Minutes of Chandra Users' Committee Meeting, December 18-19, 2000

Committee members present were: Keith Arnaud, Jill Bechtold, Megan Donahue, Jack Hughes, Shri Kulkarni, Joe Mohr, Frits Paerels, Craig Sarazin, Jane Turner, Mike Watson, and Ray White. Ex-officio members present were Fred Seward and Allyn Tennant. Alan Bunner attended from NASA HQ.
Status of Chandra

Roger Brissenden (CXC Manager) reported on the operational status of Chandra. The performance of the observatory has been excellent. Operations are now routine. The most important issue has been the complications in the grating operations due to the failure of the HETG limit switches. Solar particle events interrupted the schedule, but were handled correctly. Eclipses produced no problems. There were 5 Targets of Opportunity (TOOs) since the last meeting.

The observing efficiency (fraction of clock time with data taken) is now nearly 70%, which is very good, particularly if one considers that disruptions by solar flares have reduced it somewhat.

Data processing is going very smoothly, with delivery of data to the observer in less than 2 weeks. Reprocessing of older data is 75% complete, with expected completion in February 2001. CUC members congratulated the CXC on the excellent operations, high observing efficiency, and rapid data processing.

Radiation during normal observations (not during belt passage or during flares) is increasing the Charge Transfer Inefficiency of the ACIS-I by an apparent rate of 5% per year. Part of this is due to the varying effect of background in filling traps in the detectors. The real rate of increase is about 2% per year.

The number of warm pixels in the aspect cameras is increasing with time. This might produce a degradation in pointing, which might become important after about 5 years.

The Chandra Team was awarded the National Air and Space Museum 2000 Trophy in November.

Harvey Tananbaum (CXC Director) reported on the use of Director's Discretionary Time (DDT). DDT observations used a total of 937 ksec out of the 1000 ksec assigned to the Director.

Tananbaum was asked about the cause of the erroneous detection of a QPO in a bright X-ray source in M82. This was due to an incorrect handling of the aspect dither in the HRC timing software; this problem has been corrected. The CUC recommended that the CXC webpages list a "caveat" about this, which still might affect observers with older data sets.

Belinda Wilkes (CXC) reported the allocation of Chandra observing time by the HST review panels. They approved 10 of 32 proposals for a total of 345 ksec, less than the 400 ksec allocated to this exchange. Only one rapid TOO was accepted (HST is allowed to allocate one). In the previous Cycle, the HST panel allocated a significant fraction of the time-critical observation time for Chandra's Cycle 2, which caused problems for Chandra proposers.

Improving the CTI on the ACIS FI Chips

Paul Plucinsky (CXC) reported on efforts to improve the performance of the ACIS FI chips, and reverse some of the effects of increased CTI due to radiation damage early in the mission. The first improvement, lowering the focal plane temperature to -120 C, was implemented nearly a year ago. A considerable effort has gone into studying "squeegee mode," which is a new way of clocking the CCDs. The top rows of the CCDs are used to store and distribute charge, and are not read out. The charge is introduced into the imaging region to fill the traps produced by the radiation damage. A variety of different modes have been designed and tested. Unfortunately, the modes which could produce the greatest improvement in spectral performance produce a grade migration effect which lowers the detector efficiency. The best compromise is a mode referred to a "vanilla squeegee". However, there are at least 4 different types of traps, and squeegee only deals with the ones with the longest time scale. As a result of these complications, the present belief is that squeegee alone doesn't provide a large enough improvement to justify implementation.

There is also an effort to do event-by-event CTI correction. One relatively simple correction is to include gain variations for each column in the CCD. A computer algorithm for doing a detailed correction for CTI for each event is being developed; these techniques may be effective for the traps that squeegee misses. One concern is that the best algorithm may be too complex to be implemented on the on-board computer.

It is hoped that the combination of squeegee and event-by-event correction may allow a significant improvement in the ACIS FI CTI. The time scale for these improvements might be a year, if this effort is successful. Given the uncertainties, Cycle 3 proposals must be written assuming no improvement.

The CUC asked about the overhead cost in additional calibration time which will be required if squeegee is implemented. The current estimate is about 5% of the observing time.

