automated procedurebased on the computation of cross-correlations by Fast Fourier Transformswas used, providing the shift to apply to each image. Small relative imagedrifts are thus minimized and the SNR of the faint arc is preserved duringthe summing procedure that yields a final image for each set.In this way, 50 images were derived from the sum of the frames insideeach set, and measured as those of the previous section. The normalizationprocedure adopted was also the same. Due to poor seeing, a width of 7pixels was needed to include the entire flux. On the other hand, the goodSNR allowed to use a step of 2.25 degrees (corresponding to 2.75 pixels),in order to avoid losing information about small scale variations. However,the inspection of the resulting curves did not reveal any clear feature at thatresolution level, much smaller than the arc width smearing due to turbulence.A smoothing by running box on three points was thus applied, and the finalcurves are presented in Fig. 7. The overall brightness of the aureole results to13 igure 7: Photometric profile of the aureole during egress, obtained at the Themis solartelescope, Tenerife. The intensity is normalized as in Fig. 5. The two vertical linesrepresent the position angle of the equator (right) and the South pole (left). The curveshave been split into two families: the upper panel refers to the beginning of the egress,while the bottom one, with expanded vertical scale, refers to the central and final phases.The labels indicate the values of f associated to the corresponding curves.
14e 5% of the photospheric reference at maximum, with the shortest measuredarc. The results at Themis shows some features that are symmetric to thoseobserved at the Tourelle telescope during the entry. Here, the brightnessincreases from the equator (right bar in the plot, label ”E”) toward the pole(left bar, ”S”). Again, when the polar region is observable projected on thesky, a maximum in brightness is detected close to the pole, at about 25degrees from it. This is the last portion of the aureole that remains visible.Another interesting feature is the relative maximum at position angle ∼ ◦ ,(latitude ∼ -65 ◦ ) which is constantly present during most of the evolution ofthe arc profile. For several reasons, DOT data constitute the most valuable dataset mea-sured for this work, thanks to the optimization of the instrument for solarobservations (Bettonvil et al., 2003; Rutten et al., 2004). First of all, theimage sampling (0.07 arcsec/pixel) provides a comfortable scale for measure-ments, resulting in a Venus disk with a radius of 415 pixels. Also, imagesin four narrow bands (usually employed for solar studies) are available, allhaving a FWHM of 1 nm or less (Table 1). Four cameras were working si-multaneously in the four bands, taking every minute an image burst of 100frames at 6 frames/sec. For our measurements, all frames of each burst havebeen aligned and summed up to obtain a single image each minute. Dueto very strong turbulence the alignment process has been very critical, andwas obtained by applying a Sobel edge enhancement filter with a thresholdand then using the Hough transform (Yuen et al., 1989) to find the preciseposition of the disk of Venus in each image.Fluxes were obtained for both the aureole and the background with thesame method as with the previous data set. A step of 2 degrees along theVenus limb was used for measurements. We show in Fig. 8 the photomet-ric profile referred to the G band. This result is very similar to both theblue and red continuum images (Figures 9-10). Only in the Ca II H band(Fig. 11) some differences appear under the form of additional peaks in theprofile. Although a signature of high-contrast chromospheric structures intheir refracted image cannot be completely ruled out, those features are mostprobably related to the inner corona extending outward from the solar limband contaminating the signal of the aureole, since its flux cannot be easilyexcluded from the measurement areas.15 igure 8: Photometric profile of the aureole during egress, obtained at the DOT solartelescope, La Palma, in the G band. The intensity is normalized as in Fig. 5. The twovertical lines represent the position angle of the equator (right) and the South pole (left).The curves have been split into two families: the upper panel refers to the beginning ofthe egress, while the bottom one, with expanded vertical scale, refers to the central andfinal phases. The time interval between each curve is 60 seconds. For an easier visibilityof the bottom panel, each curve has a step of 7 × − added in intensity relatively to itslower neighbor. The labels indicate the values of f associated to the corresponding curves.Two levels of uncertainty ( σ ) related to a position angle in the range 150-250 ◦ are givenin the two panels. The upper one is applicable to values of f < f =0.7. igure 9: As in Fig. 8 in the blue continuum at 431.9 nm. igure 10: As in Fig. 8 in the red continuum at 655 nm. Relative to the other wavelengthbands, here 3 curves are missing due to corrupted data files. igure 11: As in Fig. 8 in the CaII band at 396.8 nm.
