I once read a quote in a conference abstract that stated the paper had been written to serve as “a historical guide to the asteroid literature for the perplexed”. That quote aptly sums up my desire for writing a piece on the history of asteroid classification and taxonomies, and the schemes currently in use.
When asteroid spectral types are quoted in the scientific literature they are often taken from different classification schemes. The most common taxonomies currently in use are the Bus-DeMeo, Tholen, and Gaffey schemes, but a number of other asteroid classification taxonomies have come and gone over the decades.
This article on the history of asteroid classification runs through all the earlier taxonomic systems that have informed those currently in use, from those developed in the mid 1970s, right up to date.
The quotation I referred to in the opening paragraph appeared in a XVIII LPS conference abstract by Tholen & Bell (1987). You can either read the abstract now or (what I recommend) wait until you reach the year 1987 further on in this article when you can read it in context, having read what took place in the intervening years that prompted the author to write the amusing line.
This article is organised into sections: click on a link in the list below to jump straight to that section, although I do recommend that you read the whole of this article in order from start to finish to appreciate the subtleties of this extremely fascinating, albeit extraordinarily dry, subject of asteroid classification – and I don’t mean dry as in ‘this asteroid has no water’, I mean the subject can be more than a little monotonous at times. Still, it’s a fascinating subject to delve into.
Jump to section:
• Evolution of Asteroid Classification Systems
• What Is It That’s Being Classified?
• Current Asteroid Classification Systems
• Interpreting Asteroid Spectra
• Boundaries Between Classes
• Asteroid Classification Systems: 1973 to date
• The A to Z of Classes Today
• Key Spectral Features
• Classifying Asteroids Bennu and Ryugu
Evolution of Asteroid Classification Systems
The following alluvial diagram attempts to show the evolution of the various asteroid classification systems that define the A to Z of asteroid spectral types in use today.
Click on the image for a larger view and then zoom in.
The diagram starts on the left hand side with the three original C-S-U taxonomic classes published in 1975 (named mainly after Carbonaceous and Stony-metallic meteorites, with and all other types clumped into an ‘Unclassifiable in the present system’ class). The diagram ends with the Bus-DeMeo classification scheme (published in 2009, with a minor addition in 2013) and its 25 classes and sub-types. The right hand end of the diagram summarises those 25 classes into three main groups or complexes, C-S-X and a number of smaller classes known as end members. All classes that make up the diagram from 1975 to date will be discussed in detail throughout this article.
References to the papers which define the classes are shown in the diagram. Links to all sources of data and information on the different asteroid taxonomies shown in the diagram and throughout this article are provided at the end of the article.
Note that the unequal vertical extents of each of the three strands on the left hand side of the diagram do not represent relative proportions of asteroids, but the number of sub-types that the particular class (C, S, U) splits into as more data become available over subsequent years, as you move across the diagram. For example, even though most asteroids observed belong to the C-complex (on the far right of the diagram), the early-defined S class (on the far left of the diagram) splits into more distinct classes and sub-types than does the C class as the asteroid taxonomies evolve from left to right.
What Is It That’s Being Classified?
The five main observing regions in the electromagnetic spectrum that will be referred to extensively in this article are:
(Far to Near)
|Near Infrared||NIR||0.8–2.5 µm|
|‘3µm’ Band||–||2–4 µm|
|Mid Infrared||MIR||5–40 µm|
When sunlight hits the surface of an asteroid, electromagnetic radiation is transmitted through the near-surface minerals which absorb or emit radiation at certain wavelengths which are characteristic of the particular mineral species present. The features in the processed spectrum, such as slope steepness (usually defined from 0.7–1.5µm), curvature, absorption band positions, widths and depths together imply which minerals may be present on the surface of the asteroid.
The measured (inferred) surface composition may or may not be characteristic of the composition of the asteroid as a whole, depending on the asteroid’s geological history, for example whether it is a primitive solid body, a rubble pile asteroid or a differentiated asteroid: if it’s a primitive solid body, the inferred surface composition will be characteristic of the asteroid as a whole; if it’s a differentiated body, the inferred surface composition will only be characteristic of the surface of the asteroid, not the interior; if it’s a rubble pile asteroid, the inferred surface composition is unlikely to represent the whole asteroid because a rubble pile asteroid is an aggregate of smashed up pieces of asteroid which have coalesced under gravity, rocks from collisions of different asteroids.
Current asteroid taxonomies are largely based on the presence of major features in UV to VNIR wavelength reflectance spectra together with albedo (the total amount of light reflected) and photometric color indices. Over the last four decades, the wavelength range utilized to define the classes has mostly spanned 0.3–2.45μm, covering 0.3–1.1μm in early asteroid classification schemes, and 0.45–2.45μm in more recent schemes.
Common spectral features that separate the taxonomic classes include absorptions due to the silicate minerals olivine and pyroxene, the hydrated silicates like phyllosilicates, oxides such as magnetite and spinel, sulfides such as troilite and oldhamite, and other hydrated minerals. Pure metal asteroids have near featureless VNIR spectra (or weak absorptions if silicates are mixed in) and require radar and NIR observations to identify metal (e.g. Neeley et al, 2014).
Upcoming surveys may refine VNIR asteroid taxonomies, for example the 0.35–0.9µm wavelength coverage by GAIA of 300,000 asteroids and the James Webb Space Telescope, when launched, which will provide NIR to MIR coverage of the 0.7–5.3μm wavelength region.
Observations have already been made by other surveys down to 0.1μm and up to 40μm, but not enough to officially define new classes as yet. These include:
In the NUV (0.1–0.3μm), some spectral features are associated with carbon, as well as iron-bearing silicate materials with varying degrees of space weathering (the effects of which may be evident in the UV/blue region before becoming apparent in the VNIR). NUV features may distinguish primitive asteroids because S-type slopes appear to remain redder than C-types into the NUV (e.g. Waszczak et al, 2015). For discussions of NUV spectra, see Wong et al (2019) and Hendrix et al (2016).
In the NIR, the so-called ‘3μm’ band (2–4μm) is associated with water/ice, water-bearing materials, and the OH molecule. In particular, the band around 3.1–3.2μm is associated with water/ice or ice frost and the band around 2.95μm is associated with water in minerals.
A 2.7–2.8μm band is associated with the hydroxyl (OH) molecule in phyllosilicate minerals. This band is often present when the 0.7µm band is present, but not vice versa. Phyllosilicates also have a band minimum in the MIR around 12μm, the exact position of which may indicate the degree of aqueous alteration.
Other spectral features in the NIR (1–5μm) are associated with carbonates, organics, and ammoniated minerals.
For a review of the 3μm region generally see Rivkin et al (2015; 2019), or related to primitive asteroids see Campins et al (2019). For discussions related to Trojan asteroids, particularly the distinction between types (red or less-red) see Emery et al (2011), and for 3μm spectral features of Trojans see Brown (2016).
In the MIR, most spectral features in the 5–40µm wavelength range are associated with silicate minerals. Features in the 5–14μm wavelength range include the Christensen feature and Reststrählen emission bands associated with the minerals plagioclase, pyroxene, and olivine (e.g. Donaldson Hanna & Sprague, 2009), as well as the 12μm phyllosilicate band.
Current Asteroid Classification Systems
The most current taxonomic system of asteroid classification as at 2019 is the Bus-DeMeo system published in 2009 (with a minor 2013 revision) covering the wavelength range 0.45–2.45μm, although the systems of Bus (1999) (0.44–0.92μm) and Tholen (1984) (0.3–1.04μm) are still used, as is the system of Gaffey (1993) (0.34–2.57μm) which provides a system of classifying asteroids based on mineralogical ratios.
The representative spectra that define the classes in the Tholen, Bus and Bus-DeMeo asteroid classification systems are shown below.
