13
619 Physical and Dynamical Properties of Asteroid Families V. Zappalà, A. Cellino, A. Dell’Oro Istituto Nazionale di Astrofisica, Osservatorio Astronomico di Torino P. Paolicchi University of Pisa The availability of a number of statistically reliable asteroid families and the independent confirmation of their likely collisional origin from dedicated spectroscopic campaigns has been a major breakthrough, making it possible to develop detailed studies of the physical properties of these groupings. Having been produced in energetic collisional events, families are an invaluable source of information on the physics governing these phenomena. In particular, they provide information about the size distribution of the fragments, and on the overall properties of the original ejection velocity fields. Important results have been obtained during the last 10 years on these subjects, with important implications for the general understanding of the collisional history of the asteroid main belt, and the origin of near-Earth asteroids. Some important problems have been raised from these studies and are currently debated. In particular, it has been difficult so far to reconcile the inferred properties of family-forming events with current understanding of the physics of catastrophic collisional breakup. Moreover, the contribution of families to the overall asteroid inventory, mainly at small sizes, is currently controversial. Recent investigations are also aimed at understanding which kind of dynamical evolution might have affected family members since the time of their formation. In addition to potential consequences on the interpretation of current data, there is some speculative possibility of obtaining some estimate of the ages of these groupings. Physical characterization of families will likely represent a prerequisite for further advancement in understanding the properties and history of the asteroid population. 1. INTRODUCTION The improvements of the techniques of family identifi- cation, leading to the establishment of a database of about 20 statistically reliable families recognized during the last decade (see Bendjoya and Zappalà, 2002), and the subse- quent confirmation of the cosmochemical self-consistency of these groupings from dedicated spectroscopic campaigns (see Cellino et al., 2002) has introduced new exciting per- spectives for the physical studies of asteroids. Since the early times of the first family discoveries by Hirayama at the beginning of the past century, these group- ings have been interpreted as the observable outcomes of energetic collisional events, leading to complete disruption of a number of original parent bodies, and to dispersion of the collisional fragments. In this respect, asteroid families are nothing but nice examples of the collisional phenomena that have marked the history of the solar system, although generally at much smaller energy regimes with respect to major events involving large planetary bodies (origin of the Earth’s Moon, tilt of the spin axis of Uranus, etc.). In the framework of asteroid studies, families are cru- cially important in many respects. In general terms, they represent a constraint for any attempt at modeling the over- all collisional history of the asteroid belt. In other words, the models should be able to justify the number of currently observed families, and to estimate the typical times needed to progressively erode freshly formed families in such a way to make them no longer identifiable at later epochs. The most recent investigations of this particular aspect have been published by Marzari et al. (1999). Another essential aspect concerns the information that can be obtained about the physics of the breakup events from which families originated. Understanding how aster- oids react to mutual collisions typically characterized by relative velocities around 5–6 km/s (Farinella and Davis, 1992; Bottke et al., 1994) is an essential piece of information for any attempt at modeling their overall history, as well as their internal properties. Apart from the purely theoretical issues, understanding the internal structure of asteroids is a necessary prerequisite to developing any reasonable strategy of mitigation of the near-Earth asteroid (NEA) impact hazard. In the past, most of our understanding of the way aster- oids behave when they are collisionally disrupted came from laboratory experiments, based both on hypervelocity impacts and explosive charge techniques [for reviews of this subject, see Fujiwara et al. (1989) or Martelli et al. (1994)]. The major source of uncertainty, in this respect, was the scaling problem. In other words, it was not clear how to extrapolate the results of laboratory tests involving centi- meter-sized targets, to the astronomical situation of kilo- meter-sized bodies, including relevant gravitational effects

Physical and Dynamical Properties of Asteroid Familieselements, whereas the Gaussian formulas describe the ex-pected behavior in the osculating elements space. Accord-ing to Bendjoya

  • Upload
    others

  • View
    2

  • Download
    0

Embed Size (px)

Citation preview

  • Zappalà et al.: Properties of Asteroid Families 619

    619

    Physical and Dynamical Properties of Asteroid Families

    V. Zappalà, A. Cellino, A. Dell’OroIstituto Nazionale di Astrofisica,

    Osservatorio Astronomico di Torino

    P. PaolicchiUniversity of Pisa

    The availability of a number of statistically reliable asteroid families and the independentconfirmation of their likely collisional origin from dedicated spectroscopic campaigns has been amajor breakthrough, making it possible to develop detailed studies of the physical properties ofthese groupings. Having been produced in energetic collisional events, families are an invaluablesource of information on the physics governing these phenomena. In particular, they provideinformation about the size distribution of the fragments, and on the overall properties of theoriginal ejection velocity fields. Important results have been obtained during the last 10 years onthese subjects, with important implications for the general understanding of the collisional historyof the asteroid main belt, and the origin of near-Earth asteroids. Some important problems havebeen raised from these studies and are currently debated. In particular, it has been difficult so farto reconcile the inferred properties of family-forming events with current understanding of thephysics of catastrophic collisional breakup. Moreover, the contribution of families to the overallasteroid inventory, mainly at small sizes, is currently controversial. Recent investigations are alsoaimed at understanding which kind of dynamical evolution might have affected family memberssince the time of their formation. In addition to potential consequences on the interpretation ofcurrent data, there is some speculative possibility of obtaining some estimate of the ages ofthese groupings. Physical characterization of families will likely represent a prerequisite forfurther advancement in understanding the properties and history of the asteroid population.

    1. INTRODUCTION

    The improvements of the techniques of family identifi-cation, leading to the establishment of a database of about20 statistically reliable families recognized during the lastdecade (see Bendjoya and Zappalà, 2002), and the subse-quent confirmation of the cosmochemical self-consistencyof these groupings from dedicated spectroscopic campaigns(see Cellino et al., 2002) has introduced new exciting per-spectives for the physical studies of asteroids.

    Since the early times of the first family discoveries byHirayama at the beginning of the past century, these group-ings have been interpreted as the observable outcomes ofenergetic collisional events, leading to complete disruptionof a number of original parent bodies, and to dispersion ofthe collisional fragments. In this respect, asteroid familiesare nothing but nice examples of the collisional phenomenathat have marked the history of the solar system, althoughgenerally at much smaller energy regimes with respect tomajor events involving large planetary bodies (origin of theEarth’s Moon, tilt of the spin axis of Uranus, etc.).

    In the framework of asteroid studies, families are cru-cially important in many respects. In general terms, theyrepresent a constraint for any attempt at modeling the over-all collisional history of the asteroid belt. In other words,the models should be able to justify the number of currently

    observed families, and to estimate the typical times neededto progressively erode freshly formed families in such a wayto make them no longer identifiable at later epochs. Themost recent investigations of this particular aspect have beenpublished by Marzari et al. (1999).

    Another essential aspect concerns the information thatcan be obtained about the physics of the breakup eventsfrom which families originated. Understanding how aster-oids react to mutual collisions typically characterized byrelative velocities around 5–6 km/s (Farinella and Davis,1992; Bottke et al., 1994) is an essential piece of informationfor any attempt at modeling their overall history, as well astheir internal properties. Apart from the purely theoreticalissues, understanding the internal structure of asteroids is anecessary prerequisite to developing any reasonable strategyof mitigation of the near-Earth asteroid (NEA) impact hazard.

