Astronomy & Astrophysics manuscript no.

ReportTomDaan January 17, 2011

c ESO 2011

Comet 103P/Hartley 2: nucleus rotation and coma proles

of the cometary nucleus. Methods. 103P/Hartley 2 was observed for 6 consecutive nights in October 2010 (October 12-17) with the Mercator telescope at La Palma, Spain. CCD photometry in Cousins I and Geneva V, B and U was performed with the Merope CCD camera. 103P/Hartley 2 was close to perihelion passage when observed and near its closest approach to Earth (0.121 AU on October 21). Small aperture photometry is used to lter the nucleus signal from the ux of the light reected in the coma. Results. A rotation period of P = 16.6 0.7 is derived from our measurements in the IC band, which is in good agreement with other publications, however a period of P = 25.3h cannot be excluded. From a comparison between I band observations and theory deviations from simple behavior of the coma surface brightness prole are found. A brightness gradient m1 = -1.14 0.08 for the inner coma and m2 = -1.45 0.07 for the outer coma is found. The terminal outow velocity of the grains is determined to be vgr = 191 25 m/s. We also note a possible detection of jetslike structures in the IC band.
Key words. Comets: nucleus - Comets: rotation- Comets: coma - Comets: individual: 103P/Hartley 2

1. Introduction
In the eld of solar system astronomy the understanding of the origin and nature of comets is of particular interest. This because they are tracers of our past, formed in the protosolar nebula at 10-35AU. During the migration of the outer gas giants to their present location mean motion resonances occurred. This caused these objects to be catapulted outwards, forming what is known as the Kuiper belt. Some of these objects may eventually lose their stable position and start to make a comeback into the inner solar system, reborn as a comet. In the meanwhile their chemical composition has not changed since its formation. The knowledge of their physical properties is essential in this picture, especially the nucleus is interesting. Cometary nuclei are of limited size, typically 500m to 40km, which is why they are irregular of shape. They are chemically complex bodies consisting of ices and frozen grains. These objects are among the darkest in the solar system, having a very low albedo and thus reecting just a small fraction of in falling sunlight. But even though their nucleus is small, the ion tail and coma can extend over very large distances near perihelion passage (lion 1AU, lcoma 106 km) making them both the largest and the smallest celestial bodies in our solar system. The object of this paper, 103P/Hartley 2, is a small (Rn 0.6km) periodic comet with a period of 6.47 years. Because its period is shorter than 200 years, it is called a short period comet. It has been forced into the present orbit with a aphelion distance of 5.87 AU by perturbations from Jupiter and has become a Jupiter family comet. 103P/Hartley was observed with the Mercator telescope at La Palma, Spain, for 6 consecutive nights,
Send oprint requests to: [email protected] Based on observations made with the Mercator Telescope, operated on the island of La Palma by the Flemish Community, at the Spanish Observatorio del Roque de los Muchachos of the Instituto de Astrofica de Canarias.

Table 1. Basic properties of Comet 103P/Hartley 2

starting from the night of October 12 until the night of October 17. It was close to perihelion passage at the time, which took place on October 28. Its closest approach to Earth was on October 20 when = 0.121 AU (the distance Earth-comet is typically denoted as ). This close encounter provides an excellent chance to get a close look at the internal structure of the coma and the surroundings of the nucleus. This means that 103P/Hartley 2 was easily detectable when observed, but also highly active with a pronounced coma. Table 1 lists some general characteristic properties of 103P/Hartley 2. This work covers two parts of cometary physics, rst the inner coma of 103P/Hartley 2 is studied. Studying brightness proles in the coma is useful because these give a clue about the structure of the coma when compared with theoretical coma proles and the terminal velocity of grains ejected from the nucleus can be constrained in this way (Jewitt 1991). The second part of this work studied the very smallest part of the comet, being the nucleus with a largest diameter of slightly more than 1 km (Lisse et al. 2009) this entity covers less than 1 pixel on the CCD of Mercators Cassegrain camera, MEROPE. Even though its limited dimensions it is possible to study brightness variations caused by the rotation of the nucleus by aperture photometry

T. Hendrix and D. Camps: Comet 103P/Hartley 2: nucleus rotation and coma proles

with small apertures around the nucleus. The fact that it is possible to measure the rotational period of the nucleus is due to their irregular shape. During the rotation the projected surface of the nucleus on our line of sight changes continuously, changing the amount of light reected in our direction. It is easy to show that a limited aperture is necessary to increase the amplitude of the brightness variation (Licandro et al. 2000). The brightness measured in an aperture around the cometary optocenter is given by BT () = BN + BC () (1)

