Transcript spire_scanmap_intro

NHSC SPIRE Data Processing Webinars
8th Feb 2012
Overview of SPIRE
Photometer Pipeline
C. Kevin Xu (NHSC/IPAC)
page ‹#›
PACS
NHSC SPIRE Data Processing Webinars
8th Feb 2012
Goals:
• Show how SPIRE Photometer pipeline works (functionalities of major modules).
• Explain what is new in HIPE 8.
• Brief summary of remaining issues.
• Will concentrate on scan map “user pipelines” (covering small map, large map,
SPIRE/PACS parallel mode).
page ‹#›
PACS
NHSC SPIRE Data Processing Webinars
8th Feb 2012
User Pipelines
• User pipelines (Jython scripts): Simplified version of Standard
Product Generation (SPG) pipelines.
• You can find these “user pipelines” in HIPE:
page ‹#›
PACS
NHSC SPIRE Data Processing Webinars
8th Feb 2012
SPIRE Pipeline & Data Products
5’
(in HIPE)
Our focus
(“user pipelines”)
page ‹#›
PACS
NHSC SPIRE Data Processing Webinars
8th Feb 2012
Scan Map Pipeline Flow Chart
A “reversed sequence” relative
to the chain of data acquisition
place-holder
• FFT based
• Default: concurrent deglitcher
+ wavelet deglitcher
• Alternative: σ-κ deglitcher
FFT based
(run after deglitch
& repair
to avoid ringing)
• Telescope pointing &
orientation
• SPIRE detectors positions
Choices:
• Default: scan median baseline removal
Include:
• Advanced : polynomial baseline removal
(1) Non-linearity correction
(2) Volt to Jy/beam conversion• Most advanced: an iterative destriper
Remove 1/f noise
due to T-drift
Baseline
removal
+ Mapper
• Default: naïve mapper
• Alternative: madmapper
page ‹#›
PACS
NHSC SPIRE Data Processing Webinars
8th Feb 2012
Scan-By-Scan Processing
• The pipeline processes timelines scan by scan (to ease the demand on RAM).
• Problem: ringing at the two ends of each scan due to FFT based modules.
• Solution:
(1) Before the process, attaching “turn-around” data blocks to ends of the
scan.
(2) During the process, the ringing is confined to the “turn-around” data.
(3) After the process, cut-off the “turn-around” data blocks from the scan.
Uniform scan
speed distance
Turnaround
page ‹#›
PACS
NHSC SPIRE Data Processing Webinars
8th Feb 2012
Highlights of a User Pipeline
(Jython Script)
New in HIPE 8.0
(for extended source, optional)
page ‹#›
PACS
Signal “Jump”
NHSC SPIRE Data Processing Webinars
8th Feb 2012
• Sudden (spontaneous?) jump in the timeline of a single channel.
• Mostly seen in thermistors (see below). Very rarely seen in detectors.
• The average frequency is ~ 1/day ( a few hundred instances since launch).
• The cause is still unknown.
• Effect: The pipeline uses thermistor timelines in the correction for detector signal
drift due to bath temperature drift (major source of 1/f noise for SPIRE). A
thermistor “jump” affects this correction, introducing artificial stripes in the final map.
Jump
page ‹#›
Stripe caused
by the jump
PACS
NHSC SPIRE Data Processing Webinars
8th Feb 2012
New module
Jump Detector
(automatically
identify & mask
any thermistor
affected by a jump)
Thermistor Jump Solution (in HIPE 8.0)
• a “reversed sequence” relative to the chain of
data acquisition
• scan by scan processing
Destriper
+ Mapper
Exclude masked
thermistors
page ‹#›
PACS
NHSC SPIRE Data Processing Webinars
8th Feb 2012
Status of SPIRE Scan Map Pipeline
in HIPE 8.0.
SPIRE 3-color map of NGC 5315 (a planetary nebula)
• General assessment:
In many cases, data from HSA
are already science quality!
• The official calibration accuracy is
±7% (5% from model, 2% RMS).
• An example (on the right):
The image from HSA looks good.
Caveat: detailed inspections show
thin stripes (can be improved using
a more advanced baseline remover
such as the iterative destriper).
• A major issue yet to resolve:
stripes due to bias voltage
drift (the “burp”)
(affecting a few observations)
page ‹#›
(Public data taken from HSA)
PACS
NHSC SPIRE Data Processing Webinars
8th Feb 2012
Bias Drift in “Burp” Period
• Every time when SPIRE is switched on after a
cooler recycle, the first ~6 h sees a rapid rise
of bias voltage (the “burp”).
• This effect has not been corrected in pipeline.
• It causes stripes in maps observed
during the “burp” period (see example below).
• A pipeline module is now being developed to
correct this effect (available soon).
• For point source surveys, the stripes might be
removed by polynomial baseline removal.
An example of
stripes caused
by bias drift
Resistor
voltage

After cooler recycle 1
(~40 h)
Bias
voltage
After cooler recycle 2
(~40 h)
bias drift ~10σ
“burp” ~ 6h
(Standard pipeline product)
page ‹#›
PACS
NHSC SPIRE Data Processing Webinars
8th Feb 2012
Summary
• Scan-Map pipeline covers nearly all SPIRE PHOT AOTs (small map,
large map, map in SPIRE/PACS parallel mode).
• It follows a “reversed sequence” relative to the chain of data acquisition.
• For a general user, Level0_5 should be the best starting point.
• From Level0_5 to Level 1, the pipeline processes any observation data set
scan by scan (to ease the demand on the RAM).
• The current pipeline (HIPE 8.0.1) does a good job (“science ready”) in general.
• However, the pipeline still has some open issues: that
(1) Stripes due to the “burp effect” (affecting ~10% observations).
(2) Residual stripes (can be corrected by the destriper module in HIPE).
(3) Residual glitches in maps (2nd level deglitcher being developed inside
the destriper module).
page ‹#›
PACS
NHSC SPIRE Data Processing Webinars
8th Feb 2012
• Home work:
Run Photometer Large Map user pipeline (with your own data or NGC 5315 data).
• Next: A demo of SPIA (SPIRE Photometer Interactive Analysis), a GUI based
data reduction tool (including all pipeline modules and more).
page ‹#›
PACS