Procedures for astronomical observations
Although some observing procedures may depend on the program being carried out, others are
essentially dictated by the type of data being collected. This is the case in particular
for spectropolarimetric studies, in which very small amplitude
signals (ranging typically from about 1% of the unpolarised continuum for the
largest signals down to about 10ppm for the smallest ones) are usually being looked for.
In this case, it is important to
minimise all sorts of spurious signatures that can plague the data being collected.
Optimally, one would need to record the spectra associated to orthogonal states of a given
both simultaneously (to avoid mistaking polarisation signatures with temporal variations)
and at the same place on the detector (so that pixel to pixel differences do not affect the
results). Since this is obviously impossible, the solution we adopt is to regularly swap
the role of both beams within the instrument by rotating waveplates between exposures.
This way, we make sure that both polarisation states are collected simultaneously
(although on different detector regions) within each exposure; we also ensure that the
same region of the ccd detector records both polarisation states (although not simultaneously)
to minimise all errors resulting from flat fielding procedures. This compromise, although
not ideal, has the obvious advantage of getting rid of all
systematics at first order.
This method is also useful to minimise errors caused by slight waveplates imperfections,
and in particular to correct at first order all crosstalk
between circular and linear polarisation states.
In practice, this solution consists in dividing each polarisation exposure in a
series of 4 subexposures, each taken in a different waveplate
configuration. Polarisation information
is then obtained by processing the complete series of 4 subesposures with the specific reduction
tools, while unpolarised spectra can be derived by individually processing each of the four
subexposures. These observing procedures are implemented in the
instrument control software of ESPaDOnS as scripts, chaining
automatically waveplate settings for individual subexposures along with ccd
exposure and readout tasks.
Similar procedures can be used for scientific programs interested in measuring very
small signals whose origin is not polarisation but rather temporal variations, such
as small spectral variations induced by, eg, atmospheric pulsations, wind phenomena,
activity cycles or extrasolar planets. Although the details of the observing procedure
are different, the basic principles remain the same and aim at minimising all spurious
signatures in the collected data. Such procedures are not implemented yet, but could
be added later on specific requests from users.
Similarly, it is important to run sequences of calibration exposures to ensure that
everything is setup properly for collecting stellar exposures and reducing them in
real time. Such calibration sequences are usually taken once before sunset,
and a second time after sunrise (to keep night time for stellar exposures).
A typical calibration sequence includes at least the following
Optional (and recommended) calibration exposures to be added to the series are:
(null exposure time) to evaluate the magnitude of the ccd readout noise;
(illumination from Th/Ar lamp) to determine the details of the ccd pixel to
- a series of
ten flat fields
(composite illumination from 2 halogen lamps with associated filters)
for correcting pixel to pixel response differences.
Unless radial velocity information at a precision higher than 50m/s is required, it
is not necessary to collect comparison frames throughout the night; using the
numerous telluric lines present in the collected stellar frames is usually enough to
correct for potential spectral drifts (caused mainly by thermal
and pressure fluctuations) across the night with an accuracy of a few tens m/s.
- one fabry perot
exposure to estimate the shape of the slit formed by the image slicer at
spectrograph entry with a better accuracy than with a comparison frame;
- one dark frame (no
illumination with exposure time similar to that of stellar exposures) to evaluate
the amount of background level in a typical stellar exposure;
of check exposures (with polarised Q=1 or U=1 illumination and given waveplate
configurations) to verify that the polarimetric analysis is behaving as expected;
- additional series
of flat fields in case the scientic program involves observing very bright stars
at extremely high signal to noise ratios.
Scripts designed for carrying out automatically such sequences of calibration exposures
are already implemented in the instrument control software and can be started with one
single command line or with just a few clicks.
© Jean-François Donati, last update May 24 2004