Table 2. Overview of the observations. and rh are the geocentric and heliocentric distances of the comet, respectively. Both are given in AU.
Date 10-12-2010 10-13-2010 10-14-2010 10-15-2010 10-16-2010 10-17-2010 Total n(U) 3 0 0 0 5 0 8 n(B) 1 1 2 0 0 1 6 n(V) 2 1 0 0 0 0 3 n(I) 5 2 3 3 13 3 29 0.132 0.129 0.127 0.125 0.123 0.122 rh 1.079 1.076 1.074 1.072 1.070 1.068

where BC () is the brightness contribution of the coma within the aperture of radius and BN is the brightness contribution of the nucleus. The latter is independent of the aperture radius as long as that radius is larger than the radius of the nucleus, which is always the case as mentioned earlier. In this case, the amplitude of the light curve is m() = 2.5 log BN2 + BC () . BN1 + BC () (2)

In this equation BN2 is the maximal brightness of the nucleus and BN1 the minimal brightness. Also note that it is assumed in this equation that the brightness contribution of the coma is constant over the entire time span of the observations. This is an assumption which also favors a small aperture around the nucleus because a smaller aperture minimizes the eect of possible temporal changes in activity which cause brightness variations in the coma. Another eect that has an impact on the measured magnitude in the aperature is the seeing eect (Licandro et al. 2000). The variations in the seeing during the night and over dierent nights causes the ux to be smeared out over a larger surface when the seeing is high and be more concentrated when the seeing is low. This eect is most prominent when small apertures are used, as required for a maximal amplitude eect, and thus a correction on the images is necessary. This correction implies a reduction of the seeing of all images to the seeing value of the image with the worst seeing. The complete process is explained later in this paper.

the result. Not every IC band image was useful because sometimes the pointing of the telescope went wrong or a very bright star crossed the comet in the eld of view, which nally gave us 24 usable IC images, 13 of which were made in the night of October 16. A log with an overview of all the observations is given in table 2. However, this obtained number of observations in the IC band is considerably smaller then our aim of 100 observations. Even though obtaining 100 observations in the limited timeframe that was available might have been quiet ambitious, a considerable amount of observing time was lost due to weather conditions.

3. Science Extraction
The images were reduced with a bias correction and the correction with a master ateld. For four out of six nights skyats were available, and for these nights the skyats were used. No correction for dark current needs to be made as the Merope CCD is cooled to 160K. Because the coma is much bigger than the region captured on the CCD (6.39 9.73), the background can not directly be read from the CCD values. However, a background correction was done by calculating the median value of a region in a corner of the eld (see gure 1). The shape of the region was chosen in a way that excludes the interference of smeared out background stars. The position of the region was chosen to minimize the coma contribution (i.e. in the direction opposite to that of the coma). To determine the periodicity of the nucleus rotation the magnitude of the nucleus needs to be calculated rst. The operations we performed to do so are discussed in the following paragraphs. In the last paragraph of this section the methods we used to extract a coma brightness prole from the images is described.
3.1. The Seeing Effect

2. Observations
The observations were done with the 1.2m Mercator telescope on La Palma, Spain, starting from the night of October 12 until the night of October 17. The complete scientic dataset available after the observing run consists of 46 images which accounts for approximately 3 hours of observing time. The Cousins I (IC) and Geneva V, B and U (VG, BG and UG) photometric lters available on the Merope camera (Davignon et al. 2004) were used. The Merope camera has a plate scale of 0.19 /pixel and the frame-transfer Eddington CCD (stensen 2010) has an image format of 2048 3074 pixels, resulting in a eld of view of 6.39 9.73. Our main interest was in IC band observations because these are most optimal to detect brightness variations of the nucleus and to determine the brightness prole in the cometary coma (Reyniers et al. 2009). Furthermore, the low extinction in the IC band lter enables us to do the observations without observing standard stars (see 3.4). The observing strategy was thus focused on observations in the IC band resulting in a dataset consisting of 29 IC band, 3 VG band, 6 BG band and 8 UG band images. The observations in the bands other than IC were done to have a interruption between IC observations during the available blocks of observation time, as contentious monitoring in a small amount of time would not have improved

As mentioned before, the seeing plays an important role in the interpretation of the observations, as bad seeing smears out the ux from the nucleus. Because we want to work with apertures smaller than the seeing disk to probe the variations in the nucleus, ignoring this eect might lead to a detection of the seeing in the periodicity if the seeing was variable during the observations. The necessity of this correction can also be seen from a simple argument: if we represent the signal by a Gaussian with = 0.8 and integrate with an aperture window [-0.4,0.4] we nd a ux F 0.39. If the signal is broadened to = 2, the ux drops to F 0.16. Because no standard stars were observed together with the comet, we deduced the value of the seeing from the eld stars in the comet observations. To nd the seeing we used a Markov

