シューメーカー・レヴィー第9彗星と木星の衝突の近 赤外線観測

九州大学学術情報リポジトリ

Kyushu University Institutional Repository

シューメーカー・レヴィー第9彗星と木星の衝突の近 赤外線観測

竹内, 覚

九州大学理学研究科物理学専攻

https://doi.org/10.11501/3119143

出版情報：Kyushu University, 1996, 博士（理学）, 課程博士 バージョン：

権利関係：

Near-infrared observation of the impact of comet Shoemaker-Levy 9 with Jupiter

Satoru Takeuchi

Department of Physics Kyushu University

12 May 1996

Abstract

Comet Shoemaker-Levy 9, which was discovered by C. S. Shoemaker, E. M.

Shoemaker and D. H. Levy on March 24, 1993, collided with Jupiter during July 16-22, 1994. This cometary impact was the first fortuitous event for the human-being to witness the collision between astronomical objects.

The author observed the cometary impact by near-infrared at Okayama As­

trophysical Observatory during July 16-24, as a member of the SL9 observation program. We could detect signals of the impact phenomena by the SL9 frag­

mentized nuclei C, D, and K in despite of the impact points on the far side of Jupiter seen from Earth, and revealed the complicated variations of flux asso­

ciated with the impact phenomena ; two precursors and brighter longer lasting

"main event" .

Comparing with a large amount of data by other observatories, the author concludes that the first precursor is the entry flash of the fragment into the Jovian atmosphere at higher altitude, where is visible from Earth. The second precursor is the impact-induced plutne rising above the limb of Jupiter. The motion of the plume is well approximated by a simple ballistic trajectory, be­

cause of near-vacuum density above the height of 1000 km, where the plume easily reached. The ejected speed of 17km s-1 and the ejected elevation angle of 45° can account for both the maximum height of the plume and the maximum distance of the debris from the impact point.

The main event is a most debatable subject in the impact phenomena. Many researchers and research groups have thought that the heating by re-entry (^or

splash back) of the plume gas into the lower atmosphere produced the main event.

However, the author and our group propose another explanation. In the above ballistic model the splash back of the plume starts from more than 10 minutes after the impact, although the main event started from 6 minutes af­

ter the impact. vVe propose that sudden increasing of the optical depth of the plume, by dust grain formation in the cooling plume, can induce rapid brightening of radiative flux from the plume. The author analyzed the data of Okayama observation in order to settle down this problem. For the K impact of the Okayama observation, any motion of the center position of the bright flash could not be detected at the onset of the main event, which means that what was seen in the main event is the plume itself, but the position started to move southward at 4 km s-1 from 10 minutes after the impact. Based on this results, the author conclude that the ground-based observers saw the bright flux from dust particles in the plume for the first half of the main event and that the splash back occurred and radiated for the latter half of the main event.

Contents

1 Introduction 2 J avian Atmosphere

3 Pre-Impa ct Predictions

3.1 Evolu tions and Sizes of Fragments of SL9 ... .

3.2 Impact Geometry ... .

3.3 Predictions of Comet Entry and Expanding Plume 3.4 Predictions of Impact Debris Cloud .

3.5 Predictions of Atmospheric Wave ... .

4 Observation 4

.1 Instruments ... . 4

.2 Reductions ... . 4.3 Calibration and Photometry . 4.4 Impact of Fragment C 4.5 Impact of Fragment D ... . 4.6 Impact of Fragment K . . . ^. 4.7 Impact of Fragment N and others .

5 Other Observations 5.1 Impact of Fragment A 5.2 Impact of Fragment E 5.3 Impact of Fragment G 5.4 Impact of Fragment H 5.5 Impact of Fragment L

5.6 Impacts of Fragments Q1 and Q2 . 5.7 Impact of Fragment R

5.8 Impact of Fragment S ... . 5.9 Impact of Fragment W ... . 5

.10 Impacts of Other Fragments . 5

.11 HST Observations ... .

6 Discussion

6.1 Entry, Impact Time and PC1 6.2 Comparison of Fragments 6.3 PC2 and Plume .

6.4 Main Event 6.5 Temperature 6.6 Splash Back ...

6.7 Motions of Plume and Splash Back 6.8 Dust Grain Formation ... .

1 3

8 8 9 9 10 10

20 20 21 21 2 7 2 8 29 31

40 40 41 41 42 44 46 46 48 48 49 49

53 5 3 54 57 6 6 6 7 69 69 71

6.9 Skittering and Bounce . . . ^. 6.10 Fragment Sizes and Penetration Depths 7 Conclusion

71 72 77

1 Introduction

Comet Shoemaker-Levy 9 (1993e) (hereafter referred as SL9) was discovered by Carolyn S. Shoemaker, Eugene M. Shoemaker and David H. Levy with the 0.46-m Schmidt telescope at Palomar Observatory on March 24, 1993 (1). The appearance was most unusual in that the image of SL9 by the 2.2-m telescope at Mauna Kea on March 27 showed as many as 17 separate sub-nuclei strung out like "pearls on a string" [2] . The orbital computation of SL9 by Syu-ichi Nakano indicated that a very close encounter of SL9 with Jupiter occurred during July 1992, perhaps at which time tidal forces of Jupiter broke the parent body of SL9 into numerous fragments, and that SL9 might be on a collision course with Jupiter at July 1994 [3). By the following observations and the orbital computations of SL9, the SL9 impact with Jupiter became to be certain until the end of 1993. Such a collision of astronomical objects was known indirectly by the existence of many craters on the surfaces of planets and satellites. The SL9 impact, however, was the first fortuitous event for the human-being to be able to observe such an impact scientifically under sufficient preparations.

