astro-ph0603306/ms.tex
1:  %% manuscript produces a one-column, double-spaced document:
2: 
3: \documentclass[12pt,preprint]{aastex}
4: 
5: %\documentclass[12pt]{aastex}
6: %\usepackage{emulateapj5}
7: 
8: \newcommand{\cn}{{$^{12}$C$^{14}$N}{}}
9: \newcommand{\Cn}{{$^{13}$C$^{14}$N}{}}
10: \newcommand{\cN}{{$^{12}$C$^{15}$N}{}}
11: \newcommand{\rC}{{$^{12}$C/$^{13}$C}{}}
12: \newcommand{\rN}{{$^{14}$N/$^{15}$N}{}}
13: \newcommand{\NN}{{$^{15}$N}{}}
14: \newcommand{\nn}{{$^{14}$N}{}}
15: \newcommand{\CC}{{$^{13}$C}{}}
16: \newcommand{\cc}{{$^{12}$C}{}}
17: 
18: 
19: %% If you wish, you may supply running head information, although
20: 
21: \shorttitle{DEEP IMPACT WITH UVES AT VLT AND HIRES AT KECK 1}
22: \shortauthors{JEHIN, MANFROID, HUTSEM\'EKERS ET AL.}
23: 
24: %% This is the end of the preamble.  Indicate the beginning of the
25: %% paper itself with \begin{document}.
26: 
27: \begin{document}
28: 
29: %% LaTeX will automatically break titles if they run longer than
30: %% one line. However, you may use \\ to force a line break if
31: %% you desire.
32: 
33: \title{Deep Impact : High Resolution Optical Spectroscopy \\
34: with the ESO VLT and the Keck 1 telescope}
35: 
36: %% As in the title, you can use \\ to force line breaks.
37: 
38: \author{
39: E. Jehin\altaffilmark{1},
40: J. Manfroid\altaffilmark{2},
41: D. Hutsem\'ekers\altaffilmark{2},
42: A.L. Cochran\altaffilmark{3},
43: C. Arpigny\altaffilmark{2},
44: W. M. Jackson\altaffilmark{4},
45: H. Rauer\altaffilmark{5},
46: R. Schulz\altaffilmark{6} and
47: J.-M. Zucconi\altaffilmark{7}
48: }
49: 
50: \altaffiltext{1}{European Southern Observatory, Casilla 19001, Santiago, 
51: Chile; ejehin@eso.org}
52: \altaffiltext{2}{Institut d'Astrophysique et de G\'eophysique, Sart-Tilman, B-4000 Li\`ege, Belgium; manfroid@astro.ulg.ac.be; arpigny@astro.ulg.ac.be; hutsemekers@astro.ulg.ac.be}
53: \altaffiltext{3}{Department of Astronomy and McDonald Observatory, University 
54: of Texas at Austin, C-1400, Austin, USA; anita@barolo.as.utexas.edu}
55: \altaffiltext{4}{Department of Chemistry, University of California, 1 Shields Avenue, Davis, CA 95616; wmjackson@ucdavis.edu}
56: \altaffiltext{5}{Institute of Planetary Research, DLR, Rutherfordstr. 2, 12489 Berlin, Germany; Heike.Rauer@dlr.de}
57: \altaffiltext{6}{ESA/RSSD, ESTEC, P.O. Box 299, NL-2200 AG Noordwijk, 
58: The Netherlands; rschulz@rssd.esa.int}
59: \altaffiltext{7}{Observatoire de Besan\c{c}on, F25010 Besan\c{c}on Cedex, 
60: France; jmz@dalai-zebu.org}
61: 
62: 
63: \begin{abstract}
64: 
65: We report on observations of comet 9P/Tempel\,1 carried out before, during,
66: and after the NASA DEEP IMPACT event (UT July 4), with the optical 
67: spectrometers UVES and HIRES mounted on the telescopes Kueyen of the 
68: ESO VLT (Chile) and Keck 1 on Mauna Kea (Hawaii), respectively. A total 
69: observing time of 
70: about 60 hours, distributed over 15 nights around the impact date, allowed us 
71: (i) to find a periodic variation of 1.709 $\pm$ 0.009 day in the CN and NH 
72: flux, explained by the presence of two major active regions;
73: (ii) to derive a lifetime $\gtrsim$ 5 x 10$^4$ s for the parent of the CN 
74: radical from a simple modeling of the CN light curve after the impact; 
75: (iii) to follow the gas and dust spatial profiles evolution during the 4 
76: hours following the impact and derive the projected velocities (400 m/s
77: and 150 m/s respectively); 
78: (iv) to show that the material released by the impact has the same carbon 
79: and nitrogen isotopic composition as the surface material 
80: (\rC\ = $95\pm15$ and \rN\ = $145\pm20$).
