Analysis of spectral features in TNO and asteroid spectra

1
Analysis of spectral features in TNO and asteroid
spectra
S. Erard, D. Despan, F. Merlin
2
Spectral observation of TNOs

• Dark objects ( 18th mag)
• Shallow spectral features (in the NIR)
• gt Very long exposure times required
• to access compositional information
• implication for observing strategy
• and for analysis methods

1998 Cruikshank et al.
3
Spectral observation of TNOs
Methanol is the second most easily detected ice
(after H2O) Detection requires SNR 70 for pure
ice Ammonia detection requires SNR
125 Requirements are higher if only a fraction
of the surface is covered, or mixture with other
ices
C. Trujillo, Catania 2006 meeting
8-10m telescope, mag 18 1 h exposure ltgt
SNR 100 Only 2005 FY9 has been observed with
SNR allowing detection of N2, CO, CO2, or
ethane ice (only ethane is detected)
4
Spectral observation of TNOs
The Good News - About 25 KBOs could be observed
by an international team of collaborators
using the world's largest telescopes. The
Bad News - Don't bother observing any of the
brightest 15 KBOs unless you spend at least 4
hours of exposure time on a 8m 10m telescope
in good conditions. Tips for observers -Don't
repeat objects that are already done! -Observe in
good conditions and at low airmass! -Take high
(80-100) S/N spectra!
Trujillos conclusion, Catania 2006
meeting (excerpt)
5
Spectral detection /characterization methods

• Simulation spectral fit, inversion
• - The first step is to identify the components
• - Extra components just add noise to the fits
• - Continuum is always an issue
• Spectral ratios
• - Historically important, but very crude
• MGM
• - Adapted only to specific minerals
• (pyroxenes, olivines, feldpars)
• Tetracorder, etc
• - Rely on a more or less complete data base,
• - Not really adapted to ices

Geographic mixture Geographic mixture Hapke model Hapke model Shkuratov model Shkuratov model
H2Oa - - - - 6 5µm
H2Oc 1 5µm 14 5µm 5 5µm
Carbon 83 15µm 50 15µm 68 15µm
Ice-Th 5 5µm 7 31µm 7 5µm
Tit-Th 5 5µm - - - -
Tri-Th 6 5µm 29 12µm 14 15µm
Merlin Barucci, Catania 2006 meeting
6
Multiresolution spectral analysis

• Purpose
• - Detection method adapted to low SNR situations
• Output
• - Characteristics of absorptions features
  (center, depth, width)
• - Detection thresholds in terms of S/N and
  proximity to the edges
• Basis
• - Wavelet decomposition multiscale grouping
  (based on imaging algorithms)
• - Uses a dyadic algorithm to avoid band
  reconstruction
• Performances
• - Separates bands within Rayleigh criterion (if
  slightly different)
• - Accuracy on band properties 10 for Gaussians
• - Correctly identifies bands at SNR 3 in I/F

7
Simulations  High SNR
Orthopyroxene (laboratory spectrum) The two
bands are correctly detected at all
scales Grouping and identification of a dominant
scale provides accurate band characteristics
8
Simulations  medium SNR
Simulated spectrum noise 3 wide bands and a
narrow one, correctly detected
9
Simulations  very tilted continuum
Jarosite (lab. spectrum) Many narrow bands on
varying structure, correctly detected Bands near
the edge (uncomplete) are detected with a low
statistical weight
10
Ceres, 2.1-2.4 µm
CH3OH
VLT / Naco resolved observations Bright, extended
object (mag 8)  Ice features?  Clay
features?
CH4
H2O
N2
NH3
11
Ceres, 2.1-2.4 µm
19 structures detected, mainly small telluric
(with atm. counterparts) and solar bands  No
ice absorption above 5 s          (disk centre
or pole)  Possible feature at 2.11
µm  Improvement of telluric correction pending
Erard et al., EGU 2006
12
Sedna, 1.9-2.5 µm
Observations by Barucci et al 2005   (VLT),
R3000  6 structures detected, mostly telluric
correction remants Positive detection at
2.142 µm, corresponding to N2 ice but
significantly narrower  
Erard et al., DPS 2005
13
Conclusion
 Multiscale analysis methods, coupled with noise
filtering algorithm, are very efficient in low
SNR situations TNO spectral studies require
this kind of analysis  The present one, based
on a very redundant algorithm, may still be
improved with band reconstruction Tests are
still being performed on laboratory spectra
observations  First article with full
description and tests to be submitted in 2007
(hopefully)  
14
(No Transcript)