    In the past, most of our understanding of the way aster-oids behave when they are collisionally disrupted camefrom laboratory experiments, based both on hypervelocityimpacts and explosive charge techniques [for reviews of thissubject, see Fujiwara et al. (1989) or Martelli et al. (1994)].The major source of uncertainty, in this respect, was thescaling problem. In other words, it was not clear how toextrapolate the results of laboratory tests involving centi-meter-sized targets, to the astronomical situation of kilo-meter-sized bodies, including relevant gravitational effects

  • 620 Asteroids III

    (see Holsapple et al., 2002). In this context, the availabil-ity of a number of reliable families placed at the disposalof theoreticians a number of breakup experiments that na-ture performed for us, directly in a range of masses andenergies absolutely beyond current and future laboratory ca-pabilities.

    As a consequence, a great effort has been made in recentyears to derive from the observed properties of family mem-bers clues about the physics and the kinematics of the eventsfrom which families originated. These studies have beenable to obtain relevant results, whose interpretation is stillsomewhat debated, as we shall see in the following sections.

    We should still mention the problem that the presenceof random interlopers represents for physical studies offamilies. The likely numbers of objects sharing purely bychance the same orbital proper elements of “true” familymembers have been predicted on the basis of pure statis-tics by Migliorini et al. (1995) as a function of size. At thesame time, spectroscopic observing campaigns have beenable to identify in several cases a number of likely inter-lopers, as described in Cellino et al. (2002). Interlopers aremore “harmful” when they are relatively big, since they canaffect the overall behavior of the family size distributions,and the reconstruction of the original ejection velocity fieldsof the fragments (see below). However, we can now beconfident that nearly all the biggest interlopers have beenidentified in most cases. Thus, the results described in thefollowing section are not expected to be appreciably affectedby the presence of these undesired guests.

    2. AVAILABLE DATA AND COMMONINTERPRETATIONS

    Physical studies of dynamical families are based on anal-yses of essentially three pieces of information: (1) the co-ordinates of the family members in the space of the properelements a', e', and i' (proper semimajor axis, eccentricity,and inclination respectively); (2) the size distributions offamily members; and (3) spectroscopic and spectrophoto-metric properties of the members.

    In addition to the items listed above, we should mentionthat some photometric data are also available. The majorproblem for this kind of investigation is certainly the faint-ness of most family members, apart from the generally fewlargest members of the different groupings. A few light-curves are not sufficient for any relevant statistical inference.For this reason, no definitive indications have yet come fromphotometry. The only exception is a tentative clue of a likelyyoung age of the Koronis family, based on the observeddistribution of lightcurve periods and amplitudes (Binzel,1988). This investigation found also evidence that in the caseof the Eos family the distribution of spin periods seems tosuggest collisional relaxation and a likely old age.

    Another important source of information has come fromdirect in situ explorations by means of space probes. Inparticular, the Ida-Dactyl binary system observed by theGalileo probe is a nominal member of the Koronis family.

    Moreover, 951 Gaspra, also observed by Galileo, might bea member of the Flora family, although it has not been listedas such in the most recent family classification by Zappalàet al. (1995). It is known that for the Flora family the nomi-nal membership has been likely underestimated due to thesevere statistical criteria used for identification (Miglioriniet al., 1995). The important results of the Galileo observa-tions of these two objects, including analyses of the sizedistributions of impact craters on their surfaces (conse-quences of their collisional histories), are extensively dis-cussed in Chapman (2002) and Sullivan et al. (2002).

    The family spectroscopic data (item 3 in the above list)deserve a separate analysis, and are the subject of Cellinoet al. (2002). In what follows, we then focus our attention onthe properties mentioned in items 1 and 2 of the above list.

    2.1. Ejection Velocity Fields

    The structure of the families in the space of proper ele-ments are used to derive information on the ejection veloci-ties of the fragments in family-forming events. This is pos-sible because we can interpret the differences in orbitalelements in terms of differences in ejection velocity fromthe original parent body. The conversion from velocities toorbital elements or vice versa is given by the well-knownGaussian formulas, which can be written as follows, underthe assumption (verified for families) that the ejection ve-locities are much smaller than the orbital velocity of theparent body

    δ

    δ

    a/ana e

    e f V e f V

    ee

    na

    e f

    T R( )

    ( cos ) ( sin ) ]]

    ( ) cos

    =−

    + +

    = − +

    2

    11

    1 2

    2 1/2

    2 +++

    +

    = − ++

    e f

    e fV f V

    Ie

    na

    f

    T Rcos

    cos(sin )

    ( ) cos ( )

    2

    2 1/2

    1/2

    1

    1

    1δ ω

    ee fVWcos

    (1)

    where na is the mean orbital velocity, and VT, VR, and VWare the components of the ejection velocity vector along thedirection of the motion, radial, and normal to the orbitalplane respectively. The parameters f and ω are the trueanomaly and the argument of perihelion of the parent bodyat the epoch of its disruption. These angles are a prioriunknown, and this fact has long prevented any attempt atreconstructing the original ejection velocity fields from thepresently observed locations of the members in the properelements space.

    By means of extensive numerical simulations, however,Zappalà et al. (1996) showed that very reasonable recon-structions of the velocity fields in family-forming events arepossible when the real velocity fields were not completelyrandom, but were characterized by a wide variety of possiblefield structures, including spherical, ellipsoidal, conical(cratering), and more complicated combinations of theabove fields, in which some symmetry property is (even

  • Zappalà et al.: Properties of Asteroid Families 621

    weakly) satisfied. The basic idea was to use some adimen-sional parameters, given by some combinations of the distri-butions of the velocity components of the family members,to estimate the most likely values of the unknown f and ωangles. Of course, in practical cases the reconstruction canbe attempted only when a sufficiently large number of mem-bers is available, because of statistical reliability. In manysituations a reconstruction of the fields was found to be pos-sible, as in the cases of Vesta, Dora, Merxia (Zappalà et al.,1996), and Maria (Zappalà et al., 1997). The reconstructedfields turn out to be generally symmetric and similar, interms of general structure, to what is observed in laboratoryexperiments involving much smaller targets. These resultscould be obtained in spite of the difficulty related to the factthat families are identified in the space of the orbital properelements, whereas the Gaussian formulas describe the ex-pected behavior in the osculating elements space. Accord-ing to Bendjoya et al. (1993) the overall structure of theejection velocity field is conserved in the transformationfrom osculating to proper elements. The major effect of thetransformation was found to be a parallel displacement ofthe whole set of family members, leading to conservationof the overall velocity field structure. Of course, one canexpect that over long timescales the stability of the properelements can change, making a reconstruction of the veloc-ity fields harder to obtain. On the other hand, one can alsoexpect that over long timescales families tend in any caseto disappear due to progressive collisional erosion (Marzariet al., 1995, 1999).

    An indication that aging of the proper elements is likelypresent in several cases is given by the fact that the result-ing values of the computed f and ω angles turn out not tobe homogeneously distributed within their natural range ofvariation. This is related to the well-known fact that mostfamilies appear to be more elongated in proper eccentricityand inclination with respect to semimajor axis (Zappalà etal., 1984).

    Another problem is that the computed ejection velocitiesof family members resulting from their differences in or-bital proper elements turn out to be considerably large (seesection 3.3).

    2.2. Size Distributions and Size-VelocityRelationship

    The study of the fragment size distributions is anothermajor field of investigation. Asteroid sizes are generally notdirectly measured, but they are derived from knowledge ofabsolute magnitude and albedo. Families have been foundto be characterized by very homogeneous albedo distribu-tions according to the available radiometric data, and thisfact has also been used to derive an estimate of the limit ofcompleteness of the current asteroid inventory in terms ofapparent magnitude (Zappalà and Cellino, 1996).

    The size frequencies of all the major families have longbeen known to exhibit a power-law trend, characterized bysteep slopes, much larger than the typical slopes of samples

    of nonfamily asteroids in different regions of the main belt(Cellino et al., 1991; Klaçka, 1992).

