Spectral line

data reduction

Pedro Salas
GBO postdoc

GBT observer training workshop, February 16-18, 2022

Outline

Introduction to single dish calibration.

Tracking observations:
Data reduction in GBTIDL.
Mapping observations:
Data reduction with the gbtpipeline.
High-frequency calibration.
Other tools.
 A single dish telescope measures power:

(ν) = G(ν) [Tsou (ν) + Tsys (ν)]

[cal] [cal]
P
with
[cal]
Tsys (ν) = Tatm + Tspill + Tsw + Trx [+Tcal ]

we are after

Tsou (ν)

See e.g., Winkel+2012 & O'Neil 2002

 A single dish telescope measures power:

(ν) = G(ν) [Tsou (ν) + Tsys (ν)]

[cal] [cal]
P
with
[cal]
Tsys (ν) = Tatm + Tspill + Tsw + Trx [+Tcal ]

we are after

Tsou (ν)

See e.g., Winkel+2012 & O'Neil 2002

How do we get to Tsou (ν) from P [cal] (ν) ?
From P (ν) to Tsou (ν)

= G(ν) [Tsou (ν) + Tsys (ν)]

[cal] [cal]
Pon (ν)
We assume the off position is "empty":

G(ν) [Tsys (ν)]

[cal] [cal]
Poff (ν) =

[cal] [cal]
[cal] Pon −Poff
Tsou + ΔTsys = Tsys,off [cal]
Poff

See e.g., Winkel+2012 & O'Neil 2002

From P (ν) to Tsou (ν)

[cal] [cal]
[cal] Pon −Poff
Tsou + ΔTsys = Tsys,off [cal]
Poff

[cal]
we also need to know Tsys,off

−1
Tsys,off = Tcal [ Poff − 1]
cal
Poff

All quantities are frequency dependent

See e.g., Winkel+2012 & O'Neil 2002

From P (ν) to Tsou (ν)

−1
Tsou + ΔTsys = Tcal [ Poff − 1]
cal
Poff Pon −Poff
Poff

Changes to Tcal are linear in Tsou 😃

See e.g., Winkel+2012 & O'Neil 2002

Observing strategies

Position switching: Frequency switching:

1. On-Off 1. In band
2. Beam nodding 2. Out of band
3. Sub-beam nodding

Strategy will depend on source properties and science goals.

There are other options, and you can mix these alternatives.

See e.g., Winkel+2012 & O'Neil 2002

Position switching
The science target is the "On"
position.

POn (ν) = G(ν) [Tsou (ν) + Tsys (ν)]

[cal] [cal]

An "empty" region is the "Off"

position.

POff (ν) = G(ν) [Tsys (ν)]

[cal] [cal]

*It is better to keep the elevation constant

1 # In AstrID:
2 off = Offset("AzEl", -1.0, 0.0, cosv=True)
3 OnOff( source, off, scanDuration, beamName )
Position switching
The Off region should not have emission/absorption (the Off
region should be more than a beam away from your source).
The On-Off cycle should be faster than fluctuations in the
telescope's gain (for narrow features ~few minutes <10 GHz,
faster above).
Useful for:
Observations of broad (>100 km s−1 ) spectral lines.
Observations of sources with crowded spectrum.

Drawbacks:
Lost time slewing.
Differences in Pon and Poff produce residual baselines.
Frequency switching
The LO switches the frequency generating signal/reference pairs.
Frequency switching
The LO switches the frequency generating signal/reference pairs.
Frequency switching
The Off "region" should not have emission/absorption.
Be aware of the RFI environment (you do not want to
switch to a region with RFI).

Useful for:
Observations of narrow (≤10 km s−1 ) spectral lines.

Drawbacks:
Need to know source velocity a priori.
For larger Δν , larger residual baseline.
No continuum measurements.
What is Tcal ?
Is the equivalent temperature of a noise source injected to the signal path.
A note on Tcal
By default the metadata includes a scalar value for Tcal .
However,
It is a scalar (good approximation for some receivers).
You don't know when it was measured (the temperature
of the noise diodes drifts).

➔ Perform observations of a calibrator source!

A note on Tcal
By default the metadata includes a scalar value for Tcal .
However,
It is a scalar (good approximation for some receivers).
You don't know when it was measured (the temperature
of the noise diodes drifts).

Goddy+2020

➔ Perform observations of a calibrator source!

A note on Tcal
Calibrators should:
Have a known flux density at your observing frequency.
Be stable in time (or you should know its flux density at the time of your
observation, e.g. check the ALMA calibrator database).
Be point-like (or you should have a model of their brightness distribution).

