TIPS-JIM Meeting
18 December 2003, 10am, Auditorium
1.
Multiwavelength Astrometry Calibration
for WFPC2
Vera Platais
2.
Improved ACS Geometric Distortion
Correction
Richard Hook
3.
Rules of Engagement
Peter Stockman
4.
Anomalous Scattered Light in ACS/WFC
Observations
Marco Sirianni
Next TIPS Meeting will be held on 15 January 2003.
Astrometric
Calibration for WFPC2
Vera Platais
WFPC2: 5 pix! (Holtzman et.al. 1995)
Distorted coordinates
Un-distorted coordinates
Xg=a1+a2X+a3Y+a4X2+a5XY+a6Y2+a7X3+a8X2Y+a9XY2+a10Y3
Yg=b1+b2X+b3Y+b4X2+b5XY+b6Y2+b7X3+b8X2Y+b9XY2+b10Y3
Data Set: (CAL 6941 – 1997)
- rich globular cluster _ Centauri.
F555W
F814W
F300W
Same pointing
Same orientation
Offsets:
±35 ", ±15 ", ±0.25"
Data Set: (CAL 6941 – 1997)
- rich globular cluster _ Centauri.
F555W
F814W
F300W
Same pointing
Same orientation
Offsets:
±35 ", ±15 ", ±0.25"
First Solution –
Analytical PSF fitting (IRAF/PSF) – standard error in positions
F300W - 0.08; F555W - 0.05; F814W - 0.06 pix
Metachip solution – (Holtzman-Casertano formalism):
x,y of WF cameras transformed into global coordinate system
with orientation and scale of PC1
x´=xcosθ -ysinθ + xoffset
y´=xsinθ +ycosθ + yoffset
X=x´scale
Y=y´scale
Measured
positions in each
filter, in the global
coordinate system
of PC1.
Transformed coordinates X,Y were used for least square minimization of cubic distortion solution.
A full set of coefficients aj,bj was derived.
The Problem:
Meta-chip distortion solution propagates
distortion errors of one chip into the other chips.
The residuals after
correction show
different amplitude and
phase of distortion for
each of WFPC2
cameras.
The precision of meta-chip solution is about 0.3 pix
Quo vadis?
NOVUS VIA
(A new approach)
New approach:
e_PSF (Anderson & King, 2000) measurement of stars
positions with accuracy of 0.02 pix
Astrometric flat-field (Anderson & King,2003) in F555W,
improved the geometry distortion solution with accuracy
±0.01 pix in WF and ± 0.02 pix in PC1.
Anderson-King formalism includes four independent
distortion corrections, one for each chip.
A third order polynomial was used to represent the cubic
distortion.
Histogram – under-sampled stellar profile;
Solid curve – Gaussian model;
Dotted curve – combination of Gaussian and
Lorentzian models
Effective_PSF (Anderson & King 2000)
e_PSF is entirely empirical;
e_PSF is derived from observed pixel values;
e_PSF is fitted to the pixel values by simple evaluation and
scaling without any integration.
The accuracy of e_PSF measurement of stars positions is
0.02 pixels for F555W, F814W and F300W filters.
x,y
x,y
F555W
F814W
x,y
F300W
The residuals between the
‘astrometric flat field’
–F555W positions and star
positions in F300W filter,
applying only a linear
transformation, for each
chip independently.
Astrometric flat-field (Anderson & King, 2003), the
coordinates free of distortion (Xg, Yg) in F555W were
used to derive geometric distortion corrections in X,Y
positions for F814W and F300W filters.
X=(x-425)/375
x,y
XgYg
XgYg
x,y
Y=(y-425)/375
Xg=a1+a2X+a3Y+a4X2+…+a10Y3
Yg=b1+b2X+b3Y+b4X2+…+b10Y3
XgYx,y
g
XgYg
x,y
10 set of aj bj for each chip & filter
XgYg
x,y
<aj> ,<bj > for each chip & filter
After applying the bicubicpolynomial, the residuals are
essentially flat.
The differences in the distortion correction are in the sense
“F555W-F300W”. The average increase of distortion is ~3%
in F300W and ∼1% in F814W.
Residuals are
scaled by a factor
of 300.