    For a long time such behavior has been essentially mis-understood. General models based on an assumed power-law trend coupled with simple requirements of conservationof the total mass of the parent body led Petit and Farinella(1993) to make predictions that were able to fit satisfacto-rily only a minor fraction of the whole set of known familysize distributions. In most cases, the slopes of the observedpower laws were much steeper than the predictions. A pos-sible explanation was developed by Tanga et al. (1999),based on a very simple concept first proposed by Paolicchiet al. (1996). The idea was that the fragmentation processesmust satisfy other constraints in addition to conservation ofthe parent body’s mass. In particular, the fragments areformed in a finite volume, and cannot mutually overlap. Asimple model taking into account these geometric con-straints was developed by Tanga et al. (1999), and wasfound to give excellent fits of the observed size distribu-tions of a large number of families (see Fig. 1). As a by-product of this technique, some estimates of the originalparent body size and of the actual ratio mLR/mPB betweenthe mass of the largest remnant and that of the parent bodywere also obtained. More recently, the likely role of geo-metric constraints in fragmentation phenomena has beenalso recognized by Campo Bagatin and Petit (2001), al-though the authors did not find any real proof that themodeled geometric effects are in any way related to theactual physics of collisions.

    These above-mentioned results have important and de-bated implications concerning the overall inventory of theasteroid population, since they suggest that family membersmight dominate the asteroid population at small sizes, atleast if we can believe in the reliability of extrapolations ofthe observed size frequencies to small sizes (around 1 km).This will be discussed more thoroughly in the followingsection.

    Another recent result that must be mentioned concernsthe relationship found between size and ejection velocityof the fragments in family-forming events. This problem hasbeen analyzed by Cellino et al. (1999). The result was thata relation exists between the sizes of the fragments and theirmaximum possible ejection velocities. In particular, fami-lies generally fit the predictions of a model that assumesthat the maximum amount of kinetic energy achievable byfragments in family-forming events is a constant. As a con-sequence, the maximum possible ejection velocity as a func-tion of size follows a power-law trend, and is representedby a straight line in a log-log plot of size vs. velocity, asshown in Fig. 2. In particular, the following relation holdsunder the assumptions mentioned above, and is found to fitsatisfactorily the observational data for most families

    log log 'd

    DV K= − −2

    3

    where d and D are the diameters of the ith fragment and of

  • 622 Asteroids III

    the parent body respectively, and V is the maximum pos-sible value for the ejection velocity of the fragment of size d.The K' parameter is given by

    Af E/M)KElog (= −1

    32'K

    where E/M is the specific energy of the impact (defined asthe ratio between the kinetic energy of the projectile and themass of the impacted body), fKE is the so-called anelasticityparameter, defined as the fraction of E that is converted into

    kinetic energy of the fragments, and A is an a priori un-known parameter, describing the maximum fraction of thetotal kinetic energy of the set of fragments that can be de-livered to fragments of any size.

    Apart from the most technical issues, an important re-sult of the analysis has been to recognize that not only afixed slope around (–2/3) is found to actually fit the behav-ior of most families, but also the value of the intercept K'turns out to be a linear function of log mLR/M (mLR beingthe mass of the largest family member, and M the mass ofthe parent body). This prediction is respected by most fami-

    Gefion Koronis

    Maria Themis

    00

    1

    2

    3

    4

    –1 1

    Log(D) (km)

    2 3

    10.1 10 100 1000

    Log[

    N(

    D)]

    00

    1

    2

    3

    4

    –1 1

    Log(D) (km)

    2 3

    10.1 10 100 1000

    Log[

    N(

    D)]

    00

    1

    2

    3

    4

    –1 1

    Log(D) (km)

    2 3

    10.1 10 100 1000

    Log[

    N(

    D)]

    00

    1

    2

    3

    4

    –1 1

    Log(D) (km)

    2 3

    10.1 10 100 1000

    Log[

    N(

    D)]

    Fig. 1. The size distributions of four prominent families identified in the asteroid main belt are shown, with corresponding error barsdue to the uncertainties in the sizes of the single family members. The lines represent fits of the size distributions obtained by Cellinoet al. (1999) using their model. Three different curves (practically identical) are shown for each family, corresponding to differentchoices of some random seeds determining the results of different simulations. The simulations show a very good agreement with thedata. Also shown are the limits of completeness for each family, due to the fact that the available inventory is not complete at smallsizes due to the corresponding faintness of the smallest objects.

  • Zappalà et al.: Properties of Asteroid Families 623

    lies for which mLR/M < 0.8. This fact, if fully confirmed,has some important, although not yet fully explored, im-plications for our understanding of the physics of cata-strophic breakup processes. In particular, it would suggestthat the currently observed ejection velocity fields weredetermined mostly by the impact energies (directly relatedto the mLR/mPB ratio) and that the original size-velocitytrend has been scarcely affected by subsequent dynamicaland/or physical evolution, being still visible today. Such aconclusion may be still premature, however, and what canbe said at this stage is only that the Cellino et al. (1999)size-velocity model seems to be in good agreement withavailable data for a large number of known families, andalso with some laboratory experiments.

    3. IMPLICATIONS AND OPEN PROBLEMS

    The observational facts mentioned above are generallyself-consistent, and suggest some kind of coherent scenario.In particular, it turns out that families were formed by eventsthat produced ejection velocity fields having “reasonable”structures (when we compare them with the outcomes oflaboratory experiments), although the resulting velocitieswere generally high. The fragmentation of the parent bod-ies produced huge swarms of fragments, with size frequen-cies characterized by steep power-law trends, still recog-nizable today. However, several facts must be taken intoaccount before accepting literally all the above conclusions.Moreover, the exact implications of the above-mentioned

    Fig. 2. Size-ejection velocity plots for nine of the most prominent families in the asteroid main belt. The straight lines shown in eachplot have a fixed slope equal to –2/3, and correspond to a computed best-fit value, plus or minus its uncertainty. The value K' of theintercept corresponding to the best-fit line are also written in each plot. From Cellino et al. (1999).

    log(vej) (km/s)

    lg(d

    /DP

    B)

    –1.2

    –1

    –0.8

    –0.6

    –1.2 –1 –0.8 –0.6

    log(vej) (km/s)

    lg(d

    /DP

    B)

    –1.6

    –1.4

    –1

    –1.2

    –0.8

    –1 –0.5 0

    EosK' = 1.42

    Flora

    K' = 1.50

    log(vej) (km/s)

    lg(d

    /DP

    B)

    –1

    –1.2

    –1.4

    –0.8

    –0.4

    –0.6

    –0.2

    –1–1.5 –0.5 0

    MariaK' = 1.30

    log(vej) (km/s)

    lg(d

    /DP

    B)

    –1.4

    –1

    –1.2

    –0.8

    –0.6

    –0.4

    –1.5 –1 –0.5

    log(vej) (km/s)

    lg(d

    /DP

    B)

    –1.2

    –1

    –0.6

    –0.4

    –0.8

    –0.2

    –1.5 –1 –0.5

    ErigoneK' = 1.60

    GefionK' = 1.32

    log(vej) (km/s)

    lg(d

    /DP

    B)

    –1.2

    –1.4

    –1

    –0.8

    –0.8–1 –0.6 –0.4 –0.2

    ThemisK' = 1.45

    log(vej) (km/s)

    lg(d

    /DP

    B)

    –1.6

    –1.4

    –1.2

    –1

    –0.8 –0.6 –0.4 –0.2

    log(vej) (km/s)

    lg(d

    /DP

    B)

    –1.2

    –1

    –0.6

    –0.4

    –0.8

    –1.5 –1 –0.5

    EunomiaK' = 1.65

    KoronisK' = 1.32

    log(vej) (km/s)

    lg(d

    /DP

    B)

    –2

    –2.1

    –1.7

    –1.8

    –1.9

    –1.6

    –0.6 –0.4 –0.2

    VestaK' = 2.10

  • 624 Asteroids III

    scenario must be adequately stressed and discussed. Wetouch here on a number of very delicate points affectingour overall understanding of the properties and history ofthe asteroid population.

