Observing
|
Information for Proposal Applicants
Further information
Questions regarding this page should be directed to
Kentaro Aoki ( ).
Observation procedure
- Pointing the telescope, rotating the instrument, checking the focus, & configuring spines
- Target field acquisition by making corrections to telescope
pointing and instrument rotator angle
The typical overhead to complete 1. and 2. and get ready to start
exposures is ~20 minutes, which is usually dominated by the
spine configuration time. Spine configuration can start with slewing
the telescope and rotating the instrument. The overhead will be longer
when the telescope pointing is involved with a large amount of dome
rotation and/or when the next target field is still at EL<30 degree:
Since the spine configuration should be executed at a telescope
elevation NO LOWER THAN 30 degree, the telescope elevation is
temporarily raised to 40 degree to do a spine configuration.
- Start auto guiding, and take exposures
Notes for operation:
- There are three observing methods available, Normal beam switching (NBS), Cross beam switching (CBS), and Point & stare.
- Normal beam switching --- In this mode, the
telescope is offset between "ON" and "OFF" positions, where
the fibers look at objects (sky) at the "ON" ("OFF")
position, respectively. Half of the observation time will be
spent for sky exposure.
- Cross beam switching --- In this mode, two
fibers are allocated to one object, and the telescope is
offset between two positions so that either of the two
fibers observes the object and the other observes sky. The
advantages of this method are: (1) 100% of the time can be
spent observing objects, and (2) at least in principle, sky
subtraction is not affected by time variation of sky
brightness. The disadvantage is that the maximal number of
spines allocated to objects is 200. Since the geometrical
constraint to spine allocation is strong, the actual number
of allocated spines could be even smaller in
reality.
- Point & stare --- There is no telescope offset
in this mode. Instead, some fibers need to be placed on
blank sky region and the average of the sky spectra is
applied to other fibers for sky subtraction. For example,
when most of the objects are relatively bright and the
accuracy of sky subtraction is not extremely critical, this
mode can be used.
- Operations of IRS1 and IRS2 are independent. For example,
observers can set up IRS1 for LR and IRS2 for HR if wanted. Also, as
long as the telescope stays at a certain position and the fiber
configuration stays the same, exposures by IRS1 and IRS2 do not need
to be synchronized: Duration of individual exposure and number of
exposures can be chosen independently.
- As indicated in the "basic instrument
parameters" page, currently the readout noise of the detector in
the spectrographs is not very low. In a Correlated Double
Sampling (CDS) readout, this noise will have to be fully
included to every output frame. However, the noise level can be
significantly reduced by exploiting non-destructive readouts
(i.e. ramp sampling): Given an observer wants to take an 1800
sec exposure, the pixel count can be measured N=(1800/(minimum
exposure time)) times in the process of exposure and the final frame
can be obtained by a linear fit to the N data points. This is
expected to reduce the readout noise by a factor of sqrt(N).
Observers are therefore recommended to use this ramp sampling
unless they need to repeat very short exposure.
!! Important !! : Corrections to the spine
positions during long integration
Unfortunately, even if the instrument is rotated as necessary, the
positions of objects on the focal plane gradually change as time goes
by for several reasons as below:
- Since the rotator axis is slightly misaligned with the optical
axis of the telescope main mirror, the pattern of field distortion
on the focal plane rotates as the instrument rotates.
- Strength of differential atmospheric dispersion effect changes
as the telescope elevation changes.
- Plate scale changes, e.g., when the truss temperature changes,
and subsequently the distance between the telescope primary mirror
and wide-field corrector in the PIR changes.
Consequently, before flux loss from fibers starts being significant
due to a large displacement between fiber and object positions,
the spine positions need to be re-configured at a regular interval
to keep observing the same objects with the same spine
configuration. In the recent engineering observations, it has
been confirmed that re-configuring the spine positions every 30
minutes enables the object flux to be reasonably constant. Each
re-configuration process takes about 10 minutes. The typical
observation efficiency in long integration is ~60% (i.e. the overhead
is ~40%), while the efficiency gets lower for shorter integration
because the contribution of initial configuration time (20 minutes)
becomes more significant.