Chandra Deep Fields Program

Wilkes gave an update on the Chandra Deep Fields program. An NRA for funding proposals for supporting observations and analysis is about to be released. CUC members asked about whether any of the follow up optical observations on ESO and Keck by the two PI teams would become public. This might affect plans for proposals and their review. Wilkes said she would try to determine this prior to the review.

Theory and GTO Programs

Alan Bunner (NASA HQ) asked about support for theory proposals within the Chandra data analysis (DA) budget. The NRC Decadal Report recommends that key theory programs be funded as part of the budget for any new major observatories. Bunner asked whether this might be done retroactively for Chandra. He suggested that the funds would would come from the Chandra data analysis budget (about 7% of the budget), and that the proposals would be judged during the normal Chandra observing proposal review.

The CUC members had considerable reservations about this suggestion. They felt that Chandra DA funds were already stretched to cover data analysis for new observations and archive observations. They also didn't think that Chandra observing proposal review panels could provide a fair assessment of theory proposals.

Bunner also asked about the long term future of the Guaranteed Time Observer (GTO) program. He noted that the CGRO dropped its GTO program eventually. CUC members noted that the GTO program represented a promise to the GTO teams. The CUC agreed that a review of the GTO program would be useful at some point in the future.

Observing Policy Issues

The CUC then addressed a number of questions about the observing policy which had arisen either from operational experience or from questions from users.

The first issue came about because of a successful Cycle 2 proposal for longer observations of the same targets (and with the same instrument) as observed by other PIs during Cycle 1. Current Chandra Observatory policy allows a target to be re-observed with the same instrument in a later Cycle if there is a sufficient scientific justification to do so. Examples might include source variability, or a longer exposure required to achieve an additional science goal of great importance. In this specific case, the Cycle 2 observer had relinquished her proprietary time. There is a danger that such a re-observation will reduce the proprietary time of the PI of the first observation, if the PI on the first observation requested his/her normal proprietary time, and this time has not elapsed by the start of the observing cycle in which the re-observation is to be done. The CUC recommended that, immediately following the review of proposals for each cycle, the CXC should identify any such cases. As a matter of routine policy, the CXC should delay the scheduled time for the re-observation until the proprietary time of the first PI is completed. If the re-observation has been completed before this rescheduling can be done, the CXC should hold the re-observation data, and release it to the PI of the re-observation only after the proprietary time of the first observation is complete. There was strong support for this policy. The CUC also considered whether to apply this policy only when the PI of the reobservation had given up his/her proprietary time. However, the CUC voted 7 for, 4 against to apply this policy to both cases: when the reobservation would be proprietary and when it would become public immediately. The CUC also discussed whether the CXC should discuss the situation with the PIs of the two proposals and attempt to broker an agreement between them. There was concern that pressure might be brought to bear on relatively junior PIs, and the committee voted 8 to 2 (1 abstention) against such routine contacts.

In general, observers are not allowed to change instruments on approved observations. However, during Cycle 1 this was allowed because of the radiation damage to the ACIS I. It is likely that similar circumstances will arise in the future (with the implementation of squeegee mode, for example). Problems can occur when such changes occur during the proposal writing and review period for the following cycle. A proposer might write a proposal for a complementary observation (same target, different instrument) only to find later than the observation she/he proposed was already being done. Similar problems can occur during the proposal review. CUC members argued that the CXC should try and maintain a stable observing program during the proposal writing and review period. It wasn't clear how to achieve this in practice.

The CUC reiterated the policy on the handling of reobservations due to an operational error with the original data. When this occurs, the original data should be placed in the public archive as soon as the new observations is delivered to the PI. The committee was concerned that this policy was not applied for the HRC reobservations undertaken due to the timing fault.

The CUC reiterated the standard observing policy that, in general, observers should not be allowed to transfer observing time between approved targets or from approved to nonapproved targets. In exceptional circumstances, this might be allowed if the target was in the original proposal and was not in conflict with any other proposal, and the original review panel didn't indicate a preference for the approved target.