T. Hendrix and D. Camps: Comet 103P/Hartley 2: nucleus rotation and coma proles

Fig. 1. Example of an observation. Field stars are seen as stripes due to dierential tracking. The region in the lower left corner is used for the background correction. The bar in the lower right has a length of one arcminute. The central condensation seems to be saturated

cleus in not resolved: the comet has a diameter of 1.14 0.16 km (Lisse et al. 2009), while one pixel of the Merope-Eddington CCD (0.19) would have a projected size of 17 km at the distance of the closest approach during our observations. At rst we approximated the location of the nucleus with the brightest pixel. However, this approach gave some unexpected results (e.g. the central brightness sometimes increased after we convolved with a Gaussian kernel). We then adopted the weighted mean of the pixel positions near the brightest pixel as our central value. Even though the eect of this correction is small (the center is never more than a fraction of a pixel from the brightest pixels center), this approach seemed to work much better as it solved the apparent conicts encountered when using the brightest pixel. Most programs that extract uxes work by dividing pixels in smaller entities (subpixels), and then calculate the ux by summing the uxes in all the subpixels with centers within the radius of the aperture. To increase the accuracy, especially as we wanted to use very small aperture radii, we wrote a procedure that calculates the exact analytic fraction of the area of a pixel that falls within the aperture radius. The formulae used to calculate the intersection of a circle and a pixel can be found in (Degroote 2008).
3.3. Distance Correction

During the ve days of the observations, the comet moved relative to the Earth and the sun. This has two distance eects on our observations for which we need to correct, namely the heliocentric and the geocentric distance, rh and respectively. The magnitude of a cometary nucleus can be approximated in the following way: 2 n rh . m = m0 + 2.5 log (3) () Here m0 represents the comets magnitude at r = = 1 AU, = 0, with the sun-comet-Earth angle. () is the phase function that gives the fraction of light scattered at phase angle to that at = 0. It is often approximated by the empirical relation () = 10/2.5 , (4)

Fig. 2. Seeing values obtained by tting Gaussians perpendicular to the trails of eld stars. Chain Monte Carlo program that makes a Gaussian t of the prole of the eld stars. The method seems to be stable as stars at dierent positions in the frame and of dierent brightness give similar values for the seeing. The seeing we adopted is the mean of the obtained seeing in three dierent stars on the frame. The resulting seeing values can be found in gure 2. This gure shows us that the seeing is in the range of 0.81 to 2. All observations were then convolved with Gaussian kernels to compensate for the seeing eect.
3.2. Flux Extraction

is know as the linear phase coecient, and needs to be determined empirically. Most comets have values close to = 0.035 mag/deg (Lamy et al. 2004), so we adopted this value in our correction. It can be seen from equation 3 that the dependence on the geocentric distance follows the usual inverse square law: F 2 . The nucleus usually follows the same relation for the heliocentric distance, giving n = 2. If one would want to correct for the coma, a dierent value of n needs to be adopted, because the comet increases activity as it nears the sun, generating more dust to reect the light. For the coma, a typical value would be n = 4, however values for in the range of -1 to 11 have been reported (Brandt & Chapman 2004).
3.4. Conversion to Magnitudes

To convert the found ux values F to magnitudes the relation M = ZP + 2.5 log t 2.5 log F k A (5)

With the observations all brought to the same level, we can start with the aperture photometry of the nucleus. Obviously the nu-

is used, with A the airmass, ZP the zeropoint of the instrument and k the extinction coecient at wavelength and t the exposure time. This means we still need the values of two more

T. Hendrix and D. Camps: Comet 103P/Hartley 2: nucleus rotation and coma proles

parameters to calculate the magnitude, namely ZP and k . However, as can be seen from this equation, the zeropoint only provides an oset value. Since it does not change during the observations, its true value will not inuence our results, and the value ZPI = 22.89 is adopted. The actual value of k does matter somewhat more as it is more variable. Because we did not observe standard stars we do not have values of the extinction. Luckily, the extinction in the I band is rather low, which also makes the variations smaller. Furthermore, the Carlsberg Meridian Telescope (CMT) (Evans 2001; Garca-Gil et al. 2010) at the Roque de los Muchachos provides nightly measurements of the extinction in the r band (eective wavelength = 625nm). The CMT takes 40 observations of standard frames during the night, with 30-40 calibration stars on each frame. For four out of six of our nights (October 13-16) extinction values are available. During these nights the extinction values were the same (taking the errors into account), having a value of kr = 0.091. One of the other two nights was reported to be entirely unphotometric, and as the two observations in this night showed a clear shift these observations were no longer considered when tting the periodicity. The remaining night (October 12) was only unphotometric for a limited amount of time and for these observations we assume that the extinction had the same value as the one reported during the other nights. Using values from the La Palma technical note no. 31 (King, 1985) we converted the extinction to the value in the I band, nd a value of kI = 0.032.
3.5. Extraction of Brightness Proles