Before the SL9 impact in to Jupiter, many researchers or research groups pub­

lished various predictions on phenomena associated with the impact. However it was beyond these predictions WHAT actually happened. By many infrared observations, bright and long lasting dazzling flashes, called as "main event' , were observed. The Hubble Space Telescope (HST) caught the impact-induced rising plume. At the impact sites, huge Earth-sized crescent dark clouds, which were easily observable for ground-based observers by small telescope with the aperture of only several centimeters, appeared on the morning terminator of Jupiter. We do not fully evaluate yet how valuable the informations obtained from the SL9 collision with Jupiter are on the planetary and earth science.

The author observed the SL9 impact by near-infrared at Okayama Astro­

physical Observatory (OAO), as a member of the SL9 infrared observation program, led by Jun-ichi Watanabe of National Astronomical Observatory of Japan (NAOJ) [4]. The main objectives of this observation are as follows : 1) the detection of infrared radiation from the entries of the fragments and the impact-induced plume to reveal sequential phenomena associated with the im­

pact. 2) The detection of the impact debris cloud, reflecting sunlight, in the Jovian stratosphere to reveal the optical properties of the debris dust and the stratospheric dynamics from the motion of the debris. 3) To find something that had not been able to be predicted before the impact.

Eventually we could observe the impacts of the fragments C, D and K and many debris clouds. In this paper, based on the results of these observations, the author attempts to reveal a variety of the impact phenomena, in particular the sequential phenomena on the fragment entry, the impact-induced plume and the aftermath.

This paper is constructed as follows: The structure of the Jovian atmosphere is reviewed at Section 2, and the pre-impact predictions on various phenomena

1

induced by the collision are reviewed in Section 3. Observational instruments, reduction and calibration methods, and the observational results of the impacts of the fragments C, D, and ^K are shown in Section ^4. Section 5 summarizes a large amount of results by other observatories. Section 6 gives the results of other observatories, interpretations of the observational results, and some model calculations. Section ⁷ presents the conclusions of this paper.

2 Jovian Atmosphere

In this section, the structure of the Jovian atmosphere, that is necessary for the discussion on the SL9 impact, is briefly reviewed.

Jupiter is a giant gas planet. Jupiter's "surface" shows many banded­

structures of clouds. Tracing the motions of small features in the surface cloud layer, observers have measured the wind speeds and directions of the Jovian atmosphere during more than the last 100 years. The morphology of the Jovian clouds is surnrnarized by Peek ( 1958), Hirabayashi ( 1981), and Rogers ( 1995) [5]. Fig. 1 shows the zonal flow profile, i.e., the latitudinal distribution of the east-west wind on the surface cloud, obtained from the data of two Voyager

spacecrafts [6]. Each region between the peaks of westerly and easterly winds corresponds to a belt

/

clonic shear is a belt, i.e., dark band seen for visible, and the region with the anticyclonic shear is a zone, bright band. Little north-south wind has been ob­

served on the surface cloud, except for that of the east and west edges of large vortices, e.g., the Great Red Spot

(

)

We have little informations on the vertical structure of the Jovian atmo­

sphere. The vertical profile of the temperature of the upper troposphere and lower stratosphere was measured by two Voyager experiments, the radio oc­

cultation and the infrared spectrometer

(

)

ture of the upper stratosphere is not known because of no observations. The temperature of the thermosphere was measured by the Voyager ultraviolet so­

lar

/

perature ranged from 10 bar to 0.1 mbar, synthesized by the author from Atreya et a/.(1981) and Kostiuk et a/.(1989) (8, 9]. The temperature below 1 bar level is less known because the existence of clouds prevents any observations. For many cases, the adiabatic lapse rate is assumed to estimate the temperature below

1 bar. The successful entry of the probe of the spacecraft Galileo into the Jovian atmosphere at December 7 may reveal the structures of the lower troposphere

[10].

The cloud structures of the Jovian atmosphere are also little known. A theoretical approach was carried out by Lewis(1969) and vVeidenschi/ling and Lewis

(

[11]. Assuming that the atmospheric composition is equal to the solar abun­

dance 1 they discussed what compounds are condensible in the tropospheric envi­

ronment, and concluded three layered cloud decks of ammonia

(NH3),

hydrosulfide

(NH4SH),

(H20).

sphere at the lower level with higher temperature moves upward adiabatically, the component with high condensation temperature will saturate and condense to form clouds as the temperature decreases with the height. By such a method,

Atreya and Romani(1985) predicted that the

H20

NH4SH

NH3

3

0.3-0.7 bar level, whose positions are shown roughly in Fig. 2. In the regions of no clouds, the temperature shall be adiabatic distribution and the internal gravity wave can not propagate vertically. These cloud layers are optically thick and we can not observe easily below the clouds, even below the top ammonia cloud. Ordinarily Jupiter's "surface" is defined at the ammonia cloud top level, or the 1 bar pressure level, or 100 mbar pressure level. In this paper, the au thor adopts the 1 bar pressure level as the Jovian "surface" and the origin of the vertical coordinate.

