Transcript Exoplanet Gravitational Microlensing Observations with the Deep

Exoplanet Gravitational Microlensing Observations
with the Deep Impact Flyby Spacecraft
R. K.
1*
Barry ,
2
M. Albrow ,
3
Bennett ,
4
Christiansen ,
D.
J.
6
7
8
A.Becker , D. Wellnitz , K. Klaasen
J.
5
Yee ,
1. NASA/GSFC, 2. U. Canterbury, NZ, 3. U. Notre Dame, IN, 4. NASA/Ames,
5. Harvard CfA, Sagan Fellow, 6. U. Washington, WA, 7. U. Maryland, MD, 8. NASA/JPL
Abstract: Observations conducted using the High Resolution Instrument (HRI) aboard the Deep Impact Flyby spacecraft of ongoing exoplanet
microlensing events in the Galactic Bulge, detected and monitored by ground observatories are described. A permanent HRI defocus anomaly, a
strong PSF chromaticity, and significant spacecraft pointing jitter present a particularly challenging data set that are used to explore methodologies
for crowded-field photometry. Time series photometry are first extracted of events in the highly blended field to demonstrate the efficacy of an
approach based on a standard ground observation pipeline, Pysis3, using differencing photometry. A separate approach based on a modified drizzle
algorithm to provide sub-pixel image and repeatable photometric aperture positioning is explored. Once robust photometric results are obtained, the
gravitational microlensing parallax signature for individual events observed simultaneously with an approximate 1AU baseline will be extracted.
Observations: We conducted observations of the Galactic Bulge
from May 30, 2012 to August 10, 2012 with the Deep Impact HRI
visible channel during the Demonstration Extension of the
NASA/EPOXI Extended Mission – on cruise between principal
cometary targets. Spacecraft targeting was uploaded in near real
time, depending on up-link availability, as promising highmagnification targets were detected from a network of ground
observatories (Fig. 1). Science images are in 128x128, 2 microradian pixel, sub-frame mode while 256x256 frame images are
interleaved to facilitate field recognition and post-processing. All
14-bit data were obtained in visible light through the instrument’s
CLEAR1 filter with a center wavelength of 650 nm and bandwidth
exceeding 700 nm. Data were subsequently calibrated using the
Deep Impact standard, reversible pipeline prior to the science
team’s ongoing analysis. Pipeline processing includes stripe
removal, frame-transfer smear removal, and radiance calibration.
Data were not compressed on the spacecraft prior to downlink.
Integration time for all images is between 30.5 and 180.5 seconds.
Figure 1. Ground-based image of target
field. Gravitationally lensed Galactic
Bulge source star is marked with cursor
– OGLE-2012-BLG-0406. Extremely
dense star fields present a difficult
photometric problem even when the
instrument PSF is well understood and
modeled.
The Point Spread Function: All HRI images have a welldocumented permanent defocus anomaly due to GSE calibration
errors in preflight cryogenic testing. The instrument PSF is
approximately 9.5 pixels, FWHM, and is strongly chromatic with
the central portion of the PSF being more or less ‘filled’ depending
on the color of the observed object (Fig. 2). Post processing by the
science team requires the creation of target color-specific PSFs.
We construct these using a drizzle process that co-adds PSFs from
isolated calibrator stars (Barry, et al., 2010).
Figure 2. The HRI PSF showing
defocus anomaly and chromaticity.
All images are of Canopus through
differing filters and have been
drizzled. Similar results are
obtained of stars of varying color.
CR Identification and Removal: Due to the nature of these
crowded-field observations and the needed photometric sensitivity,
the science team conducted a nearly exhaustive exploration of
schemes to flag cosmic ray strikes in images. We have tested the
following algorithms: imgclean, crfind, di_crrej, Laplacian edge
detection, and STSDAS crrej, among others. We find that
Laplacian edge detection works well on these data and use it to
flag and mask suspect CR strikes. We then stack 256x256 images
that have been so processed and median them in the Z direction to
obtain a reference image for flagging and masking of CRs in
individual 128x128 science images (Fig. 3).
Figure 3. HRI image of target field before (L) processing. The image on the right
shows a cutout of the reference image of the same field. The large, chromatic HRI
PSF poses interesting challenges to the analysis of this crowded Bulge field.
Image Registration: The team has found that image registration can
be quite challenging with these data. The FITS header files do not
provide WCS information, and, due to the crowded and blended
fields and well-documented spacecraft pointing jitter, individual
images must be separately assessed and the target – or a nearby
bright fiducial star - identified to facilitate registration.
Subsequently, the images, after an intermediate CR smoothing
step, are masked with a broad Gaussian to limit edge effects. The
relative sub-pixel shift between images are then determined using
a discrete Fourier transform up-sampling method.
Full-frame
images are subsequently drizzled together to form a reference
image, while science images are registered to the reference for
later photometric steps.
Photometry: One approach to extracting time-series photometry
from these data is through the use of new difference imaging
approach that incorporates iterative correction of image-to-image
offsets and rotations, and iterative self-determination of the flat
field (Fig. 4).
Another approach that is being explored by the team is to attempt
target color-specific PSF deconvolution of the crowded-field
images prior to registration and differencing photometry. Initial
experiments using iterative, Richardson-Lucy minimized
deconvolution have shown some promise that using this additional
color ‘dimension’ as mapped to the shape of the PSF may help
separate targets in these crowded, blended fields. Additionally, use
of our drizzled, sub-sampled PSFs rather than a single, critically
sampled image, can be used to restore images with sub-pixel
resolution and may be a critical element in registering these
images.
Figure 4. Time series photometry of event OGLE-2012-BLG-0406.
These results will be further calibrated and, when taken together with
ground-based data, a simultaneous parallax signature will be extracted.
The Untimely Demise of Deep Impact: NASA officially
terminated the Deep Impact mission one month after the Jet
Propulsion Laboratory lost contact with the spacecraft on August
8, 2013. Unfortunately, for the work outlined above, final
observations of these fields were required to establish a brightness
baseline for the source stars in the Galactic bulge. While it is no
longer possible to complete the absolute calibration of our timeseries photometry, the team will attempt to construct a new relative
calibration of the DI data. This should permit us to retrieve the
simultaneous microlensing parallax even in the absence of
instrument-specific baseline brightness information from this
exceedingly challenging data set.
Co-I Barry observing target OB0406 at Canopus
Observatory, Tasmania, while the team conducts
simultaneous observations from 1AU with the Deep
Impact Flyby spacecraft. How cool is that!?
* [email protected], 301-286-0664