    In what follows, we give (in no special order) a list ofopen problems and general implications of the results com-ing from analyses of the family data at our disposal.

    3.1. Dynamical and Physical Aging of the Families

    It is certain that, over time, asteroid families undergo anevolution modifying their originally observable features.This evolution is expected to affect both the physical prop-erties of the members and their location in the proper ele-ments space.

    In particular, the orbital proper elements a', e', and i'cannot be assumed to be really constant over time (see alsoKnezevic et al., 2002). They are expected to vary slowlyover timescales on the order of 10–100 m.y. An estimationof the range of fluctuation of the proper elements was givenby Milani and Knezevic (1992, 1994). As a consequence,the inferred kinematical properties of families (includinga reconstruction of their original ejection velocity fields)should slowly vary as a function of the time elapsed sincefamily formation. This does not prevent us from identifyingtoday a number of around 20 “robust” families in the spaceof proper elements, and to derive information on the origi-nal collisional events from which they originated. However,it is reasonable to expect that the apparent kinematical prop-erties of families can be somewhat “aged” with respect tothe original configurations at the epoch of their formation.

    Not all the proper elements behave in the same way. Thesemimajor axis seems to be the most stable parameter ac-cording to available dynamical theories (see Knezevic et al.,2002). Dynamical aging of the proper elements affectsmainly eccentricities and inclinations. This may influencethe determination of the unknown f and ω angles in theGaussian equations when the techniques of reconstructionof the ejection velocity fields are applied. As shown by thesimulations performed by Zappalà et al. (1996), this effectcan produce a spurious clustering of the inferred values off around 90°. In turn, this can affect the derived ejectionvelocities, since an enhancement of the dispersion of theeccentricities and the inclinations produces a spurious in-crease of the radial and normal ejection velocity compo-nents, leaving the transversal component essentially un-changed (as can be easily seen from equation (1)).

    Another possible source of variation of the proper ele-ments is the so-called Yarkovsky effect (Vokrouhlický andFarinella, 1998; Farinella and Vokrouhlický, 1999; Vokrouh-lický, 1999; Bottke et al., 2000; Bottke et al., 2002). Un-like the dynamical aging processes mentioned above, theYarkovsky effect produces changes of the orbital semima-jor axis, and is size dependent. Several parameters, whosevalues are currently uncertain, determine the effectivenessof the Yarkovsky effect, and the range of sizes for whichit is really important. The effect depends on the thermal

    properties of the surface, on the orientation of the spin axis,and on the obliquity angle (the angle between the spin axisand the normal to the orbital plane). In its “diurnal” vari-ant, the Yarkovsky effect can cause the orbital semimajoraxis to drift either inward or outward, depending on thesense of rotation of the body. In addition, there is a “sea-sonal” variant, due to the periodic preferential heating ofdifferent hemispheres during the orbital motion. The neteffect of this variant is a systematic decrease of the orbitalsemimajor axis.

    An estimate of the expected variation of the semimajoraxis due to the Yarkovsky effect has been recently computedby Spitale and Greenberg (2001). The effectiveness of thiskind of nongravitational force is stronger for smaller ob-jects, and should be negligible for bodies larger than kilo-meter-sized (the exact value depending on many poorlyknown parameters). Moreover, another kind of relevantlong-term dynamical effect that is expected to affect theorbital semimajor axes of all asteroids, including familymembers, is that of mutual close encounters with the mas-sive Ceres, Pallas, and Vesta objects (Carruba et al., 2000).

    In summary, dynamical aging can increase the disper-sion of the eccentricities and inclinations of family mem-bers, while the Yarkovsky effect and close encounters canalso increase the dispersion of the semimajor axes. Dynami-cal aging does not depend on the size of the body, whereasthe Yarkovsky effect is certainly negligible for objects largerthan, e.g., 10–20 km.

    The interplay of dynamical aging and the Yarkovskyeffect can be important, because during a Yarkovsky-drivendrift in semimajor axis the objects can be trapped into someof the many dynamically unstable regions crossing the aster-oid main belt (Nesvorný and Morbidelli, 1998; Morbidelliand Nesvorný, 1999), with the effect of being eventuallyremoved from the family. Of course, the timescales of dy-namical and Yarkovsky-driven evolutions must be comparedwith typical collisional lifetimes for bodies of comparablesizes. We touch here on some key problems that will belikely the subject of extensive analyses in the near future.

    In fact, collisions independently produce a “physical”aging of the families. Collisional lifetimes of main-beltasteroids are size-dependent, and range between 107 and109 yr. The exact estimates are model-dependent, since theyare derived from estimates of both the impact strength ofthe objects and the number of existing projectiles capableof disrupting an object of a given size. For a 100-km aster-oid, the average collisional lifetime should be on the orderof 109 yr. As a consequence, and according to Marzari etal. (1995), the smallest family members tend to be colli-sionally disrupted over shorter timescales with respect tothe bigger members. As a consequence, collisions not onlymodify the size distribution of families, but also producean erosion in the space of the orbital proper elements, lead-ing to progressive disappearance of the families as recog-nizable groupings (Marzari et al., 1999).

    Moreover, the possibility of secondary events involvingfirst-generation family members must also be taken into

  • Zappalà et al.: Properties of Asteroid Families 625

    account, since this may potentially produce very compli-cated structures in the space of proper elements, as possi-bly in the cases of the so-called “clans,” such as the Florafamily (Farinella et al., 1992).

    All the facts mentioned above must be seriously takeninto account when interpreting the data at our disposal.Moreover, we observe today families that likely have dif-ferent ages, and show different stages of evolution. Under-standing the complicated interplay of the different evolu-tionary tracks will certainly help in understanding the prop-erties of families, and will possibly help in determining theircurrent ages, with important implications for the studies ofthe collisional evolution of the asteroid belt.

    3.2. Size Distributions and Asteroid Inventory

    In many cases the slopes of the power-law distributionsfitting the size frequencies of asteroid families turn out tobe very steep. According to Tanga et al. (1999), the result-ing exponents of the cumulative size distributions reachoften values beyond –3. In these cases, it is easy to showthat integrations of the size distributions would lead to in-finite reconstructed masses for the corresponding parentbodies. For this reason, it is clear that the size distributionsmust change slope somewhere at small sizes. A difficultproblem is thus determining the exact value of this transi-tion size, and whether this is generally constant or variesfor different families. Some general indications come fromdifferent kinds of evidence.

    First, some families in the inner region of the main beltare complete down to sizes on the order of 5–6 km. Thismeans that any change in slope of the size distributions mustoccur at smaller sizes for these particular families. Ofcourse, this does not imply that the same is necessarily truefor all the families in the main belt. Second, Campo Bagatinand Petit (2001) explored some generalizations of the Tangaet al. (1999) model, and found analytical reasons to pre-dict that the value of the power-law exponent should as-ymptotically converge toward a fixed value, found to bearound –2.8, for any model based on a similar treatmentof the geometric constraints affecting the production offragments. The authors stressed that the above result canbe a pure artifact of the assumed geometric model, and thereis not any proof that the model is really representative ofthe actual physics of the phenomena of catastrophic disrup-tion. If proven to be adequate to represent the real world,the above result would in any case prevent the possibilityof obtaining distributions diverging to infinity in terms ofreconstructed mass.