4. Results
4.1. Coma proles

For our study of the coma of 103P/Hartley 2 we used the method described in section 3.5. All I band images were used and also some of the B band images. This was done to investigate the dierences between dierent lters. The U band images were rejected from the sample because they did not reach a sucient signal to noise ratio to perform a qualitative analysis. Our study of the V band images was very concise because their number is very limited and the wavelength range of the V band is more or less situated in between the B and I band. Meaning that the dierence between B and I will be larger than between V and I. The V band images are not mentioned any more in this paper.
4.1.1. Theoretical prole

During the development of solar system astronomy, which goes back to ancient times but really ourishes in the past 100 years, comets have been of interest for research and dierent models have been constructed to describe the structure of the coma in terms of its surface brightness prole. This because the surface brightness prole provides a useful tool from which certain aspects of the 3D structure of the coma can be deduced. A simple, but relevant model for the surface brightness prole was derived by Jewitt (1991). This model is valid for a spherically symmetric coma in a steady state case. Steady state meaning that the outow rate of particles is constant, no changes in activity occur. It is easy to show that in this case the surface brightness prole follows the simple law B () = K , (6)

The extended eld of view of the Merope ccd camera allows us to study more than just the very inner regions around the nucleus. The more exterior regions of the coma also fall onto the ccd up to just beyond 25000 km away from the nucleus. This means that coma brightness proles can be determined from the data. The rst method we used to extract a brightness prole from an image used the position of the nucleus and a given direction and then scans all pixelvalues in a 1 pixel narrow row between those points. This method exhibits one major problem, when a background star crosses the pixel row a lump in the prole appears. It was tried to construct dierent proles in dierent directions and average these, but in this manner the results are not that good. Although more or less in agreement with the next method when the prole was averaged over 8 directions. The second method we used to extract a coma prole scans the whole image up to a certain pixel distance. This prole is then binned in logarithmic steps to get a nice smooth curve. We bin in logarithmic steps because the proles are expected to have a typical shape, which is most easy visualized when plotted logarithmic (see section 4.1.1). The process of binning has the advantage that it removes the lumps due to stars or cosmics, while keeping the typical decline of the coma prole. It has to be noted that in this manner we construct a coma prole averaged over all directions away from the nucleus. This isnt a problem, the directional dependence of the proles is limited. This was checked by dividing an image in 4 dierent regions of equal size (top right, top left, bottom left and bottom right) and analyzing the binned coma prole of each region. The dierences are fairly limited and the major problem here is that the number of datapoints in each region is to limited to completely remove all stellar lumps. The standard method for our analysis is the second one, which is explained in all details in section 4.1.2.

where K is a constant and is the projected distance from the nucleus on the sky plane of the nucleus. Note here also that the prole is expected to be wavelength dependent. To check the agreement between this model and observations it is most easy to introduce the parameter m, dened as m= d log B() . d (7)

This means that an agreement with the model is reached when m = 1, making this the most logic value for the steepness of the prole and it is expected to be valid for the inner regions of cometary comae under the standard assumptions, this region is called the inner coma. Deviations of this regime occur at large because the eect of the solar radiation pressure is neglected thus far. At large distances where this eect becomes important m 1.5 as shown by Jewitt & Meech (1987), this region is the outer coma. The distance where the transition from outer to inner coma occurs is of the order of XR
2 v2 rh gr

2g sun (rh )

(8)

as shown by Jewitt & Meech (1987). In this formula vgr the terminal velocity of the ejected grains, rh the heliocentric distance of the comet, is the ratio of the acceleration of a grain due to radiation pressure to the local solar gravity and g sun is the solar gravity at rh . This means we can estimate the outow velocity of the grains if we determine the transition point from the observational proles.

T. Hendrix and D. Camps: Comet 103P/Hartley 2: nucleus rotation and coma proles

Another useful relation is the empirical BobrovmikoDelsemme relation (Delsemme 1982)

(9)

which gives an estimate of the expected grain velocity. It has to be noted that deviations from the simple m = 1/ 1.5 behavior is possible, not all coma brightness proles behave exactly the same and dierences can occur because of changes in cometary activity, strong deviations from spherical symmetry or other effects.
4.1.2. Coma proles of 103P/Hartley 2