81: 
82: \end{abstract}
83: 
84: %% Keywords should appear after the \end{abstract} command. The uncommented
85: %% example has been keyed in ApJ style. See the instructions to authors
86: %% for the journal to which you are submitting your paper to determine
87: %% what keyword punctuation is appropriate.
88: 
89: \keywords{Comets : general --- comets: individual (9P/Tempel 1) 
90: --- Solar System: formation}
91: 
92: %\clearpage
93: \section{OBSERVATIONS}
94: 
95: High-resolution spectra of comet Tempel\,1, the target of the NASA DEEP 
96: IMPACT (DI) mission, have been collected with the UV-Visual Echelle 
97: Spectrograph (UVES), of the ESO VLT in early June (UT 2, 7 and 8) and during 
98: a 10 night run around the DI event (UT July 2-11). The nights were 
99: photometric or clear and the seeing excellent to good (0.4"-1.1"). Two long 
100: exposures (of $\sim$7200s) have been secured each night using two different 
101: beam splitters, a combination which allowed us to obtain on each 
102: observing date a spectrum covering the full optical range (304-1040nm, except
103:  for a few narrow gaps). Both settings were chosen in order to include the CN
104:  (0-0) Violet band near 388\,nm. The narrow slit of the spectrograph (0.44" x
105:  10" or 288\,km x 6540\,km on July 4) provides a resolving power 
106: $\lambda/\Delta\lambda$ $\sim$ 83,000 and was generally put on the center of 
107: light of the comet (and in a few cases at about 1.0" from it). The position 
108: angle of the slit was along or perpendicular to the Sun-comet vector most of 
109: the time and an atmospheric dispersion corrector (ADC) was used to sample the
110:  same region during a given exposure.
111: 
112: On July 4, the comet was setting at Paranal Observatory at the exact time of 
113: the DI event UT 05:52 \citep{AHearn05}. The acquisition of the last UVES 
114: spectrum ended at UT 02:59 which was just a couple of hours before the impact, 
115: i.e., before the first spectrum obtained with HIRES, the High 
116: Resolution Echelle Spectrometer of the Keck 1 telescope installed 
117: on Mauna Kea (Hawaii). The comet was visible again from Paranal 17 hours 
118: later. The data sets are complementary: the UVES one nicely describes the 
119: pre- and post- impact behaviour of the comet, while HIRES data contain unique
120:  information about the direct consequences of the impact (up to 4 hours 
121: later).
122: 
123: The spectra have been reduced with special emphasis on the orders showing the
124:  CN band at 388\,nm. We used the echelle package of the IRAF software (NOAO) 
125: to calibrate and extract the spectra. The dust-reflected sunlight underlying 
126: the cometary emissions was removed by subtracting a solar reference spectrum 
127: (the Moon, in this case) after the appropriate Doppler shift, profile fitting
128:  and normalization were applied.
