Minutes of Chandra Users' Committee Meeting, June 12-13, 2000

Committee members present were: Keith Arnaud, Jill Bechtold, Adrienne Cool, 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 and Don Kniffen (NASA HQ) were connected via telephone for the first morning.

Status of Chandra

Roger Brissenden (CXC Manager) reported on the status of Chandra after almost a year in orbit. The overall performance of the observatory has been excellent. There have been only three safe mode events. The patches for safing the instruments during the radiation belt passage, for fixing the FEP0 problem, and for continuous dither are working well. The observatory was safed during a major solar event. It operated correctly through two eclipse seasons. There has been one command load error problem, but every command load was delivered on time. Observing efficiency is currently around 63%, but may improve somewhat. A limit switch on the gratings failed, but redundant hardware is being used, and the gratings can be operated without these switches if necessary.

Harvey Tananbaum (CXC Director) reported on the use of Director's Discretionary Time (DDT). There have been nine DDT observations so far. The CUC expressed the desire that DDT observations be publicized via a more prominent webpage and E-mails sent via an E-mail exploder.

HRC Timing Problem

Steve Murray (PI, HRC) reported on a hardware problem with timing on the HRC-I and HRC-S cameras. This was discovered as an apparent broadening and/or phase shift in the pulse profiles of pulsars. This is due to incorrect wiring in the detectors which assigns the time for a given event to the next event to occur. This could be fixed in software, except for the fact that the HRC has a high rate of background events, which normally are not telemetered to the ground. The background rate is high enough that it would saturate the telemetry if these events were not rejected on the spacecraft. Because the average time between events on the HRC-I is 3-4 msec, this introduces an error of this order into the timing, which can appear as lowered time resolution and/or a phase shift.

At present, all unobserved Cycle-1 HRC targets involving msec timing have been put on an indefinite hold.

Possible remediations include operating the detectors without onboard event rejection (which would lower the observing efficiency) and/or observing the source using only the center portion of the HRC-S detector. These are being tested on pulsars. The CUC recommended that the tests include at least one very rapid pulsar to demonstrate most directly that msec timing accuracy had been restored.

There are at least 8 existing observations which are strongly affected by this problem. They may be reobserved if the remedial approaches work.

This problem was discovered just prior to the deadline for Cycle-2 proposals. At that point, the possibilities to fix the problem were uncertain. A decision was made by the CXC Director not to publicize the problem prior to the deadline to avoid confusing the proposal submission process. The CUC was satisfied with this decision, but recommended that the community be notified via the web and E-mail now that the deadline is past. The CUC recommended a technical review of Cycle-2 HRC proposals to determine how they are affected by the timing problems and the potential fixes.

The CUC recommended a procedure to determine which Cycle-1 observations should be repeated (see Recommendations to Director)

Cycle-2 Proposals and Review

Fred Seward (CXC) presented statistics on Cycle-2 proposals and the draft of plans for the Cycle-2 review. The CUC discussed a draft report form for the review panels, and suggested a few minor changes. The CUC discussed the plans to review Large Proposals, Archive Proposals, and Joint CXO/HST or CXO/NOAO proposals. They recommended plans for the review of each of these new classes of proposals, which are listed in the Recommendations to the Director. The CUC concurred with the plan to have approximate total costs for the Archive Proposals available to the review.

ACIS: CTI Problem Remediation

Mark Bautz (MIT) reported on tests of the so-called "squeegee" mode of operation of the ACIS detectors, which uses charge injection to fill some of the radiation-produced traps which increase the Charge Transfer Inefficiency (CTI). It appears this this could reduce the CTI further by 30% or more, beyond the gains produced by lowering the temperature. The timescale for implementation, if all goes well, might be 3-4 months. There are several ways to implement this mode, but the requirement that the detectors be calibrated for each mode is likely to mean that that only one will be implemented. The CUC concurred on this, but also said that it was important (for grating observations and some ACIS S-3 observations) that it be possible to turn the squeegee mode off.

ACIS: Calibration

There was a free-form discussion of calibration issues on ACIS. The CUC asked about the low energy calibration, particularly for the S3. After rather complicated discussion, the CUC recommended that the CXC give this problem a higher priority, and that the users be informed more clearly about this and other calibration issues.

ACIS: Background

Maxim Markevitch (CXC) discussed the "blank-sky" background events files and software he has developed. The CUC was very happy with this. The current files are screened down to quiescent background, which is good for most observations. However, the CUC asked if unscreened events files with the background level tagged could be provided for observations with a large number of background flares.

ACIS: Spectra

At its last meeting, the CUC had requested that users be told under what circumstances to extract spectra in PI vs. PHA space. The CXC webpage on this subject is still blank. After a some complicated discussion, the CUC again asked that the users be provided with the best available information on this question.