    Another very important consideration is that furthercollisional evolution experienced by the freshly formedfamily members should lead to changes of the size frequen-cies. According to Marzari et al. (1995), the family sizedistributions should be expected to converge progressively,starting from the low-size end of the distribution, towardthe equilibrium value expected for a collisionally relaxedpopulation, being equal to –2.5 according to classical re-

    sults by Dohnanyi (1969, 1971). The convergence shouldtake place over timescales depending on the size, the small-est members being expected to run into relaxation in shortertimes (Marzari et al., 1995). The problem in this scenariocan be that the Dohnanyi-predicted value of the slope wasmodel-dependent (the outcomes of the collisions were as-sumed to be independent of size), and it seems to be un-confirmed by general analyses of the overall size distribu-tion of main-belt asteroids (Cellino et al., 1991; Jedicke andMetcalfe, 1998). In particular, it turns out that nonfamilyasteroids, for instance, exhibit a size frequency much shal-lower than the predicted Dohnanyi slope.

    Apart from causing some uncertainty about the exactvalue of the slope to which families should be expected toconverge, the above fact should also be taken into accountwhen another controversial consequence of the steepnessof the family size distributions is discussed. The subject hereis that of the possible predominance of family members inthe general asteroid inventory at small sizes. According toZappalà and Cellino (1996), family members increasinglydominate the asteroid population going toward smallersizes. This conclusion is a consequence of both the steepslopes of the family size distributions and the apparentlyshallow slopes of the nonfamily population. The situationis controversial also because family members should beexpected to become dominant at values of size (~1 km) forwhich the current asteroid inventory is severely incomplete.Many authors tend to dislike the idea that 99% of the main-belt asteroids larger than 1 km should be family members(Zappalà and Cellino, 1996). On the other hand, avoidingthis type of conclusion is not trivial, if we can believe inextrapolations (even moderate in some cases) of the ob-served family size frequencies. Allowing for some impor-tant change of slope in the size frequencies of familiessomewhere between 1 and 5 km is reasonable. However,even in this case, families could hardly become so shallowas to not give a predominant contribution to the main-beltasteroid inventory at small sizes.

    Only observations will definitely clarify the real situa-tion. Models of the general asteroid inventory based on thedominance of families down to sizes of 1 km are currentlybeing developed. Some definite predictions on the numberand apparent magnitudes of the asteroids that should befound in any given sky field can be made, and observationswill soon confirm or rule out these models (see, e.g., Tedescoand Désert, 1999).

    3.3. Ejection Velocities and Comparisonwith Hydrocodes

    The results mentioned in sections 2.1 and 2.2 concern-ing the derivation of the original ejection velocity fields offamilies, and the discovery of a defined size-velocity rela-tionship, can certainly in principle be affected by subse-quent evolutionary effects experienced by family members(section 3.1). However, it seems difficult to reconcile theexpectation of very important evolutionary changes with the

  • 626 Asteroids III

    conservation of regular patterns such those that are observedfor the velocity field structures and the size-velocity rela-tions of the most robust families. Figure 3 shows a typicalexample concerning the Eos family. The existence of de-fined, “triangular” trends in the diameter vs. proper semi-major axis, eccentricity, and inclination plots can be easilyrecognized in spite of any level of noise possibly affectingthe proper elements.

    In particular, we should note that in principle a size-dependent spreading in semimajor axis can be interpreted interms of the Yarkovsky effect. What seems more difficult toexplain, however, is the size-dependent spreading in eccen-tricity and inclination. For this, effects like Yarkovsky canbe only indirectly effective, by moving objects into narrowresonant zones. The resulting distribution of the objects inthe proper elements space should be due to a complicated

    interplay of the residence times in the resonances experi-enced by bodies of different sizes, and the relative rate ofvariation of the proper elements under the influence of thesize-dependent Yarkovsky effect and the size-independentresonant behavior. It seems fairly difficult to expect that thefinal result of the evolution should be such a regular be-havior like the one shown in Fig. 3.

    Although conclusions are certainly premature at thepresent stage, it seems that if the current interpretation ofthe data at our disposal is not grossly wrong, we shouldconclude that either the evolutionary effects are weaker thancurrently believed, or the families themselves are relativelyyoung, and still poorly affected by the above effects. Ofcourse, this conclusion would have important consequencesfor our understanding of the collisional evolution of theasteroid belt (see Davis et al., 2002).

    2.9

    1

    1.5

    Log

    d (k

    m)

    2

    2.95 3 3.05

    a (AU)

    3.1 3.15

    0.14

    1

    1.5

    Log

    d (k

    m)

    2

    0.16 0.18

    sin i

    0.2 0.22

    1

    1.5

    Log

    d (k

    m)

    2

    0.05 0.1

    e

    Fig. 3. Proper semimajor axis, eccentricity, and sinus of inclination plotted as a function of the logarithm of diameter for the Eosfamily (1645 members identified). The data refer to a new, yet-unpublished analysis for the identification of asteroid families based ona dataset of more than 30,000 asteroid proper elements computed by Milani, and available at the AstDys Web site (http://hamilton.dm.unipi.it/cgi-bin/astdys/astibo). These plots show the general existence of a well-defined trend, possibly related to the size-ejectionvelocity relationship. The same trend is recognizable in a large number of currently identified families, in spite of the existence ofdynamical and physical mechanisms that are expected to progressively modify the original locations of the family members in thespace of the proper elements.

  • Zappalà et al.: Properties of Asteroid Families 627

    From the point of view of the physics of collisions, theejection velocities that are derived from application of theGaussian formulas constitute another debated subject. Gen-erally speaking, the problem is that according to the mostsophisticated hydrocode models of catastrophic breakupprocesses (Love and Ahrens, 1996; Nolan et al., 1996;Asphaug et al., 1998), the impact energies that are appar-ently needed to disperse the family members with the ob-served speeds should have been sufficiently high as tocompletely pulverize the family parent bodies.

    In particular, the reconstructed velocity fields of fami-lies indicate that ejection velocities well beyond 100 m/sare not rare for the smallest fragments. As a comparison,typical speed values in laboratory experiments are 5–10×smaller, velocities of 100 m/s being only typical of pulver-ized fragments produced very close to the impact point(Martelli et al., 1994). This is a well-known problem. Whentrying to reproduce the reconstructed ejection velocity fieldof the Maria family using the Paolicchi et al. (1996) semi-empirical model (SEM) of catastrophic disruption events,which is generally able to give a good fit to laboratoryexperiments, Zappalà et al. (1997) found that SEM givesmaximum velocities a factor of 5 too low to reproduce thefamily data.

    According to hydrocode results, the observed velocitiesare not possible unless we assume that the observed fam-ily members are essentially reaccumulated rubble piles. Inthis scenario, all asteroids larger than some hundreds ofmeters should be rubble piles. Pisani et al. (1999), how-ever, made some quantitative tests, and concluded thatreaccumulation cannot be sufficiently efficient to justify theobserved size distributions of several known families. Inparticular, important phenomena of reaccumulation werefound to be possible for the largest remnant alone, but cer-tainly not for a large number of objects. Moreover, prelimi-nary results of the in situ exploration of the near-Earthasteroid 433 Eros, a 31 × 13 × 13-km object, thought to bea collisional fragment from the disruption of a larger par-ent body, indicate that it is very unlikely that this asteroidcan be a rubble-pile [for the meaning of terms like “rubblepile,” see Cheng (2002) and Richardson et al. (2002)].