4. Lightcurves and colors
In order to study a possible wavelength dependence of the aureole and itsvariation in time, we used the DOT images for the egress phase. The flux wasmeasured on three arc segments centered on the pole, on the brightest partof the aureole at position angle 195 ◦ (position A) and on the faintest part at218 ◦ (position B) as shown in the scheme of Fig. 12. Point B corresponds toa region at latitude 45 ◦ S.The measurements were made on arc lengths of 10 degrees, in the G band,blue and red continuum. The results are shown in Fig. 13. On all the threeplots, different wavelengths appear to follow a similar trend. This is expectedespecially for the blue continuum and the G band, which are very close inwavelength.The fastest fall of flux in red continuum beyond t ∼
12 minutes is a featurecommon to all the light curves, suggesting a certain chromaticity of theaureole. However, the related error bars are very large in this interval. Atthe same time, images around t ∼ ∼
11 min is almost certainly spurious (due to the similar wavelength) andfurther underlines the difficulty of comparing very low fluxes.In conclusion, we cannot firmly state that the images at our disposalreveal a significant departure of the aureole color from the solar spectrum.20 igure 12: Geometry at f = 0 . A corresponds to the brightest portion, and B to the faintest.
5. Modeling the aureole
The aureole observed during Venus transits can be explained by the re-fraction of solar rays through the planet outer atmospheric layers. The raysthat pass closer to the planet center are more deviated by refraction thanthose passing further out. The image of a given solar surface element isflattened perpendicularly to Venus’ limb by this differential deviation, whileconserving the intensity of the rays, i.e. the brightness of the surface elementper unit surface. This holds as long as the atmosphere is transparent, i.e.above absorbing clouds or aerosol layers.The refractive deviation of light is related to the physical structure ofthe planet atmosphere. The formalism of this deviation has been studied byBaum and Code (1953), for the cases of stellar occultations by planets. Theirapproach assumes that ( i ) the local density scale height H of the atmosphereis constant and much smaller than the planet radius R v , ( ii ) the atmosphereis transparent, ( iii ) and it has spherical symmetry.This approach is still valid in our case, but has to be modified to ac-count for the finite distance and size of the Sun. In the following we willconsider that the physical properties of the transparent atmospheric layers21 igure 13: Relative intensity of the arc brightness in the DOT images, normalized as inthe previous plots, as a function of time, where the origin of time corresponds to thethird contact (at 11:07:25 UT). The three panels refer to the Pole and to points A, B(in the order, from top to bottom). The G band, red continuum, and blue continuumare represented by the solid, dotted and dashed line respectively. The model fit is thethickest solid line in the top and bottom panel. At the B point, H=3.1 km and ∆r=26 kmprovide the best fit, while at the pole the curve corresponds to H=4.8 km and ∆r=38.5 km(see Sect. 5 for details). The cross at t ∼
15 min represents an isolated red measurementavailable. Error bars are very similar for all the three bands. H (cid:28) R v is valid here, as H is smaller than theatmospheric layer of Venus, in turn (cid:28) the planet radius. Note also thatthe locally spherical symmetry is achieved for Venus’ atmosphere. As to thetransparency assumption, it must be dropped when the rays are going deepenough for the atmosphere to become opaque.We model the aureole brightness as follows:(1) We consider a surface element dS on the Sun. A ray emitted by dS will reach the observer O after being refracted by an angle ω in Venus’satmosphere (see Fig. 14). Note that we adopt here the convention ω ≤ dS corresponds an image dS (cid:48) that will appear as an aureolenear Venus’s limb (Fig. 15).During its travel to Earth, the ray passes at a closest distance of r from theplanet center. Furthermore, dS projects itself at a distance r i from Venus’center C , as seen from O . Note that r i is an algebraic quantity which isnegative if dS and dS (cid:48) project themselves on opposite side relative to theplanet center C , and positive otherwise, see Figs. 14-15.(2) The refraction angle ω is given by Baum and Code (1953): ω = − ν ( r ) (cid:112) πr/H, (1)where ν is the gas refractivity, decreasing exponentially with r . This quantityis related to the gas number density n by ν = K · n , where K is the specificrefractivity. For CO , we have K = 1 . × − m molecule − .(3) It can be shown that surface element dS (cid:48) is radially shrunk withrespect to dS by a factor Φ = 1 / [1 + D · ( ∂ω/∂r )], where D is the distancefrom Venus to Earth. Note that since the atmosphere is assumed to have aconstant density scale height, ∂ω/∂r = − ω/H , thus:Φ = 11 − Dω/H (2)(4) It is convenient to take as a reference radius the closest approachdistance r / corresponding to Φ = 0 . ω / = − H/D . Re-arranging the variousequations given above, one finds K · n / = (cid:113) H / (cid:0) πr / D (cid:1) , where n / ' D rr i S un s u r f a c e Venus observerCdSdS' O ! " Figure 14: Geometry of the refraction of solar rays by Venus’ atmosphere, where D (cid:48) (resp. D ) is the distance of Venus to the Sun (resp. Earth). A ray emitted from the solarsurface at dS , is deviated by an angle ω (here negative) before reaching the observer O ,who observes the images dS (cid:48) just above Venus’ limb (the aureole). The observer sees dS projected at algebraic position r i in the plane going through Venus and perpendicular tothe line of sight. The origin of r i is at Venus’ center C and r i increases upward, see alsoFig. 15. All sizes and angles have been greatly increases for better viewing. r i CSun surface Venussky dr i dr’ dSdS' ll i Figure 15: Venus (dark gray disk) observed from Earth, partly against the solar disk andpartly against the sky (black background). Each solar surface element dS surrounding r i has a refracted image dS (cid:48) of length l and width dr (cid:48) , caused by Venus atmosphericrefraction. The image dS (cid:48) has the same surface brightness as dS if the atmosphere istransparent. See text for details.