The coma brightness proles of all IC images were extracted by the second method described in section 3.5. This method experienced some minor problems at the borders of all images. This was because of three dierent reasons. First the code bins the values over concentric circles but because the ccd has a rectangular shape the circles far away from the nucleus do not fall completely on the ccd reducing the number of values to be binned. This eect will be rather small. Secondly the images show some vignetting in the corners, which causes the prole to fall o rapidly but this eect is not physical, it is due to the optics. The third and most important eect is that the background has to be subtracted to retrieve the true brightness prole. But the coma spans the whole image, which makes it hard to nd the true background value. As an approximation the median value over a small bar in the left bottom of the image is used (see also gure 1). In the outer part of the prole the intensity can be very close to or even smaller than this background value causing the brightness prole to fall down very rapidly in loglog scale. Because of all these reasons the very outer part of all proles was omitted and only the points up to 15000 km from the nucleus were used. The nal set of extracted coma surface brightness proles were tted to the expected brightness distributions in both the inner and outer coma. This was done with a Markov Chain Monte Carlo (MCMC) procedure which ts a power law brightness prole, B() = 10C rm (10)

to the data. When plotted in loglog scale m is the slope of the line and C is the point of intersection with the y axis.
4.1.3. Comparison between theory and observations

for the point of inection is XR = 3600 1000 km which gives a terminal outow velocity of the grains of vgr = 191 25 m/s. These values are rather low, in the literature the inection point lies typically at about 13104 km. And the grain velocity is less than half of what is expected from the Bobrovniko-Delsemme relation (Equation 9). Figure 3 also shows a prole in the B band. This has a less steep fall of m1 = 0.90, and thus deviating from the simple m = 1 trend of the model. This value is consistent for all B band images. The deviation from an m = 1 value in the B band can probably be contributed to strong emission lines in this wavelength range. There is typically more emission in the wavelength range of the B band, eg. CN emission is detected in other comets (Reyniers et al. 2009), which makes that both continuum and emission features may contribute to the brightness prole. The presence of emission breaks the assumption of conservation of mass of a certain species, making the used model invalid. This hypothesis to explain the disagreement between theory and observations was not tested for the specic case of 103P/Hartley 2 because a spectrum in the propper wavelength range was not available1 The signicant deviation from m = 1 in the I band can be due to dierent reasons. It could be possible that some strong emission features are present in the spectrum of 103P/Hartley 2 in the wavelength range of the I lter, making the model to break down because of the same reason as before. Another possibility is that strong changes in activity occur making the steady state assumption invalid. Another, last possibility is that the eld of view of the images at the distance of 103P/Hartley 2 is to limited, that there is a true other point of inection from inner to outer coma beyond the edges of our images. This is possible because the coma is much more extended than what we see on the images, where the eld of view is fairly limited because of the close distance to earth of 103P/Hartley 2, but it is very unlikely because we do nd a point of inection in all proles in a consistent way and the outer coma found in this manner satises the expected m -1.5 behavior. Note that the very inner part of the coma ( 250 km)deviates strongly from the rest of the prole. Here, other eects become important: the acceleration zone close to the nucleus has some inuence at this small scale, but also the seeing distorts a clear brightness prole. This region was never considered for the tting procedure.
4.2. Rotation Frequency

All IC band proles were analyzed and tted to theoretical proles at the manner described in previous section. A point of inection appears to be present in all the proles. This point of inection, XR , is then dened as the point of intersection between the t to the data in both regions, this point is calculated from the m and C value in the inner and outer coma. From formula 8 it is than possible to calculate the grain velocity of all proles. The resulting values for XR , |m1 |, |m2 | and vgr are presented in Figure 4. Here |m1 | is the absolute value of the slope of the inner coma and |m2 | of the outer coma. This results in a slope of m1 = 1.14 0.08 for the inner coma prole and a slope of m2 = 1.45 0.07. This means that we nd a signicant deviation from the m = -1 value in the inner coma, while the outer coma satises a m = 1.5 behavior. The slope that is found for the inner coma is steeper than a 1/ dependence and thus the brightness falls of more rapidly. An example of such a t is given in the bottom half of gure 3. This is the prole that best agrees with the average m1 , m2 and XR values. The value found

Using the technique explained in section 3 we were able to make a time series of the inner regions of the observations. In gure 5 the result is show for an aperture of 1 pixel. In gure 6 we see the same gure, but for aperture radii of 5 and 15 pixels. The pattern of the points for an aperture of one pixel is almost the same as the one for an aperture of 5 pixels. However, the time series with an aperture radius of 15 pixels shows a dierent pattern. This dierence can most easily be seen in the rst and the last night. This eect can be expected from equation 2. To determine the rotation frequency of the nucleus we used four distinct techniques: a Markov Chain Monte Carlo (MCMC) t, a Lomb-Scargle periodogram analysis (Lomb 1976),(Scargle 1982), a Fourier transform using Period04 (Lenz & Breger
1 During the observing run with the Mercator telescope one attempt was made to make an observation of 103P/Hartley 2 with the HERMES spectrograph in order to obtain a propper spectrum. But this failed because a correct pointing appeared to be impossible.

T. Hendrix and D. Camps: Comet 103P/Hartley 2: nucleus rotation and coma proles

Fig. 3. Coma prole in B and I band.

Fig. 5. Time series of the central region of the coma. An aperture of 1 pixel was used.

Fig. 6. Time series of the central region of the coma. The result is shown for aperture radii of 5 and 15 pixels.

