pandeia: the jwst exposure time calculator the jwst exposure time calculator. 2 current version: 1.0...

Klaus Pontoppidan, on behalf of the STScI ETC teamJWST @ STScI

Proposal Planning Workshop, May 15, 2017

Pandeia: The JWST Exposure Time Calculator

2

Current version: 1.0 (January 2017)

jwst.etc.stsci.edu

for DD-ERS/GTO/GO Cycle 1

Fun fact: When an update occurs, your old workbooks will be saved as read-only

Planned update releases (version 1.1 ~June 1, including target acq + numerous updates, bug fixes and enhancements)

3

SOME HISTORY

• Originally, it was envisioned that the JWST ETC would be similar to the HST ETC.

• It became clear that using the HST ETC as a basic design was not viable.

• Works ok for imaging, but not for most of the advanced JWST modes, JWST detector noise and to support modern user interaction and collaborations

• Something more modern/complex was needed. This was ~2012.

• Imaging • NIRCam SW+LW • NIRISS • MIRI • NIRSpec TA (through MSA mesh)

• Slit spectroscopy • NIRSpec FS • MIRI LRS

• IFU spectroscopy • NIRSpec IFU • MIRI MRS

• Multi-object spectroscopy • NIRSpec MSA

• Slitless spectroscopy • NIRISS WFSS • NIRISS SOSS • MIRI LRS

• Coronagraphy • NIRCam spots + wedges • MIRI Lyot • MIRI FQPMs

• Sparse Aperture Masking Interferometry • NIRISS AMI

4

SOME HISTORY, CONT’D2012 Pandeia prototype2013 Recommendations to satisfy the ETC requirements

within a feasible framework.

1. The ETC should use a three-dimensional (two spatial and one spectral) framework.

2. The ETC should use a PSF library, for instance as generated by the WebbPSF tool.

3. The noise propagation should include correlated noise.

4. Post-observation additions of multiple exposures, subtraction of backgrounds, extraction of photometry and spectroscopy […] should be explicitly modeled under the unifying concept of “observing strategy”.

5. The ETC engine should have a stable API supporting different clients, including APT, a user web form and a standalone application.

6. ETC calculations should be sufficiently fast to support efficient comparative parameter studies, as well as inverse problems.

7. The ETC should be template-based for all three main inputs (instrument configuration, target design, observing strategy).

5

ETC enginePython library

JWST reference database

Separate from other JWST reference data

Throughputs noise properties

PSFs

Web application @

jwst.etc.stsci.edu

User interface relevant for most users of the ETC

Collaborative functionality

JWST Background Model

Currently only available through the web

application

THE PANDEIA PROJECT

6

Workbooks: Organize and save your ETC calculations

Build your own sources and scenes library

Analyze and compare different instruments and modes

Share your work with your team

WEB APPLICATION: A COLLABORATIVE WORK ENVIRONMENT

7

• JWST code project called “Pandeia”

• Under the hood: 3D sensitivity engine

• Includes PSFs, correlated noise other detector effects, effects of background subtraction and extraction.

• Pointing+time dependent JWST background model.

• 100% data driven

ETC ENGINE

Reference: Pontoppidan, Pickering, et al. 2016, SPIE

8

Python module

More functionality than available in the web application

Advanced scripting

pip install pandeia.engine

http://ssb.stsci.edu/pandeia/engine/1.0/

examples at https://github.com/spacetelescope/JWSTUserTraining2016

ETC ENGINE CODE AVAILABLE

Fun fact: The engine is a compact, general code, currently only ~12,500 lines, excluding standard libraries.

9

ENGINE ALGORITHM: SCENE PROJECTIONSImage projection

Slit spectroscopy projection

Slitless spectroscopy projection

10

Flux rate in pixel iTotal flux rate in measurement

scalar weight of pixel iVariance of measurement

covariance matrix of pixels (1,2,...,N)

ETC Strategies: Extracting signal and propagating noise

11

Strategy example: Aperture photometry

Aperture correctionPixel flux rate in aperture

A = extraction region/aperture

B = sky region

Pixel flux rate in background

Pixel weights

Negative because the background is subtracted

12

PSF LIBRARY•Uses WebbPSF to calculate theoretical point spread functions, including realistic wavefront errors. (M. Perrin, https://pythonhosted.org/webbpsf/).

•Almost 3000 individual monochromatic PSFs.

•Subsampled by integer factor of pixel size.

•Fun fact: The observed PSF depends on the color of the astronomical source.

NIRCam SW imaging PSFs

13

PANDEIA REFERENCE DATAPandeia_data Observatory (JWST)

Extinction

SED (template spectra)

Normalization (external bandpasses)

Various defaults

MIRI

NIRCam

NIRISS

NIRSpec

Telescope

Detectors(IPC+correlation)

config.json

Spectral efficiency

Dispersion

Filters

Optical efficiency

Detector QE+QY

PSFs

Fun fact: The ETC reference data is a ~1.5 GB tarball. Almost all of that are PSF libraries.