The max length for
PC is 0.18pix
The max length for
WFs is 0.25 pix
Conclusion
The precision of the geometric distortion solution
for the WFPC2 cameras depends on:
centering technique in measuring positions of
stars with under-sampled PSF;
the independent chip-to-chip solution, rather
than the meta-chip solution, excludes the error
propagation from one chip to another;
the amount of geometric distortion is a function
of wavelength.
Acknowledgements
Gabriel Brammer for help
with PowerPoint.
Anton Koekemoer and Jay
Anderson.
Colin Cox and Richard
Hook with IDCTAB .
Imants Platais for helpful
comments and suggestions
at various stages of this
project.
Improved ACS Geometrical
Distortion Correction
Richard Hook
TIPS Meeting, STScI, 18th December 2003
Credits
•
•
•
•
Early ACS distortion terminations by ACS IDT - Gerhardt Meurer,
Don Lindler and others.
Local work on distortion coefficients, including velocity aberration by
Colin Cox.
Extensive detailed further determinations made by Jay Anderson and
Ivan King as an outsourced calibration proposal.
Software support by Warren Hack (SSB) and Richard Hook.
Differences between ACS and WFPC2
geometric distortion
• Magnitude of distortion much greater (non-linear
component in corners 50 pixels compared to 5)
• The two ACS/WFC chips are rigidly mounted together.
The four WFPC2 channels are separate optical systems and
float relative to each other on many time scales.
• ACS color-dependence is expected to be smaller than for
WFPC2 as there are no refractive elements with power.
• ACS has distortion correction in the pipeline using
PyDrizzle/IDCTAB
ACS/WFC
Distortion
Timeline of ACS Distortion Determinations
• Ground calibrations based on Ronchi gratings (Cox et al.)
• PyDrizzle and IDCTAB ready at time of ACS installation.
• SMOV distortion program (Meurer, PI), April 2002.
Dithered observations of 47Tuc through F475W.
Distortion quartic coefficients for SBC/HRC/WFC. RMS
much smaller than specification of 0.2 pixels.
• Revised analysis to better determine skew terms, Cox,
October 2002. Basis of IDCTAB in current pipeline. No
filter dependence, but mechanisms in place to allow it.
• Velocity aberration effects studied and incorporated into
pipeline (Gilliland, Cox, Hack, 2003).
The Datasets used - 47 Tucanae calibration fields
Large numbers of datasets:
•Both channels (HRC/WFC)
•Most filters
•Extended time coverage
•Different orients
•Optimal stellar densities
Anderson & King’s Analysis
•
•
•
•
•
Aim for highest possible accuracy to support relative astrometry - for most GO
programs this is far higher than needed for (eg) image combination.
Data on 47 Tuc from initial distortion program (Meurer 9028), supplemental
outsourced calibration program (King 9443) and L-flat program (Bohlin
9019).
PSF fitting methods developed (originally for WFPC2) which allow stellar
positions to be determined to 0.005 pixels for bright stars in HRC (FLT, nondrizzled images used).
The data alone cannot provide scale information so the solution is relative to
the scale of the central pixel (of one chip in WFC).
Relative scale is a free parameter between epochs and varies with velocity
aberration and (smaller) breathing.
Results
•
•
•
•
•
Global distortion modeled well by 4th order polynomial (very close to
Meurer solution)
Residuals (up to +/-0.1 pixel) are systematic, not well modeled by
higher order polynomials, and stored as look-up-tables. For bright stars
in the HRC the RMS is 0.005 pixels after the LUT is included:
comparable to the measurement error.
Residuals are found to be a function of filter and to include small scale
changes and offsets with filter. Jay has supplied “residual correction
images” including the LUT information, relative to F475W.
Solutions appear to be stable over time.
Very detailed ISR on HRC available, WFC one in work.
F555W with F475W
distortion polynomial.
Same but with
distortion correction
table added.
Provisional Implementation of Anderson and King
results for ACS/WFC/UDF
•
•
•
•
•
•
•
Just GOODS/UDF filters (f435w, f606w, f775w, f850lp).
Retained polynomial coefficients from current IDC solution.
Fitted planes to Jay Anderson’s residual correction images to extract
shift/scale/skew components - Colin Cox has incorporated these into a
provisional IDCTAB.
UPINWCS - Python task to take the IDCTAB information and update
the image WCS to reflect scale changes, offsets and chip/chip offsets
(and velocity aberration too). This puts all linear effects into the WCS.