25s the gas number density at r / . The latter equation permits to derive thenumerical value of the half-light radius, once an atmospheric model of theplanet is given, that is, once the density profile n ( r ) is specified.(5) From Eq. 1, we derive: ω = − HD · e − ( r − r / ) /H (3)Simple geometrical consideration show that we also have: ω = − k ( r − r i ) /D ,where k = 1 + D/D (cid:48) ( D (cid:48) being the distance of Venus to the Sun). Combiningthis equation with Eqs. 2-3, we obtain:1 k (cid:18) − (cid:19) + log (cid:18) − (cid:19) = r / − r i H (4)This is the Baum and Code formula (apart for the correcting factor k ,which is equal to unity in the original formula since D (cid:48) = + ∞ for stellaroccultations).Thus, for each value of r i , we can calculate the corresponding value Φ( r i ),using a classical Newton iterative numerical scheme.For each value of r i , we can also calculate the corresponding closest ap-proach radius r by combining Eqs. 2 and 3: r = r / − H · log (cid:18) − (cid:19) (5)To calculate the flux dF received from an aureole element of surface dS (cid:48) with length l and width dr (cid:48) (Fig. 15), it is enough to note that the surfacebrightness of dS and dS (cid:48) are the same if the atmosphere is transparent, i.e.that dF = S (cid:12) ( r i ) ldr (cid:48) , where S (cid:12) ( r i ) is the flux received from a unit surfaceon the Sun at r i (taking into account the limb-darkening effect). Thus, wecan re-write this equation as dF = S (cid:12) ( r i )Φ( r i ) l i dr i .The aureole is not radially resolved, so we have only access to the fluxintegrated along r i , i.e. to: F = (cid:90) r i, max r i, min S (cid:12) ( r i )Φ( r i ) · l i · dr i (6)The lower bound of the integral, r i, min , corresponds to the value of r corresponding to an opaque cloud or aerosol layer at altitude r cut . The upper26ound of the integral, r i, max , corresponds the solar limb, beyond which nomore photons are emitted.By applying this model, in the following we will determine the scale height H and the half occultation radius relative to slanted opacity τ ∼ r = r / − r cut ) best reproducing the observations. In general, different portions ofthe arc can yield different values of these parameters, thus providing a usefulinsight of variations in the physical properties of the Cytherean atmosphereas a function of latitude.
6. Derivation of the physical parameters
We used the equation (4) for modeling the refraction in the atmosphereof Venus and the flux in the observed aureole. Essentially, the mathematicalmodel depends upon two parameters: the altitude of the half-occultationlevel relative to an underlying totally opaque layer (∆r) and the optical scaleheight of the atmosphere H, second free parameter of Eq. 4.