Fig. 4. XR , | m1 | , | m2 |. and vgr of all I band coma proles. 2005) and the Jurkevich-Stellingwerf PDM (Phase Dispersion Minimalisation) method (Jurkevich 1971), (Stellingwerf 1978). The rst one can be considered as the most straight foreward: we wrote a MCMC algorithm that ts a sine to the time series, tting for the amplitude, frequency, phase and oset at the same time. The Lomb-Scargle periodogram analysis ts sine waves y = a cos t + b sin t using a least square tting procedure. The frequency analysis routine used by Period04 works in three steps: rst it uses a fourier transform to calculate the frequencies. The second step does a least squares tting using the same

parameters as our MCMC model to improve the frequencies found by the Fourier transform. The nal step which compares dierent periods is a calculation of uncertainties, using the error-matrix of the least-square tting. For a description of the Jurkevich-Stellingwerf PDM method see (Aerts et al. 2010). When looking for the frequency of the rotation, we have to know in what range of frequencies we have to look. The frequency can not be too high: comets are very fragile balls of dust and ice, and too high angular velocities will result in the breakup of comets. If the nucleus of a comet is taken to be spherical, with a density of = 1000 kg m3 a minimum period of 3.3h is found (Jewitt & Meech 1988). However, comets tend to have elongated shapes (especially smaller ones like 103P/Hartley),

T. Hendrix and D. Camps: Comet 103P/Hartley 2: nucleus rotation and coma proles

Fig. 8. Reliability distribution for frequencies between 1 and 3 c d1 as obtained by Period04 Fig. 7. The most reliable frequencies, as calculated by the MCMC tting procedure. The program performed 107 tting iterations to achieve the distribution in this gure.

and most cometary densities are believed to be around 600 kg m3 (Britt et al. 2006). Together this gives us a rough maximum frequency estimate of about 4.3 cycles per day (c d1 ), or a minimal period of 5.6h. This is supported by a list of know rotation periods (Samarasinha et al. 2004). For the frequency analysis we used aperture radii of one pixel. Both the MCMC method and Period04 give comparable for the frequency identication, as can be seen in gures 7 and 8. Both of these methods identify frequencies around 1.9 c d1 and 2.9 c d1 as the most reliable frequencies (P 13h and P 8.3h). The two other methods also nd these two peaks, however they favor them in the opposite order (see gure 9). Phase diagrams for both frequencies are shown in gures 10 and 11. All four methods seem to favor their rst and second candidate with approximately the same prefence ratio. However, the remark can be made that the MCMC algorithm is somewhat less reliable as it is not designed specially for frequency analysis, contrary to the three other algorithms, which might lead to a slight preference of 2.9 c d1 as the most reliable frequency. Also, on the phase diagrams the frequency of 2.9 c d1 seems to give a much better t with the individual datapoints, where the frequency of 1.9 c d1 seems more like a t that follows the average tendens of both good datapoints and outliers. Both of these arguments incline us to believe that 2.9 c d1 is the most reliable frequency, however the frequency of 1.9 c d1 cannot be ruled out. Normally we can assume that the frequencies we nd will only be half of the rotation frequency: if the brightness dierence of the central region is caused by the dierence in reection surface of an elongated nucleus, there will be two maxima in one rotation. The rst one when we see the frontside of the comet, the other one when we see the backside of the comet. This means the actual frequency of the nuclear rotation is half of the frequency found before, giving us the nal periods P1 = 16.6 0.7 h and P2 = 25.3 0.9 h.

Fig. 9. Reliability distribution for frequencies between 1 and 4 c d1 . The upper graph represents the result of the JurkevichStellingwerf PDM method. As one might suspect, lower values mean higher reliability for this method contrary to all other methods used. The lower graph is the Lomb-Scargle periodogram.

Fig. 10. Phase diagram for frequency f=1.91 c d1 . The lower part of the graph shows the residuals of the t.

T. Hendrix and D. Camps: Comet 103P/Hartley 2: nucleus rotation and coma proles

and we therefore come to the conclusion that this must indeed be the period of nuclear rotation.
4.4. Nucleus Axis Ratio

In equation 3 we saw how the magnitude of the nucleus shows a long term variation as the comet travels trough the solar system. As described above there is also a signal superponated on this long term variation due to the rotation of the nucleus. The rst term in the right hand side of equation 3 is the magnitude term related to the physical properties of the nucleus itself, as it quantizes the amount of solar light reected from the sun at a distance rh = = 1AU and = 0. We can therefore understand that the amount of light reected must be proportional to the reective surface of the comet and its albedo: Fig. 11. Phase diagram for frequency f=2.90 c d1 . The lower part of the graph shows the residuals of the t.
Frequencies of the nuclear rotation MCMC Period04 Lomb-Scargle Jurkevich-Stellingwerf

1.87 c d1 1.89 c d1 2.90 cd1 2.91 c d1 2.87 c d1 2.93 c d1 1.91 c d1 1.91 c d1

Fcomet F R2 pv N m0 = C 5 log RN log pv

(11) (12)

Table 3. List of the strongest frequencies, obtained by using four dierent frequency analysis algorithms. The most reliable frequencies are given for each method, ranked in order of importance from left to right.