Judging the Program for Large Proposals

Starting in Cycle 2, at least 20% of the observing time is set aside for Large Proposals, defined as those requiring more than 300 ksec. At some point, a judgment needs to be made as to whether this is a good idea. After some discussion, the CUC suggested a review of the scientific impact of the Large Proposals which might result in an increase or a decrease in the allocated fraction. The first such review of the program should occur prior to the release of the NRA for Cycle 5. Criteria to measure the success of the Large Proposal program might include (1) key scientific results, (2) publications, and (3) citations (appropriately normalized). STScI undertook such a study for their large proposal program. Based on their experience, the CXC should begin tracking these metrics now. The review of the Large Proposal program should be undertaken by a panel of distinguished astronomers.

Grants Administration

There was a brief report on the distribution of funds for analyzing Chandra data, which is going very smoothly. Based on comments from a number of PIs, the CUC recommended that the performance period of Chandra grants routinely be 2 years rather than one year, or that a 1-year no-cost extension be allowed in a completely automatic fashion. The present 1-year performance duration makes it difficult to hire graduate students or postdocs. Also, based on a comments from PIs, the CUC suggested that the CXC provide a digital form for electronic submission (via E-mail or the WWW) of the annual or final technical reports on grants. STScI currently provides such a mechanism, which is likely to facilitate the timely submission of such reports.

Recently, NASA has decided to require certifications from foreign Co-Is on Chandra proposals. Since the Co-Is receive no financial support, this policy may reduce the foreign participation in Chandra science, at least officially. Don Kniffen (NASA HQ), who could not attend, sent a report on efforts to remove this requirement. The CUC strongly supported these efforts.

As part of the review of observational proposals for Chandra and other programs, PI can propose for funding of Education and Public Outreach (EPO") programs. The individual EPO grants have been small, and it is difficult to propose a substantial EPO program on this basis. Partly due to this, the EPO program has not been very widely utilized. At present, PIs are not allowed to combine EPO funds from different observational proposals and programs. The CUC had suggested that pooling these resources, which was allowed in the past, would permit PIs to propose more significant EPO programs. Kniffen sent a report indicating that there was an effort to permit the pooling of EPO proposals and grants in the future. The CUC recommended that NASA allow this as soon as possible.

Cycle 3 Proposal Submissions

Fred Seward (CXC) gave a brief report on changes in the Cycle 3 NRA.

At its June 1997 meeting, the CUC endorsed a policy for including pointing overhead in the time charged to observations. This can be particularly important for short observations. So far, this has not been done. Seward suggested adding 1500 seconds (the typical overhead time) to short observations. The committee again endorsed this policy. However, they suggested that it would be easier to include this overhead for all observations rather than only for short observations. This extra time could be built into the spread sheets, total, and estimated allocations given to each review panel.

Starting with Cycle 3, GTOs will have a guaranteed allocation of time, but must compete with General Observers (GOs) for specific targets. The allocation of the targets in conflict will be made based on the rankings of the peer review panels. Tananbaum suggested a new policy to deal with with these conflicts. The aims of this policy are to be fair to both the GTOs and GO, but also to minimize the burden of proposal writing and review. Of course, unconflicted targets will automatically be awarded to the GTOs. It would seem a waste of their time to write proposals for these targets, and it would be a similar waste of the time of the review panels to read these proposals.

Based on these arguments, the CUC voted unanimously to endorse the following policy:

GTO-GO Target Conflicts and GTO Proposals in Cycles 3 and Beyond

    a) There will be a single due date for each Cycle, with GOs sending their proposals and GTOs sending in target lists over-subscribed (in time) by a sufficient amount. Immediately following the deadline, the CXC will search the lists for conflicts. For those found between GTOs and GOs, the GTOs will be allowed to select 1 of 2 options:
    (i) Drop a target in conflict and replace the time with the next one or ones on their list, staying within their total time allocation.
    (ii) Write a proposal for the target or targets in conflict.
    b) If the GTO decides to write a proposal, this will be a standard Cycle 3 observing proposal, with a science and technical justification, and subject to the same page limits as GO proposals (4 pages, or 6 for Large Proposals). The GTO proposals must be submitted within 7 days of the notice from CXC of a target conflict. The GTOs will be provided with no information from the competing GO proposal, and the CXC will insure that GO proposals are kept in confidence during this period.
    c) The GTO proposals will be reviewed by the relevant peer panels along with the GO proposals. Possible outcomes are:
    (i) the GTO proposal scores higher than the GO proposal and is accepted; the score need not be above the panel's threshold for allocating time, since the time comes from the GTO time allocation.
    (ii) the GO proposal scores higher than the GTO and is above the allocation threshold so receives some or all of the requested time. The GTO proposal is rejected, and is replaced with the next target on the GTO list. (Note that lower priority GTO targets with conflicts will need to have proposals written and reviewed in case they are needed.)
    (iii) the GO proposal scores higher than the GTO proposal, but not high enough to be allocated time. Then neither proposal receives time, since giving time to the lower rated proposal would seem unfair.
The CUC discussed the possibility of giving the target to the GTO under case (iii) immediately above, but agreed that this would be unfair. The CUC voted 8 for, 1 against, 2 abstaining to leave the policy as stated. (As noted above, after this decision, the CUC unanimously supported the policy.)

The CUC emphasized the importance of maintaining the confidentially of the GO proposals during the time when the GTOs are writing proposals for conflicted targets. Obviously, it would be unfair if the GTOs could write their proposals either as a rebuttal to the competing GO proposal(s), or if they could include ideas from the GO proposal(s). Thus, the GTOs should receive no information other than the targets on their list which are in conflict.

The committee also felt strongly that the writing period for the GTO proposals should be limited to 1 week. (Of course, the GTO could start to write their proposals prior to knowing which targets are in conflict.) The CUC felt that this was needed to compensate for the fact that the GTOs will be able to "concentrate their fire" on the targets in conflict, whereas GO must write a strong scientific justification for all of their targets.

Peer Review of Cycle 3 Proposals

Seward discussed a number of possible changes to the peer review for Cycle 3, based on previous experience. There was also a discussion of the selection of panel members and chairs. The CUC then considered a number of suggestions for changes to the policies under which the review panels operate.

Based on a number of external suggestions, Martin Weisskopf (Project Scientist, MSFC) had asked the CUC to reconsider the conflict-of-interest rule which prohibits PIs of Chandra grants from serving on the review panel which judges their own proposal. Of course, a concern with all conflict-of-interest rules is that they make it more difficult to obtain the most competent review panels. However, the committee felt that prohibiting PIs from serving on the panel which reviews their own proposal was essential to maintain the integrity of the review process. The CUC voted unanimously to maintain this prohibition.

A panel chair during Cycle 2 had suggested that review panels be allowed to merge competing proposals for the same observational target(s). Based on previous experience, the committee did not think this was a good change, and voted 9 for, 0 against, 2 abstaining to maintain the current policy against "shotgun" marriages of proposals.

The same panel chair had also asked that panels be allowed to increase the exposure times on a target or targets if they thought this would improve the science. The CUC felt that panels should not rewrite proposals, and voted 4 for, 7 against this policy change.

Another Cycle 2 panel chair had suggested that review panels not routinely lower the grades of good proposals which are rejected because the target or targets are assigned to higher ranked proposals. The advantage of the current policy is that all rejected proposals have grades which are below some fixed cut-off grade (generally, 3.5). The negative is that the information that the proposal was good but simply lost out to a better proposal for the same target(s) is lost. The CUC asked Bunner which policy was likely to produce more protests from unhappy proposers, and he felt that it was easier to explain to a rejected proposer that she/he had lost to a better proposal, rather than explain that a good proposal had gotten a low grade because of this policy. Thus, the CUC voted 7 for, 1 against, 3 abstaining to change the policy so that the grades of proposals rejected due to target conflicts with a higher rated proposal are not artificially lowered below the acceptance threshold.