    Of course, a possible dynamically induced spreading offamilies in the proper elements space might artificially in-crease the apparent ejection velocities of family members.Taking this into account one could conclude that the realejection velocities at the epoch of formation were lower, asrequired by hydrocode results. However, there is not yet anyconclusive evidence of a dominating role played by post-formation dynamical evolution, as mentioned above. It istrue that when observations do not fit theory, scientists tendto believe that observations are likely wrong. But at thisstage we think that the self-consistency of much observa-tional data is very notable, and would tend to indicate thatsomething can be wrong in current theoretical models ofcollisional processes. Of course, this is another field in whichnew observational and theoretical activities are needed toachieve some reliable interpretation of what is observed.

    3.4. Formation of Binaries

    Strictly related to the modeling of family formation is theproblem of the possible formation of binaries. This subjectis currently very hot, after the recent discoveries of severalbinary systems among both near-Earth and main-belt aster-oids (see Merline et al., 2002). Interestingly enough, thefirst convincing discovery of an asteroid satellite came fromthe Galileo flyby of the Koronis family member 243 Ida.More recently, the asteroid 90 Antiope, a member of theThemis family, has also been found to be an outstandingexample of a binary system with components of comparablesize (Merline et al., 2000; Weidenschilling et al., 2001). Itis clear that the formation of binaries in collisional processesis an interesting field of research for which we can expectto have important results in the near future. This subject isdiscussed extensively in Paolicchi et al. (2002).

    3.5. Collision Probabilities

    The formation of big asteroid families is not only aninteresting phenomenon per se, but it also has a number ofimportant consequences. Dell’Oro et al. (2001) have foundthat a long-term increase of the collision probabilities in theregion of the main belt swept by the orbital motion of thenewly formed family members must be expected. Again,this is due to the huge numbers of kilometer-sized mem-bers that are apparently produced by these events. When alarge family is produced, its members will cross the orbitsof the majority of the main-belt population, apart fromobjects located far enough in terms of semimajor axis andeccentricity (mostly the low-eccentricity asteroids orbitingclose to the inner and outer edges of the belt). In this re-spect, it is not unlikely that major collisional events likethose that produced the big Eos and Themis families cantrigger the subsequent collisional regime over substantialtimescales. This is just another example of family-relatedmechanisms that should be taken into account by modernmodels of the overall collisional evolution of the asteroidbelt (see Davis et al., 2002).

    Another fact that should be mentioned is that accordingto recent investigations (Dell’Oro et al., 2002), the veryearly times after formation of any given family are charac-terized by a larger impact probability among members ofthe same family, with respect to the typical values control-ling the steady collisional regime in the main belt. Althoughthese epochs of enhanced interfamily collisions were foundto be short, and were on average characterized by low impactvelocities, there is still the possibility that the freshly formedsurfaces of family members can have been marked in asignificant way, with consequences on the cratering recordobserved by in situ exploration by space probes (243 Idaand 951 Gaspra were observed by Galileo). We should alsonote that, according to Weidenschilling et al. (2001), theexistence of binary asteroids like 90 Antiope (a member ofthe Themis family), characterized by components of com-parable size, might also be explained by the occurrence of

  • 628 Asteroids III

    relatively mild interfamily collisions likely occurring dur-ing the very early history of a newly born family.

    4. FAMILIES AS POSSIBLE SOURCES OFNEAR-EARTH ASTEROIDS

    Families have been found to also play an important rolein the production of near-Earth asteroids and meteorites.The first indications were found by Morbidelli et al. (1995),who convincingly showed that a number of well-knownfamilies (such as Themis) are sharply cut by the edges ofimportant mean-motion resonances with Jupiter. This quali-tatively showed two important things: First, the formationof families in several cases led to the immediate injectionof substantial numbers of sizeable fragments into resonantorbits. Many of them (mostly in the case of resonanceslocated in the inner belt, like the 3:1 mean-motion reso-nance or the ν6 secular resonance; a smaller fraction in thecase of resonances located in the outer belt) achieved near-Earth-like orbits. This process was quantitatively analyzed bymeans of numerical integrations of the orbits by Gladmanet al. (1997). The results showed that the transfer times areextremely short for resonances like 3:1 and 5:2, on the orderof a couple of million years. Based on these results and ona preliminary assessment of the numbers of objects actu-ally injected by several families at the epochs of their for-mation, Zappalà et al. (1998) found that in the past severalepisodes of paroxysmal near-Earth-asteroid production,with consequent enhancements in the collision rate with theterrestrial planets, took place, with durations on the orderof several million years in most cases. This also indicatesthat the inventory of near-Earth objects is not steady, andcan be strongly influenced by the occurrence of importantcollisional events in the main belt.

    Second, family-forming events left groupings of objectsthat are very close to the borders of several important reso-nances. This fact is important, because it can be expectedthat the normal collisional evolution experienced by theseobjects can likely produce a steady rate of production ofsmall fragments directly injected into the most efficientdynamical routes to the inner solar system (see also Zappalàand Cellino, 2001). A good example is given by the Mariafamily, whose reconstructed ejection velocity field, sharplycut by the 3:1 resonance with Jupiter, is shown in Fig. 4(Zappalà et al., 1997).

    An important result obtained by Morbidelli et al. (1995)was the discovery that several objects are currently locatedinto the 9:4 mean-motion resonance with Jupiter. This rela-tively thin resonance cuts across the big Eos family, andwas shown by Gladman et al. (1997) to give a modest butnot fully negligible contribution to the near-Earth popula-tion. The reason is that the dynamical evolution inside the9:4 resonance is relatively slow, with eccentricity beingpumped up over timescales sufficiently long to allow Marsto capture some objects before they become dominated byJupiter (with subsequent ejection from the solar system).Most of the 9:4 resonant objects cannot be recognized as

    Eos family members on the basis of their orbital elements,but it was straightforward to assume that these objects mightbe original members of this family, observed during theearly stages of a dynamical evolution that will decouplethem from the asteroid main belt. This conjecture was fullyconfirmed by subsequent observations by Zappalà et al.(2000) (see also Cellino et al., 2002).

    Another fact that should be mentioned when speakingabout the relation between near-Earth objects and familiesis the following: If direct collisional injection into some ofthe dynamically unstable regions in the main belt (such asthe 3:1 resonance with Jupiter, the ν6 secular resonance, andthe region of Mars-crossing) is assumed to be the mosteffective mechanism for supplying kilometer-sized NEAs,some constraints must be respected. In particular, directinjection cannot be accepted if it can be shown that it nec-essarily predicts the formation of too many observable fami-lies. The reason is simply that the number of known familiesin the main belt is limited. This is a strong constraint forany mechanism of steady NEA supply, mainly for sizeableobjects having diameters beyond 1 or 2 km. According toZappalà et al. (2002), it does not seem trivial to explain thecurrently observed number of NEAs larger than 2 km only bymeans of steady collisional injection into chaotic regions inthe main belt. Many candidate parent bodies exist (the mosteffective ones have been listed in Table 1), but the rangesof “permitted” impact energies that satisfy the nonfamilyformation constraint for most of them are fairly narrow.

    Fig. 4. Reconstruction of the ejection velocity field of the frag-ments produced in the disruption of the parent body of the Mariafamily. The location of the edges of the 3:1 mean-motion reso-nance with Jupiter are also indicated. The family turns out to besharply cut by the border of the resonance. From Zappalà et al.(1997).

    –0.5

    –0.5

    0

    0.5

    0.50

    Velocity (km/s)

    30 km

    3:1

    20 km

    10 km

    Vel

    ocity

    (km

    /s)

  • Zappalà et al.: Properties of Asteroid Families 629

    TABLE 1. A list of the most effective NEA candidate parent bodies found by Zappalà et al. (2001).

