Venus’ clouds as inferred from the phase curves acquired by IR1 and IR2 on board Akatsuki

Icarus 248 (2015) 213–220

Contents lists available at ScienceDirect

Icarus journal homepage: www.elsevier.com/locate/icarus

Venus’ clouds as inferred from the phase curves acquired by IR1 and IR2 on board Akatsuki Takehiko Satoh a,b,⇑, Shoko Ohtsuki c, Naomoto Iwagami d, Munetaka Ueno a, Kazunori Uemizu e, Makoto Suzuki a, George L. Hashimoto f, Takeshi Sakanoi g, Yasumasa Kasaba g, Ryosuke Nakamura h, Takeshi Imamura a, Masato Nakamura a, Tetsuya Fukuhara j, Atsushi Yamazaki a, Manabu Yamada k a

Institute of Space and Astronautical Science, Japan Aerospace Exploration Agency, 3-1-1 Yoshinodai, Chuo-ku, Sagamihara, Kanagawa 252-5210, Japan Department of Space and Astronautical Science, School of Physical Sciences, The Graduate University for Advanced Studies (SOKENDAI), 3-1-1 Yoshinodai, Chuo-ku, Sagamihara, Kanagawa 252-5210, Japan c School of Commerce, Senshu University, 2-1-1 Higashimita, Tama-ku, Kawasaki, Kanagawa 214-8580, Japan d University of Tokyo, 7-3-1 Hongo, Bunkyo-ku, Tokyo 133-0033, Japan e National Astronomical Observatory, 2-21-1 Osawa, Mitaka, Tokyo 181-8588, Japan f Okayama University, 1-1-1 Tsushima-naka, Kita-ku, Okayama 700-8530, Japan g Tohoku University, 6-3 Aramaki, Aza-Aoba, Aoba-ku, Sendai 980-8575, Japan h National Institute of Advanced Industrial Science and Technology, 1-1-1 Umezono, Tsukuba, Ibaraki 305-8568, Japan j Hokkaido University, Kita-10, Nishi-8, Kita-ku, Sapporo 060-0810, Japan k Planetary Exploration Research Center, Chiba Institute of Technology, 2-17-1 Tsudanuma, Narashino, Chiba 275-0016, Japan b

a r t i c l e

i n f o

Article history: Received 19 December 2011 Revised 18 October 2014 Accepted 19 October 2014 Available online 28 October 2014 Keywords: Venus, atmosphere Atmospheres, structure Photometry Infrared observations Radiative transfer

a b s t r a c t We present phase curves for Venus in the 1–2 lm wavelength region, acquired with IR1 and IR2 on board Akatsuki (February–March 2011). A substantial discrepancy with the previously-published curves was found in the small phase angle range (0–30°). Through analysis by radiative-transfer computation, it was found that the visibility of larger (1 lm or larger) cloud particles was significantly higher than in the standard cloud model. Although the cause is unknown, this may be related to the recently reported increase in the abundance of SO2 in the upper atmosphere. It was also found that the cloud top is located at 75 km and that 1-lm particles exist above the cloud, both of these results being consistent with recent studies based on the Venus Express observations in 2006–2008. Further monitoring, including photometry for phase curves, polarimetry for aerosol properties, spectroscopy for SO2 abundance, and cloud opacity measurements in the near-infrared windows, is required in order to understand the mechanism of this large-scale change. Ó 2014 Elsevier Inc. All rights reserved.

1. Introduction The apparent brightness of a planet is controlled by several factors: the Sun–planet distance, the planet–observer distance, the solar phase angle (the angular distance between the Sun and the observer as viewed from the planet), and the planet’s reflectivity. A phase curve can be obtained by observing the planet’s apparent brightness at a wide range of solar phase angles. Venus, as an inner planet, can be observed at all phase angles from 0° (at superior conjunction) to 180° (at inferior conjunction) from the Earth, although observation near conjunctions is not easy. ⇑ Corresponding author at: Institute of Space and Astronautical Science, Japan Aerospace Exploration Agency, 3-1-1 Yoshinodai, Chuo-ku, Sagamihara, Kanagawa 252-5210, Japan. Fax: +81 42 759 8178. E-mail address: [email protected] (T. Satoh). http://dx.doi.org/10.1016/j.icarus.2014.10.030 0019-1035/Ó 2014 Elsevier Inc. All rights reserved.

The reflection of sunlight from a cloud-covered planet, such as Venus, is a consequence of light scattering by molecules and aerosols in the atmosphere. Reflected sunlight therefore carries a wealth of information regarding the aerosols (their size and other physical properties) and the atmospheric structure (Hansen and Travis, 1974). At a given phase angle, a, the reflectivity of the planet is fundamentally influenced by the scattering phase function of atmospheric molecules and aerosols, for the corresponding scattering angle, P(h) where h = 180°  a. The phase curve therefore allows us to infer the shape of the aerosol scattering phase function, an important key to identifying the aerosols. The phase curve of Venus has been studied for several decades (Irvine, 1968). Recently, Mallama et al. (2006) performed an analysis of the forward scattering part (a > 150°, near inferior conjunction) of such curves in several different spectrum bands. They combined the data from a ground-based telescope (1999–2004)