Standard calibration sources and their properties:

Perley & Butler 2017
Ott et al. 1994
GBTIDL
GBO supported data reduction software.
Written in IDL.
Locally available: user@planck$ gbtidl
Supports spectral line data reduction.
Knows about most spectral line observing
modes, e.g., On-Off, Track, frequency
switching, beam nodding. GBTIDL -> getps, 1
GBTIDL -> getfs, 10

Code & documentation:

http://gbtidl.nrao.edu/
 GBTIDL: getting help

1 GBTIDL -> usage,'show' ; Lists optional arguments.

2 GBTIDL -> usage,'show', /verbose ; Describes the command.
3 GBTIDL -> usage,'show', /source ; Show source code.
 GBTIDL: data access

1 GBTIDL -> online ; To access the active project.

2 GBTIDL -> offline, 'AGBT16B_037_04' ; To access a project in /home/sdfits.
3 GBTIDL -> filein, 'mySDFITS.fits' ; To access an SDFITS file in
4 ; another location.
 GBTIDL: data containers
GBTIDL stores the data in containers (array like structures). There are 16 of
them, and the first (0) is called the primary data container (PDC).
1 ; Save the contents of the third row of an SDFITS
2 ; to variable `x`.
3 GBTIDL -> getrec,3 ; Read the third row into the PDC.
4 GBTIDL -> x = getdata() ; Copies the PDC into `x`.
5 ; Zero the first 500 elements of x.
6 GBTIDL -> x[0:500] = 0
Container arithmetic
1 ; Add data containers 10 and 11 and save into 12
2 GBTIDL -> add,10,11,12
3 ; Subtract data containers 10 and 11 and save into 12
4 GBTIDL -> subtract,10,11,12
5 ; Divide data containers 10 and 11 and save into 12
6 GBTIDL -> divide,10,11,12
 GBTIDL: observation information

1 GBTIDL -> summary ; Summary of loaded session.

2 GBTIDL -> header ; Metadata of container 0
3 GBTIDL -> list ; List the contents of each
4 ; row in the SDFITS.
 GBTIDL: baseline fitting

1 GBTIDL -> setregion ; Define region for baseline fitting.

2 GBTIDL -> nfit,3 ; Set polynomial order.
3 GBTIDL -> bshape ; Fit polynomials to selected ranges.
4 GBTIDL -> baseline ; Subtract baseline.
 GBTIDL: smoothing

1 ; Smooth spectrum in PDC by 5

2 ; channels, keeping every 5th
3 ; channel.
4
5 ; Using a Gaussian kernel.
6 GBTIDL -> gsmooth,5,/decimate
7 ; Using a boxcar kernel.
8 GBTIDL -> boxcar,5,/decimate
 GBTIDL: Gaussian fitting

1 ; Fit a Gaussian to the spectrum

2 ; on display. You will specify the
3 ; region to be fitted and starting
4 ; guesses using the GUI.
5 GBTIDL -> fitgauss
 GBTIDL: example 1
offline, 'TGBT21A_501_11' ; Connect to project TGBT21A_501 session 11.
summary ; List the contents of the SDFITS file.
getps,152 ; Get position switched data for scan 152.
gsmooth,5,/decimate ; Smooth the data using a Gaussian kernel 5 channels wide.
 GBTIDL: example 2
; Load a SDFITS file from your scratch area.
filein,'/home/scratch/psalas/obs_training/tutorials/rrl/AGBT20A_415_01.raw.vegas.fits'
summary ; List the contents of the SDFITS file.
for s=18,22,4 do begin for p=0,1 do begin getps,s,ifnum=1,plnum=p & accum & endfor & endfor
ave
; Fit and remove a baseline over line free channels.
setregion
nfit,1 ; Use an order 1 polynomial.
bshape ; Display the best fit.
baseline ; Remove the baseline.
; Now fit a Gaussian to the line.
fitgauss
; Save the spectrum on display.
fileout,'H93a.fits'
keep
 GBTIDL: example 2
; Load a SDFITS file from your scratch area.
filein,'/home/scratch/psalas/obs_training/tutorials/rrl/AGBT20A_415_01.raw.vegas.fits'
summary ; List the contents of the SDFITS file.
for s=18,22,4 do begin for p=0,1 do begin getps,s,ifnum=1,plnum=p & accum & endfor & endfor
ave
; Fit and remove a baseline over line free channels.
setregion
nfit,1 ; Use an order 1 polynomial.
bshape ; Display the best fit.
baseline ; Remove the baseline.
; Now fit a Gaussian to the line.
fitgauss
; Save the spectrum on display.
fileout,'H93a.fits'
keep
Mapping
The telescope "scans" an area while tracking the map center.
Mapping
There are different mapping patterns/strategies.