With F the solar ux, RN the radius of the nucleus, pv the geometric albedo and C a constant. In section 4.2 we found the most convincing period to be P = 16.6 0.7. In gure 11 a wave with this period was comared with the observations. Fitting the phase plot allows us to predict the variations in magnitude during a cycle. Doing so gives us a value of a value of m = 0.46 0.06. If we use equation 12 together with this value, and assume the albedo of the comet is constant and has the shape of a prolate spheroid we nd e0.4m = 1.20 0.03 a/b (13)

4.3. Literature Values of the Rotation

This value is a lower boundary for the axis ratio because we only observe the projection of the nucleus.
4.5. Jetlike Structure

Because 103P/Hartley was the focus of the a study with the NASA Deep Impact space probe extended mission called EPOXI it allready gained much attention long before it became close to Earth. As a preparation for the EPOXI mission, Meech et al. (Meech et al. 2009) did a quite extensive campaign to determine the rotation period between March and July of 2009, when the comet was still at a distance of r = 5.7 AU. At this distance comets are normally inactive, as they are no longer suciently heated. This makes it easier to see the nuclear brightness variation, however at this distance the nucleus is very faint: nuclear magnitudes near m 25 were expected. Observations consisted of 10 hours on the LBT (only one telescope was operative at that time), 28 hours of time on the Gemini N and S telescopes, 2 nights on the VLT 8m, 7 hrs on the GTC 10.4m and 20 hrs on the SALT telescope in S. Africa, and in addition they received 12 orbits of HST time, spanning a clock time of 1 day. Both from the data of the Gemini telescopes and the HST data they found a rotation period of 16.6h. A dierent approach was used by Knight et al. (Knight et al. 2010): they used a CN lter (e f f 387 nm with a width of 6.2 nm) to locate jets emerging from the nucleus. These jets rotate together with the nucleus, allowing to determine a nuclear rotation period. A value of 16.6 0.5h was derived. The Arecibo radio telescope made 20 observations of 103P/Hartley on October 24-27, resulting in a rotation period of 18.1 0.3h, however they could not exclude a less likely period of 13.2h. These observations also showed the comet is highly elongated. Our proposed period of P = 16.6 0.7 is in very good agreement with the ndings of both Meech et al. and Knight et al.,

On the night of the 16th of October we were able to obtain the largest number of observations: 13 of our 29 observations in the IC lter were done on this night. We combined all our observations in the IC ler to create a median frame of the inner 38 38 around the nucleus. We then subtracted that median frame from all our observations to enhance small dierences. A composite image of 12 of these observations is shown in gure 12. The center of each frame corresponds to the pixel associated with the nucleus. Starting at 03h12 a structure seems to emanate from the nucleus, seemingly rotating anti-clockwise during the rest of the night. In several of the following frames two-lobed structures are observed, consisting of a small and a big lobe. These structures might be jets as they seem to have a narrow collimation and there is a certain morphological agreement with other observations of jets. However, jets are normally looked for in narrowband lters positioned on gas emission features. Normally the light in the IC lter comes from the reection of light on dust particles. Knight et al. observed 103P/Hartley when it was at r = 1.74 AU and detected jets in the CN lter, but no structures were detected in their R lter (Knight et al. 2010). Another possibility is that the structure is an enhanced amount of dust originating from a minor breakup or sudden eruption, which moves into the dusttail during the observations. However we cannot exclude the possibility that these structures are an artifact of the process used to obtain these images, however we see no clue for this. Because we cannot conrm the origin of these structures and we only discovered them during the last stages of

T. Hendrix and D. Camps: Comet 103P/Hartley 2: nucleus rotation and coma proles

this project, we did not perform a deeper investigation on these jetlike structures.

