This paper presents the analysis of near-infrared observations of the icy s
urface of Triton, recorded on 1995 September 7, with the cooled grating spe
ctrometer CGS4 at the United Kingdom Infrared Telescope (Mauna Kea, HI). Th
is analysis was performed in two steps. The first step consisted of identif
ying the molecules composing Triton's surface by comparing the observations
with laboratory transmission spectra (direct spectral analysis); this also
gives information on the physical state of the components. Most of the ban
ds in Triton's spectrum were assigned to specific vibration bands of the CH
4,N-2,CO, and CO2 molecules previously discovered. A detailed comparison of
the frequencies of the CH4 bands confidently indicated that this molecule
exists in a diluted state in solid beta-N-2 Three new bands peaking at 5717
, 5943, and 6480 cm(-1) (1.749, 1.683, and 1.543 mu m, respectively) were a
lso observed. Laboratory experiments have shown that C2H6 isolated in solid
N-2 fits well the second band, but this would imply the appearance of unob
served bands and thus rules out this assignment. However, C2H6 may exist in
another physical state, and more experiments are necessary. No plausible c
andidate was found for these three bands when comparing with the spectra of
nine molecules (C2H2, C2H4, C3H8, NH3, SO2, HC3N, CH3OH, NO, NO2).
In view of the results of D. P. Cruikshank et al. (1993, Science 261, 742;
in preparation), the work presented here leads to two possible representati
ons of the surface of Triton. First, a two-region surface composed of a N-2
: CH4 : CO terrain, N-2 : CH4 : CO consisting of a solid solution in which
N-2 is the dominant molecule, and of a H2O + C-2 terrain, composed of a mi
xture of pure crystalline H2O and CO2 grains. The second representation is
a three-region surface composed of a N-2 : CH4 : CO terrain and two geograp
hically separated H2O and CO2 terrains.
The second step of the analysis consisted of using a bidirectionnal reflect
ance model (S. Doute and B. Schmitt 1998, J. Geophys. Res, Planets 103, 313
67). The modeling first confirms the direct spectral analysis in that CH4 i
s diluted in solid beta-N-2, giving a high degree of confidence to the conc
lusion that the N-2 : CH4 :CO terrain is in fact a solid solution. It also
provides numerical information on this terrain, namely the size of the grai
ns, the geographical abundance, and the CH4 and CO concentrations. The larg
e grain size(around 10 cm) would mean that the texture of this terrain is a
compact crystalline solid rather than granular, which is in agreement with
calculations from J. Eluszkiewicz (1991, J. Geophys. Res. 96, 19,217). In
addition, an accurate modeling of the N-2 band could suggest that the tempe
rature is greater or equal to 35.6 K.
Although undistinguishable in the spectra, a maximum of 10% surface area of
pure CH4 ice can be present at the surface of Triton, thus explaining the
high atmospheric CH4 abundance observed by Voyager 2.
Finally, the modeling showed that none of the two- or three-region represen
tations was able to fit simultaneously the K and H regions of the spectrum
of Triton. The origin of this misfit is not yet elucidated, but an instrume
ntal effect is suspected. Some questions about the physical state of the H2
O and CO2 molecules are thus raised, but unfortunately observational constr
aints are missing. New near-infrared observations could partly provide thes
e missing constraints, and would be important for detecting new molecules o
n Triton's surface. Such new data would be especially useful to identify th
e three bands at 5717, 5943, and 6480 cm(-1) (1.749, 1.683, and 1.543 mu m)
. (C) 1999 Academic Press.