Figure 16: Relative intensity of the arc brightness for H=4.8 km, ∆r ∼ f fraction of disk overlap. Since our brightness measurements are referred to different epochs, themodel must fit at the same time not only single-epoch profiles, but also their27volution. Although free parameters of Eq. 4, r / and H are quantities thatare not physically independent since they are both related to local physicalstructure of the mesosphere of Venus. We will thus check afterwards thephysical consistency of our results.The source of refracted rays (the solar photosphere) is computed as asmooth function of the radial photometric profile of the Sun (given by Hes-troffer et al. 1998), whose parameters are determined by a fit to the profilemeasured on the solar photosphere imaged close to the Cytherean disk.Fig. 16 shows a typical outcome of the model for different phases f ofVenus ingress or egress, considering the atmospheric properties to be constantall along the planet limb. When Venus appears largely overlapped to the solardisk, the refracted flux is dominated by the arc extremes, where it approachesthe solar limb. This is a consequence of a very small bending of the light pathoccurring high in the atmospheric layer involved. When f > f the signal rapidly fades. The cut-off at zero flux is a feature of the model due to the opaque layer, blockingthe refracted rays that cross the atmosphere below a given altitude.Diminishing H determines a stronger refraction that increases the amountof light coming from regions that are farther away from Venus on the solarphotosphere. The slope of the flux decrease that is observed when Venusexits the photosphere (i.e. when f is increasing) will also be less steep (Fig.17).Another feature of the model outcome is the symmetry relative to theaxis joining the center of the disks of Venus and the Sun. It is thus clear thatthe asymmetries and features visible in the measurements must be relatedto latitude variations of the physical properties in the refracting atmosphericlayers.We searched for the values of ∆r and H that better fit the time variationsreproduced in Fig.13, separately for the Pole (or, equivalently, the point A)and the point B. The procedure we adopted started with a search for thevalue of ∆r, which mainly affects the amount of aureole flux at small f (i.e.when Venus is nearly completely on the solar disk). Then, H is varied inorder to obtain the right slope for the flux variation in time. The procedure28s repeated until a convergence to the observed data is obtained.The central portion of the arc, for which the longest time series is avail-able, is best fitted by H eq =3.1 km and ∆r eq =26 km (Fig. 13). This regioncorresponds to intermediate latitudes ( ∼ ◦ - point B).We then fitted the Pole area, obtaining H pole =4.8 km and ∆r pole =38.5 km.This determination appears less satisfactory at low f values (beginning ofegress). If a better fitting of this part of the curve is forced (in particular byincreasing ∆r) the central plateau of the curve becomes more extended, andthe fit is worsened in the rest of the domain. We thus preferred to privilegethe overall better fit. This choice appears to be reasonable when comparedto the expected altitude difference of cloud top and upper haze between theequator and poles (see next section).In fact, if refraction occurs at similar absolute altitudes in the atmosphere,variations of ∆r should only correspond to changes in the altitude of theopaque layer, i.e. a slanted opacity (optical thickness) τ ∼
1. Our best fitvalues indicate that the higher flux received at the polar latitudes on Venuscan be related to a significantly lower aerosol extinction towards the poles,thus allowing more refracted rays to reach the observer. This contribution isblocked at the equatorial latitudes by the higher clouds.An indication of the formal accuracy of the determination of ∆r and Hcan be obtained by studying the fit sensitivity to changes in the parameters.We have verified that a change of ∼ ∼
100 m in H are sufficientto displace the fitting curve of an amount exceeding the extension of the errorbars all along the lightcurves. The rapid variations in the aureole behaviorfollowing changes in the parameters are illustrated in Fig.17. In particular,our results appear extremely sensitive on H. As a further consistency check,we can estimate ∆r from Eq. 3 by using a value of H ∼ f at whichthe signal from the aureole is detected) to be of the order of ω =40 arcsec. Inthis case ∆r ∼
35 km, very close to the results given above obtained by a fullmodel fit to the observed data.Eventually we can estimate the visibility of the aureole in terms of visualmagnitude. We use the simple law for the profile of the solar limb darkening(normalized to the Sun’s center brightness) by Hestroffer et al. (1998): I ( µ ) = 1 − u (1 − µ α ) (7)in which µ = (1 − r ) . . Here, r represents the distance from the center ofthe Sun, normalized to the solar radius. At a wavelength in the V-band range29 igure 17: Effects of the variation of the physical parameters controlling the aureole flux.In the top panel, H eq =4.8 km is kept constant, while the three simulated light curvescorrespond to different values of ∆r eq (indicated by the labels). On the other hand, in thebottom panel ∆r eq =38 km is constant and H is varied. λ =579.9 nm) this law fits the observed profiles when u=0.85 and α =0.8. Inthe case of our reference area the value of r is r v =0.969, so I v =0.428. Inthis region so close to the solar limb the profile approximation can be lessaccurate, but it should not be erroneous by more than 2% (Hestroffer et al.,1998).From the computed I v and using the magnitude of the Sun (V = -26.71)we obtain a surface magnitude per arcsec for the reference element m v = − .
87. By observing the plots, we can assume that the aureole has beeneasily seen and imaged when its brightness was ∼ − times the surfacebrightness of the reference area (1 arcsec at 1 Venus radius from the solarlimb), i.e. 5 magnitudes fainter. We can thus consider that the typicalmagnitude of an arc of aureole 1 arcsec in length was -5.9.One should note, however, that ample variations around this value arepresent, both along the arc and over time.