214

T. Satoh et al. / Icarus 248 (2015) 213–220

with that from the Earth-orbiting observatory, SOHO (September 2003 and June 2004). Needless to say, the advantage of a space telescope is that it is free from the contamination caused by light scattering in the telluric atmosphere when the target object (Venus) appears very close to the Sun. Conversely, for the backward scattering part (a  0°) of the phase curve, another possible means of acquiring high-quality data is to use a spacecraft which is located between the Sun and Venus. ‘‘Akatsuki’’, Japan’s Venus Climate Orbiter, happened to be in just that situation in March 2011 (Akatsuki was in inferior conjunction, as viewed from Venus) (Nakamura et al., 2014). Although the failure of the insertion into orbit around Venus in December 2010 was disappointing (Nakamura et al., 2011, 2014), we carried out a series of observations in order to acquire high-precision multiwavelength photometric data near a  0°, this information being largely unobtainable from the ground. In this paper, we report observation results in the 1–2 lm region, and their interpretation.

2. Observation and data reduction From February 2011 to May 2011, the cameras on board Akatsuki were operated so as to image Venus from significantly large distances (the closest approach being on March 16, 2011 at a distance of 1.276  107 km). Inferior conjunction of Akatsuki occurred on March 22, with a phase angle of smaller than 1°. Three cameras, UVI (Nakamura et al., 2007), IR1 (Iwagami et al., 2011), and LIR (Fukuhara et al., 2011), were operated throughout the observation period, whilst IR2 (Satoh et al., 2011) was switched off after March 25 in order for it to survive unfavorable thermal conditions during the perihelion passage. In what follows, we describe and analyze the data acquired with IR1 (1-lm infrared camera) and IR2 (2-lm infrared camera). Note that we did not use the IR1 data acquired in May 2011, at which time the Venus disk on the IR1 detector was smaller than 1 pixel. The large scatter in the data is believed to be due to the low signal-to-noise ratio,

Table 1 Summary of IR1 and IR2 observations.

a

Date (2011)

Venus-Akatsuki (104 km)

a (°)

IR1 (hh:mm)

IR2

Albedo (0.90 lm)

Albedo (2.02 lm)

February 28

1366

55.9

06:10

+3 min

1363

55.4

12:00

+3 min

0.045 0.045 0.045 0.040a 0.035a 0.025a

1361

55.0

18:00

–

March 07

1290

40.2

23:00

+3 min

March 08

1283

37.8

23:00

+3 min

March 10

1277

35.2

00:00

+3 min

March 16

1276

18.6

01:20

+3 min

March 16

1283

16.0

23:00

+3 min

March 18

1295

12.7

01:20

+3 min

March 20

1344

4.1

23:00

+3 min

March 22

1369

1.0

01:20

+3 min

March 22

1393

1.6

23:00

+3 min

0.327 0.331 0.342 0.325 0.331 0.337 0.327 0.330 0.345 0.430 0.429 0.433 0.453 0.458 0.454 0.476 0.483 0.472 0.666 0.666 0.657 0.664 0.663 0.657 0.675 0.654 0.665 0.713 0.696 0.693 0.693 0.708 0.679 0.709 0.670

March 23

1422

4.5

23:00

+3 min

March 25

1458

7.5

01:20

+3 min

March 27

1573

15.0

23:00

OFF

March 28

1619

17.4

23:30

OFF

March 29

1667

19.7

23:10

OFF

March 30

1718

21.9

23:10

OFF

Lower values due to higher temperatures of the IR2 detector (>70 K).

0.682 0.671 0.672 0.680 0.673 0.675 0.668 0.679 0.661 0.643 0.645 0.676 0.637 0.624 0.576 0.600

0.055 0.053 0.056 0.053 0.053 0.053 0.056 0.055 0.053 0.080 0.079 0.078 0.083 0.080 0.080 0.077 0.076 0.074 0.096 0.095 0.093 0.091 0.090 0.089 0.090 0.088 0.087 0.089 0.087 0.087 0.086 0.085 0.085

215

T. Satoh et al. / Icarus 248 (2015) 213–220

1

0.2

0.5 0.1

0

0

20

Scale for IR2

Albedo

Mallama et al. (2006) IR1 (0.90 μm) IR2 (2.02 μm)

0 60

40 o

Solar Phase Angle [ ] Fig. 1. Comparison of albedos as a function of phase angles (IR1, IR2, and Mallama et al.’s). Open circles: IR1, open diamonds: IR2, crosses: Mallama et al.’s I-band. The three solid diamonds represent IR2 data (February 28) that was anomalously low due to saturation of the detector caused by a slightly high temperature (>70 K). The IR1 and IR2 data are converted to albedo by the methods described in appendices and Mallama et al.’s data are scaled so that the albedo at zero phase angle is 0.57 (the geometric albedo shown in their Table 4).

Fig. 2. Venus standard cloud model compiled from ground-based, orbiter, and probe measurements (Esposito et al., 1983).

and/or the varying amount of light falling on the insensitive area (IR1’s aperture ratio is 50%). The IR1 and IR2 observation conditions for the 1–2 lm region are summarized in Table 1. For reflected sunlight observations, IR1 and IR2 are equipped with low-transmittance narrow-band filters (centered at 0.90 lm for IR1 and at 2.02 lm for IR2, the latter being in a CO2 absorption band).