Some species of molecules, which absorb the specific wavelength light, in the Jovian atmosphere have large effects for the observations of Jupiter. For example, the methane ( CH4), prevailingly existing in the Jovian troposphere and lower stratosphere, has several absorption bands in the visible and near­

and mid-infrared wavelengths, 0.89 J-Lm, 1.7 J-Lm, 2.3 J-Lm, 3.4 J-Lm, etc. For those wavelengths, the incoming sunlight into Jupiter and the reflected light from the clouds are largely absorbed by the methane in the atmosphere. Therefore Jupiter is appeared to be darker than disks for the neighboring wavelengths. If there are any dust or haze particles in the higher stratosphere, these particles can scatter the unabsorbed sunlight, and thus these particles are appeared as bright spots relative to the darker Jovian surface. Eventually we can research the Jovian stratosphere using the absorption wavelengths.

The wavelength of 0.89 J-Lm is a weak methane band in visible (more exactly in near infrared) and can be sensible by the visible CCD. For near-infrared in H-band (1.5-1.8 J-Lm) and K-band (2.0-2.4 J-Lm), there are methane bands around

1.7 J-Lm _and 2.3 J-Lm and a hydrogen band around 2.1 J-Lm. The stratosphere above

20 mbar can be sounded by these wavelengths (see Fig. 3) [12]. In the north and south polar regions, there are high altitude haze particles which appear as bright polar caps for the absorption bands. By the wavelength of 3.4 J-Lm,

at which the methane has a very strong absorption, the thermal radiation from Jovian polar auroras can be detected.

Zonal Flow Velocity (m

-1)

01 0 0 � w 0

1'\.)

c._ 0 0

<

0 ^�

CD 0

::J ....-+-

�

0 0

r Ol ^I

....-+- � ....-+-

c 0 a. ^I CD 1'\.)

0

I

w 0

I

0 �

I

01 0

-50 0 50 100 150

Figure 1: Zonal flow profile at the ammonia cloud top of Jupiter, measured by two Voyager spacecrafts. Produced from data of Limaya ( 1986) [6]. The arrow in the figure shows the latitude at the SL9 impact, which is -44° for Jovicentric

latitude.

5

-o �

(/) CD (/) c

�

-

3

0"

m

-

�

0

�

0 0

�

0 0 0

�

0 0

Temperature (K)

8 ^{120 160 200 240 280 320 360}

Figure 2: Temperature vs. Pressure profile from

10

0.1

1

(

)

[ 11].

BARS 0.001

0.010

0.100

1.000

1.50 1.75

JUPITER EXTINCTION LEVELS

'

,.--- H2 ONLY

2.0 llJl1

I I I I

2.25 2.5

Figure 3: Near-IR photon penetration levels for Jupiter. After Baines et a/.(1993) _[12]. Due to methane and hydrogen gas extinction, the light from the planet is the sunlight scattered by particles above or near the level shown. In the deepest bands near 1.7 and 2.4J.Lm, the stratosphere just above the 20 mbar level can be probed. The hatched regions in the figure represents heights of some cloud and haze layers.

7

3 Pre-Impact Predictions

3.1 Evolutions and Sizes of Fragments of SL9

Until August 1993, 21 separate fragments of SL9 had been identified in CCD (Charge-Coupled Device) images under sub-arcsecond seeing condition and were labelled numbers along the nuclei train of SL9 from the westernmost fragment (with fragment number 1) to the easternmost fragment (with fragment number 21) [13]. But this nomenclature was less prevailing, and most scientists adopt a naming convention by Sekanina et a/.(1994), which labels each fragment a letter of alphabet along the opposite direction of the above numbering convention (14].

For the latter, and familiar naming convention of the fragments, fragment A, which was also named "21", was the first impactor into Jupiter among all the fragments and fragment W, which was "1", was the last one. The letters I and 0 were not used because of the possible confusion with the number one and zero. In this paper, the letter naming is used.

But the appearance of the SL9 fragments continued to change until the impact (See Figs. 4 and 5). The fragments J and M disappeared at the end of 1993 [15]. Also Hubble Space Telescope (HST) observation at July 1, 1993, showed that the fragments P and Q had two nuclei respectively, which were designated ^as the fragments P1 and P2 and the fragments Q1 and Q2 (16].

Then between January 24 and March 30 1994, the fragment P2 broke up into two separate fragments, one of which disappeared by late June (see Fig. 6).

The fragment P1 had a "streaked" appearance on January 24 1994 and then became a barely discernible "puff" through mid-May. The P1 was not detected in subsequent observations. By March 30, also the fragment T faded to a loose puff of dust [17].

The most important information for the predictions of some impact phe­

nomena is the estimation of the fragment size, which is strongly related to the total energy released during the atmospheric entry of the fragments. Weaver et a/.( 1994) estimated the upper limits of the diameters for the 11 brightest fragments from the HST observations, in which the upper limits of the diam­

eters of the Q, having maximum brightness, and the single parent body were estimated by 4.3 km and 7.7 km, respectively [18]. Watanabe et a/.(1994) calcu­

lated the lower limits of the diameters of the 10 fragments ^as 0.7-0.9 km from their observations with the 188-cm telescope at OAO [19]. On the other hand, taking into account the tidal breakup of the progenitor of SL9 and the following separation of the fragment chain, Scotti and Melosh (1993) estimated that the mean fragment diameter was 0.74 km and that the parent body diameter was 2.3 km [20], which is consistent with the estimation by Watanabe et a/.(1994).