Each of the spectra shown above represents the flux ratio of sunlight reflected from the asteroid’s surface, relative to the sunlight incident on the surface, plotted as a function of the wavelength range shown. The faint horizontal lines shown with the Bus/Bus-DeMeo spectra represent a relative reflectance of 1, where all spectra have (by convention) been normalized to 1 at 0.55μm. That particular wavelength is chosen for normalizing to because it is the effective wavelength midpoint of a standard V (visible) band photometric filter.
The letter designation of the classes isn’t entirely arbitrary, at least it wasn’t in the early days of asteroid taxonomy. Most of the early assigned letters had some meaning often related to colour, suspected composition, or meteorite analog. This loosened as time went on and the choice of letters became more limited. Some letters have even been recycled, somewhat perplexingly, but every so often throughout this text I will provide the A to Z list that existed at the particular time.
In general, any inferred mineral assemblage (or other characteristic such as albedo) of one asteroid in a taxonomic class should be applicable to others in the same class, but it doesn’t necessarily mean that all asteroids in a class have the same composition.
You may ask yourself, why are old asteroid taxonomies still quoted in the literature? Well, it is probably a combination of many reasons:
(i) modern CCD spectrophotometry is not sensitive to measurements <0.45μm, but pre-CCD surveys extended further into the UV;
(ii) the 8-Color Asteroid Survey (ECAS) data used by Tholen (1984) analyzes spectra to 0.3μm, so even though more recent asteroid taxonomies extend into the IR, the Tholen system may contain potentially diagnostic information in the UV drop-off;
(iii) the Tholen system was a landmark in asteroid taxonomy and permeated the whole practise of asteroid classification;
(iv) the Gaffey system (see plot further below) provides a sub-classification for S-type asteroids based on an implied relative abundance of the major silicate minerals olivine and pyroxene, which is often quoted to supplement the S-class in other systems;
(v) the Bus classification system was based on the most internally consistent dataset in the 0.45–0.92μm region and largest number of asteroids, although the 0.45–2.45μm Bus-DeMeo classification supersedes it now.
Interpreting Asteroid Spectra
The interpretation of asteroid surface composition from a reflectance spectrum can be masked by compositional and non-compositional effects that can alter the spectral features such as slope and band position and shape (band centers, band depths, band widths, band area ratios), as well as having spectrally neutral effects on albedo. For a discussion of interpreting asteroid spectra, see Reddy et al (2015). Spectra can be altered by, for example, phase angle (Sun-asteroid observer angle), surface temperature, surface particle size, space weathering, addition of exogenic material from impacts, and impact shock.
In the following text, redder means increasing towards longer wavelengths (more positive), and bluer means increasing towards shorter wavelengths (more negative with increasing wavelength).
Phase angle (greater for near-Earth asteroids than for main-belt asteroids) alters spectral slope, albedo, and band depths: spectral slope becomes redder with increasing angle, or bluer with decreasing angle. Temperature (affected by distance to the Sun) alters band shapes and positions. These effects need to be corrected for to interpret mineralogy. For a discussion of how this is done, see Reddy et al (2015).
Particle size affects spectral slope, albedo and band depth: spectra typically become bluer, darker, and bands deeper with increasing particle size. Examples of blue-sloped asteroids include 2 Pallas (B-type), and the regolith-less 101955 Bennu (Cb-type) a sample of which will be returned to Earth in 2023).
Fresher, recently impact-excavated material also has bluer and darker spectra. Spectrally-neutral darkening can also occur by surface contamination with carbonaceous material, considered to be the reason why the spectrum of 4 Vesta appears less red than other V-types (e.g. Buratti et al, 2013).
Conversely, spectra may become redder and darker with increasing amounts of space weathering due to longer exposure to the space environment, showing subdued or broadened olivine absorption features (especially among certain S and A types). Space weathering depends largely on composition and in general S-types show more optical alteration than C-types. The anomalous Q-type spectra that long-baffled planetary scientists is suggested to represent fresh ordinary-chondrite-like surface compositions, whereas some S-types may represent space-weathered ordinary-chondrite-like surfaces (e.g. see Binzel et al, 2010).
The space weathering explanation on certain S-type asteroid spectra was assisted by the laboratory analysis of ordinary chondrite meteorites, observations of S-complex asteroids visited by space missions (for example 951 Gaspra, 243 Ida, 221 Eros, 25143 Itokawa) and importantly by comparison with the samples from Itokawa (Sqw-type) which were returned to Earth in 2010. And, of course, the decades of investigations of space weathering effects on Apollo lunar samples. In the most current asteroid taxonomy (Bus-DeMeo) a subscript ‘w’ is given to asteroids with similar spectral features to another class but with a higher spectral slope (as in the case of Itokawa).
For a review of space weathering on asteroids and other airless bodies, see Pieters & Noble (2016) and all references therein, but I also like Chapman (2004) for a history of the debate.
Boundaries Between Classes
Boundaries between classes of asteroids in the early (1970’s) asteroid classification schemes (the start of the alluvial diagram) were defined by a computer program which sorted spectrophotometric data into similar spectral groups, taking albedo and photometric colors into account, with spectra within a group physically examined by overlaying and comparing shapes. Spectra with large error bars or lying close to what were arbitrary boundaries at the time were assigned to one or other group, even though asteroid spectra may be continuous across a boundary.
In current asteroid classification schemes, groupings are decided by multivariate analysis of the data using principal component analysis (PCA) and assigning boundaries between the resultant clusters. Some boundaries between classes represent natural groupings of asteroid dynamical families, for example the Hungaria family, most of which are E-type asteroids like its namesake 434 Hungaria. The K class was defined for the Eos dynamical family of asteroids (and others resembling them), which show less reddening at NIR wavelengths than the S class but had originally been grouped with the S-class by expanding the S class boundaries.
Dynamical groups are a way to study the interiors of fragmented asteroids, because some parts of the surfaces of family members were originally part of the interior of a larger asteroid or planetesimal. If members of an asteroid family classify into different taxonomic groups, it could mean that the fragmented parent body was differentiated, exposing mantle or even metallic core material (like the planet-wide Rheasilvia basin-forming impact on 4 Vesta which, although not breaking the asteroid apart, did expose a layer spectrally similar to diogenite meteorites). Or different taxonomic groups within an asteroid family could just mean that an asteroid with a different composition was dynamically incorporated into the family.
Although a finite number of parent bodies produced the millions of asteroids that are thought to exist in the asteroid belt, it is not known whether there are groups of distinct mineralogical assemblages or a continuum of compositions.
Asteroid Classification Systems: 1973 to date
All classifications schemes shown in the above alluvial diagram are now described in detail below.
Note that early attempts (pre-1973) to use photometric UBV color indices and albedo as discriminators to classify asteroids are not discussed in his article.
Also note that the first generation of asteroid classification schemes (1973-1975) described below which introduced the RMF classes, are not included in the above diagram because the RMF and CSU classes are defined in a way that would make the alluvial diagram too convoluted. Instead the evolution from RMF to CSU classes are shown in a separate alluvial diagram below. You just need to picture it tagged onto the left of the main diagram above.
1973: The Chapman-McCord-Johnson classification defined 17 spectral groups for 32 asteroids observed using 24 color filters over UV-V-NIR wavelengths (0.3–1.1μm). The classification system primarily uses the overall spectral reflectance slope with sub-groups based on position (if any) of absorptions around 0.65μm and 0.95μm and the location of any UV drop-off. It distinguishes three classes (R, M, F): R = red slope (overall positive); M = medium red slope, and F = neutral (flat to bluish) with a UV drop-off. The three classes are each subdivided into four groups (R1-R4, M1-M4, F1-F4) depending on the absorption bands present. Three further R sub-types describe how red the slope is (A = very, B = moderate, C = slight) to give R sub-types R2A, R2B, R3A, R3B, R3C. The parameters in this classification system formed the basis for many later taxonomies, but not the names. For example, in this scheme, asteroid 4 Vesta is defined as M3, which does not relate to the class of M-type asteroids in current asteroid classification schemes. See Chapman et al (1973).