129: 
130: \section{CN PERIODIC VARIATIONS}
131: 
132: The relative flux in the CN band - between 386.2 and 387.5\,nm and 
133: integrated over the full slit area - was calculated for the 13 nights of 
134: UVES observations. 
135: 
136: The goal of the flux measurements was first to evaluate the variations triggered 
137: by the impact. 
138: But obvious flux variations of up to 30\% were visible from night to night. 
139: A period search analysis has been performed using Fourier series fits and other 
140: methods like the Renson algorithm \citep{Ren78} better suited for anharmonic light 
141: curves. It yielded a period of 1.709 $\pm$ 0.009 day (40.86 $\pm$ 0.05 hrs) in very 
142: good agreement with the 1.701 $\pm$ 0.014 day rotation period determined from the 
143: DI spacecraft nucleus lightcurve \citep{AHearn05}.  
144: 
145: The phase diagram of the CN flux using this period is shown in Fig.1. 
146: The flux modulation is obviously synchronized with the rotation period and was stable 
147:  over more than one month. The only deviating points are those corresponding to the 
148: first four post-impact measurements obtained on UT July 04 and 05 (phases around 0.45 and 0.1). 
149:  There is no evidence of strong sporadic activity in the data. None of the 13 UVES 
150: observing dates corresponds to the outbursts detected by the DI spacecraft or the 
151: ground based observatories.
152: After removing the periodic background, the light curve shows a net flux excess of about 
153: 20\% and 7\% respectively, 17 and 41 hours after the impact, indicating that the gas released 
154: by the ejected material had not yet completely vanished. After July 7.0 
155: (and up to July 12.0), no excess could be detected at all in the region studied. 
156: 
157: The shape of the CN phase diagram may be explained by the periodic passage of two 
158: major active regions, as well as a weaker one, into sunlight.
159: Such an interpretation was also used to explain the comet Halley photometric observations 
160: during its 1985/1986 apparition \citep{Sch90}. 
161: The brightest feature starts to produce gas around phase 0.9 (its minimum) and reaches 
162: its maximum (when the source stops its production) at phase 0.1. The second region 
163: is $\sim$20\% less active and lasts also for about 8 hours (from phase 0.6 to 0.8). 
164: They could both be located at the same latitude (they see the Sun during the same 
165: amount of time), the difference in activity resulting from a different size or 
166: sunlight illumination of the active regions due for instance
167: to different land morphology. The distance between the two features
168: is a bit more than 1/3 of the rotation. The shoulder at phase $\sim$ 0.4 could be some 
169: evidence for the presence of a third and much fainter source.
170: 
171: The impact occurred at the beginning of the strongest periodic 
172: brightening (Fig.1). It is then possible that this active region is visible close to 
173: the terminator in the images sent by the impactor. As already noted by \citet{AHearn05}, 
174: a possible candidate is the large and smooth area (labeled ``a'') in their Fig.1
175: and the scarp to the north of this feature would be an excellent candidate for a more 
176: localised, fast reacting region (the large smooth area being a kind of reservoir).  
177: 
178: It is highly significant to note that almost all outbursts observed by the DI spacecraft
179: (we have a precise knowledge of their onset time \citep{AHearn05}) correspond to minima 
180: in our gas phase diagrams (see Fig.1). Moreover 4 of the 6 outbursts are located just before the 
181: strongest brightening (phases from 0.85 to 0.95). Thus, the outbursts seem clearly
182: associated with the active regions we found (they occur at the same phases) and would be 
183: occasionally triggered when one of these regions comes into daylight. The outbursts could 
184: just be a particularly strong phase of outgassing from the same region. 
185: 
186: The flux in the NH 0-0 (A-X) band at 336 nm shows a nice correlation with the CN variations 
187: and any phase shift must be small, at most a couple of hours (Fig.1). The first post-impact 
188: measurement has an excess of 30\%, i.e., significantly more than CN, but, contrary to CN, the second 
189: one is not enhanced (or only slightly), which means that NH disapeared sooner
190: after the impact. Many more species are available in our spectra (OH, C$_3$, C$_2$, NH$_2$, etc.)