Drizzle/wdrizzle software updated to allow residual correction images
(after plane fit is subtracted) in addition to polynomials.
In use for UDF and appears to give excellent registration, eg, for SNe
searching by subtraction.
Further detailed testing in progress (Jennifer Mack) with 47Tuc data many filters, large dithers.
Residual Correction Images for F850LP - relative to F475W
X-shift
Y-shift
WFC1
WFC2
Black = -0.4 pixels, Red = +0.4 pixels
F850LP - previous IDCTAB, no residual correction image
47 Tuc, large dither, linear fit residuals, (sigma about 0.1 pixels)
(Stellar position measurement accuracy about 0.05 pixels)
F850LP - latest, with residual correction image
(sigma about 0.07 pixels)
Work remaining
• Investigate small remaining residual effects
• Improve star position measurements in tests (PSF fitting)
• Use Jay Anderson’s final polynomials and residual
correction images
• Incorporate into PyDrizzle
• Put updated software into ACS pipeline
• Extend to all filters for both ACS/WFC and HRC
• Document
JWST Rules of Engagement
November 25, 2003
2/2/04
1
Why we need ‘em
ν
Good reasons:
υ
υ
υ
υ
ν
Management reasons:
υ
υ
ν
More STScI staff are attending JWST working groups, SI teams etc.
Many requirements and ICDs are being developed that will affect the
observatory performance and the S&OC (STScI).
JWSTP has indicated that they do value STScI input.
We (STScI) need to maximize our effectiveness in raising issues and
responding to CCRs and reviews.
The Project has not yet implemented a clear review and consent process.
We do not want to commit ourselves to effort not covered under our
current contract.
Now in Phase B we can find ourselves in disagreement with the
Project.
υ
The Project is deviating from the HST lessons learned
τ
υ
A number of architectural, requirements and process problems have arisen
τ
2/2/04
The lessons from HST seem obvious to us but may often not be known to or
selected by the Project (for budgetary or other programmatic reasons).
The solutions under consideration or the current lack of a solution may have
suboptimal consequences for science or operations.
2
We have many opportunities to
affect JWST plans:
ν
Technical advice in support or leadership of
working groups and SI teams.
ν Reviews of documents prior to baselining
ν Submittal of CCRs (Change requests)
ν Response to CCB actions (baselining & CCRs)
ν Contractual changes & negotiations.
The earlier we affect a document or design, the
better.
2/2/04
3
When should we request changes
or raise issues?
ν
ν
ν
2/2/04
When a design or requirement reduces the ultimate
scientific performance of JWST (particularly with
respect to previous designs).
When a design or requirement is technically flawed (we
can show that it won’t work or carries a significant and
definable risk).
When a design or requirement is inconsistent with our
contract/development plans/operations concept (e.g.
raises our costs).
4
Who can raise an issue?
ν
Any STScI staff member can raise an issue:
υ
In support of a working group or SI team:
τ
τ
υ
In reviewing documents
τ
τ
υ
To designated lead reviewer
To WBS manager/MO
As a result of an internal study
τ
2/2/04
To chair of working group or PI
If not resolved, to WBS lead and/or Mission Office
To WBS lead
5
Working Group Etiquette
ν
ν
Know the Chair and Charter of the Working Group
Raise issues in natural course of discussion:
υ
υ
υ
υ
ν
2/2/04
Be more concerned about substance than style
Don’t harp on same concern (but feel free to email the Chair if you
think that the group misunderstands the issue or underestimates its
impact.)
Show others the respect that you would want for your own ideas.
Try to work from a position of knowledge and not opinion. If you
are not sure about a fact, take it offline as homework and
communicate by email or at the next meeting.
In general, make clear that you cannot commit the STScI
to a course of action or concurrence on a document. You
may be the most knowledgable/informed STScI staff
member but even then, there may be additional concerns,
dependencies, and contractual agreements of which you
are not aware.
6
Working Group Etiquette (2)
ν
ν
When a group seems to be deadlocked on an issue ask the chair (or
most senior GSFC person present) what the mechanism is for issue
resolution. If they don’t know, ask if there is a higher level body to
which the issue can be raised.