Because the impact speeds of the SL9 fragments were about 60 km s-1 , the diameter ranging from 0.5 km to 5 km corresponds to the total kinetic energy of the fragment from 1027 erg to 1030 erg, assuming the density of the cometary fragment is 1 g cm-3.

3.2 Impact Geometry

The impact points were predicted to be far side of Jupiter as seen from Earth and to be at the Jovicentric latitude of -43 ... 44° and 1 00 ... 1 05° west from the central meridian, seen from Earth (see Table 1 ) [2 1 ]. But because there points were behind only 5-1 0 degrees from the morning limb of Jupiter seen from Earth, the impact sites were visible from Earth within 1 0-20 minutes and were illuminated by Sun within half hour after the impact (see Fig. 7). The impact points of the fragments were predicted to be almost coincided on Jupiter, seen from the space, but due to the Jupiter's rotation the impact sites were diffused around the longitude of Jupiter on the J ovicentric latitude of 44°. During the SL9 impact at July 1994, the phase angle between Earth and Sun seen from Jupiter was 1 0.7°. One which can see the impact points directly at the impact was only the spacecraft Gali/eo, which was on the way to Jupiter (see Fig. 1 0). Figs. 8, 9, 1 0 show the geometry o f Jupiter and the impact points as viewed from Earth, and the Galileo.

3.3 Predictions of Comet Entry and Expanding Plume One of the most debatable subjects on the SL9 impact before the impact was

" how deep do the fragments penetrate into the Jovian atmosphere before the disintegrations of the fragments ? "

Sekanina ( 1993) described the entries of the SL9 fragments into the atmo­

sphere by the ablation (or bolide) model, which is originally developed for the theory of the entry of meteor into the terrestrial atmosphere, and showed that the fragments would disappear in the Jovian stratosphere [22]. But the abla­

tion model strongly depends on the ablation parameter, which represents the strength of the entry object. In contrast with Sekanina ( 1 993), who used a value of soft comet as the ablation parameter, Hasegawa and Takata ( 1 993) used the same model with the ablation parameter of a somewhat fragile meteorite and showed that the disintegration of the fragment occurred mainly below the am­

monia cloud in the troposphere [23]. Taking into account the effect of the frontal shock layer accompanied by the impactor, the penetration depth of the fragment was predicted to be 10-100 bar [24, 25], deeper than that of Hasegawa and Takata(1 993).

On the other hand, some groups used hypervelocity hydrodynamical codes in order to simulate the cometary impact. Takata et a!.( 1 994) simulated the entry of the SL9 fragment using the three-dimensional smoothed particle hy­

drodynamic (SPH) method, which is a kind of Lagrangian method, and showed that the terminal pressure level of the penetration was 1 00 bar and 200 bar for a water-ice of the diameter of 1 km and 2 km, respectively [27]. Boslough et a/.(1994) obtained similar results to that of Takata et a/.(1 994), in spite of the use of different simulation code and equation of state [29). However, Yabe et a/.( 1 994) and Mac Low and Zahnle ( 1 994) predicted shallower penetration,

9

10 bar for a water-ice of 3 km diameter and an ideal gas of 1 km diameter, re­

spectively (30, 26).

Although there are great discrepancy among the predictions on the penetra­

tion depth of the fragment, the predicted phenomena after the disintegration of the fragment are almost same, which is the expanding and rising fireball, or plume (25, 27, 29, 31). According to these predictions, the huge kinetic energy of the fragment will be deposited along the trajectory of the entry within the first 10 seconds and will heat the atmosphere and the cometary debris in the trajectory to maximum temperature of 10,000-20,000 K. The deposition of the huge energy during such a short time will induce the catastrophic explosion and the induced shock wave will be accelerated upwardly due to the conservation of momentum and the decreasing of the atmospheric density with the height. Thus the atmospheric and cometary matters also will be accelerated into the upward direction due to the upward flow accompanied with the shock wave and will form the plume. Some predictions shows that the plume will reach at a height of about 1,000 km above 1 bar pressure level, conserving the plume temperature of 2000 K within 100 seconds after the impact (27, 29, 31) and that the altitude of the plume will be able to be seen directly from Earth although the impact site is the far side of Jupiter from Earth. Whether the plume is observable or not actually strongly depends on the opacity of the plume. However, the opacity of the plume was highly uncertain.

3.4 Predictions of Impact Debris Cloud

The rising plume will deposit a considerable amount of the cometary and tro­

pospheric matters and the impact-induced chemical compounds, in the Jovian stratosphere. If any small dust will be condensed from the matters, the dust will scatter the sunlight, or radiate thermally, and may be observable for infrared, visible and ultraviolet, in particular some near-infrared wavelengths with molec­

ular absorption bands, e.g. due to hydrogen and/or methane (32). Hasegawa et a/.(1994) shows that the cometary water-ice cloud can be seen as a bright point object with enough high contrast against the background ammonia cloud for methane and hydrogen absorption bands, if the ice cloud is formed from the deposited water vapor [33]. Pryor et a/.( 1994) discussed on the dust hazes and water-ice formed from the impact and their observability [34).

3.5 Predictions of Atmospheric Wave

An other expected atmospheric phenomenon induced by the SL9 impact is gen­

eration of internal-gravity waves. Harrington et a/.( 1994) used a nonlinear dy­

namical model with five vertical layers for the Jovian atmosphere to simulate gravity waves generated by the thermal disturbance released by the impact (35).