191: and will be examined in the same way. This may provide interesting information on
192: their respective parent lifetimes. For instance, the NH parent seems to have a similar
193: lifetime to the CN parent lifetime because the phase-shift between the two species is very small.
194: With a lifetime in the range 2-6 10$^4$ s \citep{Wyc88, Fin91} for NH$_2$, 
195: similar to that of the CN parent (2-5 10$^4$ s, \citet{Fra05} and references therein), this 
196: observation is indeed in agreement with NH$_2$ being the main NH parent \citep{Fin91}.
197: 
198: Data on HCN - a possible parent candidate of CN - have been obtained with the IRAM 30-m 
199: radio telescope from May 4.8 to 9.0 and show a 1.67$\pm$0.07 day periodicity \citep{Biv05}. 
200: The beam used sampled similar regions to UVES ($\sim$10").
201: The data have been phased with the 1.709 day period and superposed with the UVES CN flux 
202: measurements. 
203: There are only five HCN measurements, with large errors, so it is not possible
204: to verify if the differences we see in shape and phase with our CN lightcurve are real.
205: If this is the case, this could be some additional evidence that HCN is not the only
206: parent of CN \citep{Fra05, Man05}.
207: 
208: 
209: \section{CN AND DUST IMPACT LIGHTCURVES}
210: 
211: Three pre-impact spectra were obtained on May 30 with the HIRES spectrograph at the 
212: Keck 1 telescope and a series of 14 relatively short exposures (10 to 30 minutes), with 
213: one right before the impact, were taken on the DI night\footnote{Those data are 
214: publicly available at http://msc.caltech.edu/deepimpact/}. 
215: The 0.86'' by 7'' slit provides a resolving power $\lambda/\Delta\lambda$ $\sim$ 48,000 
216: and samples a comparable zone (563 km x 4578 km) 
217: to that of UVES. The slit was always centered on the nucleus and aligned along 
218: the parallactic angle. The weather was photometric on July 4 and the seeing 
219: excellent ($\sim$0.6'').
220: 
221: The total CN flux in the slit has been calculated and normalized 
222: to the UVES fluxes by comparing the first, pre-impact spectrum of July 4, as well 
223: as the three May spectra with our derived lightcurve. 
224: The HIRES slit was 2x wider than the UVES one and always centered on the nucleus,
225: leading to better measurement of the dust component. The consequences of the impact 
226: are readily seen in Fig.2, both in the dust and CN flux. The CN periodic
227: variation has been removed using the UVES lightcurve and the dust background 
228: has been taken out using the pre-impact spectrum. 
229: 
230: The total CN flux reaches its maximum at UT 07:28 ($\pm$ 10min) or 5760 $\pm$ 600 seconds 
231: after the impact. It is enhanced by a factor of 2.8 compared with the pre-impact situation.
232: The decline is explained (at least partially) by the molecules starting to leave the slit
233: area, mostly through the slit length (the lifetime of the CN parent is indeed long 
234: enough \citep{Rau03} to make the escape from the slit width unnoticed).
235: The average projected speed of CN to cross the half slit is then $\sim$ 0.40 $\pm$ 
236: 0.04 km/s. This is smaller than typical molecular outflow velocities in 
237: cometary comae \citep{Com05} but it is only a lower limit due to the 
238: projection effect. 
239: The CN peak intensity of the spatial profiles is reached well before the total CN flux
240: maximum, at about UT 06:30 ($\pm$ 10min).  
241: 
242: The CN light curve contains valuable information about the dissociation lifetime of
243:  its parent molecule but a complete modeling is beyond the scope of this paper.
