spire fourier-transform spectrometer basicsherschel.esac.esa.int/.../ivan_fts_dpw_feb2012.pdf ·...
TRANSCRIPT
PACS
HSC DP workshop 20-24 Feb 2012
SPIRE Fourier-Transform Spectrometer Basics
Ivan Valtchanov (HSC) with the help Ed Polehampton, Ros Hopwood, Nanyao Lu
and the SPIRE ICC and the NHSC
PACS
HSC DP workshop 20-24 Feb 2012
Page 2
Outline • FTS basics and observations • Point-source pipeline • Mapping observations • Tentative agenda for Thursday hands-on
session
PACS
HSC DP workshop 20-24 Feb 2012
Page 3
Intro
• Please read the SPIRE Data Reduction Guide!
PACS
HSC DP workshop 20-24 Feb 2012
Page 4
SPIRE Spectrometer Fourier Transform Spectrometer (FTS): The entire spectral coverage of 194-671 µm (447-1545 GHz) is observed in one go!
(SMEC)
(194-313 µm)
(303-671 µm)
NOT USED! now at 4-5K
PACS
HSC DP workshop 20-24 Feb 2012
Page 5
Two Bolometer Detector Arrays
194 – 313 µm 1545 – 960 GHz
303 – 671 µm 990 – 447 GHz
Beam = 17”- 21” Beam = 29”- 42”
PACS
HSC DP workshop 20-24 Feb 2012
Page 6
SMEC movement (3.9 cm)
PACS
HSC DP workshop 20-24 Feb 2012
Page 7
Fourier Transform: Interferogram to Spectrum Interferogram I(x)
Source Spectrum B(σ)
Discrete Fourier Transform: B(σ) = ∑i I(xi) exp(-i2πσxi)
PACS
HSC DP workshop 20-24 Feb 2012
Page 8
Interferogram in real life
One HR scan
PACS
HSC DP workshop 20-24 Feb 2012
Page 9
Real World: Finite Interferogram
Fourier Transform
Multiplied by a top hat function
B(σ) ⊗ sin(πσ/Δσ)/(πσ/Δσ)
For unresolved lines: § A sin[π(σ-σ0)/Δσ]/(π(σ-σ0)/Δσ); § Flux = A Δσ; § FWHM = 1.207 Δσ. (Δσ = spectrometer resolution element)
SINC profile
PACS
HSC DP workshop 20-24 Feb 2012
Page 10
Spectral resolution
The spectral resolution, Δσ, depends on the maximum SMEC scan length L: Δσ = 1/(2L), where L =12.8 cm for High Res.
FTS scanning
PACS
HSC DP workshop 20-24 Feb 2012
Page 11
Sparse
Intermediate Full
2 beam spacing
1 beam spacing (4 jiggle posi3ons)
1/2 beam spacing (16 jiggle posi3ons)
FTS spatial sampling modes
Point source spectroscopy (SOF1):
Jiggle Mapping (SOF2): Raster Mapping (SOF1 or SOF2):
2’ circle
PACS
HSC DP workshop 20-24 Feb 2012
Page 12
Observing Modes
Telescope pointing: Single Pointing
Raster
Spatial sampling using an internal jiggle mirror (BSM): Sparse (2 beam spacing)
Intermediate (1 beam spacing)
Full (1/2 beam spacing)
Spectral resolution: High Res (MR): 1.2 GHz; R= 1290 – 370; ΔV = 230 – 800 km/s; Suitable for, e.g., line fluxes.
Medium Res (MR): 7.2 GHz; R = 210 – 60.
Low Res (LR): 25 GHz; R = 62 – 18, Suitable for dust continuum.
High + Low (H+L): Both High and Low scans.
Spectral resolution, Δσ, depends on the maximum SMEC scan length L: Δσ = 1/(2L), where L =12.8 cm for High Res.
L
FTS scanning
PACS
HSC DP workshop 20-24 Feb 2012
Page 13
PACS
HSC DP workshop 20-24 Feb 2012
Page 14
SPIRE FTS Observations Each observation is divided into individual Building Blocks:
§ Observations of sparse sampling:
§ All other observations (i.e., intermediate or full spatial sampling or raster):
Initialization
Move BSM
PCAL flash
FTS scans Initialization
End
PCAL flash
+
LOOP OF 1, 4 or 16
At each of the multiple telescope pointings (if a raster map):
Move BSM FTS scans
+
+
+ End
Your data is taken here!
Pipeline data processing just loops through individual FTS scan BB’s.