    Proper Elements Transfer Diameter Taxonomic ClassID Number a' (AU) e' sin i' Region (km) Tholen (1994) Bus (1999)

    304 2.404 0.122 0.257 ν6 68 C XC313 2.376 0.236 0.201 MC 96 C —495 2.487 0.118 0.045 3:1 39 — —512 2.190 0.199 0.142 MC 23 S S753 2.329 0.214 0.153 MC 24 S L877 2.487 0.142 0.059 3:1 38 F —930 2.431 0.121 0.265 ν6 36 — Ch1080 2.420 0.243 0.086 MC 23 F —1715 2.400 0.249 0.181 MC 23 — X2143 2.281 0.194 0.130 MC 19 — —

    5. FUTURE DEVELOPMENTS

    From the facts discussed in the previous sections, it iseasy to see that the physical properties of asteroid familiesplay an important role in our current understanding of thecollisional and dynamical evolution of the asteroid popu-lation. The interpretation of the data at our disposal is notstraightforward, and a coordinated observational and theo-retical effort will be needed in order to confirm or rule outa number of conclusions that are suggested by the data thatare currently available. A determination of the times neededto collisionally erode and dynamically disperse a newlyformed family will certainly be a primary field of investiga-tion in the near future. At the same time, trying to under-stand how families can achieve their steep size distributions,and how long they can keep them unchanged in the rangeof sizes covered by current datasets, will be another majorfield of research. If families evolve rapidly, and those thatare currently observed can be shown to be really young,this will deeply influence current ideas of the collisionalevolution of the main belt, and will provide new importantconstraints in addition to those currently recognized, likethe preservation of the basaltic crust of 4 Vesta; the exist-ence of metallic meteorites; the scarcity of taxonomic typesdirectly interpreted in terms of mantle material from dif-ferentiated parent bodies; the observed cratering records ofIda, Gaspra, Mathilde, and Eros, etc. In general, it will bepossible to obtain a better and more reliable estimate of theasteroid inventory at small sizes. Two hundred years afterthe discovery of the first asteroids, we are still far from fullyunderstanding these interesting objects. As the direct out-comes of mutual collisions, families certainly represent avery important piece of the asteroid puzzle.

    REFERENCES

    Asphaug E., Ostro S. J., Hudson R. S., Scheeres D. J., and BenzW. (1998) Disruption of kilometer-sized asteroids by energeticcollisions. Nature, 393, 437–440.

    Bendjoya Ph. and Zappalà V. (2002) Asteroid family identifica-

    tion. In Asteroids III (W. F. Bottke Jr. et al., eds.), this volume.Univ. of Arizona, Tucson.

    Bendjoya Ph., Cellino A., Froeschlé C., and Zappalà V. (1993)Asteroid dynamical family: A reliability test for new identifi-cation methods. Astron. Astrophys., 272, 651–670.

    Binzel R. P. (1988) Collisional evolution in the Eos and Koronisasteroid families — Observational and numerical results.Icarus, 73, 303–313.

    Bottke W. F., Nolan M. C., Greenberg R., and Kolvoord R. A.(1994) Velocity distributions among colliding asteroids. Icarus,107, 255–268.

    Bottke W. F., Rubincam D. P., and Burns J. A. (2000) Dynamicalevolution of main belt asteroids: Numerical simulations incor-porating planetary perturbations and Yarkovsky thermal forces.Icarus, 145, 301–330.

    Bottke W. F. Jr., Vokrouhlický D., Rubincan D. P., and Broz M.(2002) The effect of Yarkovsky thermal forces on the dynami-cal evolution of asteroids and meteoroids. In Asteroids III(W. F. Bottke Jr. et al., eds.), this volume. Univ. of Arizona,Tucson.

    Bus S. J. (1999) Compositional structure in the asteroid belt:Results of a spectroscopic survey. Ph.D. thesis, MassachusettsInstitute of Technology, Cambridge. 367 pp.

    Campo Bagatin A. and Petit J. M. (2001) Effects of the geomet-ric constraints on the size distributions of debris in asteroidalfragmentation. Icarus, 149, 210–221.

    Carruba V., Burns J. A., Bottke W. F., and Morbidelli A. (2000)Asteroid mobility due to encounters with Ceres, Vesta, Pallas:Monte Carlo codes versus direct numerical integrations (ab-stract). Bull. Am. Astron. Soc., 32, 1019.

    Cellino A., Zappalà V., and Farinella P. (1991) The size distribu-tion of asteroids from IRAS data. Mon. Not. R. Astron. Soc.,253, 561–574.

    Cellino A., Michel P., Tanga P., Zappalà V., Paolicchi P., andDell’Oro A. (1999) The velocity-size relationship for membersof asteroid families and implications for the physics of cata-strophic collisions. Icarus, 141, 79–95.

    Cellino A., Bus S. J., Doressoundiram A., and Lazzaro D. (2002)Spectroscopic properties of asteroid families. In Asteroids III(W. F. Bottke Jr. et al., eds.), this volume. Univ. of Arizona,Tucson.

    Chapman C. R. (2002) Cratering on asteroids from Galileo andNEAR Shoemaker. In Asteroids III (W. F. Bottke Jr. et al.,eds.), this volume. Univ. of Arizona, Tucson.

  • 630 Asteroids III

    Cheng A. F. (2002) Near Earth asteroid rendezvous: Mission sum-mary. In Asteroids III (W. F. Bottke Jr. et al., eds.), this volume.Univ. of Arizona, Tucson.

    Davis D. R., Durda D. D., Marzari F., Campo Bagatin A., andGil-Hutton R. (2002) Collisional evolution of small-bodypopulations. In Asteroids III (W. F. Bottke Jr. et al., eds.), thisvolume. Univ. of Arizona, Tucson.

    Dell’Oro A., Paolicchi P., Cellino A., Zappalà V., Tanga P., andMichel P. (2001) The role of families in determining collisionprobability in the asteroid main belt. Icarus, 153, 52–60.

    Dell’Oro A., Paolicchi P., Cellino A., Zappalà V., Tanga P., andMichel P. (2002) Primordial collisional evolution of the aster-oid families. Icarus, in press.

    Dohnanyi J. S. (1969) Collisional model of asteroids and theirdebris. J. Geophys. Res., 74, 2531–2554.

    Dohnanyi J. S. (1971) Fragmentation and distribution of asteroids.In Physical Studies of Minor Planets (T. Gehrels, ed.), pp. 263–295. NASA SP-267.

    Farinella P. and Davis D. R. (1992) Collision rates and impactvelocities in the main asteroid belt. Icarus, 97, 111–123.

    Farinella P., Davis D. R., Cellino A., and Zappalà V. (1992) Fromasteroid clusters to families: A proposal for a new nomencla-ture. In Asteroids, Comets, Meteors 1991 (A. W. Harris andE. Bowell, eds.), pp. 165–166. Lunar and Planetary Institute,Houston.

    Farinella P. and Vokrouhlický D. (1999) Semimajor axis mobil-ity of asteroidal fragments. Science, 283, 1507–1511.

    Fujiwara A., Cerroni P., Davis D. R., Ryan E. V., Di Martino M.,Holsapple K., and Housen K. (1989) Experiments and scal-ing laws on catastrophic collisions. In Asteroids II (R. P. Binzelet al., eds.), pp. 240–268. Univ. of Arizona, Tucson.

    Gladman B. J., Migliorini F., Zappalà V., Michel P., Cellino A.,Froeschlé Ch., Levison H., Bailey M., and Duncan M. (1997)Dynamical lifetimes of objects injected into asteroid belt reso-nances. Science, 277, 197–201.