In this simulation, internal gravity waves are generated by the thermal anomaly in the stratosphere, which is predicted by Sekanina (1993) [22], and propagate

horizontally with the phase speed of 400 m s-1, which is consistent with the phase speed estimated from the stratification of the Jovian stratosphere. But their result is criticized, because of impossibility to simulate the vertical prop­

agation of the gravity wave accurately due to low vertical resolution of their model.

Ingersoll et a/.(1994)

[36].

agate vertically and are trapped in the layer. The phase speed of 130 m s-1 is estimated by assuming that the composition of the Jovian atmosphere is equal to the solar abundance.

In this section, various predictions studied before the impact have been re­

viewed. Eventually, however, nobody had known WHAT actually occurred at the

SL9

11

5:5

I

•

N

Comet P/Shosmaksr-Levy 9 (1993s) January 1994

R=6

I

I

. i 1.

02=7b

I

P2:8b

N=9 L·11

I I

. T

H:14

I

,, .

E" 17 C=19

I I

• ^• •

I

I

0·18

1

8·20

•

8rlgh1 -- 8rlgh1c31

Figure 4: SL9 image by Hubble Space Telescope (HST) with the Wide Field Planetary Camera-2 (WFPC-2) on January 24-27, 1994 (top panel). The frag­

ment names are designated in the bottom panel. One fragment (W) is slightly outside of this image (to the left). Credit by JPL/NASA.

T�4 W�l U�3 SnS

I I I

• •

I

V�2

s

N

Comet ^PIS hoe maker-Levy 9 (1 993e) "'"' 1994

Re6 Q1a7a

I ^I

2 Nl9 L·11

•

• 1

1 1

i ^.I

02·7b0

1 f .L

P2·8b P1•8B

Ke12

•

I

Hs14

I G1•15a IG2•15b I

Ce19 I ^Ae21 T

^·�.

•1 • l

J·13

I

• _•

B·20

Bright Brlghtc31

, .. ' no bngct rlsldc 0 barely dl�rnablc

Figure 5: SL9 image by HST/WFPC-2 on May 17, 1994 (top panel). The fragment names are designated in the bottom panel, although the directions of the SL9 train are opposite between top and bottom panels. Credit by H. A.

Weaver, T. E. Smith (Space Telescope Science Institute (STSci) ), and NASA.

13

Figure 6: Evolutions of fragments P and Q during one year, by HST. Top panel : Image taken on July 1993. Brightest fragments are the Q (Q1 and Q2). Right­

down fragment is Q2, and left-top is Ql. Two faint fragments, right side of the Q, are the P, right fragment is P2, and left one is P 1. Also the R is seen at left side of the Q. Center panel : Image on January 1994. Bottom panel : Image on March 1994. The P1 became a barely discernible puff and disappeared. The P2 was also fragmentized two, but left, fainter one was disappeared. Observations by HST. Credit by STSci, and NASA.

14

Comet Shoemaker-Levy ⁹ Trajectory Viewed Along South Polar Axis of Jupiter

Limb as sean from Earth

Impact Point

Terminator

__...:r Earlh

-^sun

Trajectory of Comet Fragment Q

(ticks every 30 min)

P. Chodas / JPL

3/21/94

Figure ^7: Impact geometry viewed along south polar axis of Jupiter. Jupiter rotates clockwise in this figure. Credit by P. Choda.s (^JPL)^.

15

Impact of Comet Shoemaker-Levy 9 As Viewed from Earth

North

t

Jupiter

�ornlng Terminator

Impact Point Just Behind Limb

-20

Trajectory of Comet Fragment Q

(ticks every 5 minutes)

Distance from Earth: 5.2 AU

P. Choda:s / JPL

3/17/94

Figure 8: Impact Geometry as viewed from Earth. Credit by P. Chodas (JPL).

Impact of Comet Shoemaker-Levy 9 As Viewed from Earth

Terminator

Impact Pornt ^{� ',}

(On Fa"lde)

I '

Comet Passes Behind Limb Seconds Before Impact

-5 min

Tra!ectory of Comet Fragment Q (ticks every minute)

Jupiter

-3o·

-so·

P. Cnodaa / ^JPL 3/17/94

Figure 9: Impact Geometry ^as viewed from Earth. Jupiter surface between the limb and the terminator are not illuminated by Sun. Credit by P. Choda.s

(^JPL)^.

17

Impact of Comet Shoemaker-Levy 9 As Viewed from Galilee

Limb as seen from Earth

Nightside

Impact Point

-20 min

Jupiter

Terminator

Trajectory of Comet Fragment Q

(ticks every 5 minutes)

_Earth -sun

Range to Gallleo: 1.6 AU

P. Chodas / ^JPL .3/19/9•

Figure 10: Impact Geometry ^as viewed from Galileo. Galileo had a direct view of the impact point. Credit by P. Chodas (JPL).