1975: The McCord-Chapman classification extends the 1973 study above to 98 asteroids, retaining three main classes (R, M, F) but redefining the groups within them, identifying 27 significantly different spectral groups. A spectral group means a point or cluster significantly removed from others in classification space, but not necessarily different mineral assemblages. The groups were defined using nine parameters to characterise spectral variation. These are: R (16 groups: 11 with 0.95μm band, 5 without 0.95μm band); M (6 groups); F (5 groups with UV drop-off). At this point, the authors speculated from statistical analysis that they had probably identified about half of the different spectral types in the asteroid belt. See McCord & Chapman (1975 a,b).
At this point in time (1975), the parameters used for distinguishing the different classes are:
1. R/B: ratio of reflectance at 0.7μm/0.4μm (visible spectral slope), which correlates with the definitions for RA, RB, RC, M, F in the 1973 scheme.
2. BEND: visible positive curvature near 0.56μm, (R0.56 – R0.4) – (R0.73 – R0.56). A small value signify metals or opaques.
3. IR: intensity of IR to red part of spectrum, (R1.05 – R0.73). For example, olivine has a major absorption at 1.05μm, so a significant abundance gives a negative value.
4. UV Drop-off: shown by some curves with low R/B (e.g. M and F slopes). The position of the bend signifies opaques and silicates.
5. UV Slope.
6. IR Absorption due to pyroxene.
7. IR Band Centre.
8. IR Band DEPTH: ratio of reflectance at band base over highest point on short-wavelength side, related to Fe3+ absorption near 0.95μm.
9. 0.65μm band depth.
1975: The Chapman-McCord-Zellner (CMZ) taxonomy redefined the earlier RMF classes, introducing the CSU classes which formed the basis for many future taxonomies. The naming came from comparisons with meteorite spectra which suggested the C and S groups were compositionally similar to carbonaceous and stony-metallic meteorites, respectively. U means unclassifiable in this system and incorporated the distinctive M types from the RMF scheme. The classification used a dataset of 110 asteroids observed in what is now called the 24-Color Asteroid Survey covering the wavelength range 0.32–1.08μm. Classes were separated using five parameters, related to spectrophotometry (R/B, DEPTH), B-V color, albedo and polarization, but with less weight on albedo. The distinct spectra of certain RMF scheme M-types were noted at the time including 4 Vesta but specific classes were not defined for these and they were all classified under U. See Chapman et al (1975).
1976: An analysis by Zellner & Gradie is (one of) the first that introduces a formal M class for metal-rich asteroids and the E class for asteroids with high albedo attributable to pure enstatite. The authors suggested M types may represent extremely reduced examples of C or S types. See Zellner & Gradie (1976). At this point in time, the broad defined classes are C, S, M, E, U.
1977: The Zellner & Bowell analysis extends the classification to eight groups (C, S, M, E, O, T, I, U) based on observations of 359 asteroids, and an algorithm based on parameters related to UBV color, spectrophotometry, albedo and polarization. There are two new short-lived classes O and T. Classes are described in terms of composition: O (ordinary chondritic, metal poor, i.e. LL chondrites), T (Trojan of unidentified composition), C (carbonaceous), S (silicaceous), M (metal- rich), E (metal-free, enstatite). They also use I (indeterminate or inadequate data) and U (unclassifiable or unusual, basically none of the other defined classes, and included 2 Pallas, 4 Vesta, and 221 Eos). See Zellner & Bowell (1977).
1978: The Bowell taxonomy drops O and T and introduces a new R class for the reddest UBV color asteroids (this R class is not the same as the earlier one in the RMF scheme and is related to the dropped O class). This makes six classes C, S, M, E, R, U based on observations of 523 asteroids. It uses an algorithm based on seven parameters of spectrophotometry (R/B, BEND, DEPTH), UBV color, albedo and polarization. Of the 523 asteroids, C = 36%; S = 27%; M = 2%, E = 0.5%, R = 0.5% and 34% are ambiguous or unclassifiable in this system. The number of objects classified with this algorithm was extended to 752 asteroids soon after, with no changes to the classes. See Bowell et al (1978); Zellner (1979).
1979: Another Chapman classification also defined the six C, S, M, E, R, U classes based on observations of 277 asteroids from the 24-Color Asteroid Survey, with recalibrations and re-averages of their previous published data, but still with less weight given to albedo. This system recognised about 80 different spectral types among the asteroids surveyed (C = 30 types, S = 22 types, M/E = 7 types, R/U = 20 types). They also recognised 221 Eos and family as one of the few groups that might be mixtures of C and S. See Chapman et al (1979).
1979: Degewij & Van Houten found that 30% of sampled Hildas, Trojans and outer Jovian satellites are reddish and dark (compared to C-types) and referred to them as RD types. The RD class has low albedo (2–4%) and steep spectral slopes around 0.7–0.9μm. A new but short-lived T class (for Trojan) was referred to in this paper. See Degewij & Van Houten (1979).
1981: Zellner et al introduced a new X class for M-like spectra with low albedo. This X class was briefly referred to by others as DM for dark-M, later renamed pseudo-M or PM (and later referred to simply as P class). The earlier RD class was renamed D class. See Zellner et al (1981); Hartmann et al (1981).
1982: A Gradie & Tedesco classification places greater emphasis on albedo to define new classes F and P. Albedos are derived from 10μm and 20μm radiometry and the spectra from the 0.3–1.1μm 8-Color Asteroid Survey (ECAS) data. The F class is a flat spectrum (as per the McCord-Chapman 1975 taxonomy). The P class (renamed from PM or pseudo-M) has spectral characteristics 0.3–1.1μm that are indistinguishable from M types, but with an albedo similar to C types (<0.065) rather than M types (0.07–0.23). They note that 2 Pallas and 4 Vesta are still not classifiable in this scheme and 1 Ceres is an unusual C type. See Gradie & Tedesco (1982).
At this point in time (early 1980s), descriptions of the classes, in terms of albedo and reflectance spectra (slope and absorption bands 0.3–1.1μm) are:
C – Low albedo (< 0.065). Neutral slope, weak band <0.4μm.
D – Low albedo (< 0.065). Very red >0.7μm.
E – Very high albedo (> 0.23). Featureless, sloping up into red.
F – Low albedo (< 0.065). Flat.
M – Moderate albedo (0.07–0.23). Featureless, sloping up into red.
P – Low albedo (< 0.065). Featureless, sloping up into red.
R – Very high albedo (> 0.23). Very red, bands deeper than S.
S – Moderate albedo (0.07–0.23). Reddened, band typically 0.9–1.0μm. d1ecdf
U – Unusual or Unclassifiable in the current system.
I – Indeterminate or inadequate data.
1983: A photometric analysis by Veeder at UBV (0.36–0.55μm) and JHK (1.25–2.2μm) wavelengths introduces the A class for very red asteroids. Except for 349 Dembowska, all members previously assigned to the R class are reassigned to the S and A classes based on colour and albedo. See Veeder et al (1983).
