Information for Proposers
- Instrument Description
- Proposal Guidelines
and Restrictions
- Mailing List
- Raw Sensitivity
- Observing Modes and
Scan Strategy
- Pointing and
Calibration
- Observing Efficiency
- Data Archiving
- Analysis Software
- Revision History
Back to BolocamWebPage
Instrument Description
Bolocam, a large-format bolometric camera for observations at 1.1 and
2.1 mm, is open to proposals. The camera will in general be
available for only one of these wavelengths in any given
semester. The camera has 115 working pixels
with 8 arcmin FOV (approximately circular). The beam FWHM is 30
arcsec at 1.1
mm and 60 arcsec at 2.1 mm. At all wavelengths, the pixel spacing
(nearest neighbors of hexagonal close-packed array) is 38 arcsec.
wavelength unless otherwise indicated:
1.1 mm
Proposal Guidelines
and Restrictions
Proposals that overlap the instrument team's key projects (blank-field
surveys at 1.1 mm for dusty extragalactic point sources, blank-field
surveys at 2.1 mm for clusters using the Sunyaev-Zeldovich effect, and
pointed observations of galaxy clusters for dusty extragalactic point
sources and the Sunyaev-Zeldovich effect) will only be considered if in
collaboration with or approved by the instrument team.
All other science topics are open to all proposers, though
collaboration with the instrument team is strongly encouraged.
While analysis software will be made available publicly via this web
site, the lack of funding for support limits the instrument team's
availability for providing technical support on the software to those
projects with instrument team members involved as collaborators.
Prospective proposers are encouraged to contact members of the
instrument team with questions or to investigate possibilities for
collaboration:
Jamie Bock (jjb@astro.caltech.edu)
Jason Glenn (jglenn@origins.colorado.edu)
Sunil Golwala (golwala@astro.caltech.edu)
Phil Mauskopf (philip.mauskopf@astro.cf.ac.uk)
Prospective proposers should endeavor to make such contacts or requests
well in advance (> 1 month) of proposal deadlines; the instrument
team cannot guarantee immediate response to such requests.
Mailing List
As of March, 2004, we have started a mailing list for Bolocam
proposers/users who would like to be notified of major web page
updates, especially with regard to sensitivity calculations and
analysis software. Please contact href="BolocamSupport.html">the
Bolocam support person
if you would like to be added to this list.
Raw Sensitivity
Due to the short history of observations with Bolocam, exhaustive
sensitivity numbers are not available. We give below those
numbers
that are available.
At 1.1 mm:
Observations taken during the early 2003 El Nino give the following
per-pixel instantaneous sensitivity percentiles at 1.1 mm:
unchopped mode
25%: 68 mJy √sec
50%: 78 mJy √sec
75%: 92 mJy √sec
chopped mode:
25%: 102 mJy √sec
50%: 112 mJy √sec
75%: 167 mJy √sec
Cumulative distributions are shown here (NOTE: the horizontal axis must be scaled by
3.4!): pdf
The chopped observations took place partially after the end of the El
Nino and give more representative sensitivities. For the purpose
of
proposals, we suggest the use of the following 1.1 mm sensitivities:
unchopped mode: 100 mJy √sec
chopped mode: 143 mJy √sec
For survey mode, one can calculate a mapping speed from
MS = Npix Ωbeam / sensitivity2
Npix = 115
Ωbeam = 0.32 arcmin2
giving (approximately)
unchopped mode: 13 arcmin2
hr-1 mJy-2
chopped
mode: 6 arcmin2 hr-1
mJy-2
The above results are based on an aggressive sky subtraction. For
"blank-field" data, this aggressive subtraction seems to have a
noticeable but small effect on source fluxes (~20% currently).
For bright and/or extended sources, such an aggressive sky removal
cannot be applied if accurate recovery of flux from bright and/or
extended sources is desired. A much less aggressive sky
subtraction preserves such structures but also results in poorer
sensitivity. Therefore, we ask that proposers use the following
mapping speeds depending of their type of observation and observing
mode:
- unchopped, blank-field: 10
arcmin2
hr-1 mJy-2
- unchopped, bright and/or extended objects: 1 arcmin2
hr-1 mJy-2
- chopped: 5 arcmin2
hr-1 mJy-2
We quote the same number for all kinds of fields when chopping because
we assume the proposer is uninterested in extended structures because
chopping will remove them.
Observations not taken in survey mode should begin with the above
instantaneous sensitivity numbers and then correct for the number of
pixels on-source
at any given time as discussed below.
Regardless of observing mode, proposers should take account of various
observing efficiency factors listed below
in estimating the amount
of time needed for a given observation.
Observing Modes and
Scan Strategy
Observations with bolometric cameras require some sort of temporal
modulation of the astronomical signal in order to separate such signal
from sky noise and instrumental 1/f. This need is reflected in
the available scanning modes.