Table 1: Predicted Impact Parameters for Fragments of P /Shoemaker-Levy 9

Fragment Impact 1-sig J ovicentric Meridian Angle Date Time Unc. Lat. Long. Angle E-J-F July (UTC) (min) (deg) (deg) (de g) (de g) A = 21 16 19h59m40s 5.5 -43.13 178 64.48 98.72 B = 20 17 02h54m13s 4.1 -43.16 70 63.82 99.18 c _{= 19 17 07h02m14s 3.7 -43.37 218 65.24 98.12} D = 18 17 11h47m00s 4.7 -43.45 30 65.58 97.85

E = 17 17 15h05m31s 3.1 -43.47 150 65.76 97.72

F = 16 18 00h29m21s 4.0 -43.56 132 64.52 98.57 G = 15 18 07h28m32s 3.1 -43.59 23 66.63 97.07 H = 14 18 19h25m53s 3.1 -43.73 96 66.83 96.89 K = 12 19 10h18m32s 3.1 -43.80 275 67.76 96.21 L = 11 19 22h08m53s 3.4 -43.91 344 68.18 95.88

N _{= 9 20 10h20m02s 4.9 -44.29 66 67.77 96.10}

P2= 8b 20 15hllm55s 4.6 -44.61 244 66.68 96.80 Q2= 7b 20 19h31m43s -44.35 39 68.78 95.37 Q1= 7a 20 19h59m10s 4.5 -44.06 55 69.20 95.12 R=6 21 05h25m56s 4.6 -44.07 37 69.44 94.94 S =5 21 15h10m22s 4.4 -44.16 30 69.80 94.67 T =4 21 18h03m45s 11.5 -44.99 137 67.34 96.26

u _{= 3 21 21h48m30s 12.8 -44.43 272 68.81 95.33}

v _{= 2 22 04h16m53s 8.1 -44.43 146 69.50 94.83}

w _{= 1 22 07h57m36s 5.2 -44.15 278 70.44 94.21}

Predicted impact times by P. W. Chodas and D. K. Yeomans (JPL/Caltech) as of 1994 July 16, last predictions before the impact [21]. Fragments J =13, M=10, and P1=8a are omitted because they have faded from view. The March'94 HST images show that P2=8b and G= 15 have split; there are not sufficient data to obtain independent predictions for the sub-components. The impact date/time is the time the impact would be seen at Earth, if the limb of Jupiter were not in the way (i.e., the time listed is the time of impact plus the light travel time to Earth); the date is the day in July 1994; The impact time uncertainty is a 1-sigma value in minutes. The impact latitude is J ovicentric (latitude measured at the center of Jupiter); the Jovigraphic latitudes are about 3.84 deg more negative. The impact longitude is System III, measured westwards on the planet. The meridian angle is the J ovicentric longitude of impact measured from the midnight meridian towards the morning terminator. This relative longitude is known much more accurately than the absolute longitude. At the latitude of the impacts, the Earth limb is at meridian angle 76 deg and the terminator is at meridian angle 87 deg. Angle E-J-F is the Earth-Jupiter-Fragment angle at the impact; values greater than 90 deg indicate a far-side impact. All impacts will be just on the far-side as viewed from Earth; later impacts will be closer to

the limb. 19

4 Observation

4.1 Instruments

The author joined the infrared observation program of the SL9 impact at the Okayama Astrophysical Observatory (OAO), which is a branch of National As­

tronomical Observatory of Japan (N AOJ). The observational group consists of the following members : Jun-ichi Watanabe of N AOJ, the team organizer, Hi­

toshi Hasegawa of ASTEC, Inc., Masanao Abe of Institute of Space and Astro­

nautical Science, Yuka Hirota of Tokyo Gakugei University, Takuya Yamashita of N AOJ, the chief of the OASIS, his three students of graduate course, Eiji Nishihara, Shin-ichiro Okumura, Atsushi Mori, and the author.

We used the 188-cm telescope of OAO, which is located at Mount Tikurinji, Kamogata town, Okayama, Japan (E133°36', N34°34', 372 m above the sea surface), to observe during July 15-24 on 1994. The imaging observation was carried out with a near-infrared camera, OASIS (Okayama Astrophysical System for Infrared imaging and Spectroscopy), which was developed by T. Yamashita and his group, attached to the Cassegrain focus (f/18) of the 188-cm telescope.

The OASIS uses NICMOS3 (HgCdTe) array, which can detect near-infrared wavelength from 1.0 pm to 2.5 pm. The details on the OASIS are described by Yamashita et a/.(1995) [37). This imaging system has a field of view as 4.1 arcmin x 4.1 arcmin, covered by 256 pixels x 256 pixels, and one pixel of the detector corresponds to 1.0 arcsec, which corresponds to about 3600 km at the distance of Jupiter for this observation. But the real spatial resolution was about 3 arcsec due to convolution effects of seeing.) The OASIS was at final stage of its marking at the SL9 impact and therefore there were some problem on the operation of the OASIS. For example, the integration (exposure) time occasionally varied due to some instabilities of the control system of the OASIS, and smear of electrical noise occasionally appeared in the frames.

The objects of this observation are already described in Section 1. We used some filters centering the wavelengths of the methane and/or hydrogen molec­

ular absorption bands in order to make certain of the detection of impact phe­

nomena. The observational wavelengths are listed for Table 2.

At the SL9 impact, July 1994, Jupiter rose above horizon at about 13h and fell at about 24h every day for local time at Okayama. Good condition for the observation was between 20h (the end of twilight) and 22-23h, at which Jupiter fell down to the lower angle limit of the telescope operation. However, we attempted the observation during the daytime, when some SL9 fragments collided to Jupiter for the daytime of Okayama after 14h. Because of the favorite weather during July 15-24, we obtained more than 3500 frames of Jupiter in total.

4.2 Reductions

The raw data include various noises, which must be removed. The raw data is represented as :

(1)

where Iraw denotes the raw data, Iobj the flux from the object, l3ky the flux from the sky, f3en the flat factor, or the sensitivity of each pixel of the detector,

!dark the dark and bias flux. In order to obtain the flux from the objects, we must correct the dark, the flat factor and the sky flux.