1984: A landmark in asteroid classification history, the Tholen taxomony uses principal component analysis and minimal tree clustering to define 14 broad classes: A, B, C, D, E, F, G, M, P, Q, R, S, T, V, and I (for Inconsistent). The number of classes is determined by the length of the tree branch to nearest neighbors in the clustering (reducing the branch cut-off size will result in more classes). The taxonomy is based on the highest quality 0.34–1.04μm wavelength spectra in the 8-Color Asteroid Survey (ECAS) dataset (405 of 589 asteroids), supplemented by radiometric albedo to improve separation of the classes (for example between E, M, P, and between B, C). There are five new classes (B, G, Q, T, V) and one class (R) is reintroduced having been dropped by Veeder the year before. The analysis indicates that 95% of the variation in the 8-Color data lies in the first two principal components related to two absorption features (one in the UV region, the other in the NIR). The C class is split into (B, C, F, G) groups of asteroids based on spectral variation at UV/blue wavelengths, where B, F, G are similar to but outlie C. The T class means Tentative (not the 1979 T for Trojan class) and lies between S and D. The P class lies between C and D and within the X (EMP) group. Two new classes are introduced for distinct asteroids: Q for 1862 Apollo and V for 4 Vesta, and the R class is resurrected for 349 Dembowska. The I class (Inconsistent) replaces U (Unclassifiable in earlier systems) which is now used for Unusual. See Tholen (1984).
At this point in time (mid 1980s), descriptions of the classes, in terms of albedo and reflectance spectra (slope and absorption bands 0.3–1.1μm), are:
A – High albedo. Very red <0.7μm. Strong band near 1.05μm.
B – Moderately low albedo. Flattish. Weaker absorption <0.4μm.
C – Low albedo. Flat to reddish >0.4μm. Absorption <0.4μm.
D – Low albedo. Featureless. Neutral to reddish <0.55μm. Very red >0.55μm, levelling out >0.95μm.
E – Very high albedo. Spectrum like M, P.
F – Low albedo. Flat to bluish featureless. Weaker absorption <0.4μm.
G – Low albedo. Flat >0.4μm. Stronger absorption <0.4μm.
M – Moderate albedo. Spectrum like E, P. Flat to reddish, featureless.
P – Low albedo. Spectrum like E, M. Intermediate to C and D.
Q – Moderately high albedo. Strong absorption <0.7μm. Strong band near 1μm.
R – High albedo. Strong band <0.7μm, strong band near 1μm (broader than V, deeper than S).
S – Moderate albedo. Absorption strong <0.7μm, weak or none >0.7μm.
T – Low albedo. Absorption <0.85μm. Flat >0.85μm.
V – High albedo. Strong absorption <0.7μm. Strong band near 0.95μm.
X – EMP where no albedo available.
U – Unusual spectrum.
I – Inconsistent data.
1987: Time for a short break. I mentioned at the start that in researching the information for this article on asteroid classification schemes, I came across a paper by Bell which was stated to serve as “a historical guide to the asteroid literature for the perplexed.” I just love that. And it fed by desire to write this article. The quote and the following diagram appeared in a XVIII LPS conference abstract (Tholen & Bell, 1987) and in Asteroids II (p.299). The diagram contains the same information as in my alluvial diagram from 1975 to 1984 — and being far simpler than mine, I suppose it must be better.
1987: The Barucci taxonomy re-analyses 438 asteroids from the 8-Color data used by Tholen combined with IRAS albedos. The analysis is carried out using G-mode multivariate clustering. The result gives nine classes (containing 18 types): A0, B0-B3, C0, D0-D3, E0, G0, M0, S0-S3, V0. Three of the classes (B, D, S) are each composed of four sub-units, where 0 to 3 represent increasing albedo. The number of classes defined is determined by the confidence level chosen (setting a lower confidence level will result in more classes). The total of 18 groups are obtained by reducing the confidence level from 99.7% to 97.5%. The Tholen class T is sub-unit D3; F is grouped into C0 and B1; R, Q, V are reduced into V0; and P is grouped into C0 and considered to be the tail end of C types that reside beyond 3.5 AU. The analysis was re-run in 1990 at 95% confidence resulting in different groupings, the main differences being eight S sub-units, EMP all reduced into a D sub-unit, and T remaining as a separate class. I’ve not shown the re-run groups in my alluvial diagram. See Barucci et al (1987) for the analysis at higher confidence levels, and Barucci & Fulchignoni (1990) for the analysis at the lower confidence level.
1988: Using spectrophotometric data from the 52-Color Asteroid Survey (0.8–2.5μm), the K class was introduced to account for the atypical spectra of the Eos dynamical family which have visible spectra similar to S types and flat IR spectra similar to C types. The K class asteroids may be the source of CV/CO chondrites. See Bell (1988) and Bell et al (1988).
1989: This Tedesco classification is a brief detour away from reflectance spectra. It uses visual examination of stereo pairs of 2D projections of 3D data of U-V color indices, ECAS v-x color indices and IRAS albedos. It aims to make a point that a similar class clustering can be obtained without using spectral reflectance. It characterises 357 previously classified asteroids into 11 taxonomic classes (A, C, D, E, F, G, K, M, P, S, T). Except for the unique classes Q, V, R (and B because of 2 Pallas), it places 96% of the asteroids into the same classes as Tholen’s principal components analysis. It sounds from reading the paper that the analysis was done in part to prove a point about the pitfalls of combining classifications derived from different methods (spectral reflectivity, photometry, radiometry, polarimetry). See Tedesco et al (1989).
1991: In an extension to Tholen’s 1984 analysis, Burbine conducted an 8-Color + 52-Color PCA on a subset of the asteroid spectra used in Tholen’s 8-Color PCA and found no difference in the clustering relationships. This is not plotted in my alluvial diagram because no reclassification of classes was proposed. See Burbine (1991).
1993: The Gaffey system for S-type asteroids is based on the relationship between the 1μm absorption band centre and the 2μm to 1μm band area ratio, providing sub-groups for the S class based on the olivine/pyroxene ratio and pyroxene composition. This asteroid classification system is still in use today. Seven classes S(I) to S(VII) are defined for olivine-dominated to pyroxene-dominated compositions, respectively. It is based on 39 of the 144 S-type asteroids used in the Tholen classification analysed over the spectral range 0.35–2.55μm by combining datasets from three surveys: 24-Color (0.33–1.1μm), 8-Color (0.32–1.1μm) and 52-Color (0.8–2.5μm). See Gaffey et al (1993).
1994: The Howell taxonomy used an artificial neural network with a self-organising map (SOM) hidden layer to cluster reflectance spectra of 539 asteroids over the range 0.34–2.57μm. The analysis combined 8-Color (0.34–1.1μm) and 52-Color (0.8–2.57μm) survey data, with each combined spectrum resampled to 65 data points or colors. After unsupervised training and convergence of data into clusters for well-characterized spectra, the network was supervised pre-assigning Tholen’s class names to clusters, with further classes post-assigned based on the network cluster suggestions. The neural network approach is less sensitive to data noise than some of the other methods used so far in this discussion. Using lower signal-to-noise data, Howell’s 8-Color SOM results are still consistent with Tholen’s 8-Color PCA, issues mainly arising for classes separable by albedo (i.e. E,M,P and C,B). With the combined data (65 colors), the Howell taxonomy shows two compositionally meaningful S sub-groups: So and Sp, for olivine-rich/reddest and olivine-poor/least red, respectively. The C class is split into two sub-groups: Cv and Cx based on continuum slope curvature (v=concave down, reflectance weakly increasing or strongly decreasing with wavelength; x=convex down, reflectance strongly increasing with wavelength). The B and F classes are combined into a single B+F class. Indeterminate classifications include BCv (2 Pallas), CvB (1 Ceres), CvP (some C types), and SoT and TSo. See Howell et al (1994).
1995: Around this time, Rivkin proposed a new W class for hydrated M type asteroids. An IRTF spectrophotometric survey of 16 EMP class asteroids showed absorption features diagnostic of hydrated minerals in the 3μm region across all three EMP classes. A later observation of 20 M type asteroids suggested that larger M types may be relatively primitive material (E chondrites or salty C chondrites) with iron meteorites derived from smaller anhydrous M type asteroids. See Rivkin et al (1995; 2000).
