Transcript Long time-series photometry on temperate sites

Long time-series photometry on
temperate sites
and what to gain from a move to Antarctica
ARENA workshop,
“Time-series observations from Dome C”
Catania
T.Granzer & K.Strassmeier,
Sep 17th, 2008
Outline:
Long-term stellar photometry:
Spot modelling
Cycle variations
Astroseismology
Transit searches,…
Robotic observations
Needs/gains
The perfect observation
Thermal/Antarctic
(Direct) Spot modelling:
Continuous, covering at least a
single rotation
*
Complementary to Doppler-Imaging
Strassmeier K. G., et al., 2002
Activity cycles:
Extremely long time scales, like
decades.
Constant data quality.
*
Olah, et al., 2008
Activity cycles cont‘d
Stellar activity
cycles like Sun.
Bright targets.
Data obtained with
75cm, photoelectric
robotic telescope
Transit searches:
Continuous observations (unknown
parameter space)
*
High precision on many targets.
Can be done in white light.
Winn, Holman &
Fuentes, 2006
*
Astroseismology:
Uninterrupted data sets to resolve
entire frequency spectrum.
*
Two colors.
Short exposure times.
29 frequencies found in BI CMi (Breger, et al., 2002)
Astroseismology (cont‘d):
‘Whole Earth Telescope’ to beat
day/night cycle.
Highest duty times with robotic
telescopes.
All APT observations with a
single, robotic telescope!
Fairborn Observatory
Washington Camp, Arizona,
1560m
14 robotic telescopes, 0.1-2m
First installation world-wide
Mainly Photometry
Twin-telescope STELLA
Tenerife / Teide
2400m Altitude
2x 1,2m telescopes
WiFSIP: 4kx4k imager
SES: high-R Echelle
STELLA
STELLA-I Instrumentation
Fiber-fed Echelle
spectrograph, fixed
format, fiber
entrance 50µm
(2.1"), R42000
STELLA-II Instrumentation
4kx4k CCD, 22’ FoV,
whole Strömgren, Sloan
& Johnson filter set +
H
Task: Feed light into fiber
STELLA-I Acquisition unit
Beam-splitter diverts 4% on guider
CCD (KAF-0402ME, uncooled).
Mirror around fiber entrance.
Optic wheel with flat mirror for
calibration light, glass pyramid for
focus.
Task: Feed light into fiber
Fiber entrance
At acquire, bring stellar image onto
fiber position
Hold it there during science exposure
Image from mirror around fiber
Image through beam splitter
Flat field exposure, guider image
Task: Pointing
Guider field-of-view ~2.5 arcmin
Pointing accuracy STELLA-I currently 15.8 arcsec
Classic pointing model
7-parameter model (alt/az mount), automatically
determined in STELLA at predefined intervals:
A  Aoff  AN sin A tan E  AE cos A tan E  N PAE tan E  BNP sec E
E  Eoff  AN cos A  AE sin A  TF cos E
AN,AE…
NPAE…
BNP…
TF…
tilt of az-axis against N,E
non-perpendicularity of alt to az axis
non-perpendicularity of opt. axis to alt axis
tube flexure
Consequences
A stable mount is required for good pointing.
Temperature drifts in some parameters already on
rocky grounds.
Drifts of the ice will not be completely planeparallel and thus introduce drifts in the pointing
model with time.
Cannot use only the science observations, they
introduce bias.
Task: Acquire
Read-out stripes (shutter-less system)
Acquire on beam-splitter image
At acquire, 2-5 images are required.
Depending
Image from mirror
around fiberon star brightness, this
translates to ~10-40 sec.
Mirror image shows fiber
Beam-splitter causes the images to
be elongated in y-direction.
Image from beam-splitter
Acquire (cont.)
Acquire frames are bias and dark corrected.
A truncated gauss is used for star detection
(similar DAOfind).
Stars are discriminated from cosmics by
their elongation and sharpness.
Elongation criterion must be weak due to
beam-splitter.
Stars identified at prob.  0.443
Probability function defined by manual identification of stars on ~100
acquire frames
Task: Closed-loop guiding
Guiding is done on beam-splitter image
51 Peg, 20 min, ~1200 guider frames, average
Magnitude difference on added guider frames
allows estimate of light loss
Here: 32%
30 min @ LQ Hya, Gauss-filtered
Closed-loop guiding (cont.)
Each guider frame gives a single offset for
the two telescope axis
Up to ten single offsets are averaged
(target brightness depending).
This average offset is fed into a PID-loop
The PID output is applied to the telescope
at f=1/5 Hz.
Problems with high wind gusts.
Dependency of optimal PID parameters on seeing and
guider dead-time, from a telescope model
Currently, three PID parameter set per axis are used,
Task: Focus
A focus pyramid in the beam splits the
image into four parts.
At correct focus, the images have a
certain distance.
Pyramid is out-of focus, when star is in
focus (different optical path).
Measure diagonals or
Measure side length.
Not a perfect square, but distances
highly reproducible.
For STELLA-I, Δs=1px for
Δf=0.03933mm
5-20sec. for focusing.
Task: Scheduling
Scheduling currently simple, a few science
targets plus RV and flux standards.
Each run starts at solz > 0 with bias, followed by
flat-fields and ThAr.
During night, a ThAr plus an RV standard is taken
every 2h.
Approach:
Dispatch scheduling:
Picks target according to actual conditions.
Must run in real-time, but N
Allows easy reaction to weather changes.
Used on most robotic systems.
Robotic/Remote:
Robotic: (Almost) no human
interference.
Low bandwidth sufficient.
Unattended observations,
autonomous reaction to unforeseen
events (bad weather).
STELLA and many other projects show that it works!
What can we gain from polar
sites:
A simple example: Take a 75cm telescope from
Arizona to Dome-C.
The perfect observation:
No read-out noise, etc.
Ignore seeing (2nd order effect in photometry)
Remaining error sources: Scintillation, Photon noise,
Background noise.
Scintillation: ²~sec(Z)³N²T3/2 (Davids et al., 1996)
Photon noise: ²~N (Poisson statistics)
Background with Moon. Use a perfect comparison star.
Take one month around 21st Dec.
Take an object that passes the zenith.
Observe all night with hsun<-18°.
Model of a perfect time-series:
10 sec.exposures
Scintillation noise
limited
Periodogram:
Same for Dome C:
Use same scintillation law
(probably much better!)
Zenith-passing object now z<30°
Observe at hsol < -12°
51092 vs. 98692 measures:
Periodogram:
Detection probability:
The geographic uniqueness alone offers
profound advantages over low-latitude
sites for time-series observations.