Using IRAF (Image Reduction and Analysis Facility) of NOAO (National Optical Astronomy Observations) for the imaging reductions and analyses, the reduction was proceeded as follows. After the diurnal observations of Jupiter, we took dark frames for each exposure time and flat frames for the filters used.

Ten dark frames, including bias, with same exposure time were averaged or chosen a median to be a mean dark frame for its exposure time. We subtracted the mean dark frame with same exposure time for each image from all the raw tmages.

For the correction of pixel-to-pixel variations in the sensitivity of the array, we got the flat frame by two methods, sky flat or dome fiat. The sky was taken with one of the filters and subtracted the mean dark frame with same exposure to the sky frame. Ten of the subtracted sky frames were averaged to be a mean sky frame, which was normalized to an average value of 1.0 to be made the sky-flat frame for each filter. On the other hand, a white sheet, which was illuminated or not by an illumination lamp, on inner side of the dome of the telescope for each filter was taken to make the dome fiat. Ten of the light-on frames and ten of the light-off frames were averaged respectively, and we subtracted the mean light-off from the mean light-on and normalized the resulting frame to an average value of 1.0, to make the dome-flat frame for each filter. We obtained one of the sky- and dome-flat frames every night.

The dark frames with same exposure time to the object frames were first subtracted from all the object raw frames, and the dark-subtracted frames were divided by the fiat frame with same filter to the object. There are still many pixels, which do not correctly respond to the incoming flux, in the flat-fielding frames. These bad pixels are corrected by interpolating from the neighboring normal pixels. After the reductions, the reduced frames include only images of Jupiter, his satellites, the impact radiation, and the sky radiation.

4.3 Calibration and Photometry

Because we observed a comparison star only one time for every night, it is difficult to correct dependency of flux on the air-mass, which absorbs the light from the objects, for absolute photometry. Thus we used another method for absolute photometry.

21

As mentioned earlier, the OASIS has the wide field of view as 4.1 arcmin

x 4.1 arc min and a nominal spatial resolution of 1.0 arc sec. Jupiter had the diameter of 38 arcsec at the SL9 impact. Although the resolution is not enough to resolve fine details in Jupiter's disk, this wide field capability allows us to take images of Jupiter with at least one of Galilean satellites in the same frame. This Galilean satellite can be used as a reference for calibration. We used Europa for the C impact (impact of the fragment C), and Io for the D and K impacts. The flux from the satellite, which is almost reflected sunlight for nearby wavelength of 2 J.Lm, can be estimated as follows (38]:

F = 7r _R2 F$ S

�

where F is the reflected flux from the satellite, A the phase albedo of the satel­

lite (39], 1r F3 the solar flux at the distance of 1 AU, R the distance of Sun and Jupiter in the AU unit, 5.41 AU at July 1994, � the distance of Earth and Jupiter, 5.17 AU (7.75 x 108 km), S the cross section of the satellite. The radii of Io and Europa are 1821 km and 1565 km, respectively. The solar flux at the distance of 1 AU is 56.2 wm-2J.Lm-1 for the wavelength of 2.35 J.Lm,

80.2 Wm-2J.Lm-1 for 2^.17 J.Lm, and 202.2 Wm-2J.Lm-1 for 1.70J.Lm (40]. The albedos of the Galilean satellites are known to change according to the position of the satellite around Jupiter (39, 41]. We estimate the albedos of the satellites from Sill and Clark ( 1982) according to the satellite positions around Jupiter at the impacts (41]. The position of the satellite around Jupiter is presented as position angle, or orbital longitude, which is the angle between the satellite and the superior (outer) geocentric conjunction point toward the orbital direction in the Jovian equatorial plane.

One example of the obtained images is shown in Fig. 11, which is the image of the impact of fragment K for July 19, 1994. Fig. 11 shows Jupiter, which can be seen only north and south polar hazes, some debris clouds by previous impacts of fragments H, E, A and C (from right to left) above the south polar haze, a very bright spot from the K impact at lower-left limb of Jupiter, and two Galilean satellites, Io (left) and Ganymede (right). The author measured the flux of the impact brightness and the comparison satellite by the aperture photometry using IRAF. The aperture size is determined to be able to cover all photons from the object. It is found that the aperture with the radius of 6 pixels normally cover the unresolved object.

For the photometry of the object, the center position of the aperture is placed at the centroid of the object. When the center of the object is well determined, the count distribution against the distance from the center shows a smoothed Gaussian function, as shown in Fig. 12.

The sky background is estimated by averaging pixel count outside the aper­

ture. The outside region is an annulus, whose inner radius is 10 pixels and outer radius is 20 pixels. The sky background is subtracted from the total counts within the aperture to calculate the total count of only the object. The total

brightness of the object and the error of the measurement are calculated as :

Nobj = Ntotal - J_,kySap , (J"obj =

J

(3) (4)

where Nobj is the total brightness of the object, Ntotal the total count within the aperture, I.,ky the sky background averaging within the annulus, Sap the area of the aperture or number of pixels within the aperture, (]" obj the standard deviation of the total brightness of the object, i.e., error of the Nobj, (J"_,ky the standard deviation of the sky background within the annulus.