On-the-fly (see e.g., Mangum+2000):

RaLongMap, DecLatMap
Daisy
Grid:
PointMap
Others:
Spider
Z17
Mapping
If there are regions free of
emission in the mapped area, you
can use these as Off positions.

O'Neil 2002
 GBTGRIDDER
Observatory supported gridding software.

1 gbtgridder -h # get help on input options.

2 # Example:
3 gbtgridder --channels 81:4014 --average 1 \
4 --output mycube \
5 my_reduced_sdfits.fits another_reduced_sdfits.fits

Source code at: https://github.com/GreenBankObservatory/gbtgridder

GBT pipeline
Calibrates and grids the data.
It uses the gbtgridder for gridding.

Works for receivers with noise diodes.

Designed for processing KFPA observations.
Default values suitable for KFPA observations.

Documentation:
https://safe.nrao.edu/wiki/bin/view/GB/Gbtpipeline/PipelineRelease
Source code:
https://gbt-pipeline.readthedocs.io/en/latest/#
 GBT pipeline
Some options:
-i # Input SDFITS.
-m # Mapping scans.
--refscan # Scans to use as reference.
-w # Spectral window to process.
-c # Channels to grid.
--beam-scaling # Multiply Tcal by this value.
--imaging-off # If you do not want to grid.

Example:
gbtpipeline -i my.sdfits.raw.vegas -m 14:24 --refscan 13,26
Map scans: 14 to 24, reference scans 13 and 26
 GBT pipeline: example 1

1 gbtpipeline -i "/home/scratch/dfrayer/DATAdemo/TGBT17A_506_11.raw.vegas" \
2 -m "14:26" --refscan 27 -w 0
 Questions?

Send feedback to: [email protected]

[email protected]
[email protected]

The Green Bank Observatory is a facility of the National Science Foundation

operated under cooperative agreement by Associated Universities, Inc.
Hot & cold loads
ARGUS and the W-band receiver do not use noise diodes, they use hot and cold
loads.
The W band receiver can see two loads: Tamb & Tcold .
ARGUS can see one load : Tamb

W band receiver calibration wheel.

Hot & cold loads
W-band

(Tamb −Tcold )
G= (Pamb −Pcold )

Tsys = GPoff
Pon −Poff
Ta = Tsys Poff

Frayer 2019, GBT memo #302

Hot & cold loads
ARGUS
∗ Pon −Poff
Ta∗ = Tsys Poff
∗ Tcal
Tsys (t) =
( P −1)
Pamb
off

Tcal ≃ (Tatm − Tbg ) + (Tamb − Tatm )eτ0 A

Tcal ≈ Tamb
GBTIDL data reduction scripts: /home/astro-util/projects/Argus/PRO
Frayer 2019, GBT memo #302
Temperature scales
Ta : Antenna temperature.
Ta′ =Ta eτ0 A : Antenna temperature corrected for atmosphere.
′
Ta∗ = Tηal : Forward beam brightness temperature.
Ta′
Tmb = ηmb : Main beam antenna temperature.

For the GBT:

ηl ≃ 0.99, GBT memo #16 & #19
ηmb = 0.44 ± 0.04 @ 86 GHz, GBT memo #302
ηmb = 0.94 @ 5 GHz
SDFITS

Single Dish FITS files.

Default data i/o for GBTIDL and gbtgridder.
VEGAS and DCR data stored as SDFITS.
Definition: https://fits.gsfc.nasa.gov/registry/sdfits.html
Details: https://safe.nrao.edu/wiki/bin/view/Main/SdfitsDetails

Note: even with a well documented definition there are multiple flavors of
SDFITS, e.g., Parkes vs GBT.
Community developed
data reduction tools
GAS: KFPA ammonia survey (https://gas.readthedocs.io/en/latest/)
DEGAS: ARGUS survey (https://github.com/GBTSpectroscopy/degas)
TMBIDL: general use (https://github.com/tvwenger/tmbidl)
groundhog: general use (https://github.com/astrofle/groundhog)
SDgridder: gridder (https://github.com/tvwenger/sdgridder)
HCGrid: gridder (https://github.com/HWang-Summit/HCGrid)
sdpy: ??? (https://github.com/keflavich/sdpy)

"It takes a community to develop robust data reduction tools" - ancient proverb
 Questions?

Send feedback to: [email protected]

[email protected]
[email protected]

The Green Bank Observatory is a facility of the National Science Foundation

operated under cooperative agreement by Associated Universities, Inc.