Minutes of Chandra Users' Committee Meeting, January 20-21, 2000


Committee members present were: Keith Arnaud, Jill Bechtold, JP Caillault, Megan Donahue, Craig Sarazin, Jane Turner, and Ray White. Mike Watson and Shri Kulkarni were connected for a portion of the meeting by telephone.

Ex-officio members present: Fred Seward and Allyn Tennant.

Alan Bunner was present from NASA HQ.


Status of Chandra

Harvey Tananbaum, CXC Director, reported on the status of Chandra after six months in orbit. The overall performance of the observatory has been excellent.

Operating software for enhanced safe modes to protect the ACIS from radiation damage either in the radiation belts or due to events at other times has been tested and uplinked in December. So far, it has not been activated. Software for enhanced dither mode (continuous dither) has been tested and will be uplinked soon. A patch to the operating software which will detect times when the ACIS Front End Processor 0 (FEP0) malfunctions and switch from 6 chips to 5 chips has been uplinked, and the normal operating mode will be to observe with 6 chips (except when FEP0 malfunctions) starting in late January. Small current glitches in one of the gyros have been detected, but are not believed to represent a danger to the observatory lifetime.

The Users' Committee had been concerned that the 1 year term of GO grants, the Chandra launch delays, and the uncertainty in the observation schedule have made it difficult for PIs to hire graduate students and post-docs. Tananbaum reviewed the grants policy and noted that 1-year no-cost extensions would be granted by CXC routinely, based on a request from the PI, a very brief progress report, and a statement that such an extension was needed to complete the analysis of the Chandra data. The CUC suggested that PIs be informed of this policy immediately via E-mail.

Tananbaum reported on the use of Director's Discretionary Time so far during the mission. Four such observations were approved.

Tananbaum summarized a large number of improvements to the interaction between the CXC and observers (both GOs and GTOs). One improvement is to assign a single member of the CXC scientific staff as a contact point for questions about data processing and software. Many of the other improvements are described below.

There were reports, from two deep field observations, of nonuniform distributions of X-ray sources, which raised a concern about a "hole in the center of ACIS-I". More detailed analyses and observations of the Coma cluster (which fills the detector) have not shown any obvious anomalies, but this question is still being addressed. There is also a concern about odd grade distributions in some regions of some of the ACIS detectors, but this appears to be the result of non-X-ray background events.


ACIS Detector Status

Mark Bautz (MIT) reported on the status of the ACIS detectors. A very considerable effort has gone into the CTI (Charge Transfer Inefficiency) increase produced by radiation damage in the FI (Frontside Illuminated) chips. Initially, the effort was to prevent further damage. Subsequent work has been aimed at identifying the causes of the deterioration, modeling the detector damage, and reducing the effects of the damage so as to achieve the best possible scientific performance. Since 20 September 1999, there has been almost no further degradation. The enhanced safe mode and the use of other spacecraft (particularly ACE) to monitor soft proton events should reduce the chances of more damage.

The damage is probably due to low energy protons (< 1 MeV) which are focussed by the mirrors on Chandra. The protons produce "traps" for electrons in the detectors. The physical nature of all of the traps produced is uncertain.

There are several strategies to reduce the effects of the damage. First, the temperature of ACIS will be reduced soon, and this will significantly improve the spectral resolution of the FI chips. The nominal frame time will be increased from 3.3 to 10 seconds, which will also reduce the effects of CTI. There are also schemes for unusual clocking of the FI ccds or charge injection, which are being studied and might be used within several months. There are plans to use the positions of cosmic-ray particles, which are rejected by the onboard software, to improve the determination of the CTI. Row-dependent grade rejection criteria maybe used with the onboard software.

Keith Arnaud (CUC) asked about the evidence that lowering the focal plane temperature might reduce the spectral resolution of the BI (backside illuminated) chips. Bautz showed data which indicated that this effect was either very small, or actually went the other way in the temperature range of interest (from -110 K to -123 K).