- simple raster scanning,
unchopped: If chopping is not employed, the telescope must be
scanned sufficiently quickly that sky noise is mitigated to the point
where it can be subtracted using the array format. As noted
above, good sensitivities are achieved at 1.1 mm when a scan speed of
60 arcsec/sec is used. We do not recommend changing the scan
speed
until the sensitivity as a function of scan speed has been
characterized. While scanning slower will result in better
observing efficiency, the degradation in sensitivity more than cancels
any such gains. Faster scanning has not been explored in detail;
the pointing stability will likely degrade at faster scan speeds and
the degradation has not yet been characterized. See the XRASTER_SCAN command in the CSO
manual
for available coordinate systems and options.
- raster scanning + chopping:
When the secondary chopper is used, much slower scan speeds are
possible because chopping removes the bulk of the sky noise.
However, the system remains unstable at DC and so the telescope must be scanned. Good
sensitivity has been demonstrated at a scan
speed of 5 arcsec/sec. Higher scan speeds in chopped mode are not
advised because the individual bolometer sampling of the target field
becomes
poorer. Slower scan speeds are susceptible to slow drift in the
chopper-synchronous offset. We recommend that drastically
different scan speeds not be used until the sensitivity at different
scan speeds has been characterized. Only azimuthal chopping
combined with azimuthal raster scanning has been used. Observers
may use other chop directions at their own risk. See the XRASTER_SCAN and SECONDARY commands for further
possibilities.
- jiggle mapping, a.k.a. chopping
+ nodding: Not currently available, may be implemented.
- more complicated scan modes:
The LISSAJOUS and BOX_SCAN modes used by SHARC-II are
available to
Bolocam but have not been characterized. The analysis software
also cannot deal with these modes yet. Observers may use them
at their own risk and with the understanding they may have to modify
the pipeline (substantially). Detailed information about these
scan modes can be found on the SHARC II web page, available via the CSO
main page, http://www.cso.caltech.edu.
Important additional notes on scan strategy:
- Sky subtraction: To first
order, sky subtraction removes the average signal seen across the
array. It thus attenuates or
removes all astronomical structure comparable in size to the FOV and
larger. We expect that more advanced analysis methods may
allow recovery of some of this information, but no guarantees can be
made at this point.
- Striping: For large
fields, observers should be aware that striping due to residual 1/f
can be a problem. As noted above, sensitivities have been
calculated for point-source searches; larger-scale structure will
likely be observed at poorer sensitivity. To mitigate the effects
of striping, observers should use cross-linked scan
strategies. These can consist of alt-az scans coupled with field
rotation, or equatorial-mode raster scans at two different
angles. The instrument team is developing optimal destriping
methods, but these can only work when the maps are cross-linked in the
above manner.
- Point-source photometry:
The most efficient observing mode for photometry of single sources
small compared to the FOV is a jiggle-map mode. Such a mode has NOT YET been implemented for
Bolocam, though we expect it will at some point. Instead of
jiggle-mapping, observers will have to employ either a fast, unchopped
raster scan or a slow, chopped scan.
Proposers must mention their
intended observing mode(s) so the CSO TAC
may judge whether the desired observations are consistent with the
constraints imposed by a bolometric camera.
Pointing and
Calibration
Pointing observations must
be done frequently with Bolocam. It is strongly suggested that
observers use multiple pointing sources near the field to be
observed (within 10 to 20 deg), checking pointing once every 2 hours or
so, more frequently if near
zenith and the field's local coordinates are changing quickly.
The pointing sources should be spread out so that the pointing on the
field of interest can be interpolated, not extrapolated, from the
grid. Pointing sources with fluxes of 2 Jy and higher are easiest
to use, though lower-flux sources can be used if necessary.
Pointing sources effectively become relative flux calibration
observations by dint of doing them frequently, but they must be tied to
primary or secondary flux calibrators at some point during your
observing.
Absolute flux calibration on a primary or high-quality secondary
calibrator
should be done once or twice per night, more frequently if observing
conditions are highly variable. Bolocam has continuous monitors
of sky loading (the bolometer resistances) that can be used to do a
running relative flux calibration correction, but sufficient data must
be
accumulated to measure check that this relationship maintains its
standard form. In general, your pointing observations will
serve this purpose.
A beam map/flux calibration of the full
array should be done once every few nights using a primary calibrator
such as Mars (Jupiter and Saturn are too bright) or a bright secondary
calibrator (Uranus, Neptune, or bright quasars or galactic
sources). Even for short (1-2 night) runs, at least one beam map
should be done.
Relative gains of the array elements can be monitored using sky noise.
More information on doing calibration observations is provided on the PreparingForObserving page.
Observing Efficiency
In calculating time requirements, five important observing efficiency
factors should be taken into account:
- Turnaround Efficiency
At high scan speeds and for small fields, a significant amount of
observing time may be spent in turnarounds. We have not optimized
the turnaround time, but we have found that the standard turnaround
time of 10 sec (5 to slow down, 5 to speed up) provides the
sensitivities given above. Shorter turnaround times may be
possible. To be conservative, observers should assume 10 sec
turnaround time and all turnaround time is lost. This contributes
an efficiency factor
etaturn =
(l/vscan) / [
(l/vscan) + tturn ]
where l is the
scan length, vscan
is
the scan speed, and tturn
is the turnaround time.