1995: The SMASS survey (Small Main-Belt Asteroid Spectroscopic Survey) observed 316 asteroids over the wavelength range 0.44–0.92μm. Initial analysis of 80 asteroid CCD spectra suggested two new classes: J and O. The J class was proposed for some Vesta family members which had VNIR reflectance spectra separated from 4 Vesta in PCA space, similar to the HED diogenite meteorite Johnstown (but future confirmation was needed that there was no NIR feldspar absorption). The O class was proposed for 3628 Božněmcová which was found to have a VNIR spectrum similar to L6/LL6 ordinary chondrite meteorites. See Binzel & Xu (1993); Binzel et al (1993); Xu et al (1995).
1999–2002: The Bus taxonomy builds on the framework of the Tholen taxonomy to define 26 classes. Importantly it uses the internally consistent SMASSII (Small Main-Belt Asteroid Spectroscopic Survey II) dataset of visible wavelength CCD spectra (0.44–0.92μm) of 1447 asteroids sampled. Classes were defined using principal component analysis of spectra formed of 48 data points (input channels) analyzed with a correlation matrix. The larger sample size and higher resolution revealed more structure in the data producing more sub-classes. The problematic EMP classes, previously only separable by albedo, are now split into sub-classes. The 26 classes are the dark C-complex: C, Cb, Cg, Ch, Cgh; the brighter S-complex: S, Sa, Sk, Sl, Sq, Sr; the X-complex: X, Xc, Xe, Xk; and the others, which are generally referred to as end members and outliers: A, B, D, K, L, Ld, O, Q, R, T, V. See Bus (1999) and Bus & Binzel (2002).
2009: The Bus-DeMeo taxomony extends the Bus taxonomy into the NIR covering a range of 0.45–2.45μm, using IRTF/SpeX observations of 371 asteroids over the range 0.8–2.45μm and 321 asteroids from the SMASSII dataset (0.44–0.92μm). Classes were were defined using principal component analysis of spectra formed of 40 data points (input channels) analyzed with a covariance matrix. This taxonomy has 25 classes, getting rid of some Bus classes and adding some new ones. Out are Ld, Sk, Sl. In are Sv, Xn, and a ‘w‘ subscript added for reddened (space-weathered) spectra. The C-complex remains as C, Cb, Cg, Ch, Cgh; the S-complex is now S, Sa, Sq, Sr, Sv; the X-complex is now X, Xc, Xe, Xk, Xn; and the end members and outliers are now A, B, D, K, L, O, Q, R, T, V. See DeMeo et al (2009; and rev. 2013).
So that’s it. We’re now up to date with the current asteroid classification systems.
The following diagram is the short summary version of my original opening diagram, showing the evolution from the original three CSU classes through the widely adopted Tholen classes to the Bus classes, and finally the Bus-DeMeo classes and sub-types and main CSX complexes recognised today. The next section will concentrate on descriptions of the complexes, classes and sub-types in the Bus-DeMeo system.
The A to Z of Classes Today
The following descriptions of the classes in the VNIR region (0.4–2.45μm) are based on information in the papers referred to throughout this article but in particular that given in DeMeo et al (2009; and rev. 2013):
A – High albedo. Very red slope longward of 0.7μm. Strong band near 1.05μm +/- shallow 2μm band. Originally named VR (very red). Analog: olivine-rich achondrites, brachinites.
C-Complex: B, Cb, C, Cg, Cgh, Ch:
B – Moderately low albedo. Flat or Blue slope, overall negative. Weak absorption at ~0.4μm, bump at ~0.6μm. Some show 1μm absorption attributed to magnetite. Some show 1–2μm concave up curvature. Has subsumed the old Tholen F class.
Cb – Low albedo. Flat with slight positive slope beginning at 1.1μm.
C – Low albedo. Flat to reddish longward of 0.4μm, absorption shortward of 0.4μm, +/- bump at ~0.6μm, +/- feature at 1μm and slightly positive slope beginning at 1.3μm. Originally named for assumed similarity to Carbonaceous chondrites.
Cg – Low albedo. Slight positive slope beginning at 1.3μm with a pronounced UV drop off (i.e. shortward of 0.4μm). Naming signifies like a C-type, with similarity to the old G-type, the founding member of which was 1 Ceres.
Cgh – Low albedo. Slight positive slope beginning at 1μm with a pronounced UV drop off. Broad, shallow absorption centered near 0.7μm. Naming signifies like a C-type, with similarity to the old G-type and hydrated.
Ch – Low albedo. Slight positive slope beginning at 1.1μm with a slight UV drop off. Broad, shallow absorption near 0.7μm. Naming signifies like a C-type and hydrated.
D – Low albedo. Linear and featureless. Neutral to reddish shortward of 0.55μm, very steep red slope longward of 0.55μm, levelling out around 0.95μm +/- slight curvature around 1.5 μm. Originally named RD (red and dark). Includes the red-type Trojans. Not to be confused with the current class R.
I – Not a class: refers to Inconsistent with the particular data.
J – High albedo. Sharper peak than V, and deeper, narrower 0.9μm absorption shifted to shorter wavelength. Analog: Johnstown HED diogenite (from deeper in the crust than V-types).
K – Between C and S classes (lying in the dividing gap between C/X and S in principal component space). Low to moderate albedo. Deep absorption band shortward of 0.75μm, shallow and wide band at 1μm with straight sides. No 2μm band. Analog: CV/CO chondrites, like the Eos dynamical family.
L – Between K and S classes. Low to moderate albedo. Steep visible slope levelling out 0.7μm, sometimes a concave down curvature at 1.5 μm, +/- 2μm band. Analog: CAI-rich, spinel-bearing meteorites.
O – Flat shortward of 0.8μm, deep, broad rounded absorption at 1μm and another absorption feature at at 2μm. Named for Ordinary chondrites with class specifically defined for 3628 Božněmcová. Analog: L6/LL6 ordinary chondrites.
Q – Between V and S. Moderately high albedo. Strong absorption shortward of 0.7μm. Strong band near 1μm, others near 1.3μm and 2μm. Type example is the near-Earth asteroid 1862 Apollo. Current analog: fresh (non-space-weathered) ordinary chondrites.
R – High albedo. Strong band shortward of 0.7μm. Strong bands near 1μm (narrower than Q, broader than V, deeper than S) and near 2μm. Originally named for Red slope, originally VR (very red). Class defined for 349 Dembowska.
S-Complex: S, Sa, Sq, Sr, Sv (and weathered types Sqw, Srw, Svw):
S – Moderate albedo. Moderate absorption at 1μm, and at 2μm which varies in depth between objects. Originally named for Silicaceous or Stony-iron meteorites. Current analog: ordinary chondrites. Subsumes some of the old Sk class objects now spectra are extended into the NIR.
Sa – Moderate albedo. Deep, broad absorption at 1μm. Features similar to A class but less red. This class subsumes the old Sl class now spectra are extended into the NIR.
Sq – Moderate albedo. Wide absorption at 1μm. Features at 1μm and 1.3μm similar to Q class, but shallower 1μm band. Subsumes some of the old Sk class objects now spectra are extended into the NIR. Current analog: ordinary chondrites,
Sr – Moderate albedo. Features at 1μm and 2μm similar to R class, but shallower 1μm band.
Sv – Moderate albedo. Features at 1μm and 2μm similar to V class, but shallower 1μm band.
T – Low albedo. Linear and featureless, moderately red slope with absorption feature shortwards of 0.85 µm. Flat longward 0.85μm, often gently concaving down. Originally referred to Trojan (which now fall into current D, P, C classes) and later named for Tentative or requiring follow-up.
U – Not a class: refers to Unusual, and previously referred to Unclassifiable in the particular system of the time.