14

SATURATION

Two types of saturation:1. Partial saturation — the ramp saturates before it finishes, but more than the minimum reads are unsaturated

2. Full saturation — the ramp saturates before the minimum number of reads is reached

15

Zodiacal and Galactic backgrounds

• JWST uses a dynamic “background model generator” (BMG).

• Zodi+ISM heritage from Spitzer’s operational system (IPAC).

• Data are from COBE/DIRBE.

• ISM “cirrus” uses fixed spectrum, including PAHs. Intensity set by Schlegel et al. extinction maps, which were fits to IRAS + DIRBE/COBE data.

• Wainscoat et al.1998 star counts model.

• The Zodi and Galactic emission is calculated for a given RA & DEC, in one of 2 ways:

1. for a given RA, DEC, date: “Dated background”

2. as a percentile of the backgrounds for those coordinates, over the visibility. “Dateless background”

• Fun fact: Benchmark sensitivities computed for (J2000, 17:26:44 -73:19:56) on June 19, 2019. — “minzodi*1.2”

• Fun fact: If using the ETC engine “stand-alone”, you should still use the web app to create a background.

Zodiacal

ISM

Stars

A. Noriega-Crespo

16

Stray light• The background model calculates the

stray light for a given position and date, following Lightsey 2016 SPIE.

• Radiance Transfer Functions (RTFs) from Lightsey, at 0.7, 1, 2, 3, 5, 10, 15 micron.

• Sky map from Galactic + Zody engine (for that position, for that date)

• Interpolate RTFs at other wavelengths.

(Lightsey 2016 SPIE)

17

Thermal self-emission

• Thermal emission from JWST itself is the dominant background at ~>15um.

• Sum of 20 blackbodies (P. Lightsey, J. Rigby).

• Produces backgrounds of 0.74 and 174 MJy/SR at 10 and 20 micron. (Reqs. are 3.9 and 200 MJy/SR). Temperatures are as expected: T(SS)=89K, T(PM)=54K (hottest segments).

• Blackbody model agrees with Lightsey calculations at 15, 25 micron to within 0.5%. Thermal-only curve (black line) + Glasse et al. 2015

SPIE (red and blue solid lines.) Zodiacal model of Glasse (dashed lines.) Figure from J. Rigby

18

Fun fact: You can download all the ETC inputs/outputs in a convenient format from the downloads tab

19

NIRCam F250M/F335M/F460M

MIRI F560W/F770W/F1000WFun fact: You can also use the ETC python module to script

complex, large scenes

USING THE ETC: IMAGING

Red bulge with star-forming knots at z=0.5

20

MIRI F560W/F770W/F1000W

Fun fact 1: In spectroscopic modes extraction apertures are one wavelength pixel wide.

USING THE ETC: SPECTROSCOPIC MODESMIRI LRS

NIRSpec MSA

Fun fact 2: You can upload your own spectra, and add lines.

21

ETC calculation of a MIRI MRS observation of a protoplanetary disk model.

12 individual calculations (one for each MRS channel+sub-band)

Fun fact 1: The ETC will slow down for more finely sampled input spectra (improved in v1.1).

Fun fact 2: All IFU calculations (NIRSpec+MIRI) will perform a dither to subtract background (pick in-field or off). (simplified in v1.1)

22

BENCHMARKING THE ETCETC requirement: 10% in signal-to-noise ratio for 90% of calculations

Compared limiting sensitivities (10 sigma in 10,000 s) for all major instrument modes to independent calculations provided by the JWST instrument teams.

In comparing a 3D calculation with 1D calculations 10% is a reasonable threshold.

Using the same data and same assumptions, get within 10%.

Algorithms are appropriate and accurate Dashed curves are the benchmarks

23

MIRI

NIRISS

NIRSpec

NIRCam

VALIDATION COMPARISON

24

REAL LIFE ETC RESULTS VERSUS STANDARD SENSITIVITIES

There are many parameters affecting ETC sensitivities

That’s the point of having a versatile tool! Comes with some responsibility.

• Think about your background subtraction • Think about what extraction aperture is optimal (point or extended sources) • Is your background correct for your target? • Different read-out patterns change read noise • The ETC does cut some corners: No distortion

for instance (except for NIRISS SOSS).

25

HOW TO GET HELPThe ETC is powerful = a lot of parameters

Documentation: jwst-docs.stsci.edu Known issues: Check the ETC help menu JWST help desk: jwsthelp.stsci.edu Have patience with us - the team is working hard (400+ open tasks)

…more to come