SIRTF-IRS software

Fred LahuisLeiden

18 Nov 2002

(presented by Ewine van Dishoeck and Adwin Boogert,c2d meeting Dec. 11, 2002)

IRS Instrument

Low resolutionR~60-120imaging spectrograph1st and 2nd orderSL Si:As 5.3 - 14.2 micronLL Si:Sb 14.2 - 20.0 micron

High resolutionR~600 echelle spectrographorder 11 to 20SH Si:As 10.0 - 19.5 micronLH Si:Sb 19.3 - 37.0 micron

IRS images (128x128)

Simulated single exposure of an extended source with an M82 + Circinus-like spectrumfrom 10-19.5 mm, includingcosmic rays.

Simulated IRS SL image

Sample IRS calibration files

IRS-ISO example

IRS Low-Res

IRS High-Res

IRS dataflow

IRS dataflow (cont’d)

BCD products

Lights-offScience pipeline

Lights-onScience pipeline

Science by legacy teamTEXAS

SSC

Best effortFinal delivery

SSC Pipeline

BlackboxProducts2d imagesflatfielded BCDapplied flatfield

1d spectrumnonimal extraction

Flat-field is only controllable stepOptimal extraction in SMART

SMART

SMART functionality

Read SSC data and handle complex projects (data volumes)View and evaluate individual and series of filesBasic pixel arithmetic: add/subtract/multiple framesCombine a series of images according to their qualitySpectral extraction for point and extended sources(PU image analysis)Combining spectra from different orders, modules, runs, …Most of the ISAP features (line/BB/zodi fitting, photometry, line identification, dereddening)Template fitting (source class identification, redshift)Spectral maps (line ratio maps)Interactive defringing

C2D Interactive Analysis

SMART (legacy release @ launch)project handlerIRS image analysisoptimal spectral extractionIRS calibration tools and data

Defringingflatfield modification and robust sine wave fitting

Modelinggas-phase molecules, disks, ices, solid state features

C2D science pipelinescripts and data

Spectral analysis/reduction toolszodiacal light, line-fitting, ....inspired by SWS IA, SMART-IDEA, .......

.......?

Catalogue

1. Spectra flux vs wavelength

2. Identification of features, where feasible

3. Flag for off-source emission, where feasible

4. Strengths of features

5. Classification of objects, cross refs

6. Complementary data

Only 1-3 will be in catalog/deliveries to SSC; 4-6 TBC for final delivery

Instrumental fringes

Fabry-Perot effectson plane parallel surfaces within the instrumentinside filtersPlane mirrorsdetector surfaces

Separations of a few mm are most efficient

Correction methods:FFT, badRobust sine fittingFlatfield matching

Flatfield modification

Minimize fringe residualsHow? by modifying flatfieldFringe pattern changes per observationsource offset in the slitshifts wavelengthshifts fringe pattern

source extentreduces effective resolution

Stands or falls with the quality of the flatfield(s)

Flatfield modification II

Determine phase shift and smooth/enhance factors for all orders

Use smooth polynomial fit for

phase shift and/or smooth/enhance

factor?

Modflat pipeline issues #1

Flat needs to be available, and in 1D approach extracted exactly the same way as science target.

In 1D approach: maximum sampling extracted spectrum crucial, in particular for highest echelle orders.

Smoothing fringe amplitude flat easier than enhancing. Flat used for Modflat should have highest possible resolution (point source).

Ideally, flat is available for each position in the slit, and best match with science observation is chosen from flat field 'database'.

For extended sources 2D version of Modflat may be considered: determine phase shift and smooth/enhance factors for each column.

Adwin Boogert, IRS Fringing meeting Cornell, 30 August 2002

Modflat pipeline issues #2

In case of extended source in combination with point source? Ideally extract spectrum of background and point source, and defringe independently, or simply subtract sky from point source.

Return header keywords: phase shift and amplitude reduction factor for each order.

2D or not? May be better for extended source: apply modflat on each column. Point sources not expected to improve in 2D, what really matters is extraction at high sampling.

Adwin Boogert, IRS Fringing meeting Cornell, 30 August 2002

Defringing: sine waves

Why use sine functions'Simple' and robust algorithmAllows better estimation of fringe frequency than FFTNot limited by sampling or gaps in the spectrumAutomatic mode using Bayesian evidence prevents overfitting

Goal: fringe artifact < 1%Why?

Most interesting science is in the weak features

Theory

I tT 2

1R 2

Io

1 4R

1R 2sin 2 wD

ItA cos 2 wDa

ItAcos 2wD Bsin 2 wD

?D2ndnrefractive index

TtransmissionRreflectance

dthickness

@low reflectance

Computationallyhandy formula

Implement awavelengthdependence ofR and n and ax,y dependenceof d

Can we

Implementation

Isolate fringesfrom continuum Find the strongest

fringe components

Compute the totalfringe spectrum andreturn the defringedspectrum

Fringe characteristics

Fringe parameters (frequency, amplitude and phase) relate to true physical parameters(refractive index, wafer thickness,reflectance and wavelength (source slit position))

? ..... what's next

Next

"Full" optical/physical model?Parameters need to be extracted from the data2-D: requires optimal extraction and source modeling....Will the data/calibration be good enoughWe'll just have to try

Good (self)calibration is crucial

Waterloo Conference

First (to be) published IRS defringing paper

ISO-SWS Orion KL, Boonman et al. 2002

....