All observations were performed by executing time-line (TL) commands that invoked pre-programmed observation macros. For photometry, an image of Venus was taken, preceded and followed by a dark image. The averaged dark image was subtracted from the Venus image and the resulting Venus image was recorded in the on-board data recorder. This was performed 3 times on every day of observation. We did not average or take the median of the 3 daily Venus images, as such processing might have degraded photometric accuracy. Instead, all 3 images on each day were used individually to measure Venus’ apparent brightness. The IR2 exposure commands were issued, by the observation macro, 3 min after each of the IR1 exposures (as indicated in Table 1). After applying standard image processing (removal of bad-pixels, and flat-field correction), aperture photometry was performed on each image using the Image Reduction and Analysis Facility (IRAF) developed by the National Optical Astronomical Observatory (NOAO). Venus’ disk subtended 3–4 pixels in the IR1 and IR2 images taken during the February–March observations. We therefore integrated the pixel counts in a circular area (Fcircle) of diameter 20 pixels. Just outside it, an annulus 5-pixels wide was used to estimate the average sky (or remaining dark) background (Asky). Here, the net flux from Venus is calculated as:

F Venus ¼ F circle  Asky  Ncircle where Ncircle is the number of pixels in the circular area. Then obtained FVenus is converted to its albedo as described in Appendices (A for IR1 and B for IR2). The resulting phase curves (albedos as a function of phase angles) are shown in Fig. 1, together with Mallama et al.’s (2006) I-band (0.90 lm) curve for comparison. Note that these phase curves are independent of distances (either the Sun to Venus, or Venus to the observer) as albedo is unitless and describes how much light is reflected compared to a perfectly-reflecting body. The albedo at phase angle 0° is so-called geometric albedo. Since, as described in A, the absolute calibration of IR1 involves substantial uncertainty due to discrepancy of pre-flight and post-flight calibration, the IR1 curve is multiplied by an additional factor so that the curve in the 30–60° range matches that of Mallama et al. (2006). The effect of this to the model parameters will be evaluated in the later section. The phase curves of IR1 and IR2 are noticeably different: the IR1 curve exhibits a prominent shoulder near a  20°, while the IR2 curve is fairly flat. The discrepancy between the IR1 curve and Mallama et al.’s I-band curve is also substantial, despite both of them being in a similar wavelength band (0.90 lm). In the following section, we perform radiative-transfer computations in order to interpret the causes of these differences. 3. Radiative-transfer analysis We refer to Esposito et al. (1983) for the standard cloud model (Fig. 2 and Table 2): the clouds are multi-layered and each layer

Table 2 Standard cloud model (Esposito et al., 1983). Region

Altitude range (km)

Optical depth, s (at 0.63 lm)

Mean diameter (lm)

Average number density (N cm3)

Upper haze Upper cloud

70–90 56.5–70

0.2–1.0 6.0–8.0

Middle cloud

50.5–56.5

8.0–10.0

Lower cloud

47.5–50.5

6.0–12.0

0.4 Mode Mode Mode Mode Mode Mode Mode Mode

500 1500 50 300 50 10 1200 50 50

1: 2: 1: 2: 3: 1: 2: 3:

0.4 2.0 0.3 2.5 7.0 0.4 2.0 8.0

216

T. Satoh et al. / Icarus 248 (2015) 213–220

Table 3 Aerosol parameters (Grinspoon et al., 1993). Particle mode 1 2 20 3 a

Modal radius, r (lm) a

0.1 1.0 1.4 3.65

Variance, r

Effective radius, reff (lm)

Effective variance,

1.56 1.29 1.23 1.28

0.2 1.2 1.6 4.3

0.33 0.07 0.04 0.06

veff

Made smaller so that the effective radius is 0.2 lm.

Fig. 3. Wavelength dependencies of aerosol optical thickness as computed from Mie theory. The solid curve is the total optical thickness, i.e., modes 1 plus 2 plus 3, if the population is tri-modal. The dashed line represents modes 1 plus 2, and the dotted line represents mode 1 only. Note that for mode 1 optical thickness decreases as the wavelength gets longer, whereas there is little change for modes 2 and 3.

has an aerosol size distribution of uni-modal (mode 1), bi-modal (modes 1 and 2) or tri-modal (modes 1, 2, and 3) type. To simplify the model, the aerosol properties used throughout these layers are taken from Grinspoon et al. (1993) (see Table 3). A modification is made for mode 1 particles: the modal radius is changed from 0.3 lm to 0.1 lm so that the effective radius (reff) is 0.2 lm, being in agreement with those in Esposito et al.’s cloud model and other observations (Kawabata et al., 1980; Wilquet et al., 2009). The model parameters are summarized in Tables 2 and 3. The single scattering phase function and extinction cross section is computed based on Mie theory (Fig. 3) and the intensity of reflected sunlight from Venus is computed with the doubling-adding radiative transfer code that accounts for the effect of multiple light scattering