7. Comparison to Venus Express observations
Venus’ mesosphere, which extends typically from 70 to 110 km, appearsto be a transition region between the troposphere (0–70 km), dominatedby the ∼ , SO , H O and chlorine compounds lead to the for-mation of sulfuric acid, which is the main component of the cloud and upperhaze, associated with low H SO vapor pressure (Zhang et al., 2010). At the limit between the troposphere and the mesosphere, the uppercloud structure reveals distinct dynamical structure in the latitudinal direc-31ion, with convective, wave-dominated zonal flow in the lower latitude rangenear Venus’s equator, and a significant transition to less convective bandedflow between 45 ◦ up to about 70 ◦ –80 ◦ . The temperature field forms a bulgeof cold air at 60 ◦ –80 ◦ latitude called the ”cold collar region”, which verti-cally extends up to 75 km. The torus-like structure encloses a vast vortexseveral thousand kilometers across, with a slower rotation period ( ∼ ∼ bands at 1.6 µ m measured by VIRTIS (Ignatiev et al. 2009), and CO+CO gaseous ab-sorption in the 4.5-5 µ m range using VIRTIS and VeRa temperature profiles(Lee et al. 2010) while Luz et al. (2011) brought the first extensive charac-terization of the vortex dynamics and its precession motion. Although theabsolute cloud top altitude poleward and equatorward of the polar collardiffers when applying the two modeling techniques (63–69 km / 74 ± ◦ –70 ◦ near the South polar limb in DOT telescope data (seeFig.8a of Ignatiev et al. 2009). Interestingly, similar agreement can be tracedback to Russell’s drawings of 1874 (See Link, 1969 and Fig. 1 of Pasachoffet al., 2011). The Venus upper haze (70–90 km) was first evidenced by measurementsfrom Pioneer Venus orbiter limb scans at 365 and 690 nm at northern mid-latitude. It is mainly composed of submicron sulfuric acid (H SO ) aerosolparticles with typical radii from 0.1 to 0.3 µ m (Lane and Opstbaum, 1983; Sato et al., 1996). Wilquet et al. (2008) also present the first evidencefor a bimodal particles distribution, similar to the two modes in the upperclouds, at the latitudes probed during the observation of solar and stellaroccultations by ESA’s Venus Express, with typical radii of mode-2 particlesbetween ∼ µ m. These measurements were performed close to thepolar regions. From the results of Wilquet et al. (2008) the absorption at dif-ferent wavelengths as a function of the altitude can be deduced. In general,32he absorption levels are found to be 6 ± µ m.As an integrated aerosol optical depth ∼ r / well above the upper haze.Quantitatively, the altitude of the aureole’s half–occultation level in the polarregion will be found by adding the value of ∆r pole =38.5 km to the altitudewhere aerosols slanted opacity τ ∼
1. Recent VEx/SOIR results (Wilquet etal., 2011, personal communication) place that altitude at 73 ± µ m band, to be further increased by 6 ± / ∼ . ± ∼ ± ◦ S and 55 ◦ S. Therefore, as∆r eq =r / − r cut is significantly smaller than ∆r pole the resulting value ofr / at mid-latitude differ from our calculation in the polar region by abouttwo scale heights (Table 2). We can thus consider that our results onlyshow marginal latitudinal variation of the altitude for the half-occultationlevel along the terminator, although in a region of important temperaturevariation which need to be independently assessed. The same SPICAV/SOIRresults show that, under reasonable assumptions, the refraction index atvisible wavelengths is fairly constant, thus confirming that the process is nota source of relevant chromatic effects in the aureole. Only scarce measurements of vertical temperature profiles have been per-formed above 100 km altitude, where temperature fields, especially in thepolar region, as well as their time variability, are still debated (see, e.g.,Vandaele et al. 2008, Clancy et al. 2008, 2011; Piccialli et al. 2008, 2011).Inverted equator-to-pole temperature gradient on isobaric surfaces above 75km, were first reported by NASA’s Pioneer Venus Infrared Radiometer andradio occultation measurements (Taylor et al., 1983; Newman et al., 1984),Venera-15 Fourier Spectrometry data (Zasova et al., 2007) and more recentlyESA’s Venus Express, e.g. Grassi et al. (2008) ; Piccialli et al. (2008) us-ing VIRTIS -M observations ; Tellmann et al. (2009) based on VeRa radiooccultations on board VEx. The collar region also divides the atmospherevertically. Below the collar, the atmosphere cools with increasing latitude.33 able 2: Results of modelling and retrieved half-light altitude r / , scale height H (in km)and temperature along the Lomonosov’s arc, assuming a locally isothermal atmosphere.Explanations of symbols in the text. T iso = m(z) g(z) H/k. T mod (K) from Hedin et al.1983. Point Lat. ∆r H r cut r / T iso (K) T mod (K)A 68 ◦ S 38.5 ± ± ± ± ◦ S 26.0 ± ± ± ± ∼
115 km region (see Table 2) as isothermal. Thisis an acceptable hypothesis since the diurnal average of dT/dz on the back-ground atmosphere (Hedin et al., 1983) is negligible, although significantlocal-time and daily variability has been retrieved (Clancy et al. 2008, 2011;Sonnabend et al. 2008, 2011). In particular, the elevated value of T iso =212K at 120.5 ± ◦ S is in agreement with high kinetic temperaturesderived from heterodyne mid-infrared spectrum of non-LTE emission at ter-minator, see in particular Table 1 of Sonnabend et al. 2011. This resultsuggests a warmer or rapidly variable temperature condition in the altituderange.