Analyzing ACIS Spectra

Martin Elvis (CXC) discussed efforts to include the effects of the CTI increase in the ACIS FI chips in the analysis of X-ray spectra. There is a substantial calibration and software effort to partially reduce the effects of the CTI increase with improved gain corrections, revised redistribution matrices, and grade reconstruction of events. The spatial variation of the spectral response and the number of different operating temperatures have resulted in a very large number of calibration files.

There is another problem associated with analyzing spectra from the BI S3 chip, which is being used more often than expected due to the CTI deterioration of the FI chips. Spatial variations in the spectra response mean that there are now 1024 response files for this detector, one for each 32x32 pixel region. This greatly complicates the analysis of the spectra of extended sources. Even for point sources, it can be difficult to determine the correct calibration file, and dithering may cause the source to be extended over several regions on the detector.

Elvis said that a script was being tested which would allow the automatic selection of the correct response file for each point on the detector. The ACIS temperature will eventually be included as a keyword in the fits headers of data files, allowing the automatic selection of the correct set of temperature-dependent response files.

Work is ongoing on a CIAO tool to construct weighted RMF files for extended sources or point sources dithered across node boundaries. Keith Arnaud has developed such a script for analyzing data with FTOOLS. The CIAO weighted RMF tool and the automatic selection of FEF calibration files will be released as a patch to CIAO 1.1 as soon as possible.

The CUC applauded these efforts, and recommended that they be given a very high priority. They also expressed concern that users needed advice on what to do in the interim, and also under what circumstances the very large number of calibration files are needed, and when a single response file is sufficient. Also, users have been given conflicting advise about whether spectra should be extracted in PHA or PI space. Much of the data distributed to observers or in the archive was derived with incorrect gain files, causing the PI values in the data to be incorrect.

Elvis noted that, whenever the instrument parameters (e.g., temperature) are changed, new calibration data must be acquired and new calibration files constructed. This takes several weeks. Thus, there will be a period when the pipeline processing of data uses the older incorrect gain file. This means that users will need to reprocess the events to generate correct PI values using the tool acis_process_events. Elvis and the CUC both agreed that improved instructions for doing this should be provided to users as soon as possible.

In response to the questions from the CUC about when all of the response matrices are needed and what users should do until a tool for constructing weighted RMFs is available, Mike Wise (CXC) showed the results of fits of spectra and analysis. His results showed that, for extended sources on the S3 chip, a single response matrix was adequate in most cases as long as spectra were analyzed in PI space and the correct gain file was applied. He also argued that the same was true to small regions of the FI chips.

The CUC recommended that clear advice on the use of PHA vs. PI spectra, the use of acis_process_events to correct the PI values, and the conditions under which a single response file is adequate be provided to users on the CXC webpage as soon as possible.


LETGS/HRC-S Issues

Jeremy Drake and Brad Wargelin (CXC) presented information of two instrumental issues associated with the LETGS and HRC-S. First, the quantum efficiency of the HRC-S at wavelengths greater than 50 Angstroms is about 1.6 times lower than the nominal value expected. Second, the total background in the detector is about 100 times larger than expected, mainly due to the failure of the anticoincidence shield. The background has a wider pulse height distribution than the X-ray events, and this can be used to reduce the background by about a factor of ~2 (with a loss of ~2% of the X-ray photons) or ~4 (with a loss of ~10% of the X-ray photons). Some other more complex BG filtering algorithms also show promise.

The increased background leads to near saturation of the telemetry rate for the instrument, and this can produce large dead time corrections to the flux. The present understanding of the dead time leads to errors in the corrected flux of about 10%.
Plans of AO-2 Release

Fred Seward presented the current time table for the AO-2 release for Chandra. The release date for the NRA has been delayed further to March 1, 2000. This delay is due to the time needed to review the instrument documents and insure that they reflect the actual behavior of the instruments in orbit. On this schedule, the deadline would be about May 31, and the peer review would occur in August. The list of approved targets would be available in early September. This schedule is rather tight. At the present observing efficiency, the list of approved AO-1 targets would be exhausted on approximately October 1. If some delay occurred in the review process, Chandra could start Cycle 2 by observing the GTO targets, which will be known shortly. However, Dan Schwartz (CXC) noted that it would be better if all of the targets were known at least one month before the start of Cycle 2, as this would permit the generation of a more efficient observing schedule.