(Hansen and Travis, 1974). Due to the multiple light scattering in Venus’ atmosphere, it is impossible to uniquely determine all the cloud parameters. In particular, information from deeper levels (below several optical depths) is completely masked. In contrast, the information in the upper few optical depths can still be obtained. 3.1. Standard model against Mallama et al.’s curve Firstly, we adopt the standard cloud model for wavelength 0.90-

lm. The aerosol optical thicknesses at 0.90-lm are estimated based on the wavelength dependency of the extinction cross section, as shown in Fig. 3. Note that Venus’ atmosphere at

217

T. Satoh et al. / Icarus 248 (2015) 213–220

aerosols are stratified, not the altitudes at which they are located). The phase curve computed from the model is compared with Mallama et al.’s curve for validation (Fig. 4). The agreement between the observed data and model computation is found to be fairly good, although there is substantial scattering in the observed data at small phase angles. The model curve indicates a shallow minimum around a  10°, which was identified as a glory feature in I band (García Muñoz et al., 2014).

1

Albedo

Standard Model Upper Haze τ x0.1 Mallama et al. (2006)

3.2. IR1 curve analysis

0.5

0 0

40

20

60

o

Solar Phase Angle [ ] Fig. 4. Validation of the standard cloud model with Mallama et al.’s phase curve.

1

Albedo

Standard Model Best-Fit Model Mallama et al. (2006) IR1 (0.90 μm)

0.5

0

0

20

40

60 o

Solar Phase Angle [ ] Fig. 5. The best-fit model curve is plotted with the IR1 data and Mallama et al.’s data. The standard model curve is also shown, as a dashed line.

wavelength 0.90-lm is so transparent that we do not have to be particularly careful about the exact altitudes of the aerosol layers (in other words, from the 0.90-lm data, we can only tell how

While the standard model reproduces Mallama et al.’s curve well, this model cannot reproduce the Akatsuki/IR1 curve. We tried modifying a minimal number of parameters in the standard model. The first step is to decrease the optical thickness of the upper haze (consisting of mode 1 particles only). Due to the rapid decrease of the extinction cross section of mode 1 particles, s = 0.6 at 0.63 lm (the mean value of Esposito et al.’s 0.2–1.0) translates to s = 0.3 (0.03) at 0.90 lm (2.02 lm), as plotted in Fig. 3. The IR1 data, especially the shoulder-like section at a  20°, seems to favor a much smaller optical thickness. This low optical thickness (<0.1) is required to effectively expose the scattering characteristics of larger particles (bumps in backward scattering) in the underlying layer. We arbitrarily set this to 0.03, one tenth of the standard, and plotted the model computed curve in Fig. 4. In order to reproduce the high plateau of IR1 curve at small phase angles (<20°), a portion of the mode 2 particles in the upper cloud layer were replaced with mode 3 particles. The ratio between mode 2 and mode 3 is not very well constrained but a 0.4:0.6 mixture seems to work well. Next, in the same layer, the size of the mode 2 particles was slightly increased, making them into the mode 20 particles in Table 3. This better reproduces the shoulder at a  20°. Test computations with these modifications yield a curve for the 2.02-lm wavelength that is too bumpy. We therefore introduced a small amount of mode 20 particles into the upper haze layer so that the bumpiness is slightly masked. Optical thicknesses of 0.1–1.0 (every 0.1 step) were tested and 0.3 is found to yield the best result. The curve for the resulting model shows an excellent fit with the observed data (Fig. 5). The best-fit model parameters are summarized in Table 4. Note that we firstly performed parameter optimization in coarse grid and then fine-tuned the parameter within a limited parameter range. However, our values may be sufficient to provide an idea of which parameters improve the fit of the model curve to the data. As is mentioned before, uncertainties in photometric calibration of IR1 data are substantial. In our analysis, we assumed that the IR1 phase curve in the phase angle range 30° to 60° is the same with Mallama et al.’s curve. One may also assume that the IR1 albedo at phase angle 0° is the same with Mallama et al.’s (15% dimmer). Such decreased albedo can easily be reproduced by slight increase of absorption (changing the single scattering albedo from 0.999 to 0.990) in the haze and upper cloud layers. No change in aerosol mixing ratio is required. Therefore, we are confident that the shape

Table 4 Best-fit cloud model for IR1 and IR2 phase curves. Region

Aerosol: s (at 0.90 lm)

Upper haze Mode 1: 0.03 Mode 20 : 0.3 Upper cloud

Middle cloud Lower cloud

Mode 1: 0.7 Mode 20 : 2.7 Mode 3: 4.0 Not constrained Not constrained

CO2 absorption, s Standard (0.30) (0) Standard (0.7) (6.0) (0) Not constrained Not constrained

Altitude range

0.09

Standard (0.25)

Above 75 km

1.4

Standard (2.85)

Up to 75 km

Not constrained Not constrained

218

T. Satoh et al. / Icarus 248 (2015) 213–220

varying from 25 to 40. Satoh et al. (2009) constructed a cloud model with a total optical depth of 30–55 in order to reproduce the VIRTIS-M 1.74 lm spectra (March 2007) for regions of various brightness. Hence, Case 3 can likely be rejected, and although Case 2 may be acceptable, our preference is for Case 1 (s1 = 0.09, and s2 = 1.4), which places the top of the upper cloud layer at an altitude of 75 km. The best-fit parameters for Case 1 are summarized in Table 4 and Fig. 7. The aerosol optical thicknesses in Table 4 are those corresponding to the CO2 absorption optical thickness of s2 = 1.4. Note that there are sources of uncertainty. Uncertainties in the absolute photometry of the IR2 data are currently estimated to be 20%. We have tested the sensitivity of s1 (the absorption optical thickness of the upper haze layer) to such uncertainties and have obtained a range of s1 from 0.17 (20% darker) to 0.03 (20% brighter). The corresponding cloud top altitude range is 72–80 km.