Be aware of documents or CCRs that will be submitted for to the CCB
and bring these to the attention of your WBS lead and/or MO so that
they can have wider review within the STScI and comments returned to
the author/working group chair before CCB submittal.
υ
υ
υ
2/2/04
This step should be facilitated by the working group chair/PI/instrument
manager but often isn’t.
NASA will often assume that your presence and participation guarantees a
STScI rubberstamp approval. It doesn’t.
It is much better to indicate impacts and concerns before the CCB process
than during it. At a minimum, it reduces the elements of surprise and
frustration.
7
What do you do if a new capability
is allocated to the S&OC?
ν
If you are participating in a meeting wherein a capability is being
allocated to the S&OC that exceeds the scope of the S&OC
contract/requirements/architecture (e.g.via requirements or interface
definition) then:
υ
υ
υ
υ
υ
2/2/04
identify to the group that there is an impact to the S&OC,
ensure that the requirement/request for the capability is documented
(Specification, ICD, OCD, CCR)
if you believe the impact to be small allow the group to proceed with their
assumption that the S&OC will provide the new capability,
if you believe the impact is large encourage the group to develop
alternatives as the proposed change to the S&OC may not be approved,
report the situation to your JWST direct report (Team Lead, WBS
Manager or JWST Mission Office).
8
External Review Etiquette
ν
ν
If you are a member of a board, you will be advised of your
responsibilities by the chair of the board and the board’s charter.
If you are in the audience:
υ
υ
υ
Maintain a professional decorum. (snickering, muttering, groaning, etc. is
discouraged)
Generally, you are at the review to support another speaker or to learn.
You should interject only when there is a significant factual error
presented (raise your hand and be recognized by the speaker or the chair).
NASA has discouraged the submission of RIDs or RFAs by project
personnel (including the STScI) at external reviews.
τ
τ
υ
Nevertheless, if you are sure that an important technical issue has been
missed or misrepresented, we suggest the following steps:
τ
τ
2/2/04
Gives the impression of poor communication within the Project (may be true!)
Should have been another opportunity for raising the issue (see working group
etiquette).
Speak with the chair or a member of the board at a break. They may ask you to
write up the concern so that they can discuss it later.
For some reviews, the MO will provide a procedure for submitting a concern
through a designated STScI representative to the Board/Project.
9
Sometimes a strategic retreat is
the best option:
ν
If a working group/PI etc. is dedicated to a course of action
with which you disagree, there are other opportunities to
mitigate the impact on science and/or the STScI.
υ
υ
υ
ν
2/2/04
Issues can be brought to the Project’s attention through the MO
monthly reporting process.
Risks can be tracked within the STScI and in the Project Risk
database.
Ultimately, impacts to the STScI can be addressed with additional
funding by the CCB/Contract modification process (and this
process can result in the course of action or CCR being denied
because of its overall financial impacts.)
Take some solace in the overall scientific power of JWST
and the healing power of additional funding (even if it is
not the optimal solution).
10
Remember
ν
ν
ν
You probably have the most operational and scientific
expertise on your working group.
The STScI is expected to represent the ultimate scientific
user of the observatory.
Our SOW explicitly requests that we endeavor to:
υ
υ
ν
2/2/04
Minimize total mission costs (including operations)
Maximize ease of observatory operations
We must earn the respect of the Project. We have the
respect of the HST project because we earned it but it is
not transferable to JWST; it needs to be earned anew.
11
How do we track issues, risks,
assumptions?
ν
ν
ν
2/2/04
Within the STScI, issues and risks are presented and
tracked through the MSR and MA-01 process.
Within JWSTP, issues are tracked by the JWST
engineering team and at the Top-Issues meetings. Risks are
tracked in the Risk management database and Risk Board
reviews.
Contract assumptions are tracked by the MO and will be
reported as part of our semi-annual planning responses and
in negotiations.
12
Anomalous scattered light
in ACS/WFC Data
Marco Sirianni
D. Van Orsow, A. Welty, T. Wheeler
R. Gilliland, R. van der Marel
ACS/WFPC2 group
INS Division
Marco Sirianni
TIPS 12/18/2003
Anomalous scattered light in ACS/WFC Data
On Dec 03 two PIs reported an anomaly on their ACS/WFC data
A
B
WFC-1
WFC-2
C
D
Marco Sirianni
TIPS 12/18/2003
Anomalous scattered light in ACS/WFC Data
Initial clues:
Both programs used F814W and were executed on Nov 11 2003
The shape of the pattern is the same
The count-rate is similar in different images (~ 0.18 e-/sec)
The photometry of the objects in the contaminated region is consistent with an
additive background.