244: To first approximation, the CN impact light curve may be interpreted by assuming 
245: a competition between the CN creation resulting from the dissociation of a parent 
246: molecule ($\propto (1-\exp(-t/\tau))$ where $\tau$ is the parent lifetime) and the 
247: exit from the slit area ($\propto t^{-1}$ after filling the slit width and 
248: $t^{-2}$ after filling the slit length).
249: A precise determination of the parent lifetime from our data would require a good
250: knowledge of the CN velocity distribution. The radial profiles (Fig.3) indicate a range of 
251: projected velocities from 250 to 650 m/s, in good agreement with the mean velocity 
252: of 400 m/s deduced from the light curve. Adopting 400 m/s, the slow decline of the light
253: curve indicates that CN creation still occurs after filling the slit area, requiring 
254: relatively long parent lifetimes in agreement with those determined using the Rosetta 
255: spacecraft data ($\tau$ $\sim$50000 s) \citep{Kel05}.
256: 
257: The excess of the CN flux observed with UVES until two days after the impact
258:  indicates that the decline might have occured in two phases, a fast one documented by 
259: HIRES, and a slow one lasting over more than a full rotation of the comet. 
260: A linear extrapolation of the flux decrease observed by HIRES after the impact shows 
261: that the extra CN should have been gone from the studied area about 8 hours after the impact. 
262: The excess measured in UVES data could be explained by remnant activity of the 
263: crater or the presence of another minor CN parent with a longer lifetime.  
264: Unfortunately, data are missing to describe the lightcurve between MJD 53555.4 and 53556.0. 
265: 
266: Taking the observed time evolution of the CN excess during 
267: the impact as the response of the comet to an elementary event, we tried to model the 
268: out-of-impact CN lightcurve as the response to a series of such elementary pulses. 
269: The observations can be accurately reproduced with just three bursts (or jets, corresponding 
270: to the observed bumps) (see Fig.1). These are simulated by convolving the elementary response 
271: with a simple temporal variation representing the instantaneous production of each source.
272: There is no room for a ``background" CN flux, i.e., all CN is produced by these periodic 
273: jets starting to release gas at phases 0.25, 0.45 and 0.85 (intensity of, respectively, 
274: 0.2, 0.7 and 1.0).
275: The full width at half maximum duration of the bursts is 8 hours assuming a 
276: quasi-symmetrical triangular shape to the intensity of the jets. This behaviour equally 
277: explains NH and presumably all gases. 
278: 
279: The variations of the dust after the impact are much faster, with a very steep brightening. 
280: The maximum is already reached  at about 
281: UT 06:18 $\pm$ 10\,min, the dust being at that time enhanced by a factor $\sim$ 8.5 with 
282: respect to the pre-impact spectrum. Contrary to CN, the intensity peak in the dust profiles 
283: is reached at about the same time as the spatially integrated flux.
284: The decline of the dust emission may also be interpreted by the escape of dust from 
285: the slit. This would give a projected dust velocity of $\sim$ 0.18 $\pm$ 0.05 km/s,
286: as, in this case, the escape from the slit width will be the dominant factor.  
287: However the slow and quasi linear decline would require a rather broad range of velocities 
288: -- slower than the gas component (Fig.3) -- and/or complex processes like destruction 
289: of highly reflective icy grains by sunlight.
290: The radial profiles show that the dust is expanding at a much slower pace 
291: than CN, at about 0.13 $\pm$ 0.03 km/s during the first 1.5 hour after the impact. 
292: This is consistent with the value given above and what others have measured \citep{Kel05,
293: Sug05}.
294: 
295: There is a slight but clear increasing shift of the position of the CN intensity peak in
296: the radial profiles with respect to the dust (nucleus dominated) peak (Fig.3). The shift
297: of $\sim$350\,km (at 07:10 UT, and at a mean PA=215$^\circ$) is in the direction of the
298: plume ejecta and is most certainly associated with it. It may indicate a preferential ejection 
299: direction for the gas with respect to the dust, the gas being emitted towards the Sun.