0.12 τ 1 = 0.09, τ2 = 1.4 τ 1 = 0.20, τ2 = 0.70 τ 1 = 0.30, τ2 = 0.35 IR2 (2.02 μm)

0.1

Albedo

0.08

0.06

0.04

4. Summary and discussion 0.02 0

40

20

60

o

Solar Phase Angle [ ] Fig. 6. The best-fit model curve compared to the IR2 data. Note that the scale of the ordinate is different from that in Fig. 5. Three representative combinations of (s1, s2) are plotted, and show only marginal differences.

Table 5 Possible combinations of haze/cloud parameters.

CO2 absorption, s1 Top of upper cloud CO2 absorption, s2 Extent of upper cloud Cloud s (if extended down to 56.5 km)

Case 1

Case 2

Case 3

0.09 75 km 1.4 75–61 km 14

0.25 70 km 0.70 70–64 km 28

0.35 68 km 0.35 68–65 km 56

of phase curves, even if the absolute values are less accurate, are sensitive enough to study the aerosol sizes and mixing ratios in the upper atmosphere. 3.3. IR2 curve analysis In contrast to the IR1 data, the IR2 2.02-lm data is very sensitive to the altitude of the uppermost aerosols due to strong CO2 absorption. The CO2 absorption cross section was computed by referring to the line parameters from CDSD (Carbon-Dioxide Spectroscopy Database) 296 and the HITRAN database. An optical depth of unity occurs at the 63 km level in Venus’ atmosphere. It is found that the standard model yields a planetary disk which is too dark when compared to that observed. The most influential parameter is the amount of CO2 in the upper cloud layer (absorption optical thickness s2  2.85). The next is the amount of CO2 in the upper haze layer (s1  0.25). Three representative combinations of s1 and s2 are shown in Fig. 6 and Table 5. The goodness of fit is almost identical for these three cases. Since the opacity of the upper cloud is large enough to hide its lower part from our view, we cannot determine the actual location of the base of the upper cloud. If it remains at the nominal altitude (56.5 km) and the gas–aerosol mixing ratio is constant throughout the layer, the total aerosol optical thickness amounts to 14 for Case 1, 28 for Case 2, and 56 for Case 3 (Table 5). Because more clouds are expected to exist in the deeper levels, the upper cloud is probably not too opaque. Grinspoon et al. (1993) concluded, from analysis of the Galileo NIMS data (February 1990), that the brightness variations at 1.7 and 2.3 lm can be explained by the cloud optical depth

In this section, we compare our results with those from previous studies. Firstly, a glory feature in disk-integrated photometry data (Mallama et al., 2006), similar to our IR1 data, is identified by García Muñoz et al. (2014). In this paper, we focused on Mallama et al.’s I-band data (the same wavelength with the IR1 data) but the glory in their data was more noticeable in shorter wavelengths (García Muñoz et al., 2014). Secondly, we compare our results with the recent results from ESA’s Venus Express (Svedhem et al., 2007). The most prominent feature that we found in the Akatsuki IR1/IR2 data is the shoulder in the IR1 phase curve at a  20° (glory). This is interpreted as being due to the enhanced visibility of the larger (1 lm or larger radii) cloud particles. It is interesting that a similar result is reported from analysis of the Venus Express/VMC data (Markiewicz et al., 2011), the authors suggesting that the low-latitude and near-equatorial regions tend to be dominated by larger particles (up to a few microns) while smaller particles dominate in the high-latitude region. Because the sub-spacecraft latitude, or most sensitive region at the time of the Akatsuki observations, was always near the equator, the observed brightness is expected to be strongly modulated by

Standard Model

This Study

Fig. 7. Schematic representation of the vertical cloud structure obtained. The most prominent difference from the standard model is that the upper cloud layer is raised by 5 km, although the location of its base (the boundary with the middle cloud layer) is not well constrained due to the large optical thickness of the aerosol. The 2.02-lm CO2 absorption optical depth is also shown, as a solid line.