V – High albedo. Strong band near 0.95μm and narrow band near 1μm. Class defined for 4 Vesta. Analogue: HED achondrites.
W – Moderate albedo. Spectrum like M but with a 3μm hydration feature.
w – Sub-types of S and V with redder slopes due to space weathering: Sw, Sqw, Srw, Svw, and Vw defined so far.
X-Complex: X, Xc, Xe, Xk, Xn:
X – Linear with medium to high slope. Used to refer to the old EMP types when no albedo was available. Other than some E types, there is no particular correlation between the new X classes and the old EMP classes. Analogs include iron meteorites, enstatite meteorites, and other primitive types.
Xc – Low to medium slope, slightly curved, concave down. Low albedo. Transitional between X and C types. Some of the old P-types (which have a spectrum intermediate to C and D and included the less-red type Trojans) are now classified as Xc now that spectra are extended into the NIR.
Xe – Low to medium slope, similar to Xc and Xk, and transitional between X and E types. Absorption feature around 0.49μm, like the old E-type. High albedo. Analog: enstatite chondrites.
Xk – Low to medium slope, slightly curved, concave down, similar to Xc. Weak feature 0.8 to 1 μm. Transitional between X and K types.
Xn – Low to medium slope. Class defined for 44 Nysa (which was previously classified as E and Xc).
Y – Not used.
Z – Not used.
Key Spectral Features
The key features that are used to classify asteroids under the different asteroid classification schemes are as follows:
UV drop-off. Absorption feature due to strong Fe2+–Fe3+ intervalence charge transfer transitions.
0.49μm band: Associated with sulphides in E/Xe-type asteroids.
Slope longward of 0.55µm: magnitude depends of the presence or absence of reddening agents such as Fe-Ni metal or organics.
0.60-0.65μm band: Absorption due to Fe3+ in Fe alteration minerals.
0.7μm band: Intervalence charge transfer transition in oxidized iron (Fe2+–Fe3+) in phyllosilicate minerals, with band center varying from 0.59–0.67µm for saponite to 0.70–0.75µm for serpentine, for example. This band may be a proxy for the 2.7μm overtone when NIR data are unavailable.
Slope longward of 0.7µm: calculated over the region 0.7–1.5µm. May be used to distinguish between two types of Trojans: red (like D-types) and less red (intermediate C- to P-types).
0.8-0.9μm band: Absorption due to Fe3+ in Fe alteration minerals.
1µm band: Olivine. A composite of three absorption bands associated with electronic transitions of Fe2+ between structural sites. The band centre moves to longer wavelengths with increasing Fe2+ content.
1µm (Band I), 2µm (Band II): Pyroxene. Associated with crystal field transitions of Fe2+ between structural sites, the band centres being at:
0.9µm, 1.9µm: low-Ca pyroxenes (orthopyroxenes).
1.05μm, 2.35μm: type B high-Ca pyroxenes (lower % wollasonite).
0.9μm, 1.15μm: type A high-Ca pyroxenes (higher % wollastonite).
1.4μm band. Absorption due to H2O vibrational overtone in hydrated minerals.
1.9μm band. Absorption due to OH vibrational overtone in hydrated minerals.
2μm grand divide (in principal component space): A gap between the S-complex (spectra that have a 2μm absorption) and the C- and X- complexes (spectra that don’t).
2.5µm to 3.5µm. Absorptions due to bound H2O and structural OH in hydrated minerals.
2.7μm band: Fe2+–Fe3+intervalence charge transitions due to structural OH in phyllosilicates. The band center is not observable from Earth. When NIR data are not available, the 0.7μm band may be a proxy for the 2.7μm band.
3µm band. Absorptions at 2.9µm and 3.1µm due to interlayer and adsorbed H2O stretch modes, and H2O ice.
Classifying Asteroids Bennu and Ryugu
The near-Earth asteroid 101955 Bennu is currently being visited by NASA’s OSIRIS-REx mission.
Prior to the spacecraft’s arrival in December 2018, the asteroid was classified as F/B-type from ground-based observations (as summarised in Lauretta et al, 2015). The F classification is from the Tholen asteroid taxonomy; the B classification is from the Tholen and Bus/Bus-DeMeo taxonomies (the F class having been dropped in favour of B and Cb types in the Bus/Bus-DeMeo taxonomies).
Measurements by the OSIRIS-REx on-board instruments indicate that the asteroid has a negative slope and featureless visible spectrum, except for a possible magnetite absorption at 0.55μm (see Lauretta et al, 2019). At NIR wavelengths, Bennu shows a broad 2.74μm absorption attributed to hydrated clay minerals (Hamilton et al, 2019).
As of 2019, Bennu remains classified as B-type.
The near-Earth asteroid 162173 Ryugu (below) was visited by JAXA’s Hayabusa2 spacecraft from June 2018 to November 2019.
Prior to the spacecraft’s arrival, the asteroid had been classified as Ch/Cgh-type due to its slightly positive spectral slope and a weak 0.7μm hydration feature observed by one ground-based observation. Other ground-based observations did not observe the hydration feature and classified it as C/Cg-type (see Perna et al, 2017).
After arrival, the globally averaged data obtained by Hayabusa2’s on-board instruments found only weak hydration features in the VNIR spectra, with no 0.7μm absorption and a narrow 2.72μm absorption attributed to hydrated clay minerals (see Watanabe et al, 2019; Sugita et al, 2019; Kitazato et al, 2019).
As of 2019, Ryugu has been re-classified as Cb-type.
Both Ryugu and Bennu may originate from the same dynamical family of asteroids (Polana or Eulalia) (see de León et al, 2018) although maybe from different generations as their spectra indicate different histories. The spectra of both asteroids show similarities with laboratory spectra of carbonaceous chondrite meteorites, however Ryugu may have experienced more heating and shock than Bennu. Ryugu shows similarity to thermally metamorphosed CI-type and shocked CM-type chondrites, whereas Bennu shows similarity to aqueously altered CM-type chondrites. This is discussed in the 2019 papers cited above for each asteroid.
That’s it for my review of the history of asteroid classification schemes and taxonomies. See the section below for a list of all the papers quoted throughout this article, including a link to the software used to produce the alluvial diagrams.
For a visual overview of all the asteroids (and comets) that have been visited by spacecraft, see the post Asteroids and Comets Visited by Spacecraft, and for a discussion on the difference between asteroids and comets, read the article Asteroid vs Comet.
For a non-rigorous analysis of the asteroid on steroids in the 1998 film Armageddon, read the article It’s the Size of Texas, and for a list of every asteroid and comet movie ever made, read Making an Impact: Lights, Camera and Asteroid!
If you come across any links that don’t work, please let me know and I’ll fix them asap.
1. Classes defined from 1973 to date (as shown in alluvial diagram) (date order)
2. Other papers referred to (alphabetical order)
3. Bennu and Ryugu (alphabetical order)
4. Graphing software for alluvial diagrams
5. Asteroid textbooks
1. Classes Defined from 1973 to Date (as shown in ALluvial Diagram):
Chapman, C.R., McCord, T.B., & Johnson, T.V. (1973). Asteroid spectral reflectivities. The Astronomical Journal, 78, pp.126-140. http://adsabs.harvard.edu/full/1973AJ…..78..126C.
McCord, T.B. & Chapman, C.R. (1975a). Asteroids: Spectral reflectance and color characteristics. The Astrophysical Journal, 195, pp. 553-562.
McCord, T.B. & Chapman, C.R. (1975b). Asteroids: Spectral reflectance and color characteristics II. The Astrophysical Journal, 197, pp. 781-790. http://articles.adsabs.harvard.edu/full/1975ApJ…197..781M.