Comparing the total brightness of the impact radiation with that of the satellite, the absolute flux of the impact radiation is obtained as follows:

Nimpact

Fimpact ⁼ N F_,atellite ,

"'ate/lite

(5)

where F3atellite is the calculated flux from the satellite by equation

(2)

Fimpact the absolute flux from the impact. The mean error of the flux ratio of the impact to the satellite is estimated to be about

5

23

Figure 11: An example of near infrared image. This image was taken at 10h 26m 15s UT, for wavelength of 2.35 pm during the observation of the impact of fragment K on July 19, 1994 at OAO/OASIS. This image shows Jupiter, whose north and south polar hazes only can be seen, the debris clouds by previous impacts of the fragments H, E, A and C (from right to left) above the south polar haze, a very bright spot from the K impact at lower-left limb of Jupiter, and two Galilean satellites, Io (left) and Ganymede (right).

--- -�--

<1)

:::;,

....

=--

<1)

>- v>

c: <1) c:

1000 750 500 250 0

NOAO/IRAF V2.10EXPORT takeuchi1sl9 Fri 19:02:12 15-0ec-95 jdf7190037c: Radial profile at 171.49 185.89

0 2 ⁴

Radius

6 ⁸

NOAO/IRAF V2.10EXPORT takeuchiisl9 Fri 19:08:36 15-0ec-95 Center: xc=171.52 yc=185.97 xerr=0.03 y^err^=0.03 Sky: value=3446.77 s^igma⁼38^.88 skew=11.97 nsky=871 nrej=68

Photo�etry: �a^xaJ,^ert=6^.00 mag=15.671 merr=0.037 I�ge: jdf71 cr0037c Sf�^r 1: 172066 ^185.

6js

4400 4200 4000

3800

3600

3400

10 15 20 25

Radial Distance (lower-pixels, upper-scale units)

Figure ^12: An example of aperture photometry. Using NOAO IRAF, the flux of the ^K impact in Fig. ¹¹ is measured. (Top) Radial plot of bright flash of the ^K impact, which appears to be a point like object. When the center of the object is well determined, radial plot shows smoothed sharp of Gaussian function, like this. (Bottom) Aperture photometry. All counts within radius of ⁶ pixels are summed to be total count. The total count is subtracted all sky count within the aperture. The sky level is estimated as a average within outer annulus.

25

Table 2: List of filters used in the observations. FWHM is bandwidth (Full Wide at Half Maximum) of filter. The wavelength of 2.35 pm was mainly used for the observations reported in this paper.

Wavelength

(pm)

1.50 1.58 1.70 2.165 2.35

FWHM

(pm)

0.05 0.01 0.05 0.022 0.05

Remarks

Weak ammonia absorption Continuum

Methane absorption

Methane and Hydrogen absorptions Strong methane absorption

4.4 Impact of Fragment C

The impact of the fragment C was the first impact observable from Japan, and its predicted impact time was 07h02m UT, July 17 1994, (16h02m JS T (Japan Standard Time) ) [21], which means that the C impact was a daytime event in Japan. Because the sky flux is very high in daytime, we used the longest wavelength, 2.35 ^J..Lm, for the observation in order to darken the sky.

However, the detector was saturated around the center of the image of the Galilean satellite Io, as shown in Fig. 13. Thus we used the satellite Europa for calibration.

At the C impact the position angle of Europa was 150-155°, Therefore we must use the average of albedos of the leading and trailing side of Europa.

However, the albedo of the leading side of Europa is unknown. The author estimated the albedo for the C impact as 0.12 from the albedo of the trailing side, 0.11, taking into account the albedo ratio of the leading value to the trailing one for the wavelength of 1.0 ^J..Lm (41]. Thus the flux from Europa was estimated to 9.4 x 10-13Wm-2 J..Lm-1 from the equation (2).

The observation of the C impact event was continued from 06h51m to 07h50m UT. The integration time was chosen 1 sec and the time interval be­

tween the integrations was about 10 sec. The accuracy of absolute timing is

± 1s. (In this paper the start times of integrations are referred as the observa­

tional times for our observations.) For this observation a debris cloud produced by the impact of fragment A, which occurred one Jovian rotation earlier, was situated on the south-west limb of Jupiter, and the observation system could not resolved two sites of the A and C impacts, as shown in Fig. 14. Thus we can measure only the combined flux from the C impact event with the reflected flux by the A cloud.

The calibrated lightcurve for the C impact event is shown in Fig. 15. Before the C impact, the flux originated with reflecting sunlight from the A debris cloud. The flux increased from 07h12m07s to 07h14m. This precursor is the first signal of the C impact in our observation. The peak flux of the precursor is estimated to be about 1 x 10-13Wm-2J..Lm-1 (0.18Jy) at 13m21s, subtracted by pre-impact level of about 1 x 10-13Wm-2J..Lm-1. ( 1 Jy is 10-26Wm-2Hz-1.)

Then the flux rapidly increased from 07h17m00s, 5 minutes after the precursor, and the detector was saturated from 07h19m14s to 07h24m22s due to this bright signal. The bright flash continued about 10 minutes and finished at 07h27m00s.

In decreasing phase of flux of the C impact a plateau (or shoulder) appeared from 07h27m to 07h33m. After the plateau, 07h33m, the flux originated with mainly reflected light by the A and C clouds.

The C impact was observed by other observatories, the 3.9-m Anglo Aus­

tralian Telescope (AAT) with IRIS (InfraRed Imaging Spectroscopy) at Siding Spring Observatory (Australia) (42], and the NASA IRTF (InfraRed Telescope Facility) 3.0-m telescope (Mauna Kea, Hawaii) [43] and others. By the AAT

27