T. Satoh et al. / Icarus 248 (2015) 213–220

light scattering from aerosol particles at the low latitudes. The results obtained independently by the two spacecraft are therefore consistent with each other. The cause of such a change in particle size is unknown, as the clouds are the product of complicated chemistry and microphysics, as reviewed in Esposito et al. (1983). One hint with regard to this may be the long-term variability of the abundance of SO2 in Venus’ upper atmosphere since possible correlation between the SO2 abundance and the optical thickness of polar haze (sub-micron particles above the cloud) is suggested in the Pioneer Venus era (Esposito, 1985). In more recent years, an increase in SO2 (the material of the clouds), after decades in which there was a decreasing trend (the PVO era), has been reported based on SPICAV/SOIR observations (Belyaev et al., 2008). If this has triggered the increase in larger particles, it may also have increased the cloud opacity detectable in the 1.7 lm and 2.3 lm windows. As mentioned in the previous section, the total optical thickness of clouds found by Satoh et al. (2009) from the March 2007 observations was larger (30–55) than that determined by Grinspoon et al. (1993) from the February 1990 observations (25–40). If such a difference is meaningful, it might be the consequence of a very low SO2 abundance in 1990 (Na et al., 1994). After the arrival of Venus Express at Venus, we just have accumulated continuous measurements of SO2 abundance as well as the flux of radiations in near-infrared windows. Further correlative studies with phase curve measurements, spectroscopy for SO2 abundance, and cloud-opacity measurements in the near-infrared windows should be performed on a continuing basis. Our analysis indicates the existence of mode 20 particles above the cloud top (altitude 75 km). A small amount (s  0.3) of mode 20 particles is favored: it enhances the shoulder of the IR1 phase curve at a  20°; it helps maintain the brightness of the 2.02-lm curve at a reasonable level; and it avoids the 2.02-lm curve from becoming too bumpy. From the SOIR and SPICAV-IR observations (August 2007), Wilquet et al. (2009) reported the detection of 1-lm particles at an altitude of 74–77 km. Integration of the number density of their mode 2 particles through the 74–77 km range (the SPICAVIR profile for orbit 485 in their Fig. 9) yields a column density of 2  106 (cm2). By multiplying this by the corresponding extinction cross section, an optical thickness of 0.3 is obtained. Therefore, our result is quite consistent with the SPICAV-IR observations made in August 2007. Wilquet et al. (2009) also found that the number density of mode 1 particles were between 10 and 30 (cm3) below 90 km, the values being 1 order of magnitude lower than those in Esposito et al.’s standard model (Table 2). They pointed out that their values were more consistent with the measurements from Pioneer Venus limb scans. We have also found that a smaller haze optical thickness, which was arbitrarily set to 0.03 at wavelength 0.90 lm, is favored so as to reproduce the IR1 curve well. There have been reports of the haze optical thickness at a similar level based on the Pioneer Venus/OCPP observations: Kawabata et al. (1986) analyzed the 935-nm polarimetry data in a period from December 1978 to the better part of 1980 and concluded that the haze optical thickness, in the equatorial region, was 0.02 ± 0.01. From analysis of the data set for a longer period (1978–1990), Braak et al. (2002) obtained the disk-averaged haze particle column density, which exhibits a clear decreasing trend from 0.7 (lm2) in 1980 to 0.2 (lm2) in 1989. The corresponding haze optical thickness is approximately within 0.02 ± 0.01. Altogether, it seems that the sub-micron haze of optical thickness 0.02 ± 0.01 covering the equatorial and low-latitude regions is a common feature for the Pioneer Venus era and for recent years from 2007 to present. We do not know what kind of status the upper haze was at the time of Mallama et al.’s observations (1999– 2004). Their phase curve seems to prefer a thicker haze (s  0.3) to a thinner one (s  0.03), as plotted in Fig. 4. A similar analysis needs to be performed on their data at shorter wavelengths.

219

There have been interesting suggestions that may explain the coincidence of larger cloud particles and smaller number density of the haze. Toon et al. (1982) suggested that small haze particles (Mode 1) may serve as condensation nuclei for cloud droplet formation. Using a one-dimensional microphysical cloud model, Imamura and Hashimoto (2001) showed that a decrease of the prescribed photochemical production of condensation nuclei in the upper cloud can lead to larger droplet sizes of up to several micrometers due to a large amount of H2SO4 condensation onto each nucleus. Ignatiev et al. (2009) performed cloud-top altimetry in the CO2 absorption band at wavelength 1.61 lm (the data acquired with VIRTIS-M/-H on board Venus Express). They reported that, apart from in the polar regions, the cloud top is located at 73–74 km in both hemispheres, although the data for the northern hemisphere was limited due to Venus Express’ orbit. Our result (Case 1) is also consistent with theirs. The cloud top altitude in the Pioneer Venus era was also at a similar altitude, 73 km (Braak et al., 2002). It seems that the general condition of Venus’ atmosphere and aerosols in 2006–2011, as observed by Venus Express and Akatsuki, can be described by a single scenario that includes: (1) high SO2 abundance in the upper atmosphere; (2) upward extension of mode 2 particles above the cloud top (75 km) amounting to s  0.3 at 0.90 lm; (3) mode 3 particles within the upper cloud layer, which is usually just bi-modal (modes 1 and 2); (4) possibly higher total opacity of the clouds than in the period of low SO2 abundance. It is fortunate that we have Galileo’s flyby of Venus in early 1990, the low-SO2 period, which produced a good set of data. Further studies comparing 1990 with the present should be of great interest. 5. Conclusion Although its insertion into orbit around Venus failed (December 2010), Akatsuki and its on-board instruments are functioning normally and have demonstrated their capabilities by performing photometric observations of Venus (February–May 2011). The phase curves obtained at 0.90 lm (IR1) and at 2.02 lm (IR2) at the small phase angles have enabled us, through radiative-transfer analysis, to detect increased visibility of larger cloud particles. To further confirm this and determine whether (and how) the clouds return to their previous state, monitoring observations either from space or the ground are required. Acknowledgments The authors are grateful to W. Markiewicz for useful discussions. The observations would not have been possible without the selfless assistance of the engineering staff and operations-support staff. Comments from A. Mallama and another knowledgeable reviewer were of great help to improve the manuscript. This study was carried out with the Grants-in-Aid for Scientific Research (A) (20227937) from the Ministry of Education, Culture, Sports, Science, and Technology (MEXT). This study is also supported in part by the Research Fellowships of the Japan Society for the Promotion of Science for Young Scientists (for S.O.). Appendix A. Calibration of IR1 data Pre-flight laboratory calibration of IR1 was performed and it is reported that 37 (mW cm2 lm1 str1) input results in 633 (ADU/s) output in 0.90-lm (for Venus dayside) filter (Iwagami et al., 2011). The total counts for Venus disk at the smallest observed phase angle (March 22) was 25,000 (ADU) in 6 s integration. The area of Venus disk on IR1 CCD was 13.6 (pix2) as the