PACS
HSC DP workshop 20-24 Feb 2012
Page 15
FTS pipeline flowchart SPIRE DRG Figure 6.31
PACS
HSC DP workshop 20-24 Feb 2012
Page 16
Level 1 products: • unmodified interfergrams • average spectrum (apodized) • average spectrum (unapodized)
Interferograms (stored in Level 1)
Level 0.5 products: • detector time lines • scan mirror time line • house keeping time lines
The FTS Pipeline – Overall Flow Chart SPIRE Common Pipeline
1. Modify Detector Timeline V(t)
2. Create Interferogram V(x)
3. Modify Interferogram V(x)
4. Fourier Transform V(x)->S(σ)
5. Modify Spectra S(σ) (extended source flux cal.)
6. Create Level-2 products
Spectral cube if a mapping obs, or a point-source spectrum for the central detectors (SSWD4/SLWC3)
PACS
HSC DP workshop 20-24 Feb 2012
Page 17
1st Level Deglitching
Clipping Correction
Time-domain Phase Correction
Bath Temperature Correction
V(t)
V(t)
V(t)
Pipeline Step 1: Modify Timelines
V(t)
V(t)
Level 0.5 Timelines
Modified Level 0.5 Timelines
Non-linearity Correction
V(t)
Bolometer Nonlinearity Table
Bolometer temp. drift corr. product
Detector/electronic time constants
V(t), Voltage as a function of time
PACS
HSC DP workshop 20-24 Feb 2012
Page 18
Create Interferograms Once time domain processing is complete, the detector signals and SMEC positions can be merged to create interferograms. The created “unmodified” interferograms are also stored in Level 1 in case users want to do their own interferogram-to-spectrum process.
V(t)
Pipeline Step 2: Create Interferograms
Level 0.5 Timelines
V(t)
Unmodified Interferograms V(x)
(Stored in Level 1)
SMEC Positions
x(t')
Pointing P(t'')
PACS
HSC DP workshop 20-24 Feb 2012
Page 19
Baseline Removal
2nd Level Deglitching
Phase Correction
(Default apodization)
V(x)
V(x)
V(x)
Pipeline Step 3: Modify Interferograms
V(x)
V(x)
(Level 1) Interferograms
Modified Interferogram Products (both unapodized and apodized)
Correct for asymmetry between the two optical paths in the interferometer
Norton Beer Order-1.5 function
PACS
HSC DP workshop 20-24 Feb 2012
Page 20
Fourier Transform
Apply the Fourier Transform to each interferogram to create a set of spectra for each spectrometer detector. The spectra are in units of V/GHz, not yet flux calibrated.
Pipeline Step 4: Fourier Transform
Spectra
Modified Interferograms
V(x)
PACS
HSC DP workshop 20-24 Feb 2012
Page 21
Signal in Spectral Domain
@80K @4-5K
PACS
HSC DP workshop 20-24 Feb 2012
Page 22
Instrument Background Emission
At about 4-5K, instrument emission is only significant at the long wavelength end of SLW (BB peak at 2898 µm.K / T ~ 580 µm – Wien’s law).
Instrument temperature varies with time: Blue: before instrument subtraction
Red: after instrument subtraction
PACS
HSC DP workshop 20-24 Feb 2012
Page 23
Telescope Emission Sketch of the telescope spectrum as if it is a point source
Your observations are most likely dominated by the telescope emission!
Telescope temperatures vary with time:
PACS
HSC DP workshop 20-24 Feb 2012
Page 24
Telescope Background Emission: A Typical Case
Telescope + Source
Source only
Telescope model is good to ~ 0.2%, or 1-2 Jy.
PACS
HSC DP workshop 20-24 Feb 2012
Page 25
Flux calibration Scheme Level-1 spectrum
Brightness in W/m2/Hz/sr assumes extended emission
f = Cpoint I Level-2 spectrum
Flux Density in Jy assumes point-like emission
Telescope RSRF Instrument model and RSRF important for SLW (T ~ 4-5 K)
Telescope model
Point source conversion factor (= Rtel/Rpoint)
I = [S - R instMinst] - Mtel Rtel
1
RSRFs are empirically derived by observing a source with a known spectrum and dividing by a model:
Rtel: Dark Sky (= the telescope) Rpoint: Uranus
25
Signal in Spectral domain
Page
PACS
HSC DP workshop 20-24 Feb 2012
Page 26
Telescope Background Removal
Vsrc + Vinst + Vtelescope
Pipeline Step 5: Modify Spectra Spectra
Level-1 Spectrum Product
Extended RSRF Extended Flux Calibration
Instrument RSRF + instrument temp. Instrument Background Removal
Telescope RSRF + Telescope temp.
Spectra are all in extended-source calibration at Level 1 (W m-2 Hz-1 sr-1)
RSRF = Relative spectral response function, or flux conversion factor.