    Holsapple K., Giblin I., Housen K., Nakamura A., and Ryan E.(2002) Asteroid impacts: Laboratory experiments and scalinglaws. In Asteroids III (W. F. Bottke Jr. et al., eds.), this vol-ume. Univ. of Arizona, Tucson.

    Jedicke R. and Metcalfe T. S. (1998) The orbital and absolutemagnitude distributions of main belt asteroids. Icarus, 131,245–260.

    Klaçka J. (1992) Mass distribution in the asteroid belt. Earth MoonPlanets, 56, 47–52.

    Knezevic Z., Lemaître A., and Milani A. (2002) The determina-tion of asteroid proper elements. In Asteroids III (W. F. BottkeJr. et al., eds.), this volume. Univ. of Arizona, Tucson.

    Love S. G. and Ahrens T. J. (1996) Catastrophic impacts on grav-ity dominated asteroids. Icarus, 124, 141–155.

    Martelli G., Ryan E. V., Nakamura A. M., and Giblin I. (1994)Catastrophic disruption experiments: Recent results. Planet.Space Sci., 42, 1013–1026.

    Marzari F., Davis D. R., and Vanzani V. (1995) Collisional evolu-tion of asteroid families. Icarus, 113, 168–187.

    Marzari F., Farinella P., and Davis D. R. (1999) Origin, aging anddeath of asteroid families. Icarus 142, 63–77.

    Merline W. J., Close L. M., Dumas C., Shelton J. C., Menard F.,Chapman C. R., and Slater D. C. (2000) Discovery of com-panions to asteroids 762 Pulcova and 90 Antiope by directimaging (abstract). Bull. Am. Astron. Soc., 32, 1017.

    Merline W. J. et al. (2002) Asteroid satellites. In Asteroids III(W. F. Bottke Jr. et al., eds.), this volume. Univ. of Arizona,Tucson.

    Migliorini F., Zappalà V., Vio R., and Cellino A. (1995) Interlop-ers within asteroid families. Icarus, 118, 271–291.

    Milani A. and Knezevic Z. (1992) Asteroid proper elements andsecular resonances. Icarus, 98, 211–232.

    Milani A. and Knezevic Z. (1994) Asteroid proper elements andthe dynamical structure of the asteroid main belt. Icarus, 107,219–254.

    Morbidelli A., Zappalà V., Moons M., Cellino A., and Gonczi R.(1995) Asteroid families close to mean motion resonances: dy-namical effects and physical implication. Icarus, 118, 132–154.

    Morbidelli, A. and Nesvorný D. (1999) Numerous weak reso-nances drive asteroids toward terrestrial planets orbits. Icarus,139, 295–308.

    Nesvorný D. and Morbidelli A. (1998) Three-body mean motionresonances and the chaotic structure of the asteroid belt. Astron.J., 116, 3029–3037.

    Nolan M. C., Asphaug E., Melosh H. J., and Greenberg R. (1996)Impact craters on asteroids: Does gravity or strength controltheir size? Icarus, 124, 359–371.

    Paolicchi P., Verlicchi A., and Cellino A. (1996) An improvedsemi-empirical model of catastrophic impact processes. Icarus,121, 126–157.

    Paolicchi P., Burns J. A., and Weidenschilling S. J. (2002) Sideeffects of collisions: Spin rate changes, tumbling rotation states,and binary asteroids. In Asteroids III (W. F. Bottke Jr. et al.,eds.), this volume. Univ. of Arizona, Tucson.

    Petit J. M. and Farinella P. (1993) Modelling the outcomes of high-velocity impacts between solar system bodies. Cel. Mech. Dyn.Astron., 57, 1–28.

    Pisani E., Dell’Oro A., and Paolicchi P. (1999) Puzzling asteroidfamilies. Icarus, 142, 78–88.

    Richardson D. C., Leinhardt Z. M., Melosh H. J., Bottke W. F. Jr.,and Asphaug E. (2002) Gravitational aggregates: Evidence andevolution. In Asteroids III (W. F. Bottke Jr. et al., eds.), thisvolume. Univ. of Arizona, Tucson.

    Spitale J. N. and Greenberg R. (2001) Numerical evaluation of thegeneral Yarkovsky effect: Effects on semimajor axis. Icarus,149, 222–234.

    Sullivan R., Thomas P., Murchie S., and Robinson M. (2002)Asteroid geology from Galileo and NEAR Shoemaker data. InAsteroids III (W. F. Bottke Jr. et al., eds.), this volume. Univ.of Arizona, Tucson.

    Tanga P., Cellino A., Michel P., Zappalà V., Paolicchi P., andDell’Oro A. (1999) On the size distribution of asteroid families:The role of geometry. Icarus, 141, 65–78.

    Tedesco E. F. and Désert F. X. (1999) The ISO faint asteroidsurvey (abstract). Bull. Am. Astron. Soc., 31, 1091.

    Tholen D. J. (1984) Asteroid taxonomy from cluster analysis ofphotometry. Ph.D. thesis, Univ. of Arizona, Tucson. 150 pp.

    Vokrouhlický D. (1999) A complete linear model for the Yar-kovsky thermal force on spherical asteroid fragments. Astron.Astrophys., 344, 362–366.

    Vokrouhlický D. and Farinella P. (1998) The Yarkovsky seasonaleffect on asteroidal fragments: A nonlinearized theory for theplane-parallel case. Astron. J., 116, 2032–2041.

    Weidenchilling S. J., Marzari F., Davis D. R., and Neese C. (2001)Origin of the double asteroid 90 Antiope: A continuing puzzle

  • Zappalà et al.: Properties of Asteroid Families 631

    (abstract). In Lunar and Planetary Science Conference XXXII,Abstract #1890. Lunar and Planetary Institute, Houston (CD-ROM).

    Zappalà V. and Cellino A. (1996) Main belt asteroids: Present andfuture inventory. In Completing the Inventory of the Solar Sys-tem (T. W. Rettig and J. M. Hahn, eds.), pp. 29–44. ASP Con-ference Series 107.

    Zappalà V. and Cellino A. (2001). Formation of asteroid familiesand delivery of NEO showers from the main belt. In CollisionalProcesses in the Solar System (M. Marov and H. Rickman,eds.), pp. 303–321. Kluwer, Dordrecht.

    Zappalà V., Farinella P., Knezevic Z., and Paolicchi P. (1984)Collisional origin of the asteroid families: Mass and velocitydistributions. Icarus, 59, 261–285.

    Zappalà V., Bendjoya Ph., Cellino A., Farinella P., and FroeschléC. (1995) Asteroid families: Search of a 12487 asteroid sampleusing two different clustering techniques. Icarus, 116, 291–314.

    Zappalà V., Cellino A., Dell’Oro D., Migliorini F., and Paolicchi P.(1996) Reconstructing the original ejection velocity fields ofasteroid families. Icarus, 124, 156–180.

    Zappalà V., Cellino A., Di Martino M., Migliorini F., and PaolicchiP. (1997) Maria’s family: Physical structure and possible im-plications for the origin of giant NEAs. Icarus, 129, 1–20.

    Zappalà V., Cellino A., Gladman B. J., Manley S., and MiglioriniF. (1998) Asteroid showers on Earth after family break-upevents. Icarus, 134, 176–179.

    Zappalà V., Bendjoya P., Cellino A., Di Martino M., Dores-soundiram A., Manara A., and Migliorini F. (2000) Fugitivesfrom the Eos family: First spectroscopic confirmation. Icarus,145, 4–11.

    Zappalà V., Cellino A., and Dell’Oro A. (2002) A search for thecollisional parent bodies of large NEAs. Icarus, in press.