As of Feb 2011, we are still in the process of characterization and
optimization and this frequency of re-configurations may be reduced in
future as the instrument characteristics are better understood.
However, the observers are strongly recommeneded to follow this
sequence (i.e. executing the re-configuration process every 30
minutes) for long integration.
Data acquisition for flat fielding and calibration
- Flat fielding
Dome-flat frames are taken in the evening and/or morning by
closing the dome and using the dome-flat lamp. In doing this, the
Echidna spines are configured to the same as for a scientific
exposure. If observers have more than one target field/spine
configuration to observe, then they would need to take the same
number of sets of dome-flat frames with the spine configuration
changed as appropriately.
- Wavelength calibration
A Th/Ar lamp is available in the buffle structure of the
tertiary mirror pointing to the focal plane at the prime focus.
Using this lamp, CAL frames are taken in the evening and/or morning
(spine configuration at the time of this data acquisition should not
matter for the accuracy of the calibration). A line list will be
provided on this web site in the near future.
- Telluric absorption correction and flux
calibration
What has been usually done so far is to assign a few fibers to
observe faint stars simultaneously with science targets. A guideline
of the brightness of stars for this method is 15-18 mag (AB) in
JH. The brighter limit of this range is set to minimize the effect
of ghost features after the typical exposure time of an individual
frame (i.e. 15 min), which tends to appear at a
three-orders-of-magnitude (i.e. 7.5 mag) fainter level than the
original brightness. The fainter limit is set to keep S/N of these
stellar spectra high enough for calibration. The spectral types
preferred are F, G, and K early dwarfs (A stars can be handled by
the reduction package but are not recommeneded). Broad-band colors
are expected to be useful to select them in advance, while it is
also possible to estimate spectral type from observed spectrum,
given the instrument throughput.
In theory, such correction/calibration is possible if one star is
observed per spectrograph, but it is strongly recommended to observe
a few to several stars possibley of different types so that the
result of the correction/calibration can be cross-checked.
An alternative method is to observe a standard star in a different
field before and/or after observing a science target field.
Observers would need to prepare an S2O file for a standard star
observation separately from those for science target field
observations, where field center, CCS, & GS are necessary in the
same way as science fields. This method would be recommended
e.g. when a target field is at a low Galactic latitude and standard
stars available there as well as science targets are highly affected
by Galactic extinction.
Operation of high resolution mode
In the High Resolution mode (HR), the FMOS spectral coverage (0.9-1.8
μm) is divided into four pre-defined bands ("J-short", "J-long",
"H-short" & "H-long") with a band width of ~0.25 μm, and one of
them is observed at one exposure with a spectral resolution of ~5A.
Please check this page for the basic
parameters and this page for the
sensitivity information.
There will be a few restrictions to the HR
operation as follows, and applicants/observers should keep
them in mind about the use of HR:
- The FMOS spectral coverage (0.9-1.8 um) is divided into four
bands in HR: J-short (0.92-1.12 μm), J-long (1.11-1.35
μm), H-short (1.40-1.60 μm) & H-long (1.60-1.80
μm), and one of them is covered by a single
exposure. HR observation is carried out by
using these pre-defined bands only.
- It takes about one hour to chnage one mode to another one
(e.g. LR → HR, HR → HR but a difference band).
To minimize the additional overhead, we recommend to keep the same setting of IRS1 & IRS2 from
beginning to end of a night.
- It would be acceptable to use IRS1 and IRS2 in different
observation modes (e.g. LR in IRS1 and J-long in IRS2),
while it is strongly recommended to use IRS1 and IRS2 in the
same mode to accurately position Echidna spines/fibers: Due
to the effects of atmospheric dispersion and chromatic
aberration, the position of an object on the focal plane is
slightly different as a function of wavelength, and the
spine-to-object (s2o) allocation software cannot allocate spines
according to different requests (e.g. observation wavelength) to
spines for IRS1 and those for IRS2. This means, if one tried
observation in J-long with IRS1 and in H-long with IRS2 and set
observation wavelength for spine allocation to 1.3 μm, more
light would tend to be lost at the fiber entrance for the H-long
observation.