In order to check the consistency of our results with the known physicsof Venus atmosphere, we can compute the expected value of the density34cale height H at the altitude of the refracting layer. The temperature scaleheight is defined as H = kT ( mg ) − , where k is the Boltzmann’s constant,while T, m and g respectively represent temperature, molecular mass andgravitational acceleration at the altitude considered. As discussed above, weconsider a locally isothermal atmosphere. For our computation we derive thevalue of g as a function of altitude. We derived an altitude for the opaquelayer from SOIR aerosol transmittance measurements near 3 µ m (Wilquetet al., 2011 ; Wilquet, personal communication), and added 6 ± is the dominant gas onthe dayside below 160 km and on the nightside below 140 km, replaced athigher altitudes by O. Mean molecular mass remains close to 44 a.m.u. below130 km, where CO dominates, and decreases towards higher altitudes (Hedinet al., 1983; Mueller-Wodarg et al., 2006). Although day–night differencesin composition are substantial, we also may consider longitudinally averagedvalues at the altitude considered. Concerning m , since the layer is in theheterosphere, a decrease due to fractionation has to be taken into account,due to the decrease in CO relative to other components, in particular atomicoxygen O, molecular nitrogen N and He. As a result, at 120 km we use anaverage molecular mass of 42.9, and a day-night longitudinal average T=164K, obtaining H=3.7 km. An equal value is obtained at 115 km. Both arein remarkable agreement with the average one obtained from the aureolemeasurements, H=3.95 km (see Table 2). Finally, we noted that our resultsonly show marginal latitudinal variation of 1-2 scale heights for the altitudefor the half–occultation level along the terminator, near 117 km.
8. Conclusions
Our results represent the first successful model of the aureole of Venus,observed during solar transits. We are able to reproduce the main featuresof the lower mesosphere observational constraints : longitudinally averagedthermal structure near 117 km, slanted opacity of aerosols and their merid-ional variation, with consistent physical parameters.35he results obtained by this first set of measurements are encouragingand suggest that more accurate planning will produce more precise data.Improvements are also possible concerning the refraction model, which couldbe completed by a more realistic, gradual transition from transparency toabsorption at the cloud deck level based on most recent results on the latitu-dinal distribution of upper haze aerosols. Also, the contribution of scatteredlight in addition to refracted light could be included. In fact, the absorp-tion curves in Wilquet et al. (2008) Fig. 3 are wavelength dependent, i.e.the opaque layer altitude in our model can be function of the color. As aresult the aureole should not be perfectly ”grey”, suggesting that future ob-servations capable of accurate multicolor photometry should investigate thispossibility.Even a grey absorption would cause the solar flux in the aureoleto look dimmer than without aerosols, which would be wrongly interpretedas a drop in scale height, i.e. in local temperature at tangent point altitude,in the inversion code.In the two hemispheres, the cold collar temperature structure and thepolar regions are very similar (Tellmann et al., 2009), so the latitudinal vari-ations obtained for the 2004 transit around the South polar region mighttentatively apply to the North polar data we expect to acquire in June 2012.Recent interpretations also attest a general N-S symmetry in the latitudinalvariation of aerosols extinction (Wilquet et al., 2011). Given the differencein brightness between the arc and the solar photosphere, accurate measure-ments of this phenomena are challenging. The degree of turbulence, themagnification, the amount of light scattering in the optics are all elementsthat contribute in determining the effective visibility of the aureole in a giveninstrument, and the accuracy of the measurements. The transit of June 2012will be our last opportunity for observing the aureole using this spatiallyresolved technique, until the next pair of transits of Venus, which will be inthe ascending node, on 11 December 2117 and 8 December 2125.