Vsrc + Vtelescope
Ssrc + Stelescope
Ssrc
PACS
HSC DP workshop 20-24 Feb 2012
Page 27
Point Source Flux Conversion (based on Uranus)
Pipeline Step 6: Create Level-2 Products
Point-source spectrum in Jy for central detectors
(SSWD4/SLWC3)
Level 1 Spectra
S(σ) All Other Modes Sparse, Single
Pointing Mode
Spectral Cube Creation
Level 2 Spectral Cube I(σ)
Average spectra
PACS
HSC DP workshop 20-24 Feb 2012
Page 28
Flux Calibration Accuracy (point source)
• Calibration Accuracy: – 4 - 6% compared to planet models.
• Repeatability: – 1.5 - 4% for line flux; – 5 - 7 km/s for line velocity.
• Cross comparisons with HIFI show up to 20% flux agreement for line fluxes. • But, continuum flux suffers an additive term of 1-2 Jy uncertainty due to an imperfect telescope emission subtraction.
PACS
HSC DP workshop 20-24 Feb 2012
Page 29
Flux Calibration
Note: Model fluxes are much more accurate for planets than for asteriods
PACS
HSC DP workshop 20-24 Feb 2012
Page 30
Extended vs. Point Source Flux Calibration Semi-extended source True point source Fully extended source
Brig
htne
ss
Flux
Den
sity
PACS
HSC DP workshop 20-24 Feb 2012
Page 31
Beam Profile
Sharp changes in size where feedhorn modes cut in
HIFI beam size
Airy fit
Gauss 2D fit
SSW
SLW Gauss 1D fit
Gauss 1D (thick black line) are the only values given to users in the cal tree
You are good if you have a point source or truly extended source. Tricky if you are in between.
PACS
HSC DP workshop 20-24 Feb 2012
Page 32
FTS mapping observations
PACS
HSC DP workshop 20-24 Feb 2012
Page 33
FTS mapping pipeline
PACS
HSC DP workshop 20-24 Feb 2012
Page 34
SSW SLW
SLW SSW
1 2 3 4
Jiggle Patterns
Intermediate ~1 beam spacing
35” beam
16” spacing
Full ~1/2 beam spacing
2’ FOV from HSpot
19” beam
28” spacing
14” spacing 8” spacing
35” beam
1 2 3 4
16 15 14 13
PACS
HSC DP workshop 20-24 Feb 2012
Page 35
Fully Sampled Positions on Sky
• Fully sampled observation gives a random looking scatter on the sky
• Some gaps due to dead detectors in SSW • Only unvignetted detectors are shown
SSW separation of points ~8” SLW separation of points ~14”
PACS
HSC DP workshop 20-24 Feb 2012
Page 36
Naive Projection
Positions observed on sky (intermediate sampled 2x2 raster):
Grid with 19” squares (for SSW which has beam size ~17-20”)
19”
Naive Projection is the standard algorithm currently used in the pipeline
(the same as used by the SPIRE Photometer)
PACS
HSC DP workshop 20-24 Feb 2012
Page 37
Intermediate
Full
SLW SSW
35”
17.5”
19”
9.5”
Coverage
PACS
HSC DP workshop 20-24 Feb 2012
Page 38
In the final spectral cube:
“flux” = mean of all spectra within square “error” = standard error on the mean “coverage” = number of spectra within square
19”
19”
Each position may have many spectra (repeated scans were not averaged beforehand)
Different detectors can occur within the same grid square
Naive Projection
Empty grid squares are set to NaN
PACS
HSC DP workshop 20-24 Feb 2012
Page 39
Problems with Naive Projection
• The Naive algorithm works well for the SPIRE Photometer – high redundancy – each grid square contains data from many
detectors • However, there is little (no!) redundancy in our spectrometer
jiggle pattern – often only one point per grid square – then the error reflects the noise from one detector (std error over
repeated scans) – However, in the few squares with multiple points (different
detectors), the error also reflects inhomogeneity in source distribution
• Also, the algorithm does not perform well for recovering the flux of point sources or structure in the map
PACS
HSC DP workshop 20-24 Feb 2012
Page 40
Spectral map vs Photometer map Photometer 250 µm map contours
Very good continuum match, shown at 1120 GHz
NGC 7023
PACS
HSC DP workshop 20-24 Feb 2012
Page 41
Thursday plan (hands-on) • Advanced topics, reprocessing
– Tools for data exploration and analysis • SDS Explorer • SpectrumToolbox, line fitting, continuum removal, matching the
photometer points, noise estimation • Cube analysis toolbox (maps), line maps
– Optimised background subtraction (faint sources) – Mapping extended sources, gridding methods
• Round table, collecting information about each particular FTS project (observations, challenges etc)
• Case-by-case discussion • Hands-on • Wrap up