Seward suggested some possible changes in the divisions of subject areas for the review panels. Based on complaints from panelists during the Cycle 1 and 2 reviews, the committee recommended that the subject divisions separate normal stars from SNRs, given the vastly different physics between these two classes of objects. In order to reduce the possibility that the review of proposals would depend very strongly on which panel they were assigned to, the CUC recommended that parallel panels (i.e., panels dealing with the same topic) dealing with large number of proposals in related areas should have as similar a mix as possible. They also suggested that it might be useful to have one panel to handle miscellaneous classes of objects in subject areas with a small number of proposals. There was some discussion of broadening the areas covered by panels (say, to just galactic vs. extragalactic), but the consensus was that this would make it too difficult to obtain enough reviewers with sufficient expertise in specific areas on each panel.

Changes in the Review of Large Proposals and the Merging Panel

Seward proposed a number of changes to the structure and rules for the Merging Panel which also handles Large Proposals. One common complaint from the Cycle 2 review was that the merging panel was too large and unwieldy. Seward suggested that the Merging Panel be reduced to about 15 members, consisting of the topical review panel chairs, and 3 scientists with broad research interests brought in just for the Merging Panel. The CUC endorsed this idea. They felt that these 3 overall members of the Merging Panel should not be members of any of the topical panels. The committee also recommended that the Merging Panel have a chair appointed in advance who is a distinguished astronomer.

One concern with the Cycle 2 merging panel was that a lot of time was spent on discussing "grey area" small proposals (proposals at the border of acceptance/rejection). It is difficult to compare these to the "grey area" large proposals, and this comparison is not fair to the small proposals, since one "grey area" large proposal generally would require as much time as many of the small proposals combined. The CUC recommended that the merging panel should compare the strongest subset of gray area small proposals, taken as a single suite, against the gray area Large Proposals.

Current Chandra policy does not allow topical review panels to cut time or targets from Large Proposals; they may make recommendation to the Merging Panel. The CUC received a request from a panel chair from the Cycle 2 review to reverse this policy. However, the committee continued to unanimously support not cutting Large Proposals in the individual topical panels.

Changes in the Phase II Cost Review

Seward suggested that the Phase II Cost Review of proposals be simplified, since most proposers ask for funding at a level consistent with the overall funding levels, and most of the decisions made in the review are straightforward. It would be more efficient if the review could concentrate on cases where the funding requests were higher than average because of special circumstances. The CUC suggested a simpler means to deal with the majority of funding requests. With the call for Phase II cost proposals, the CXC should provide proposers a "fair share" allocation based on 90% of the available funds, using a sensible algorithm including the observing time, number of targets, ranking of the proposal, and review panels assessment of the difficulty of the analysis. It would be expected that proposals which request this fair-share allocation or less would generally be approved, subject to proper budget justification. Proposers may request higher funding levels, but these proposals will undergo more detailed review. The CUC voted 7 for, 1 against, 3 abstaining to recommend this change.


Larry David (CXC) gave a presentation on overall issues involving the calibration and of the mirror point-spread-function (PSF).

Richard Edgar and Norbert Schulz (CXC) discussed the calibration of the sensitivity and spectral response of the ACIS detectors. The CIAO 2.0 software now includes the nonlinear gain function at low energies. The many FEF files have been repackaged into a few Super-FEF files which will be released very soon. Calibration observations using the HETG have shown that there is an extended low energy tail to the spectral response which affects the quantum efficiency as well. New calibration files which include this should be available in several months. The BI S3 chip and FI S2 chip disagree on the quantum efficiency by about 15% below 1.2 keV, and this issue needs to be resolved. The FEFs will be extended to include energies from 0.1-11 keV, although the outer parts of this range are very uncertain. New response matrices for the ACIS-S will be available roughly in April 2001. For the ACIS-I, the current matrices are believed to be reliable for energies above 1.5 keV, or above 0.8 keV if one corrects for the effects of CTI.

Scott Wolk (CXC) reported on efforts to calibrate the optical sensitivity of the ACIS detectors. The ACIS detectors have optical blocking filters to reduce their sensitivity to optical photons. These filters are thinner for the ACIS-S in order to not reduce the soft X-ray sensitivity of the BI chips. The pre-launch measurements indicated that the optical sensitivity was very low. During early on-orbit commissioning, an observation of Vega suggested that the sensitivity of the S3 was 4 times greater than expected. However, a later observation of the region around Jupiter produced an image of Jupiter on the bias frame, and indicated that the sensitivity might be 90 times higher than the pre-launch estimates. A September 2000 calibration observation of Betelgeuse indicated that the sensitivity of the S3 is 16 times higher than expected. Laboratory work suggests that these differing result are due to increased transmission of the optical blocking filter near 8500 Angstroms (a red leak).