220

T. Satoh et al. / Icarus 248 (2015) 213–220

spacecraft was some 13.7 million km away from the planet. Therefore, a pixel yielded 25,000/6/13.6 = 310 (ADU/s) and the corresponding intensity is estimated as 37 ⁄ (310/633) = 18 (mW cm2 lm1 str1). It should be noted, however, that the post-flight sensitivity of IR1 was found to be noticeably lower than pre-flight measurement (Iwagami and Ohtsuki, 2011, 2012). The Earth image, acquired using 0.90-lm dayside filter, suggest the sensitivity is 2/3 of pre-flight estimate. On the other hand, star measurements in Sagittarius suggest the sensitivity is 1/3 although the filter is different (0.90-lm Nightside). The reason of this has not yet been understood but this is the area where the largest uncertainty of calibration of IR1 data exists. Therefore, the observed intensity at the smallest observed phase angle can be any value from 27 to 54 (mW cm2 lm1 str1) at this moment (this, of course, needs refinement when arrived at Venus). The solar flux at 0.90 lm to Venus orbit is 180 (mW cm2 lm1) (derived from that at the Earth orbit tabulated in Appendix 8 in Houghton, 1986). If observed at zero phase angle and geometric albedo is A, then the intensity from Venus is 57A (mW cm2 lm1 str1). By comparing this with the above IR1 observation, the albedo A can 0.47–0.94, including Mallama et al.’s (2006) value, 0.57, in the range. In favor of matching the slope of phase curve to Mallama et al.’s in 30–60° phase angle range, we applied a factor of 2.16 after calibrating IR1 data using pre-flight calibration constant. This yields the geometric albedo in IR1 data A = 18  2.16/57 = 0.68. Appendix B. Calibration of IR2 data In contrast to IR1 (operated at room temperature), IR2 was not calibrated in laboratory because of necessity of cooling, but it uses post-flight observations of celestial objects as photometric references. One of such is the Earth–Moon images (3 in a sequence) acquired on October 26, 2010 (Satoh et al., 2011, 2012). The spacecraft was 0.201 (AU) away from the Moon. The counts on the Moon pixels are summed and the average of 3 images, 477 (ADU) with ±20% uncertainty, is obtained. Then, this number is converted to 19.1 (ADU) as if the Moon were located 1 (AU) from the Sun and from the spacecraft. The same conversion is done to the Venus counts, 21,000 (ADU) at the smallest phase angle, yielding 93.9 (ADU). The difference between these two values reflect different diameters and different albedos of the Moon and Venus. For the phase angle 12.9° and the wavelength 2.02 lm, we refer to Velikodsky et al. (2011) and estimated the lunar albedo of 0.22 for this condition. Using this, we obtain the Venus albedo observed at the smallest phase angle as 0.09 (diameters of Venus and the Moon, distances from the Sun and from the spacecraft are considered). This seems to be reasonable value for the albedo in 2.02 lm CO2 absorption band. We also have performed another check using the same Earth– Moon images. The Earth is assumed to be covered by 70% ocean and 70% of the globe is cloudy. By referring to Gottwald et al. (2006), the disk averaged albedo of the Earth is estimated as 0.065. By following the same procedure as the above, we obtain the Venus albedo as 0.10 which is consistent with the value obtained using the Moon as reference. Therefore, we believe the estimate using the Moon, 0.09, is accurate within ±20% or better.