Seward suggested that prior human technical reviews be dropped for AO-2. This would save about 2 weeks of time between the AO-2 deadline and the peer review. Because four of the CUC members could not attend this meeting, the CUC decided to vote on this again via E-mail. The vote was 2 for, 5 against, 4 abstaining. Thus, the CUC now recommends that prior human technical reviews NOT be done by the CXC for AO-2. The CUC asked that human technical reviews be available upon request by either the reviewers or the panel chairs for the period after the proposals have been sent to reviewers and prior to the panel meetings, and during the panel meetings themselves. They also suggested that the CXC provide the reviewers with information on typical technical problems and up to date information on the instruments at the same time that the proposals are sent to the reviewers. During the meetings of the review panels, the panels should be instructed that they should not reject a proposal on purely technical grounds without consulting with a technical expert from the CXC.

Ray White (CUC) had expressed a concern about the effects of science conflicts during the peer review process. Such conflicts would occur when a panelist is a PI or Co-I on a proposal with the same major science objectives as a proposal being reviewed, although different targets are being observed. While there will often be some level of conflict between proposals in the same area associated with the secondary aims of the proposals, White's concern were about major conflicts where the primary aims of the proposals are the same. By a vote of 4 for, 1 against, 2 abstaining, the CUC decided to recommend that such conflicts be considered during the review. First, significant science conflicts would be self-identified by the panelist or identified by the panel chair. The affected panelist should be asked to leave the room briefly near the end of the discussion of a particular proposal, while the panel discusses the possible affect of this conflict. The affected panelist will be allowed to vote.


CIAO Data Analysis Software

Elvis presented the schedule for new releases of CIAO. As discussed above, a patch with tools for automatic selection of ACIS response files and for the analysis of extended source ACIS spectra will be produced as soon as possible. The next major release of CIAO (2.0) is nominally planned for the summer of 2000, but may be delayed due to the time pressure for other tasks. Among the new features in CIAO 2.0 would be the automatic identification of line features in spectra. Another was fortran wrappers for tools which would allow outside development of software in fortran. There is a long list of possible improvements to CIAO, but the finite resources available mean that they must be prioritized carefully and the most critical ones done first.

Although not currently incorporated into CIAO, timing tools for barycentric corrections and time conversion (from s/c to MJD, for example) are available from CXC upon request.

The CUC discussed their reactions to CIAO 1.1, which were generally very favorable. People were quite impressed by the power of the tools, and the facilities for complex filtering and binning of data.

There was a brief discussion of compatibility of CIAO with FTOOLS and XSPEC. The fix for the TLMIN and TLMAX keywords will ready soon. An upgraded version of dmextract which includes a WMAP is planned. Sarazin (CUC) noted that the BACKSCAL keywords was not being set correctly by dmextract, at least in some cases.

Two areas where the CUC felt effort should not be expended were porting CIAO to IRAF, or maintaining compatibility with PROS X-ray software.


CIAO Documentation

Belinda Wilkes (CXC) discussed the documentation on Chandra data processing which was released with CIAO 1.1. There is a Beginner's Guide for CIAO, and manuals for most of the major software packages. At this point, there is only a draft of the Calibration Products Guide, and the Chandra Data Products Guide hasn't been produced. However, a basic description of the Chandra data files is given in the Beginner's Guide.

The CUC was very concerned that "walk throughs" are needed for all of the basic initial data analysis procedures (e.g., extracting a spectra from ACIS data). Jane Turner (CUC) was particularly concerned that it was very difficult to understand how to start analyzing grating data. Another area where people were having trouble was in reprocessing events with the correct gain file (see above) with acis_process_events.

The CUC strongly recommended that illustrative "walk throughs" be compiled for the dozen or so basic data analysis threads.

During this discussion and others between the CXC staff and the CUC, it became clear that a major problem was that the CXC staff had not had any time to use the CIAO software to analyze data. Thus, writing the documentation (particularly the walk throughs) could not arise as a natural part of using the software to do science. The tremendous time pressures on the CXC at the start of the mission have made it very difficult for staff to take the science time which is part of their contract. The CUC felt that it was important that the science staff have the science time to use the Chandra software to work on data.