SCATTERED LIGHT ?
First investigations on all ACS data between Nov 10 and Nov 12:
Not all F814W images were contaminated.
14 contaminated images (only F814W and F606W, no contamination in F555W).
F606W images show a similar pattern but a much lower count-rate
Some images show a lower count rate than others taken with the same filter
LAMP CONTAMINATION ?
VERY RED LAMP ?
Marco Sirianni
TIPS 12/18/2003
Anomalous scattered light in ACS/WFC Data
Analysis of telemetry/commanding:
No Anomaly in ACS Lamps
No Lamp Activity in STIS or NICMOS
WFPC2 was taking a series of 1800 sec internal flat fields
(Tungsten lamps + F390N filter) in those days.
The comparison of the ON-OFF cycles of the WFPC2 INTFLAT lamps and the
timing of the ACS anomaly gives a perfect match:
when ACS/WFC images show the bright pattern the WFPC2 lamps are ON
when the count rate is lower (images with the same filter) the lamp has
been turned on during the exposure
(the count rate is uniform if the real exposure to the lamp is taken into account).
The telemetry data allowed us to find more contaminated images
CONTAMINATION FROM WFPC2: INTFLAT+F390N
Marco Sirianni
TIPS 12/18/2003
Anomalous scattered light in ACS/WFC Data
Brightness and shape of the pattern are consistent
with the following scenario:
WFPC2 optical configuration:
From OTA
B
Fold
Fold
mirror
mirror
To
ToCassegrain
Cassegrain
++CCD
CCD
B
Pick-off
Pick-off
mirror
mirror
SOFA
Lamps SOFA
Lamps
Shutter
Shutter
A
A
Pyramid
Pyramid mirror
mirror
Light leakage from WFCP2
shutter (≤ 1 %)
to
WFPC2 POM
to
ACS
TO ACS
any other light path (OTA secondary or glint near ACS
entrance aperture) would produce much fainter illumination
Marco Sirianni
TIPS 12/18/2003
Anomalous scattered light in ACS/WFC Data
For a direct path from WFPC2 a simple ray tracing model
predicts a bright image of ACS WFC IM1 mirror
Edges of ACS entrance aperture ?
shifted diagonally toward WFPC2
Edges of IM1 Mirror
Multiple images due to WFC CCD
windows reflections
Marco Sirianni
TIPS 12/18/2003
Anomalous scattered light in ACS/WFC Data
Immediate actions:
- All WFPC2 INTFLAT programs on hold
- More detailed investigation of the problem
Occurrence
WFPC2 filter dependence
Are other channel/instrument contaminated?
WFPC2 VISFLAT potential problem?
Marco Sirianni
TIPS 12/18/2003
Anomalous scattered light in ACS/WFC Data
Preliminary results:
Problem since SM3B, but it is quite rare circumstance
(~ 40 images so far, mostly parallel, 4 GOs programs)
No contamination in ACS/HRC STIS or NICMOS
No correlation with the WFPC2 Filter ( -> direct leakage)
Clear Correlation with WFPC2 Shutter Blade:
Approximate average count rate e-/sec
WFCPC2 SHUTTER
WFC FILTER
A
B
F606W
0.007
n.a.
F775W
0.080
0.025
F814W
0.180
0.051
Marco Sirianni
TIPS 12/18/2003
Anomalous scattered light in ACS/WFC Data
Summary / Conclusion:
• WFPC2 shutter leakage contaminates ACS/WFC data during INTFLATs
• The level of contamination depends on:
• WFPC2 TUNGSTEN LAMP CYCLE
• WFPC2 SHUTTER BLADE
• ACS/WFC FILTER
• ACS/WFC EXPOSURE TIME
• The PI of the programs with corrupted data will be contacted.
• The problem is fully predictable and therefore can be avoided with
an appropriate scheduling strategy
• WF3 should be a better neighbor
(we’ll plan a specific SMOV test for SM4)
Marco Sirianni
TIPS 12/18/2003