Acknowledgments
We thank the referees for useful comments on the manuscript, M. Fras-sati and Unione Astrofili Italiani for the use of the drawing reproduced inFig. 2. We also thank Val´erie Wilquet, Ann-Carine Vandaele and ArnaudMahieux of the Belgium Institute for Space Aeronomy for sharing their workin progress on the longitudinally averaged optical extinction of mesospheric36erosols on Venus. R. Hammerschlag (Dutch Open Telescope, La Palma)provided valuable help and comments.
ReferencesReferences
Baum, W.A., Code, A.D., 1953. A photometric observation of the occultationof σ Arietis by Jupiter. Astron. J. 58, 108-112.Bettonvil, F.C.M., Hammerschlag, R.H., Sutterlin, P, Jagers, A.P.L., andRutten, R.J., 2003. Multi-wavelength imaging system for the Dutch OpenTelescope. In Keil, S.L, Avakyan, S.V., Innovative Telescopes and Instru-mentation for Solar Astrophysics, Proc SPIE 4853, pp. 306-317.Bougher, S.W., Alexander, M.J., Mayr, H.G., 1997. Upper atmosphere dy-namics: global circulation and gravity waves. In: Bougher, S.W., Hunten,D.M., Phillips, R.J. (Eds.), Venus II, Tucson, pp. 259-291.Bouguer, S.W., Rafkin, S., Drossart, P., 2006. Dynamics of the Venus upperatmosphere : Outstanding problems and new constraints expected fromVenus Express, Planet. Space Sc. 54, 1371-1380.Clancy, R.T., Sandor, B.J., Moriarty-Schieven, G. H., 2008. Venus upperatmospheric CO, temperature, and winds across the afternoon/eveningterminator from June 2007 JCMT sub-millimeter line observations, Planet.Space Sc. 56, 1344-1354.Clancy, R.T., Sandor, B. J., Moriarty-Schieven, G. 2011, Thermal structureand CO distribution for the Venus mesosphere/lower thermosphere : 2001-2009 inferior conjunction sub-millimeter CO absorption line observations,Icarus, in press.Dollfus, A., E. Maurice, 1965. ´Etude de l’allongement des cornes du croissantde V´enus. Comptes Rendus de l’Academie des Sciences, 260, 427-430.Grassi et al. 2008. Retrieval of air temperature profiles in the Venusian meso-sphere from VIRTIS-M data: Description and validation of algorithms, J.Geophys. Res., 113, E00B09. 37urwell, M.A., Melnick, G.J., Tolls, V., Bergin, E.A., Patten, B.M., 2007.SWAS observations of water vapor in the Venus mesosphere. Icarus 188,288-304.Hedin, A.E., Niemann, H.B., Kasprzak, W.T., 1983. Global empirical modelof the Venus thermosphere, J. Geophys. Res. 88, A1, 73-83.Edson, J.B., 1963. The twilight zone of Venus, in: Kopal, Z. (Ed.)Advancesin Astronomy and Astrophysics (London and New York), 2, 17-31.Hestroffer, D., Magnan, C., 1998. Wavelength dependency of the Solar limbdarkening, Astron. Astrophys. 333, 338-342.Howell, S.B., 1989. Two-dimensional aperture photometry: signal-to-noiseratio of point-source observations and optimal data-extraction techniques.Publ. Astron. Soc. Pacific 101, 616-622.Ignatiev, N. I. et al. 2009. Altimetry of the Venus cloud tops from the VenusExpress observations, J. Geophys. Res., 114, E00B43.Knollenberg, R. G., Hunten, D. M., 1980. The clouds of Venus - A synthesisreport. J. Geophys. Res., 85, 8059-8081Kuiper, G. P. 1954. Determination of the pole of rotation of Venus. Astro-phys. J. 120, 603-605.Lane, W.A., and Opstbaum, R., 1983. High Altitude Venus Haze from Pio-neer Venus Limb Scans, Icarus 54, 48-58.Lee, Y.J., Titov, D., Tellmann, S., Piccioni, G., Piccialli, A., Paetzold, M.,Drossart, P., 2010. Vertical structure of the Venus cloud top from the VeRaand VIRTIS observations onboard Venus Express, EGU General Assembly2010, 12, EGU2010-11522-1 (abstract).Lellouch, E., Clancy, T., Crisp, D., Kliore, A.J., Titov, D., Bougher S.W.,1997. Monitoring of Mesospheric Structure and Dynamics, in