Proposal checklist
- There are three observing methods available and the
main differences are summarized in a table below. Have you chosen one
for your observation with them considered?
Observing method |
Fraction of fibers for science targets |
Fraction of on-source integration time (1) |
Fraction of on-source integration time (2) |
Normal beam switching |
~100% |
50% |
~30% |
Cross beam switching |
~50% |
100% |
~60% |
Point & stare |
~80-90% |
100% |
~60% |
(1): This percentage represents 100 x
([On-source Integration Time]) / [Total Integration Time]).
(2): This percentage represents 100 x
([On-source Integration Time]) / [Total Observing Time (including overhead)]).
- [Operational mode] From S12A, the High
Resolution mode (HR) as well as the Low Resolution mode (LR) are
available on both IRS1 and IRS2.
- [Number of fibers] There are 12 fibers that
are not available for observation of science target, mainly because
the spines move not well enough (so most of them are still useful to
take sky spectra). Therefore the total number of fibers available to
observe science target is 388.
- [High Resolution mode] In the operation of
HR, a few important restrictions will be applied. Applicants should
take the following notes into account in making proposals:
- The FMOS spectral coverage (0.9-1.8 um) is divided
into four bands in HR: J-short (0.92-1.12 μm), J-long
(1.11-1.35 μm), H-short (1.40-1.60 μm) & H-long (1.60-1.80
μm), and one of them is covered by a single exposure.
HR observation is carried out by using these
pre-defined bands only.
-
It takes about one hour to chnage one mode to another one
(e.g. LR → HR, HR → HR but a difference band).
To minimize the additional overhead, we recommend to keep the same setting of IRS1 & IRS2 from
beginning to end of a night.
- It would be acceptable to use IRS1 and IRS2 in
different observation modes (e.g. LR in IRS1 and J-long in IRS2),
while it is strongly recommended to plan to use IRS1 and IRS2 in
the same mode to accurately position Echidna spines/fibers: Due
to the effects of atmospheric dispersion and chromatic aberration,
the position of an object on the focal plane is slightly different as
a function of wavelength, and the spine-to-object (s2o) allocation
software cannot allocate spines according to different requests
(e.g. observation wavelength) to spines for IRS1 and those for
IRS2. This means, if one tried observation in J-long with IRS1 and in
H-long with IRS2 and set observation wavelength for spine allocation
to 1.3 μm, less flux would fall into the fiber for the H-long
observation.
- [Overhead estimation] Have you calculated the
total amount of time to request with taking overhead into
account appropriately? Typically, ~20 minutes is necessary for
telescope pointing, dome rotation, Echidna spine configuration, target
field acquisition, and focusing. In addition, for long
integration, Echidna spine re-configuration which takes ~10
minutes is necessary every ~30 minutes. Resultantly, the
typical observation efficiency in long integration is ~60%
(i.e. the overhead is ~40%).
[See this page for more
details.]
- [Magnitude zeropoint] If you justify
on-source integration time based on a certain magnitude in the
proposal, have you clarified which zeropoint, Vega or AB,
should be applied to the magnitude? In NIR, the difference between
Vega mag and AB mag against a given flux density is 1-2 mag, which is
significant.
- [Guide stars and coordinate calibration
stars] In the target fields, are there a number of bright stars
suitable for guide stars (R = 12-16 mag) & coordinate
calibration stars (R = 12-15 mag) with accurate relative
astrometry to science targets?
[See this page for more details.]
- [OH mask] Is the spectroscopic feature you
need to observe expected to be robust enough against the effect of OH
masks? You can check the list of OH masks
here.
Also, FMOS spectrum simulator may be
helpful to see the effect.
Last updated: July 23, 2012
|