www.sciencemag.org/
cgi/
content/
full/
science.
aa
p
9
85
5
/DC1
Supp
lementary
Material
s
for
A radio counterpart to a neutron star merger
G.
Hallinan
,
*
A.
Corsi
,
K. P.
Mooley
,
K.
Hotokezaka
,
E.
Nakar
,
M. M.
Kasliwal
,
D. L.
Kaplan
,
D. A.
Frail
,
S. T.
Myers
,
T.
Murphy
,
K.
De
,
D.
Dobie
,
J. R.
Allison
,
K. W.
Bannister
,
V.
Bhalerao
,
P.
Chandra
,
T. E.
Clarke
,
S.
Giacintucci
,
A. Y. Q.
Ho
,
A.
Horesh
,
N. E.
Kassim
,
S. R.
Kulkarni
,
E.
Lenc
,
F. J.
Lockman
,
C.
Lynch
,
D.
Nichols
,
S.
Nissanke
,
N.
Palliyaguru
,
W. M.
Peters
,
T.
Piran
,
J.
Rana
,
E. M.
Sadler
,
L. P.
Singer
*
Corresponding author. Email:
gh@astro.caltech.edu
Published
1
6
Octo
ber
20
1
7
on
Science
First Release
DOI:
10.1126/science.
aa
p
9
85
5
This PDF file includes:
Materials and Methods
Supplementary Text
Figs. S1 to S5
Table S1
References
1. Radio Data Reduction
1.1. VLA
Radio observations of the EM170817 field were carried out with the Karl G. Jansky Very Large
Array (
36
) in its C, CnB, and B configurations, under our target of opportunity programs
(VLA/16A
-
206; PI: A. Corsi; VLA/17A
-
374; PI: K. Mooley). All observations reported here were
carried out in
C
-
band (nominal center frequency of 6 GHz, with a bandwidth of 4 GHz
and
S
-
band
(nominal center frequency of 3 GHz, with a bandwidth of 2 GHz) with the Wideband
Interferometric Digital Architecture (WIDAR) correlator. We used QSO J1258
-
2219 (
C
-
band)
and QSO B1245
-
197 (
S
-
band) as our phase calibrator sources, and 3C
286 or
3C
147 as flux and
bandpass calibrators.
VLA data were calibrated and flagged for radio frequency interference (RFI) using the VLA
automated calibration pipeline which runs in the Common Astronomy Software Applications
package (CASA,
37
). When necessary,
additional flags were applied manually after calibration.
Images of the observed field were formed using the CLEAN algorithm (
38
), which we ran in the
interactive mode. When deconvolving bright background sources, care was taken not to include
any clean co
mponents near the emission of EM170817.
Observations carried out after 2017 September 1 UTC were intermittently affected by rapidly
fluctuating phases (timescale ~ minutes) caused by the on
-
going enhanced solar activity and
potentially related activity in
the troposphere. To mitigate phase errors in the
S
-
band, a single
round of phase
-
only self
-
calibration (including all sources brighter than 50 μJy in the initial model)
was performed. At
C
-
band, uncorrected phase errors resulted in a lower amplitude for a
ll sources
in the field of EM170817 during the first week of September. We therefore calculated a correction
factor to the measured flux densities of the radio transient by estimating the average fractional
change in the flux density of unresolved backgrou
nd radio sources in the field for each affected
epoch. These corrections amount to ~30% of the flux for the most affected epoch (UT 2017
September 5) and to ~17% of the flux on the date of first discovery in C
-
band (UT 2017 September
3). A revised strategy
of rapid phase calibration was implemented for observations in both bands
on and after UT 2017 September 7, which greatly reduced the impact of the solar activity.
The results of our VLA follow
-
up campaign of EM170817 are reported in Table S1. Flux
measur
ement errors are calculated as the quadratic sum of the map root
-
mean
-
square (rms) noise
plus a 5% fractional error on the measured flux which accounts for inaccuracies in the flux density
calibration. Where necessary, an amplitude correction factor is app
lied for known phase error
issues in
C
band data. For non
-
detections, upper
-
limits are calculated as the measured flux at the
position of EM170817 plus 2× the map rms. An example map with a high signal
-
to
-
noise detection
(11σ) of EM170817 is shown in Figur
e S1.
1.2. ATCA
We observed EM170817 on 2017 August 18, 21, 28 and September 05 using the Australia
Telescope Compact Array (ATCA) under a target of opportunity program (CX391; PI: T.
Murphy). During the Aug 18 observation the array was in the EW352 conf
iguration, for all other
observations it was in the 1.5A configuration. The Aug 18, 21 and 28 observations used two 2
GHz
frequency bands with central frequencies of 8.5 and 10.5 GHz, while for the Sept 05 observation
we centered these two frequency bands
on 5.5 and 9.0 GHz. For all epochs, the flux scale and
bandpass response was determined using the ATCA primary calibrator PKS B1934
-
638, and