· method note

Albedo as a weak taxonomic discriminant / 反照率判据

139,582 catalogued asteroids have a measured diameter. 2,094 of them now have a measured spectral class — 1,414 from SBDB (Bus-DeMeo) and 680 recovered from the PDS Asteroid Taxonomy archive (Bus-DeMeo 13 / Tholen 494 / SMASS 173). The remaining 136,378 bodies with an albedo have no measured class of any kind. This note asks one narrow question: can their geometric albedo tell us anything?

Short answer: only a little, and never enough to name a type. Our rule is 71.6% correct on the very data used to build it (vs 48.7% for always guessing the majority class). The X-complex is only 28.1% correct, and M-type has zero measured examples. So on a body's detail page the rule may down-weight one implausible branch — it never removes a branch, never prints a single type, and never narrows the price bracket.
简单说:反照率只能给某个分支降权,不能认定类型。判据自测准确率仅 71.6%,M 型零样本。页面永远保留 C/S/M 全部三个分支与完整估值区间。

1 · Where the numbers come from

The four works below establish that albedo and taxonomic class are statistically associated and that the albedo values themselves are well calibrated. They do not supply the numeric thresholds. We could not retrieve their full texts, so quoting a threshold from them would be a fabrication. Every number on this page is computed from this site's own 1,403 bodies with both a measured Bus-DeMeo class and an albedo, and is reproducible from the catalogue data.

The four DOIs establish that albedo and taxonomic class are statistically associated. They DO NOT supply the numeric thresholds below: we could not retrieve full texts, so we refuse to quote thresholds we have not read. All numeric values are computed from THIS SITE'S OWN 1,398 measured-class bodies and are reproducible from public/data/catalogue-assessed.json.

2 · Measured albedo by class

Classnminp25medianp75p90max
C3420.0160.0430.0540.0710.1171
S5650.05490.210.2530.3130.3790.952
MNO measured bodies in this family — SBDB spec_B folds M-type into the X-complex, so albedo CANNOT identify metallic M bodies. This is a hard limit, not a missing sample.
X2530.0180.0470.1120.2010.3441

M-type has ZERO measured examples in spec_B: Bus-DeMeo folds M into the X-complex. The discriminant can therefore never certify or exclude M from albedo alone — the M branch MUST remain in every bracket.

3 · The derived rule

Thresholds are midpoints of the C / X / S medians (0.054, 0.112, 0.253) — no class is privileged.

Albedo pVInferred bandMeaningEngine type
[0, 0.083)Cdark / carbonaceous-likeC
[0.083, 0.182)Xambiguous X-complex — NOT priced as Mnot priced as M
[0.182, 1]Sbright / silicate-likeS

midpoints of the C/X/S sample medians. X-complex maps to no engine type: it is ambiguous, not metallic.

4 · How well it actually works

leave-one-out over the same measured sample (optimistic: same data used to derive thresholds). Weak classifier (71.6% on its own training data). X-complex is only 28% correct, and M has zero examples. It may EXCLUDE an implausible class, NEVER certify one. Every body still gets the full C/S/M bracket.

ClassCorrectHits / n
C81.3%278 / 342
S85.3%482 / 565
M—0 / 0
X28.1%71 / 253

5 · Coverage

136,378 unclassified bodies have a usable albedo: C 71,976 · X 29,178 · S 35,224. These are band assignments, not classifications; every one of them still receives the full C/S/M bracket on its page.

6 · Literature (DOIs verified against CrossRef)

KeyDOICitation & what it supports
masiero201410.1088/0004-637X/791/2/121Masiero et al. (2014), Main-belt asteroids with WISE/NEOWISE: near-infrared albedos, ApJ 791, 121
Large measured albedo sample per taxonomic class; establishes that albedo and class are statistically associated.
demeo201310.1016/j.icarus.2013.06.027DeMeo & Carry (2013), The taxonomic distribution of asteroids from multi-filter all-sky photometry, Icarus
Taxonomic classes have distinct photometric/albedo distributions; provides the class taxonomy used here.
pravec201210.1016/j.icarus.2012.07.026Pravec et al. (2012), Absolute magnitudes of asteroids and a revision of asteroid albedo estimates, Icarus
Calibration of the albedo values themselves (H + thermal), so the input quantity is trustworthy.
bus200210.1016/S0019-1035(02)00066-0Bus & Binzel (2002) / distribution of taxonomic classes in the main belt, Icarus
Class frequencies differ by population — a reminder that a prior derived from the main belt is not universal.
Use albedo ONLY to down-weight an implausible C branch (very bright body) or S branch (very dark body), and always display the error rate and the full bracket.
Forbidden: Printing a single type or a single price from albedo alone · Claiming the four papers supply the thresholds · Treating X-complex as metallic M · Removing the M branch (zero measured examples — cannot be excluded) · Hiding the cross-validation error rate

7 · Download & cite

albedo-taxonomy.csv (41 rows) · /public/data/albedo-taxonomy.json (full contract).

Cite: Helios Prospects (2026). Albedo as a weak taxonomic discriminant. https://www.heliosprospects.com/albedo-taxonomy.html

Locked by _albedo-verify.mjs (DOI metadata, reproducible thresholds, the M-zero-sample limit, and the error rate). See also spectral priors and methodology.