Chapman, C.R., Morrison, D., & Zellner, B. (1975). Surface properties of asteroids: A synthesis of polarimetry, radiometry, and spectrophotometry Icarus, 25(1), pp. 104-130. https://doi.org/10.1016/0019-1035(75)90191-8.
Zellner, B., & Gradie, J. (1976). Minor planets and related objects. XX: Polarimetric evidence for the albedos and compositions of 94 asteroids. The Astronomical Journal, 81, pp. 262-280. http://articles.adsabs.harvard.edu/full/1976AJ…..81..262Z
Zellner, B., & Bowell, E. (1977). 2: Asteroid Compositional Types and their Distributions. International Astronomical Union Colloquium, 39, pp. 185-197. https://doi.org/10.1017/S0252921100070093.
Bowell E., Chapman, C.R., Gradie, J.C., Morrison, D., & Zellner, B. (1978). Taxonomy of asteroids. Icarus, 35(3), pp. 313-335. https://doi.org/10.1016/0019-1035(78)90085-4.
Zellner, B.T. (1979). Asteroid taxonomy and the distribution of the compositional types. In: Asteroids, pp. 783-806. http://adsabs.harvard.edu/abs/1979aste.book..783Z.
Chapman, C.R. & Gaffey, M.J. (1979). Reflectance spectra for 277 asteroids. In: Asteroids. (A80-24551 08-91) Tucson, Ariz., Univ. Ariz. Press, pp. 655-687. NASA-supported research. https://ui.adsabs.harvard.edu/abs/1979aste.book..655C/abstract.
Degewij, J. & Van Houten, C.J. (1979). Distant asteroids and outer Jovian satellites. In: Asteroids. (A80-24551 08-91) Tucson, Ariz., Univ. Ariz. Press, pp. 417-435. NASA-supported research. https://ui.adsabs.harvard.edu/abs/1979aste.book..417D/abstract.
Zellner, B., Tedesco, E.F. & Tholen, D.J. (1981). Highlights from the Eight-Color Asteroid Survey. In: Bulletin of the American Astronomical Society, 13, p. 717. http://adsabs.harvard.edu/full/1981BAAS…13..717Z.
Hartmann, W.K., Cruikshank, D.P., Degewij, J. & Capps, R.W. (1981). Surface materials on unusual planetary object Chiron. Icarus, 47(3), pp 333-341. https://doi.org/10.1016/0019-1035(81)90181-0
Gradie, J. & Tedesco, E. (1982). Compositional structure of the asteroid belt. Science, 216(4553), pp. 1405-1407. https://doi.org/10.1126/science.216.4553.1405. Note: The paper that introduces F and P classes is referenced as Tedesco & Gradie (1982) submitted to Icarus, but it doesn’t look like it was ever published. The two classes are however referred to in Gradie & Tedesco (1982).
Veeder G.J., Matson D.L., & Tedesco E.F. (1983). The R asteroids reconsidered. Icarus, 55(1), pp. 177-180. https://doi.org/10.1016/0019-1035(83)90058-1.
Tholen, D.J. (1984). Asteroid taxonomy from cluster analysis of photometry. Doctoral thesis, University of Arizona. https://repository.arizona.edu/handle/10150/187738.
Tholen, D.J. & Bell, J. F. (1987). Evolution of Asteroid Taxonomy. In Lunar and Planetary Science Conference, 18, pp. 1008-1009. http://articles.adsabs.harvard.edu/full/1987LPI….18.1008T.
Barucci, M.A., Capria, M.T., Coradini, A. & Fulchignoni, M. (1987). Classification of asteroids using G-mode analysis. Icarus, 72(2), pp. 304-324. https://doi.org/10.1016/0019-1035(87)90177-1. There is an extension to this paper (not included in the alluvial diagram as the taxonomy is unchanged from the 1987 version) which can be found at Fulchignoni, M., Birlan, M., & Barucci, M.A. (2000). The extension of the G-mode asteroid taxonomy. Icarus, 146(1), pp. 204-212. https://doi.org/10.1006/icar.2000.6381
Bell, J.F., Owensby, P.D., Hawke, B.R., & Gaffey, M.J. (1988). The 52-Color Asteroid Survey: Final Results and Interpretation, Abstracts of the Lunar and Planetary Science Conference, 19, p. 57. http://articles.adsabs.harvard.edu/pdf/1988LPI….19…57B.
Bell, J.F. (1988). A probable asteroidal parent body for the CO or CV chondrites. Meteoritics, 23, pp. 256-257. http://articles.adsabs.harvard.edu/pdf/1988Metic..23..256B
Tedesco, E.F., Williams, J.G., Matson, D.L., Weeder, G.J., Gradie, J.C., & Lebofsky, L.A. (1989). A three-parameter asteroid taxonomy. Astronomical Journal, 97, pp. 580-606. Research supported by USAF. http://articles.adsabs.harvard.edu/full/1989AJ…..97..580T
Burbine, T. (1991). Principal component analysis of asteroid and meteorite spectra from 0.3 to 2.5 microns. Master’s thesis, Univ. of Pittsburgh, Pittsburgh, Pa.
Gaffey, M.J., Bell, J.F., Brown, R.H., Burbine, T.H., Piatek, J.L., Reed, K.L. & Chaky, D.A. (1993). Mineralogical variations within the S-type asteroid class. Icarus, 106(2), pp. 573-602. https://doi.org/10.1006/icar.1993.1194.
Binzel, R.P. & Xu, S. (1993). Chips off of asteroid 4 Vesta: Evidence for the parent body of basaltic achondrite meteorites. Science, 260(5105), pp. 186-191. https://doi.org/10.1126/science.260.5105.186.
Binzel, R.P., Xu, S., Bus, S.J., Skrutskie, M.F., Meyer, M.R., Knezek, P. & Barker, E.S. (1993). Discovery of a main-belt asteroid resembling ordinary chondrite meteorites. Science, 262(5139), pp. 1541-1543. https://doi.org/10.1126/science.262.5139.1541.
Howell, E.S., Merenyi, E., & Lebofsky, L.A. (1994). Classification of asteroid spectra using a neural network. J. Geophys. Res. : Planets, 99(E5), pp. 10847-10865. https://doi.org/10.1029/93JE03575.
Xu, S., Binzel, R.P., Burbine, T.H. & Bus, S.J. (1995). Small main-belt asteroid spectroscopic survey: Initial results. Icarus, 115(1), pp. 1-35. https://doi.org/10.1006/icar.1995.1075.
Rivkin, A.S., Howell, E.S., Britt, D.T., Lebofsky, L.A., Nolan, M.C. & Branston, D.D. (1995). 3-μm spectrophotometric survey of M-and E-class asteroids. Icarus, 117(1), pp. 90-100. https://doi.org/10.1006/icar.1995.1144.
Bus, S.J. (1999). Compositional Structure in the Asteroid Belt: Results of a Spectroscopic Survey. Doctoral thesis, Massachusetts Institute of Technology. https://dspace.mit.edu/handle/1721.1/9527.
Rivkin, A.S., Howell, E.S., Lebofsky, L.A., Clark, B.E. & Britt, D.T. (2000). The nature of M-class asteroids from 3-μm observations. Icarus, 145(2), pp. 351-368. https://doi.org/10.1006/icar.2000.6354.
Bus, S.J. & Binzel, R.P. (2002). Phase II of the small main-belt asteroid spectroscopic survey: A feature-based taxonomy. Icarus, 158(1), pp. 146-177. https://doi.org/10.1006/icar.2002.6857.
DeMeo, F.E., Binzel, R.P., Slivan, S.M., & Bus, S.J. (2009). An extension of the Bus asteroid taxonomy into the near-infrared. Icarus, 202(1), pp. 160-180. https://doi.org/10.1016/j.icarus.2009.02.005, https://hal.archives-ouvertes.fr/hal-00545286 and flowchart revision (2013): http://smass.mit.edu/busdemeoclass.html.