Bougher,S.W., D.M. Hunten, R.J. Phillips eds., Venus II, Tucson, pp. 295-324.Lellouch , E., Paubert, G., Moreno, R., Moullet, A., 2008. Monitoring Venus’mesospheric winds in support of Venus Express: IRAM 30-m and APEXobservations, Planet. Space Sci. 56, 1355-1367.38ink, F., 1969. Eclipse Phenomena in Astronomy, Springer-Verlag (Berlin).Luz, D., Berry, D. L., Piccioni, G., Drossart, P., Politi, R., Wilson, C. F.,Erard, S., Nuccilli, F., 2011. Venuss Southern Polar Vortex Reveals Pre-cessing Circulation, Science, 332, 6029, 577.Marov, M.Y., 2005. Mikhail Lomonosov and the discovery of the atmosphereof Venus during the 1761 transit in Kurtz, D.W., Bromage, G.E. (Eds.),Transits of Venus : New Views of the Solar System and Galaxy,IAU Colloq.196, Cambridge University Press, pp. 209-219.Mahieux, A., Vandaele, A. C., Neefs, E., Robert, S., Wilquet, V., Drummond,R., Federova, A., Bertaux, J. L., 2010. Densities and temperatures in theVenus mesosphere and lower thermosphere retrieved from SOIR on boardVenus Express: Retrieval technique. J. Geophys. Res. (Planets), 115, E14,E12014.Mueller-Wodarg I.C.F., Forbes J.M., and Keating G.M., 2006. The thermo-sphere of Venus and its exploration by a Venus Express AccelerometerExperiment, Planet. Space Sci. 54, 1415-1424.Newman, M., Schubert, G., Kliore, A. J., Patel, I. R., 1984. Zonal Winds inthe Middle Atmosphere of Venus from Pioneer Venus Radio OccultationData, J. Atmos. Sci., 41, 1901-1913.Pasachoff, J.M., Schneider, G., and T. Widemann, 2011. High-ResolutionSatellite Imaging of the 2004 Transit of Venus and Asymmetries In theCytherean Atmosphere, Astronomical Journal, 141, 112.Pasachoff, J.M., Scheehan, 2012. Lomonosov, the Discovery of Venus’s Atmo-sphere, and Eighteenth-century Transits of Venus, J. History and Heritageof Astronomy, submitted for publication.Piccialli, A., Titov, D.V., Grassi, D., Khatuntsev, I., Drossart, P., Piccioni,G., Migliorini, A., 2008. Cyclostrophic winds from the Visible and In-frared Thermal Imaging Spectrometer temperature sounding: A prelimi-nary analysis, J. Geophys. Res., 113, E00B11.Piccialli, A., Tellmann, S., Titov, D.V., Limaye, S.S., Khatuntsev, I.V.,P¨atzold, M., H¨ausler, B. 2011, Dynamical properties of the Venus meso-sphere from the radio-occultation experiment VeRa onboard Venus Ex-press, Icarus, in press. 39iccioni, G., et al., 2007. South-polar features on Venus similar to those nearthe north pole, Nature 450, 637-640.Russell, H.N., 1899. The atmosphere of Venus. Astrophys. J. 9, 284-299.Rutten, R.J., Hammerschlag, R.H., Bettonvil, F.C.M., Sutterlin, P., andde Wijn, A.G., 2004. DOT tomography of the solar atmosphere, Astron.Astrophys. 413, 1183-1189.Sandor, B.J., Clancy, R.T., 2005. Water vapor variations in the Venus meso-sphere from microwave spectra, Icarus, 177, 129-143.Sato, M., L. D. Travis, and K. Kawabata, 1996. Photopolarimetry analysisof the Venus atmosphere in polar regions, Icarus, 124, 569-585.Schroeter, J.H., 1791. Selenotopographische Fragmente zur Genauern Ken-ntniss der Mondfl¨ache, ihrer Erlittenen Ver¨anderungen und zeichnungen,Universit¨ats G¨ottingen.Schubert, G. et al. 1980. Structure and circulation of the Venus atmosphere,J. Geophys. Res., 85, 8007-8025.Sicardy, B., Colas, F., Widemann, T., Beisker, W., Kretlow, M., Ferri, F.,Lacour, S., Lecacheux, J., Lellouch, E., Pau, S., Renner, S., Roques, F.,Fienga, A., et al. 2006. The two Titan stellar occultations of 14 November2003, J. Geophys. Res., Vol. 111, E11S91Sonnabend, G., Sornig, M., Scheider, R., Kostiuk, T., Delgado, J., 2008.Temperatures in Venus upper atmosphere from mid-infrared heterodynespectroscopy of CO around 10 µµ