J/MNRAS/496/4276 Spin parity of spiral galaxies II (Tadaki+, 2020)
Spin parity of spiral galaxies II: a catalogue of 80k spiral galaxies using big
data from the Subaru Hyper Suprime-Cam survey and deep learning.
Tadaki K.-I., Iye M., Fukumoto H., Hayashi M., Rusu C.E., Shimakawa R.,
Tosaki T.
<Mon. Not. R. Astron. Soc., 496, 4276-4286 (2020)>
=2020MNRAS.496.4276T 2020MNRAS.496.4276T (SIMBAD/NED BibCode)
ADC_Keywords: Galaxies, optical ; Morphology ; Models
Keywords: techniques: image processing - catalogues - galaxies: spiral
Abstract:
We report an automated morphological classification of galaxies into
S-wise spirals, Z-wise spirals, and non-spirals using big image data
taken from Subaru/Hyper Suprime-Cam (HSC) Survey and a convolutional
neural network (CNN)-based deep learning technique. The HSC i-band
images are about 36 times deeper than those from the Sloan Digital Sky
Survey (SDSS) and have a two times higher spatial resolution, allowing
us to identify substructures such as spiral arms and bars in galaxies
at z>0.1. We train CNN classifiers by using HSC images of 1447
S-spirals, 1382 Z-spirals, and 51650 non-spirals. As the number of
images in each class is unbalanced, we augment the data of spiral
galaxies by horizontal flipping, rotation, and rescaling of images to
make the numbers of three classes similar. The trained CNN models
correctly classify 97.5 per cent of the validation data, which is not
used for training. We apply the CNNs to HSC images of a half million
galaxies with an i-band magnitude of i<20 over an area of 320deg2.
37917 S-spirals and 38718 Z-spirals are identified, indicating no
significant difference between the numbers of two classes. Among a
total of 76635 spiral galaxies, 48576 are located at z>0.2, where we
are hardly able to identify spiral arms in the SDSS images. Our
attempt demonstrates that a combination of the HSC big data and CNNs
has a large potential to classify various types of morphology such as
bars, mergers, and strongly lensed objects.
Description:
This work is based on data from the second public data release (PDR-2)
of the HSC-SSP for the Wide layer (Aihara et al. 2019PASJ...71..114A 2019PASJ...71..114A).
For morphological classification, we use i-band images, which have
reached an exposure time of about 20min.
We use galaxies with 5.7arcsec aperture magnitudes i<20 so that we can
visually classify their morphology. It is still possible to identify
spiral arms in HSC images even for galaxies with i∼21, allowing for
morphological classification of galaxies at higher redshift. However,
as such cases are rare, the inclusion of fainter objects makes it
difficult to identify a larger number of spiral galaxies for training
CNNs.
We chose the XMM-LSS field (Pierre et al. 2004JCAP...09..011P 2004JCAP...09..011P) over an
area of ∼28deg2 to make a training data set for CNNs. Among 56787
objects, we confirm 1447 spiral galaxies with clear S-wise spiral
structure and 1382 with clear Z-wise spiral structure by visual
inspection. Additional 1177 and 1131 galaxies are identified to have
S-wise and Z-wise spiral structure with somewhat reduced confidence
level. To define as clearly as possible spirals, we classify these
galaxies into a category of unclear/dubious, which is not used for
training CNNs. The remaining 51650 galaxies are non-spiral galaxies.
We apply the trained CNN models to a large data set in other HSC-SSP
fields, where 561251 galaxy images are available over an area of
∼320deg2. We use five CNN models made from the cross-validation to
derive average predicted probabilities of each class. We identify
37917 S-spirals and 38718 Z-spirals and provide the catalogue
including the predicted probabilities in Table 3.
File Summary:
--------------------------------------------------------------------------------
FileName Lrecl Records Explanations
--------------------------------------------------------------------------------
ReadMe 80 . This file
table3.dat 37 76635 A spin parity catalogue of spiral galaxies
--------------------------------------------------------------------------------
Byte-by-byte Description of file: table3.dat
--------------------------------------------------------------------------------
Bytes Format Units Label Explanations
--------------------------------------------------------------------------------
1- 17 I17 --- ID Object identifier (1)
19 I1 --- Class [1/2] Class flag (1:S-spiral, 2:Z-Spiral)
21- 25 F5.3 --- p0 Predicted probability of non-spiral
27- 31 F5.3 --- p1 Predicted probability of S-spiral
33- 37 F5.3 --- p2 Predicted probability of Z-spiral
--------------------------------------------------------------------------------
Note (1): Object identifier is the same in the PDR-2 of the HSC-SSP for the
wide layer (Aihara et al. 2019PASJ...71..114A 2019PASJ...71..114A)
--------------------------------------------------------------------------------
History:
From electronic version of the journal
References:
Iye et al., Paper I 2019ApJ...886..133I 2019ApJ...886..133I
Iye et al., Paper III 2021ApJ...907..123I 2021ApJ...907..123I, Cat. J/ApJ/907/123
(End) Ana Fiallos [CDS] 19-Jul-2023