2. other papers referred to:
Binzel, R.P., Rivkin, A.S., Stuart, J.S., Harris, A.W., Bus, S.J. & Burbine, T.H. (2010). Observed spectral properties of near-Earth objects: results for population distribution, source regions, and space weathering processes. Icarus, 170(2), pp. 259-294. https://doi.org/10.1016/j.icarus.2004.04.004.
Buratti, B.J., Dalba, P.A., Hicks, M.D., Reddy, V., Sykes, M.V., McCord, T.B., O’Brien, D.P., Pieters, C.M., Prettyman, T.H., McFadden, L.A. & Nathues, A. (2013). Vesta, vestoids, and the HED meteorites: Interconnections and differences based on Dawn Framing Camera observations. Journal of Geophysical Research: Planets, 118(10), pp. 1991-2003. https://agupubs.onlinelibrary.wiley.com/doi/pdf/10.1002/jgre.20152.
Campins, H., de León, J., Licandro, J., Hendrix, A., Sánchez, J.A. & Ali-Lagoa, V. (2018). Compositional diversity among primitive asteroids. In: Primitive Meteorites and Asteroids, pp. 345-369. Elsevier. https://doi.org/10.1016/B978-0-12-813325-5.00005-7.
Clark, B.E., Hapke, B., Pieters, C. & Britt, D., (2002). Asteroid space weathering and regolith evolution. Asteroids III, 585, pp. 90086-2.
Donaldson Hanna, K.D. & Sprague, A.L. (2009). Vesta and the HED meteorites: Mid‐infrared modeling of minerals and their abundances. Meteoritics & planetary science, 44(11), pp. 1755-1770. https://doi.org/10.1111/j.1945-5100.2009.tb01205.x.
Emery, J.P., Burr, D.M. & Cruikshank, D.P. (2011). Near-infrared spectroscopy of Trojan asteroids: Evidence for two compositional groups. The Astronomical Journal, 141(1), pp. 25. https://doi.org/10.1088/0004-6256/141/1/25.
Hendrix, A.R., Vilas, F. & Li, J.Y. (2016). The UV signature of carbon in the solar system. Meteoritics & Planetary Science, 51(1), pp. 105-115. https://doi.org/10.1111/maps.12575.
Neeley, J.R., Clark, B.E., Ockert-Bell, M.E., Shepard, M.K., Conklin, J., Cloutis, E.A., Fornasier, S. & Bus, S.J. (2014). The composition of M-type asteroids II: Synthesis of spectroscopic and radar observations. Icarus, 238, pp. 37-50. https://doi.org/10.1016/j.icarus.2014.05.008.
Pieters, C.M. & Noble, S.K., (2016). Space weathering on airless bodies. Journal of Geophysical Research: Planets, 121(10), pp. 1865-1884. https://doi.org/10.1002/2016JE005128.
Reddy, V., Dunn, T.L., Thomas, C.A., Moskovitz, N.A. & Burbine, T.H., (2015). Mineralogy and surface composition of asteroids. Asteroids IV, pp. 43-63. http://www.mtholyoke.edu/~tburbine/reddy.asteroidsIV.2017.pdf.
Rivkin, A.S., Campins, H., Emery, J.P., Howell, E.S., Licandro, J., Takir, D. & Vilas, F. (2015). Astronomical observations of volatiles on asteroids. In: Asteroids IV, pp. 65-87. https://arxiv.org/abs/1502.06442.
Rivkin, A.S., Howell, E.S. & Emery, J.P. (2019). Infrared Spectroscopy of Large, Low‐Albedo Asteroids: Are Ceres and Themis Archetypes or Outliers? Journal of Geophysical Research: Planets. https://doi.org/10.1029/2018JE005833.
Waszczak, A., Ofek, E.O. & Kulkarni, S.R. (2015). Asteroids in GALEX: Near-ultraviolet Photometry of the Major Taxonomic Groups. The Astrophysical Journal, 809(1), p.92. https://doi.org/10.1088/0004-637X/809/1/92.
Wong, I., Brown, M.E., Blacksberg, J., Ehlmann, B.L. & Mahjoub, A. (2019). Hubble Ultraviolet Spectroscopy of Jupiter Trojans. The Astronomical Journal, 157(4), p.161. https://doi.org/10.3847/1538-3881/ab0e00.
3. Bennu and Ryugu:
Lauretta, D.S., Bartels, A.E., Barucci, M.A., Bierhaus, E.B., Binzel, R.P., Bottke, W.F., Campins, H., Chesley, S.R., Clark, B.C., Clark, B.E., & Cloutis, E.A. (2015). The OSIRIS‐REx target asteroid (101955) Bennu: Constraints on its physical, geological, and dynamical nature from astronomical observations. Meteoritics & Planetary Science, 50(4), pp. 834-849. https://doi.org/10.1111/maps.12353
Perna, D., Barucci, M.A., Ishiguro, M., Alvarez-Candal, A., Kuroda, D., Yoshikawa, M., Kim, M.J., Fornasier, S., Hasegawa, S., Roh, D.G., & Müller, T.G. (2017). Spectral and rotational properties of near-Earth asteroid (162173) Ryugu, target of the Hayabusa2 sample return mission. Astronomy & Astrophysics, 599, p. L1. https://doi.org/10.1051/0004-6361/201630346
De León, J., Campins, H., Morate, D., De Prá, M., Alí-Lagoa, V., Licandro, J., Rizos, J.L., Pinilla-Alonso, N., DellaGiustina, D.N., Lauretta, D.S. and Popescu, M. (2018). Expected spectral characteristics of (101955) Bennu and (162173) Ryugu, targets of the OSIRIS-REx and Hayabusa2 missions. Icarus, 313, pp. 25-37. https://doi.org/10.1016/j.icarus.2018.05.009
Hamilton, V.E., Simon, A.A., Christensen, P.R., Reuter, D.C., Clark, B.E., Barucci, M.A., Bowles, N.E., Boynton, W.V., Brucato, J.R., Cloutis, E.A., & Connolly, H.C. (2019). Evidence for widespread hydrated minerals on asteroid (101955) Bennu. Nature Astronomy, 3(4), p. 332. https://doi.org/10.1038/s41550-019-0722-2
Kitazato, K., Milliken, R.E., Iwata, T., Abe, M., Ohtake, M., Matsuura, S., Arai, T., Nakauchi, Y., Nakamura, T., Matsuoka, M., & Senshu, H. (2019). The surface composition of asteroid 162173 Ryugu from Hayabusa2 near-infrared spectroscopy. Science, p. eaav7432. https://doi.org/10.1126/science.aav7432
Sugita, S., Honda, R., Morota, T., Kameda, S., Sawada, H., Tatsumi, E., Yamada, M., Honda, C., Yokota, Y., Kouyama, T., & Sakatani, N. (2019). The geomorphology, color, and thermal properties of Ryugu: Implications for parent-body processes. Science, 364(6437), p. eaaw0422. https://doi.org/10.1126/science.aaw0422
Watanabe, S., Hirabayashi, M., Hirata, N., Hirata, N., Noguchi, R., Shimaki, Y., Ikeda, H., Tatsumi, E., Yoshikawa, M., Kikuchi, S., & Yabuta, H. (2019). Hayabusa2 arrives at the carbonaceous asteroid 162173 Ryugu—A spinning top–shaped rubble pile. Science, 364(6437), pp. 268-272. https://doi.org/10.1126/science.aav8032
4. Graphing Software (alluvial diagrams):
5. Asteroid Textbooks:
For further reading about the asteroid classification systems, the history of asteroid classification and everything else you could possibly want to know about asteroids, I recommend the five definitive publications on the subject:
Asteroids II (1989)
Asteroids III (2002)
Asteroids IV (2015)