300: 
301: 
302: \section{ISOTOPIC RATIOS}
303: 
304: Measuring the isotopic ratios in 9P/Tempel\,1 before and after the impact was a unique 
305: opportunity to check whether the material buried several meters below the surface and released
306: by the impact is different or not from what we usually observe.
307: 
308: It was important to obtain a high quality pre-impact spectrum in order to compare it
309: with post-impact data. This was achieved with UVES thanks to the 10 hours obtained 
310: in June and the 8 hours during the two pre-impact nights. The individual CN (0-0) Violet 
311: band spectra were combined after extraction with an optimal weighting scheme in 
312: order to maximize the overall signal-to-noise ratio. Synthetic spectra
313: of the different CN isotopes were computed for each observing circumstance following 
314: the scheme described by Zucconi and Festou \citep{Zuc85}. The isotope mixture was then 
315: adjusted to best fit the observed final spectrum 
316: \citep{Arp03, Jeh04}. The same was done for the UVES post-impact data, and as no difference 
317: was found (which is not too surprising as those spectra were only slightly affected 
318: by the impact (Fig.1)), all the spectra were combined to produce a single 50 hours spectrum.
319: The best fit is obtained for an isotopic mixture \rC$=(95\pm15)$ and \rN$=(145\pm20)$.
320: 
321: From the HIRES impact lightcurve (Fig.2) we extracted the spectra showing a brightening 
322: of more than a factor two (the 10 spectra from 06:33 to 09:11 UT). Thus in those spectra 
323: half the CN flux might come from the fresh material released by the impactor. 
324: Despite the shorter exposure time (about 4 hours of total exposure time) and the
325: lower resolving power of the HIRES spectrum compared to the UVES out of impact spectrum, 
326: the large collecting area of the Keck 1 telescope and the relative brightening of the 
327: comet allowed us to determine the following values \rC$=(95\pm15)$ and \rN$=(165\pm30)$ 
328: for the carbon and nitrogen isotopic ratios.
329: Those values are compatible with the values determined out of impact. It appears that 
330: the \rN\ ratio of the burried material is still most probably below the solar value \citep{Arp03}.
331: 
332: After comet 88P/Howell, comet 9P/Tempel\,1 is the second Jupiter-family 
333: comet to have a \rN\ ratio determination. Both values are in excellent agreement and are
334: similar to the ratios measured in half a dozen Oort-cloud comets \citep{Hut05}. 
335: The fact that the ejected material - supposed to be pristine as it was never exposed to 
336: the solar radiation and cosmic rays - has the same isotopic composition, 
337: favors an isotopic homogeneity between the two populations of comets, despite the fact 
338: that they are expected to have formed at very different locations in the solar system \citep{Wei99}. 
339: This is a strong argument in favor of a primordial origin of the high content of
340: $^{15}$N in cometary volatiles. This peculiarity is not the result of some alteration process 
341: of the comet's surface material and was already present in the protosolar nebula 
342: before the accretion process, which gave birth to the comets and planets, took place. 
343: In case the N isotopic ratio changes within the solar nebula, this is a new argument in 
344: favor of the volatile ices in Jupiter-family and Oort-cloud comets 
345: originating in a common region of the protoplanetary disk \citep{Mum05}.  
346:  
347: \acknowledgments
348: Based on observations carried out at the European Southern Observatory (ESO) 
349: under prog. 075.C-0355(A). This program was part of a joint European initiative in 
350: support of the NASA DEEP IMPACT mission to comet 9P/Tempel\,1 by ground-based 
351: observations at La Silla and VLT Paranal. 
352: Some of the data presented herein were obtained at the W.M. Keck Observatory, 
353: which is operated as a scientific partnership among the California Institute 
354: of Technology, the University of California and the National Aeronautics and 
355: Space Administration. The Observatory was made possible by the generous 
356: financial support of the W.M. Keck Foundation.
