Lunar and Planetary Science XLVIII (2017)
1706.pdf
GLOBAL CLASSIFICATION MAP OF ABSORPTION SPECTRUM OF LUNAR REFLECTANCE
OBSERVED BY SPECTRAL PROFILER / KAGUYA. Makoto Hareyama1, Yoshiaki Ishihara2, Hirohide Demura3, Naru Hirata3, Chikatoshi Honda3, Shunichi Kamata4, Yuzuru Karouji5, Jun Kimura6, Tomokatsu Morota7,
Hiroshi Nagaoka8, Ryousuke Nakamura9, Satoru Yamamoto10 and Makiko Ohtake2, 1St. Marianna University
School of Medicine, 2Japan Aerospace Exploration Agency, 3University of Aizu, 4Hokkaido University, 5Tokyo
Metropolitan University, 6Osaka University, 7Nagoya University, 8Waseda University, 9National Institute of Advanced Industrial Science and Technology, 10National Institute for Environmental Studies.
the K-means and TiO2 division was classified more by
Introduction: A geologic map is important for
ISODATA of an unsupervised classification method.
clarifying the formation of land. Many geologic maps
As a result, the whole Moon was classified into 66
were made for the Moon. Previous maps classified
classes in total. The detailed analysis was described in
geologic units based on terrain features (e.g. [1]), the
reference [7].
reflectance spectrum (e.g. [2]), and the elemental contents (e.g. [3]). However, they were for specific areas
Table 1: Classification sequence. The columns of Kof interest to each researcher. Though a global geologmeans and
TiOThe
express
class
name and
2 division
ic map of the Moon was reported by Wilhelms (1987)
Table 1: Classification
history.
columns of
K-meanstheir
and TiO
2 division express their class
the
column
of
ISODATA
expresses
the
number
of
[4], that was reconstructed from several geologic maps
name and the column of ISODATA expresses number of classes. names in TiO2 division indiclasses.
Each
name
in
TiO
division
means
amount
of
2
that were made under different or unknown classificacate amount of TiO2 , VHT: 10 wt%⇠, HT: 7.5⇠10
wt%, IMT: 4.5⇠7.5 wt%,LT: 1.0⇠4.5 wt%
TiO2 (VHT: 10 wt% ~, HT: 7.5 ~ 10 wt%, IMT:
tion criterions depending on the research.
and VLT: ⇠1.0 wt%.
4.5 ~ 7.5 wt%, LT: 1.0 ~ 4.5wt%, and VLT: ~1.0wt%).
This study seeks to make a global geologic map of
classification
K-means
TiO2 division
ISODATA
the Moon under unified classification conditions for
VHT
1
the entire Moon based on hyper spectrum data. This
HT
1
work employed the method of unsupervised classificaK1
IMT
6
tion such as K-means and ISODATA for classifying
LT
10
lunar absorption spectrum obtained by the Spectral
VLT
3
Profiler (SP) onboard Kaguya (SELENE) [5]. This
K2
–
8
report presents a global classification map of lunar
VHT
1
class name
HT
3
absorption spectra under unified conditions for the
or
K3
IMT
4
entire Moon and discusses the relation between spec# of class
LT
4
tral groups and their corresponding area.
Observation and Data: The SP observed lunar reflectance of the entire Moon with 296 bands in the
range of 512.6 to 2587.9 nm in wavelength with a
footprint of 500 m × 500 m.
This work analyzed the data called the SP-Cube
Depth within ±72.5 deg. of latitude. The SP-Cube was
lunar reflectance spectra with 160 bands between
512.6 and 1676.0 nm, which was averaged in the area
of 0.5 by 0.5 degrees of latitude and longitude. The SPCube Depth was the absorption spectra deducted continuum from the SP-Cube. In addition, this work used
TiO2 contents in weight percent derived from the
Multiband Imager (MI) on Kaguya [6] to classify mare
region because of poor sensitivity for TiO2 of the SPCube Depth.
Analysis: Table 1 presents the sequence of classification. The SP-Cube Depth was classified into seven
classes by using the K-means of an unsupervised classification method to reduce calculation time. Some
classes including mare region were then divided into
small classes by TiO2 contents. Finally, each class after
VLT
2
VHT
1
HT
1
IMT
1
LT
2
VLT
4
K5
–
8
K6
–
2
K7
–
3
K4
Total # of class
7
19
65 (+1)⇤
* A class was classified by the class interpolation in section 3.5, not ISODATA.
* A class was classified from the data of unsuitable
pixels for applying the unsupervised classifications.
Result and Discussion: Our unified classification
procedure yield 66 spectral types, and we found that
those types can be categorized into the following five
groups.
Mare (M) group. This group includes 38 classes
located mainly on maria as shown in Fig. 1. The spectra of this group have 5relatively deep absorption
around the 1µm band and indicated materials of volcanic origin such as basalts in maria, a volcanic dome
Lunar and Planetary Science XLVIII (2017)
1706.pdf
at the Marius hill, and pyroclastic deposits at the Aristarchus plateau. At the Aristarchus region, the map of
our classification was similar to the result of Zhang et
al. [3] that classified the basalt units in this region.
Highland (H) group. This group includes four
classes distributed on the highland region (Fig, 2).
Nearside and farside have different classes, and their
distributions are similar to FHT-O and FHT-An reported by Jolliff et al. [8]. Featureless spectra are classified
into this group, and its distribution is consistent with
that reported by Yamamoto et al [9].
Fig.1 Distribution of Mare group classes
South Pole-Aitken (S) group. This group consists
of nine classes distributed on the SPA, around Mare
Frigoris and Copernicus crater (Fig. 3). The spectra
indicate the presence of low calcium pyroxene. The
classification map inside the SPA is consistent with the
result reported by Ohtake et al. [10].
Fig.2 Distribution of Highland group classes
Boundary (B) group. This group consists of six
classes located around maria and the SPA (Fig. 4).
Some parts of this group are located far from maria
and most of them corresponded to cryptomare defined
by Whitten & Head [11].
Fresh ejecta in highlands (F) group. This group
consists of nine classes mainly located on fresh craters
and their ejecta in the highlands (Fig. 5). Each class
seemed to be classified by types of mixing material
such as M-, B-, H-, and S-groups.
References: [1] Whitford-Stark, J. L. (1981) Icarus,
48, 393-427. [2] Papike, J. & Vaniman, D. (1978) Geophy. Res. Lett., 5, 433 – 436. [3] Zhang F. et al..
(2014) Icarus, 227, 132–151. [4] Wilhelms D. (1987)
USGS Professional Paper 1348. [5] Matsunaga, T. et
al. (2008), Geophys. Res. Lett., 35, L23201. [6] Otake,
H. et al. (2012) 43rd LPSC, #1905. [7] Hareyama, M.
et al. (2016) 47th LPSC, #1390. [8] Jolliff, B. et al.
(2000) J. Geophys.Res., 105, 4197–4216. [9] Yamamoto, S. et al. (2015) J. Geophy. Res., 120, 2190–220.
[10] Ohtake, M. et al. (2014) Geophy. Res. Lett., 41,
2738–2745. [11] Whitten, J. & Head, J. (2015) Icarus,
247, 150–171.
Fig.3 Distribution of South Pole-Aitken group classes
Fig.4 Distribution of Boundary group classes
Acknowledgement: The present study is supported by a Grant-in-Aid for Scientific Research (B)
(26287107) (P.I. Makiko Ohtake) from the Japan Society for the Promotion of Science (JSPS).
Fig.5 Distribution of Fresh ejecta in highland group
classes