5. Conclusions
From our limited dataset obtained from our observing run we were able to investigate the structure of the coma, the rotation of the nucleus and possibly discovered some jetlike structures. The analysis of the coma surface brightness proles in the IC lter lead us to a m value of 1.14 0.07 for the inner coma, and m = 1.45 0.07 for the outer coma. Thus the structure of the inner coma clearly deviates from the expected m = 1 value that is obtained from theoretical considerations. Further observations, such as spectroscopy and photometric observations in wavelength ranges where only continuum emission is important, could help to clear out this discrepancy. The outer coma structure is, within the margin of error, in agreement with what is expected from the literature. The point of inection from inner to outer nucleus, XR , is found to be situated at 3600 1000 km, which is particularly close to the nucleus. This nally gives a terminal grain ejection velocity of 191 25 m/s, which is signicantly lower than expected from an empirical law. Possibly indicating that the activity of 103P/Hartley 2 is less than what is normal for a comet close to perihelion or otherwise indicating that the structure in the inner coma of 103P/Hartley 2 is dierent from a spherically symmetric, steady state model and the theory does not make sense anymore. Further observation could perhaps exclude one of the possibilities. Using small aperture photometry of the region around the nucleus and correcting for the seeing we were able to detect amplitude variations due to the rotation of the nucleus. Our frequency analysis resulted intwo possible rotation periods, of which P1 = 16.6 0.7 h was found to be the most reliable. This value is in very good agreement with other publications. No other observations seem to come close to our second period of P2 = 25.3 0.9 h. This period might have been caused by outliers due to our limited amount of observations. A larger amount of observations might have made it possible to rule out this period.
Acknowledgements. We would like to thank Maarten Reyniers for providing us the idea for this research project, and Pieter Degroote for his masters thesis which provided us a solid background on cometary physics. We are grateful to Professor Van Winckel for his support and usefull remarks during both researchproject II and III, and to Peter Papics for helping us out during the observations.

Knight, M., Schwieterman, E., & Schleicher, D. 2010, IAU Circ., 9163, 2 Lamy, P. L., Toth, I., Fernandez, Y. R., & Weaver, H. A. 2004, The sizes, shapes, albedos, and colors of cometary nuclei, ed. Festou, M. C., Keller, H. U., & Weaver, H. A., 223264 Lenz, P. & Breger, M. 2005, Communications in Asteroseismology, 146, 53 Licandro, J., Serra-Ricart, M., Oscoz, A., Casas, R., & Osip, D. 2000, AJ, 119, 3133 Lisse, C. M., Fernandez, Y. R., Reach, W. T., et al. 2009, PASP, 121, 968 Lomb, N. R. 1976, Ap&SS, 39, 447 Meech, K. J., Hainaut, O., Weaver, H. A., et al. 2009, in AAS/Division for Planetary Sciences Meeting Abstracts, Vol. 41, AAS/Division for Planetary Sciences Meeting Abstracts, 20.07+ stensen, R. H. 2010, ArXiv e-prints Reyniers, M., Degroote, P., Bodewits, D., Cuypers, J., & Waelkens, C. 2009, A&A, 494, 379 Samarasinha, N. H., Mueller, B. E. A., Belton, M. J. S., & Jorda, L. 2004, Rotation of cometary nuclei, ed. Festou, M. C., Keller, H. U., & Weaver, H. A., 281299 Scargle, J. D. 1982, ApJ, 263, 835 Stellingwerf, R. F. 1978, ApJ, 224, 953

References
Aerts, C., Christensen-Dalsgaard, J., & Kurtz, D. W. 2010, Asteroseismology, ed. Aerts, C., Christensen-Dalsgaard, J., & Kurtz, D. W., 348350 Brandt, J. C. & Chapman, R. D. 2004, Introduction to Comets, ed. Brandt, J. C. & Chapman, R. D., p35 Britt, D. T., Consolmagno, G. J., & Merline, W. J. 2006, in Lunar and Planetary Institute Science Conference Abstracts, Vol. 37, 37th Annual Lunar and Planetary Science Conference, ed. S. Mackwell and E. Stansbery, 2214+ Davignon, G., Blecha, A., Burki, G., et al. 2004, in Presented at the Society of Photo-Optical Instrumentation Engineers (SPIE) Conference, Vol. 5492, Society of Photo-Optical Instrumentation Engineers (SPIE) Conference Series, ed. A. F. M. Moorwood & M. Iye, 871879 Degroote, P. 2008, K.U.Leuven Master Thesis, Faculty of Science Delsemme, A. H. 1982, Icarus, 49, 438 Evans, D. W. 2001, Astronomische Nachrichten, 322, 347 Garca-Gil, A., Mu oz-Tu on, C., & Varela, A. M. 2010, PASP, 122, 1109 n n Jewitt, D. 1991, in Astrophysics and Space Science Library, Vol. 167, IAU Colloq. 116: Comets in the post-Halley era, ed. R. L. Newburn Jr., M. Neugebauer, & J. Rahe, 1965 Jewitt, D. C. & Meech, K. J. 1987, ApJ, 317, 992 Jewitt, D. C. & Meech, K. J. 1988, ApJ, 328, 974 Jurkevich, I. 1971, Ap&SS, 13, 154

10

T. Hendrix and D. Camps: Comet 103P/Hartley 2: nucleus rotation and coma proles

Fig. 12. Observations of the 16th of October in the IC lter after extracting a median of these observations. Observation time is given in the bottom right corner of every frame. After 03h12 a structure seems to emerge from the nucleus. Each frame has a eld of view of 38 38.