357: JM is Research Director and DH is Research Associate at FNRS (Belgium).
358: 
359: 
360: %\clearpage
361: 
362: \begin{thebibliography}{}
363: \bibitem[A'Hearn et al.(2005)]{AHearn05} A'Hearn, M. F., et al. 2005, Science, 310, 258
364: \bibitem[Arpigny et al.(2003)]{Arp03} Arpigny, C., Jehin, E., Manfroid, J., 
365: Hutsem\'ekers, D., Schulz, R., St\"uwe, J.A., Zucconi, J.-M., \&  Ilyin, I. 
366: 2003, Science, 301, 1522
367: \bibitem[Combi et al.(2005)]{Com05} Combi, M. R., Harris, W. M., \& Smyth, W. H. 
368: in {\it Comets II}, M. C. Festou, H. U. Keller, H. A. Weaver, Eds. 
369: (The University of Arizona Press, Tucson, 2005), pp.523-552.
370: \bibitem[Biver et al.(2005)]{Biv05} Biver, N., Bockel\'ee-Morvan, D., Colom, P.,
371:  Crovisier, J., Lecacheux, A., \& Paubert, G. 2005, IAUC, 8538
372: \bibitem[Fink et al.(1991)]{Fin91} Fink, U., Combi, M.R., \& DiSanti, M.A. 1991, \apj, 
373: 383, 356
374: \bibitem[Fray et al.(2005)]{Fra05} Fray, N., B\'enilan, Y., Cottin, H., Gazeau,
375:  M.-C., \& Crovisier, J. 2005, Planet. Sp. Sci., 53, 1243
376: \bibitem[Hutsem\'ekers et al.(2005)]{Hut05} Hutsem\'ekers, D., Manfroid, J., Jehin,
377: E., Arpigny, C., Cochran, A., Schulz, R., St\"uwe, J. A., \& Zucconi, J.-M. 2005, \aap, 440, L21
378: \bibitem[Jehin et al.(2004)]{Jeh04} Jehin, E., et al. 2004, \apj, 613, L161
379: \bibitem[Keller et al.(2005)]{Kel05} Keller, H., et al. 2005, Science, 310, 281
380: \bibitem[Manfroid et al.(2005)]{Man05} Manfroid, J., Jehin, E., Hutsem\'ekers, D., Cochran, A., 
381: Zucconi, J.-M., Arpigny, C., Schulz, R., \& St\"uwe, J. A. 2005, \aap, 432, L5
382: \bibitem[Mumma et al.(2005)]{Mum05} Mumma, M. J., et al. 2005, Science, 310, 270
383: \bibitem[Renson(1978)]{Ren78} Renson, P. 1978, \aap, 63, 125
384: \bibitem[Schleicher et al.(1990)]{Sch90} Schleicher, D. G., et al. 1990, \aj, 100, 896
385: \bibitem[Sugita et al.(2005)]{Sug05} Sugita, S., et al. 2005, Science, 310, 274
386: \bibitem[Rauer et al.(2003)]{Rau03} Rauer, H., et al. 2003, \aap, 397, 1109
387: \bibitem[Weissman(1999)]{Wei99} Weissman, P. R. 1999, Space Sci. Rev., 90, 301
388: \bibitem[Wyckoff et al.(1988)]{Wyc88} Wyckoff, S., Tegler, S., Wehinger, P.A., Spinrad, H., 
389: \& Belton, M.J.S., 1988, \apj, 325, 927
390: \bibitem[Zucconi \& Festou(1985)]{Zuc85} Zucconi, J.M., \& Festou, M.C. 1985, 
391: \aap, 150, 180
392: \end{thebibliography}
393: 
394: 
395: %\clearpage
396: 
397: %---------------------------------------------------------------------
398: \begin{figure}
399: \plotone{f1.eps}
400: \caption{
401: The UVES CN and NH gas emission rotational phase diagrams. Zero phase is at the time of 
402: impact as observed from Earth : MJD 53555.2445 (July 4, UT 05:52:02). 