When in slow raster + chopping
mode, the efficiency is greatly increased due to the slow scan speed.
- "Sampling" Efficiency
The Bolocam array does not
instantaneously provide full sampling of the sky. The array has a
hexagonal close-packed format, with pixels spaced by 38 arcsec.
The beam FWHM is 30 arcsec at 1.1 mm and 60 arcsec at 2.1 mm, so the
effective pixel spacing is 1.3 f lambda and 0.6 f lambda, respectively. Full
sampling requires a rectangular array with 0.5 f lambda pixel spacing, so clearly
instantaneous full sampling is not achieved.
However, the camera can be rotated relative to the scan direction such
that full sampling is achieved after one full pass of the
array across a line perpendicular to the scan direction. While
the sky is fully sampled in
this mode, it is not necessarily uniformly
sampled due to the distribution of bad
pixels in the array. When the sampling is not uniform, the noise
will vary across the map. Observers must therefore realize that
one pass across the source will not suffice to provide a uniform
map! For most observers, who will make many scans across a field,
the scan strategy can be arranged to provide fairly uniform (to better
than a few percent) sampling after many scans. But observers
proposing to observe a field smaller than or comparable to the field of
view for only a short amount of time should consult the PreparingForObserving page to
determine whether they will need to observe a source for longer than
expected in order to build up a uniformly sampled map.
- "Mapping" Efficiency
The coverage in any scan mode
will decrease at the edges of the map because the entire array does not
sweep over the edges of the field defined by the scan strategy.
The field edges defined by the scan strategy correspond to the motion
of the center of the array. If the center of the array scans to a
certain point, only half the array has scanned past that point, so the
coverage is reduced to approximately 50% of its value at the center of
the field. The coverage will grade down linearly from 100% to 50%
over the last FOV/2
=
4' at each edge of the map. To
conservatively estimate this
inefficiency, observers should assume that an edge of width FOV/2 =
4' is lost around the edges of the map.
For example, for a square map of
size 30' x 30' (dimensions indicate the path that the telescope
boresight follows), the full coverage region will be only 22' x 22',
corresponding to a mapping efficiency of only 0.54. In general,
the mapping efficiency is
etamap = (lx
- FOV) (ly -
FOV) / lx ly
where lx
and ly are the
dimensions in the x and y directions of the scan. Alt-az rasters
over long periods (where field rotation has an effect) will yield
uniform coverage only over a circle whose diameter is the smaller of lx - FOV and ly - FOV,
corresponding to
lower mapping
efficiency. These formulae become inaccurate when lx and/or ly
become comparable to the FOV
or smaller.
- Point-Source Photometry Mode
Observations of a single object small compared to the
FOV, as opposed to operation in survey mode, can be quite
inefficient
because only a few pixels are on-source at any time. For
photometry of point sources, effectively only 1 pixel is on-source at
any given time and so one should assume the per-pixel instantaneous
sensitivities given above rather than the survey-mode mapping
speeds. For distributed sources small compared to the field of
view, one must similarly correct, calculating a mapping speed based
only on the number of on-source pixels.
- Pointing and Calibration
The time spent on pointing and
calibration should be included in calculation of observing time.
Typical pointing and flux
calibration observations only require 5 min + slewing time because they
only require the center bolometers of the array cross over the
source. Full-array beam maps
and flux calibration scans require large amounts of time (30 min in
unchopped mode, 60 min to 2 hrs in chopped mode!), but need only be
done once every few nights.
Data Archiving
See the DataArchiving page.
Analysis Software
An archive of the Bolocam analysis pipeline (written in IDL) with some
example execution scripts will be made
available via the AnalysisSoftware
page. Directions for operating the pipeline are provided there
also. A more detailed manual will be available in the
future. The pipeline is installed in the bolocam account on kilauea.submm.caltech.edu, one
of the summit computers, for use during your observing run.
We emphasize that this pipeline is
neither foolproof nor fully automated. While we expect the
pipeline to improve with time (and hopefully user contributions!), it
is important that proposers realize that bolometer camera data is,
unlike any other kind of astronomical data, a nontrivial exercise in
timestream signal processing and noise correlation analysis.
Proposers should judge whether the time investment is worth the
expected science return. The instrument
team will not be available for on-demand technical support except in
collaborative relationships.
Revision History
- 2003/10/20 SG
First Web version
- 2003/10/22 SG
Correct MAJOR error in mapping speeds!
Sensitivities were 3.4 times too good, mapping speeds were 10 times too
good!
- 2004/01/31 SG
Add link back to main page, add links to SCUBA calibrator pages,
correct error in mapping efficiency formula (have FOV/2 instead of FOV
in the numerator).
- 2004/03/18 SG
Post reduced mapping speed for bright/extended source observations, add
section on mailing list.
- 2004/04/07 SG
Move some of the Sampling Efficiency discussion to the
PreparingForObserving page, clarify details on Mapping Efficiency,
reduce turnaround time to 10 sec, other minor changes.
- 2004/05/09 SG
Link to DataArchiving page, modify AnalysisSoftware section.
Questions or
comments?
Contact the Bolocam support person.