References Belyaev, D. et al., 2008. First observations of SO2 above Venus’ clouds by means of solar occultation in the infrared. J. Geophys. Res. 113, E00B25 (1–10pp.). Braak, C.J., de Haan, J.F., Hovenier, J.W., 2002. Spatial and temporal variations of Venus haze properties obtained from Pioneer Venus Orbiter polarimetry. J. Geophys. Res. 107 (E5), 5032 (1–13pp.). Esposito, L.W., 1985. Long term changes in Venus sulfur dioxide. Adv. Space Res. 5, 85–90. Esposito, L.W., Knollenberg, R.G., Marov, M.Y., Toon, O.B., Turco, R.P., 1983. The clouds and hazes of Venus. In: Hunten, D.M., Colin, L., Donahue, T.M., Moroz, V.I. (Eds.), Venus. The University of Arizona Press, Tucson, Arizona, pp. 484–564. Fukuhara, T. et al., 2011. LIR: Longwave Infrared Camera on board the Venus Orbiter, Akatsuki. Earth Planets Space 63, 1009–1018. García Muñoz, A., Pérez-Hoyos, S., Sánchez-Lavega, A., 2014. Glory revealed in diskintegrated photometry of Venus. Astron. Astrophys. 566 (L1), 1–6. Gottwald, M. et al., 2006. SCIAMACHY, Monitoring the Changing Earth’s Atmosphere. DLR, Germany. Grinspoon, D. et al., 1993. Probing Venus’ cloud structure with Galileo NIMS. Planet. Space Sci. 41, 515–542. Hansen, J.E., Travis, L.D., 1974. Light scattering in planetary atmospheres. Space Sci. Rev. 16, 527–610. Houghton, J.T., 1986. The Physics of Atmospheres, second ed. Cambridge University Press. Ignatiev, N.I. et al., 2009. Altimetry of the Venus cloud tops from the Venus Express observations. J. Geophys. Res. 114, E00B43 (1–10pp.). Imamura, T., Hashimoto, G.L., 2001. Microphysics of venusian clouds in rising tropical air. J. Atmos. Sci. 58, 3597–3612. Irvine, W.M., 1968. Monochromatic phase curves and albedos for Venus. J. Atmos. Sci. 25, 610–616. Iwagami, N., Ohtsuki, S., 2011. IR1 camera. In: The Report of Flight Operation of PLANET-C. SES-TD-10-003, SES Data Center, ISAS/JAXA, pp. 155–161 (in Japanese). Iwagami, N., Ohtsuki, S., 2012. IR1 camera. In: The Report of Flight Operation of PLANET-C (Part 2). SES-TD-11-006, SES Data Center, ISAS/JAXA, pp. 137–142 (in Japanese). Iwagami, N. et al., 2011. Science requirements and description of the 1 lm camera onboard the Akatsuki Venus Orbiter. Earth Planets Space 63, 487–492. Kawabata, K. et al., 1980. Cloud and haze properties from Pioneer Venus polarimetry. J. Geophys. Res. 85, 8129–8140. Kawabata, K., Lane, W., Sato, M., Travis, L.D., 1986. Local variation of haze thickness on the equatorial region of Venus. In: Proceedings of the Fifteenth International Symposium on Space Technology and Science, Tokyo, Japan, pp. 1829–1834. Mallama, A., Wang, D., Howard, R.A., 2006. Venus phase function and forward scattering from H2SO4. Icarus 182, 10–22. Markiewicz, W.J. et al., 2011. 2000 Earth days around Venus: imaging with Venus Monitoring Camera on Venus Express. EPSC Abstracts, vol. 6, EPSC-DPS2011464-2. Na, C.Y., Esposito, L.W., McClintock, W.E., Barth, C.A., 1994. Sulfur dioxide in the atmosphere of Venus. Icarus 112, 389–395. Nakamura, M. et al., 2007. Planet-C: Venus Climate Orbiter mission of Japan. Planet. Space Sci. 55, 1831–1842. Nakamura, M. et al., 2011. Overview of Venus Orbiter, Akatsuki. Earth Planets Space 63, 443–457. Nakamura, M. et al., 2014. Return to Venus of the Japanese Venus Climate Orbiter AKATSUKI. Acta Astron. 93, 384–389. Satoh, T. et al., 2009. Cloud structure in Venus middle-to-lower atmosphere as inferred from VEX/VIRTIS 1.74 lm data. J. Geophys. Res. 114, E00B37 (1–13pp.). Satoh, T., Ueno, M., Uemizu, K., 2011. IR2 camera. In: The Report of Flight Operation of PLANET-C. SES-TD-10-003, SES Data Center, ISAS/JAXA, pp. 162–171 (in Japanese). Satoh, T., Ueno, M., Uemizu, K., 2012. IR2 camera. In: The Report of Flight Operation of PLANET-C (Part 2). SES-TD-11-006, SES Data Center, ISAS/JAXA, pp. 143–148 (in Japanese). Svedhem, H., Titov, D.V., Taylor, F.W., Witasse, O., 2007. Venus as a more Earth-like planet. Nature 450, 629–632. Toon, O.B., Turco, R.P., Pollack, J.B., 1982. The ultraviolet absorber on Venus: Amorphous sulfur. Icarus 51, 358–373. Velikodsky, Yu.I. et al., 2011. New Earth-based absolute photometry of the Moon. Icarus 214, 30–45. Wilquet, V. et al., 2009. Preliminary characterization of the upper haze by SPICAV/ SOIR solar occultation in UV to mid-IR onboard Venus Express. J. Geophys. Res. 114, E00B42 (1–13pp.).