403: In this figure, and in the following ones, the  fluxes are given in arbitrary units.
404: The dashed line models of the gas "lightcurves" are based on a three harmonic least
405: squares fit in which each data point has the same weight. 
406: The fit of the NH data is independent of the CN one and the similarities of the two curves 
407: should be noted.
408: Both data sets are phased to P = 1.709 day and reveal the existence of two main active 
409: regions.
410: The observations can be accurately reproduced (continous line) with just three periodic bursts 
411: (or jets, corresponding to the observed bumps) starting to release gas at phases 0.25, 0.45 
412: and 0.85 and with a duration of 8 hours (see text).
413: The enhanced post impact data for UT July 4 and 5 (around 23:00) are marked as circles and
414: correspond to phases 0.44 and 0.49 and 0.06 and 0.11, respectively. 
415: The phase of the outbursts detected by the DI spacecraft are indicated as tick marks
416: at the bottom of the figure. 
417: The scatter in the data comes from the centering and orientation of the slit, 
418: sky transparency or intrinsic variability of the sources. The NH 0-0 (A-X) emissions at 336 nm 
419:  appear only in the bluest setup so we have only one data point per night for that species.
420: }
421: \end{figure}
422: % ------------------------------------------------------------------
423: 
424: 
425: %\clearpage
426: 
427: %---------------------------------------------------------------------
428: \begin{figure}
429: \plotone{f2.eps}
430: \caption{
431: The HIRES CN and dust impact lightcurves.
432: Time 0.0 is the time of the impact as observed from Earth : MJD 53555.2445 
433: (July 4, UT 05:52:02). 
434: Each data point represents for each spectrum (at mid exposure) the 
435: total (spatially integrated) flux in the CN (0-0) violet band (388\,nm) and in the dust 
436: continuum underlying the CN features (empty and filled squares respectively). 
437: The fluxes are normalised to the exposure time, corrected for airmass, and the respective 
438: backgrounds are subtracted. The CN curve is scaled (x3.5) to the dust curve for 
439: easier comparaison. 
440: }
441: \end{figure}
442: % ------------------------------------------------------------------
443: 
444: %\clearpage
445: 
446: %---------------------------------------------------------------------
447: \begin{figure}
448: \plotone{f3.eps}
449: \caption{
450: The HIRES CN and dust radial profiles during the first hour following the impact.
451: The profiles of the first 6 post-impact spectra (respectively at mid exposure time
452: 06:00:18.0, 06:11:12.0, 06:24:35.0, 06:40:29.0, 06:56:22.0 and 07:12:16.0 UT, July 4) 
453: are extracted along the 7'' slit of the spectrograph and background subtracted.
454: The pixel scale in spatial direction is 0.239'' or 155 km. 
455: The slit was set along the parallactic angle with the consequence that the position angle (PA) 
456: on sky was changing during the observations (20.1$^{\circ}$, 24.8$^{\circ}$, 29.7$^{\circ}$, 
457: 35.5$^{\circ}$, 40.6$^{\circ}$ and 48.9$^{\circ}$ for, respectively, the first 6 post-impact spectra).
458: %The slit was set along the parallactic 
459: %angle and changing from position angle PA=20$^{\circ}$ to 50$^{\circ}$ during the observations. 
460: This direction is close to the axis of the expanding cloud (PA=225$^{\circ}$, \citet{AHearn05}). 
461: Note that the PA of the extended Sun-comet radius vector is 111$^{\circ}$.
462: The profiles are recentered on the dust peak to compensate for the shift along the
463: slit induced by the atmospheric refraction. The different behaviour
464: of the gas and the dust is obvious.
465: }
466: \end{figure}
467: % ------------------------------------------------------------------
468: 
469: 
470: \end{document}
471: 
472: