J/ApJ/899/15 Parameters for the 58 τHI(v) sightlines (Murray+, 2020)
Extracting the Cold Neutral Medium from HI Emission with Deep Learning:
Implications for Galactic Foregrounds at High Latitude
Murray C.E., Peek J.E.G., Kim C.-G.
<Astrophys. J., 899, 15 (2020)>
=2020ApJ...899...15M 2020ApJ...899...15M
ADC_Keywords: Interstellar medium; H I data; Galaxies; Spectra, radio;
Keywords: Interstellar medium ; Interstellar atomic gas ;
Interstellar absorption ; Cold neutral medium ; Milky Way Galaxy ;
Convolutional neural networks ; Radio astronomy
Abstract:
Resolving the phase structure of neutral hydrogen (HI) is crucial for
understanding the life cycle of the interstellar medium (ISM).
However, accurate measurements of HI temperature and density are
limited by the availability of background continuum sources for
measuring HI absorption. Here we test the use of deep learning for
extracting HI properties over large areas without optical depth
information. We train a 1D convolutional neural network using
synthetic observations of 3D numerical simulations of the ISM to
predict the fraction (fCNM) of cold neutral medium (CNM) and the
correction to the optically thin HI column density for optical depth
(RHI) from 21cm emission alone. We restrict our analysis to high
Galactic latitudes (|b|>30°), where the complexity of spectral
line profiles is minimized. We verify that the network accurately
predicts fCNM and RHI by comparing the results with direct
constraints from 21cm absorption. By applying the network to the
GALFA-HI survey, we generate large-area maps of fCNM and RHI.
Although the overall contribution to the total HI column of CNM-rich
structures is small (∼5%), we find that these structures are
ubiquitous. Our results are consistent with the picture that
small-scale structures observed in 21cm emission aligned with the
magnetic field are dominated by CNM. Finally, we demonstrate that the
observed correlation between HI column density and dust reddening
(E(B-V)) declines with increasing RHI, indicating that future
efforts to quantify foreground Galactic E(B-V) using HI, even at high
latitudes, should increase fidelity by accounting for HI phase
structure.
Description:
To trace HI emission throughout the local interstellar medium (ISM),
we use the Galactic Arecibo L-band Feed Array Survey (GALFA-HI) at the
Arecibo Observatory. GALFA-HI is the highest angular resolution (∼4'),
highest spectral resolution (0.18km/s), large-area (13000deg2)
Galactic 21cm emission survey to date.
As a sensitive probe of the absorption properties of HI in the local
ISM, we assemble a sample of available 21cm optical depth spectra from
the literature. From the 21cm Spectral Line Observations of Neutral
Gas with the Karl G. Jansky Very Large Array (VLA) survey (21-SPONGE)
we select the 30 spectra in our region of interest. In addition to
high-latitude spectra from 21-SPONGE, we include spectra from the
Millennium Arecibo 21cm Absorption-Line Survey. These spectra have
lower optical depth sensitivity than 21-SPONGE (στHI=0.01
per 0.18km/s channels). We select the 28 spectra unique relative to
the 21-SPONGE sample in our region of interest that do not show
spurious spectral artifacts.
File Summary:
--------------------------------------------------------------------------------
FileName Lrecl Records Explanations
--------------------------------------------------------------------------------
ReadMe 80 . This file
tabled1.dat 87 58 Parameters for the 58 τHI(v) sightlines used for
verifying the CNN model
--------------------------------------------------------------------------------
See also:
J/ApJS/145/329 : Millennium Arecibo 21-cm Survey (Heiles+, 2003)
J/ApJ/793/132 : Perseus cloud sources Gaussian parameters (Stanimirovic+,2014)
J/A+A/594/A116 : HI4PI spectra and column density maps (HI4PI team+, 2016)
J/ApJS/234/2 : The GALFA-HI survey data release 2 (Peek+, 2018)
J/A+A/633/A14 : GaussPy+ decomposition of Galactic Ring Survey (Riener+, 2020)
--------------------------------------------------------------------------------
Byte-by-byte Description of file: tabled1.dat
--------------------------------------------------------------------------------
Bytes Format Units Label Explanations
--------------------------------------------------------------------------------
1- 8 A8 --- Source Source name
10- 10 I1 --- Ref [1/2] Reference (1)
12- 18 F7.3 deg RAdeg [10.2/350] Right Ascension (J2000)
20- 25 F6.3 deg DEdeg [1.29/35] Declination (J2000)
27- 31 F5.2 10+20/cm2 NHI-thin [0.84/12] HI column density in the
optically-thin limit, Equation 4
33- 36 F4.2 10+20/cm2 e_NHI-thin [0.04/0.8] Uncertainty in NHI-thin
38- 42 F5.2 10+20/cm2 NHI-iso [0.84/15] Total HI column density,
Equation 3
44- 47 F4.2 10+20/cm2 e_NHI-iso [0.04/0.7] Uncertainty in NHI-iso
49- 52 F4.2 --- fCNM [0/0.77] Fraction of Cold Neutral Medium
observed, Equation 8
54- 57 F4.2 --- e_fCNM [0/0.3] Uncertainty in fCNM
59- 62 F4.2 --- fCNM-CNN [0.01/0.33] Fraction of Cold Neutral Medium
predicted by the CNN
64- 67 F4.2 --- e_fCNM-CNN [0.01/0.1] Uncertainty in fCNM-CNN
69- 72 F4.2 --- RHI [1/1.56] Column Density Correction Factor,
Equation 5
74- 77 F4.2 --- e_RHI [0/0.01] Uncertainty in RHI
79- 82 F4.2 --- RHI-CNN [1/1.2] Column Density Correction Factor
predicted by the CNN
84- 87 F4.2 --- e_RHI-CNN [0.02/0.07] Uncertainty in RHI-CNN
--------------------------------------------------------------------------------
Note (1): References as follows:
1 = Murray+, 2018ApJS..238...14M 2018ApJS..238...14M
2 = Heiles+, 2003, J/ApJS/145/329
--------------------------------------------------------------------------------
History:
From electronic version of the journal
(End) Prepared by [AAS], Coralie Fix [CDS], 25-Oct-2021