Observation Scheduling

The operation of observation scheduling and interaction with the PIs has been improved recently, as discussed by Dan Schwartz and Bill Forman (CXC). The Long Term Schedule is now relatively stable. Observations which appear in the schedule, which is accessible from the web, will almost certainly be observed at essentially the time shown in the schedule. In addition, the schedule contains a "pool" of unconstrained targets which can be observed at almost any time. The scheduled targets in the Long Term Schedule only use up a fraction of the observing time, and the rest is filled with "pool" targets to maximize observing efficiency. The "pool" target are several large groups at the end of the Long Term Schedule, which appear as if they were very long weeks of observing.

The CUC members said that they had not understood that these targets at the end of the Long Term Schedule were "pool" targets, and were not really scheduled for the end of Cycle 1. These "pool" targets could be observed at any time, generally much earlier than the end of Cycle 1. The CUC recommended that this be made clear to PIs.

Forman and Schwartz said that all PIs should now look at the schedule, determine when their targets were scheduled or if they were "pool" targets, and look over the observation parameters for their targets.

A contact for the CXC will contact each PI, generally several months before the observation. They will discuss the observing parameters with the observer, and may recommend (but not dictate) instrument configurations. The CXC will inform the PI when the observation has been scheduled, if the scheduled observation is postponed, and when the data is on the ground.

Fred Seward (CXC) asked the CUC members about several web tools for determining the status of observations. The CUC members noted that the Java OBSCAT had not worked properly for several months. There is a nice tool for checking observing parameters and viewing the instrumental field on a DSS optical image or ROSAT X-ray image. One problem is that the observing parameters are not always updated very quickly. Thus, a PI who has changed the parameters for a target may become concerned that the observation hasn't been changed.


Data Processing and Archiving

Pepi Fabbiano (CXC) described the process of pipeline data processing and archiving, and presented statistics on the results. The data receipt process is current, and is generally working well. The Automated Processing (AP) of the data has been fairly stable recently. The elapsed time between receipt and processing is typically 1-2 weeks. Processing speed is five times real time when AP is working well. After AP, the data is ingested into the archive. Science Verification and Validation (V&V) typically takes less than 1 week. There is an effort to reduce the V&V time by involving all of the Science Data System scientists in V&V, and by writing scripts to automate the paperwork and routine functions. Following V&V, propriety data is distributed to the PI and public data is accessible.

In February, the CXC will begin reprocessing all of the older data. It will require about 2 months to complete the reprocessing. One concern is that the reprocessing will double the V&V load, and this will strain the time of CXC staff further.

A new web tool to determine the processing status of all observations received in telemetry is accessible from the CXC webpage. Automatic E-mail notification to the PI of the receipt of data will start next week.


Webpage Improvements

Belinda Wilkes and Andrea Prestwich (CXC) presented improvements to the CXC webpage which are are aimed at making up to date information of observations, instruments, and data analysis more accessible to astronomers. These included a reorganization of the information of Chandra Target and Scheduling Info, a "Things to Watch Out For" page, and a "What's New" page. The "Things to Watch Out For" page is divided into two sections for observation planning and for data analysis, and lists instrumental anomalies, problems with the data, major bugs with analysis software, and other things which an observer needs to know as he plans his/her observation or starts to reduce his/her data. The "What's New" page lists the changes to the Chandra webpages, so that a user can quickly determine if there is something new he/she needs to check. The CUC was generally very pleased with these changes.


Chandra Fellows Program

Nancy Evans (CXC) gave a brief update on the Chandra Fellows program. The Selection Panel had just finished its meetings a few days before the CUC met. The CXC hosted the Chandra Fellows Symposium in the fall, which was very successful. The program continues to attract some of the most talented young astronomers at U.S. institutions.