The higher optical sensitivity can affect observations of planets, bright stars, and other optically bright objects. Work arounds include doing the bias observations during the slew (now standard procedure), using shorter frame times, using the ACIS-I, changing the event threshold to eliminate low energy events (due to a number of optical photons), and using the zeroth order image with a grating in place (but this reduces the effective area).

Herman Marshall (CXC) reported on calibration of the HETGS. A number of systematic wavelength errors have been eliminated. The disagreement in the effective area measurements between the BI and FI chips is ~ 15%. The CUC were concerned by the outstanding problem of a significant but unexplained discrepancy evident between some HETG and LETG spectra and the simultaneous ASCA data.

Hank Donnelly (CXC) reported on the calibration of the HRC. The fix to the wiring problem for the timing has restored the timing resolution to about 15 microseconds. The quantum efficiency of the HRC is uncertain at low energies.

The CUC continued to be concerned that information on important calibration issues in not reaching many of the users. The committee recommended that the CXC staff involved in calibration should provide a short list of the major calibration issues which might affect the analysis of observations in a substantial manner, with a brief description of the problem and a crude estimate (one month, six months, one year, ...) of the time scale for resolution. This is likely to be an increasing problem with the public release of initially proprietary data.

CIAO 2.0 and Beyond

Martin Elvis (CXC) gave a short presentation on the newly released CIAO 2.0, and on plans for the future. CIAO 2.0 includes improvements to the algorithms for fitting and error analysis in Sherpa. The scripting language S-lang is provided. There is a beta version of the line identification tool GUIDE, which is an evolution of ISIS. The source detection routines deal with exposure variations across the field.

There also are improvement to the documentation. There is a new ahelp interface with "about" (an "apropos" feature). There is a new dictionary, which should be particularly helpful for graduate students and others new to X-ray astronomy and/or Chandra. There are a very extensive new set of software threads and scripts for many routine tasks.

CIAO 2.1 might be released in February 2001. It will include an enhancement of dmextract to extract counts from regions (similar to "imcounts" in IRAF/PROS).

The CUC was happy with the large number of new CIAO scripts, and the major improvements to documentation. In addition to some specific comments on a few tools, the committee recommended that an effort be made to keep the "major bug list" up to date, as this can save a great deal of effort by observers and by CXC staff in responding to questions by CIAO users. The CUC also noted that some of the tools require that the user pass parameters and values from the data files to the tools directly. These parameters should be read from the data files as the defaults (still giving the user the option to accept these specific defaults or change them).

Public Data Archive

Arnold Rots (CXC) reported briefly on the first public release of initially proprietary data, the rules for the public release of data, and the plans for the data archive. The CUC congratulated the CXC on the timely release of the first originally proprietary data to go public. At present, the best way to learn about data which has or is about to go public is to use the "Processing Status" tool. Since many users may not realize this, the CUC recommended that this be highlighted several places on the CXC webpages, particularly on the archive page.

At present, the access to the data archive is through a provisional interface, which doesn't allow very complex data searches. The regular archive tool should be released in March 2001. It will be a Java-based tool. Since such Java-based tools have sometimes created problems for users running unusual or outdated software systems, the CUC recommended that some sort of "bare bones" non-Java interface be maintained. This might be the current tool, or it might be done through the HEASARC.

New Members and Chair

Under the standard staffing plan for the CUC, four members rotate off each year. This year, these four members would be Jill Bechtold, Craig Sarazin (Chair), Mike Watson, and Ray White. The committee discussed the general plan for recruiting new members and selecting a new chair.

Next CUC Meeting and Telecon

The CUC decided to meet again in about six months, roughly in June 2001. Because some issues tend to come up requiring a quicker response, a telecon for February 2001 was also tentatively scheduled.