Emission Line Project

Nancy Brickhouse (CXC) reported on the Emission Line Project. The project accepted 15 Level 2 proposal for participation. Level 2 participation includes funding and is limited to US institutions. A number of Level 1 proposals (no funding, but participation in the ELP effort) were accepted, including foreign institutions.


Press Relations and Outreach

Kathy Lestition (CXC) discussed the efforts at public and press relations since the time of the launch. If anything, Chandra has been more successful at getting press coverage than expected, thanks in part to the very spectacular images. From the public calibration images, 10 enhanced versions were released with 10 accompanying press releases. There was a NASA HQ press conference for the first light Cas A image, and a Space Sciences Update conference on the Crab image. The Cycle 1 GTO and GO observations have so far yielded 8 public relation images, with 6 accompanying press releases and 2 press conferences at the January AAS meeting. A story from a AAS Chandra press conference appeared on the front page of the New York Times, and there have been numerous stories within the Times, Washington Post, and Life magazine.

The Chandra public relations and outreach websites have won a large number of awards; these awards are useful, since they can lead to the websites being linked on collections of the best websites which increases their visibility.


TOO Observations by GTOs in Cycle 2:

Just prior to the meeting, the CUC and CXC received a proposal by George Ricker (MIT). He wanted to use a portion of his Cycle 2 GTO time to do follow-up TOO observations of GRBs detected with HETE II. The current Chandra Observing Policy does not allow GTOs in Cycle 1 or 2 to propose TOO observations. This policy was adopted with the very strong support of the Users' Committee. The concern was that GTO select their targets first. Since the specific TOO targets are not decided in advance (only the triggers for selecting them) and the total number of TOOs is very limited because of the problems they posed for efficient observation scheduling, the CUC felt that GTO TOOs would allow GTO to lock up entire classes of observations with almost no possible competition from GOs. During Cycles 1 and 2, GTO could propose TOO observations to the normal GO observation selection. After Cycle 2, GTOs and GO compete together for all targets, so there is no restriction against GTOs proposing TOO observations.

However, the CUC was very interested in the proposal by Ricker, because he proposed that the Chandra data in question would be made public immediately, and the HETE II data will be public as well. Moreover, HETE II is a time-critical resource. Thus, the CUC voted unanimously to recommend that the CXC make a exception to the observing policy. In fairness, the CUC suggested that the CXC should consider other Cycle 2 GTO TOO proposals. However, the CUC very strongly recommended that any accepted proposals should make the Chandra data public immediately. Because of the very limited number of allowed TOO observations, the CUC also recommended that the total number of GTO TOO targets be limited to the fraction of total TOO targets which is equal to the proportion of Cycle 2 time which goes to GTOs. In practice, this would lead to at most 3 GTO TOOs.


Several Controversies

Tananbaum (CXC Director) discussed the 50 ksec HRC-I observations of Cas-A which was scheduled as a follow-up to the discovery of a point source at the center of the remnant in the Chandra first light image. This observation was requested by Steve Murray (CfA), in collaboration with a group of astronomers at several institutions. The primary purpose was to search for pulsations from the point source; this can't be done very well with the ACIS observations because of the frame time in the ccd. A secondary purpose was to compare the higher resolution image of filaments in the remnant with the longest ACIS images, which are being made in an observation with Steve Holt (GSFC). Holt is a Co-I on the HRC-I Cas A image. Tananbaum decided to schedule the HRC-I observation because of the great scientific importance in determining the nature of the central source in Cas-A. The time was taken from Murray's Cycle 1 GTO time, rather than being done with Director's Discretionary Time. Tananbaum noted that there was no approved Cycle 1 HRC-I observations of Cas-A, and that Murray could have selected this observation as part of his Cycle 2 GTO time with no competition from GOs. Thus, the main effect of the Director's decision to schedule this observation was to provide the data this year rather than next. Marty Weisskopf (MSFC, Mission Scientist) concurred with Tananbaum on this decision.

In general, the CUC agreed with the scientific value of this observation. They suggested that it might have been good to have informed the CUC about this decision, since it has aroused some controversy.

There was also a brief discussion of some controversies concerning the AAS press conference on resolving the X-ray background with Chandra. The CXC was not directly involved in this matter.