At its January 2000 meeting, the CUC had expressed a very strong concern the difficulty of extracting spectra of extended sources with ACIS because of the very large number (in some cases, 1024) of response files needed for each chip. Although it is in principal possible for the users to extract 1024 spectra and compute 1024 response matrices for an extended source, scripts to do this in an automatic way for extended source regions are desperately needed. Martin Elvis (CXC) said that other time commitments had prevented the CXC from making any progress on this problem. The CUC and Elvis noted that some users (particularly Alexey Viklinhin) have written scripts to do the extended source spectral extraction and response matrix generation. Some CXC staff argued that the Viklinhin scripts were not optimal, but the CUC insisted that some solution to this problem must be provided. They asked that the CXC either immediately develop their own tools, or that they very actively support the distribution of tools written by users for this basic and vital task.

Aspect Issues

Tom Aldcroft (CXC) gave a report on number of pointing and aspect problems that had occurred with Chandra data so far. In each case, a fix was known and either had been applied in custom processing or would be applied in reprocessing of the data.

Jonathon McDowell (CXC) reported on the randomization (+/- 0.5 pixels or 0.25 arcsec) of pixel positions of ACIS events which is applied during standard processing of observations. This is done during the transformation from chip to sky coordinates in order to avoid aliasing effects. At present, ACIS event locations are only determined by the central chip pixel value and not by the grade distribution, so they are quantized at the one pixel level. Some users had expressed a concern this randomization would lower the spatial resolution of Chandra ACIS observations. The CUC agreed with the CXC that this randomization was appropriate for most observations and observers. The randomization can be removed by the user through reprocessing the data with acis_process_events. A brief description of the method to do this is given on the CXC website. The CUC suggested that this thread be expanded to include all parameters for acis_process_events, as this is a very powerful tool which many users find intimidating.

Data Analysis Software

Martin Elvis (CXC) reported on the plans for new releases of CIAO. CIAO 2.0 is about to start testing, with release likely in October 2000. The CUC asked about adding a script to handle the functions of IMCNTS in IRAF/PROS. The basic function is to determine the number of counts from a source, but IMCNTS handles multiple regions and background regions, so that it can be used for multiple sources, surface brightness profiles, etc. The CIAO tool DMSTAT can perform this basic function, but this use is not obvious to most users. The CUC noted the the light curve tools also apparently contains this functionality, but again is is not obvious how to use it for this purpose. The CUC suggested that a simple script or science thread be written to allow users to extract background corrected counts from one or many source regions.

User Support

Tananbaum reported on the effects of the reorganization of the User Support functions at CXC, which seems to have improved this area. Andrea Prestwich (CXC) gave a brief report on HelpDesk responses and on changes to the CXC website to improve information for users.

Chandra Grants

Alexandra Rollins (CXC) gave a brief report on the Chandra grants program, which seems to be going well. Most grants arrive within a few weeks after the data is distributed to the PI. One early complication was that it was found that the CXC (through the SAO) could not fund directly PIs at NASA institutions, and that these grants had to be distributed by MSFC. Craig Sarazin (CUC) asked if the Federal Demonstration Project, which simplifies the administration of grants, could be extended to the Chandra grants. Rollins said that this was not possible, as the SAO was not really a federal agency.

Setting Final Observational Parameters

Pat Slane (CXC) reported on the problems which have occurred because a small fraction of Chandra observers do not respond in a timely manner to requests for the final observations parameters for their observations. This has often required that CXC staff go to great lengths to try to reach these observers. Of greatest concern is the delays in the construction of the command loads for the CXO and the danger of errors associated with last minute changes. As noted above, the CXC hasn't ever missed a command upload deadline, but delays and last minute changes could result in a significant loss of observing time. They already place a terrible burden on the CXC operations staff. The CUC was appalled to hear about these cases of irresponsibility on the part of observers.

Slane suggested a new policy for setting final observational parameters. Each PI would be contacted well in advance of the observation (several months for observations in the long term timeline, and at the start of the timeline for pool observations). Their final observing parameters would be requested, and they would be given an absolute final deadline for submission of these parameters. If they failed to sign off on their final parameters by this deadline, their observation would be removed from the observing timeline. The observation might be delayed for one year, since the same target would generally be observable at that time. Users could appeal this decision to the CXC Director, who would only grant appeals under extraordinary circumstances.

The CUC supported this draconian but apparently necessary change. They requested that the new policy and deadline be made very clear near the beginning of any communication to the PI.

Sarazin expressed a concern about ACIS pool targets which are extended sources. In many cases, observers can optimize the placement of the source on the detector if the roll angle (which depends on the date of the observation) is known. If these observers are required to provide their final observational parameters before the observing date is known (even approximately), this optimization would not be possible. In the end, no really satisfactory way to deal with these cases was suggested, unless they were limited in number and noted in the comments area of the proposal target forms.

Data Pipeline

Pepi Fabbiano (CXC) reported on the pipeline processing of observations. This is now proceeding smoothly, and the time between the receipt of data at CXC and its distribution to the PI is about 3 weeks on average. This can probably be shortened further. The bottleneck in V&V, which requires a lot of human effort, has been greatly reduced, but may reappear once reprocessing starts.

Reprocessing of all Cycle-1 data has been split into two stages. The first stage which deals with Level 0 processing of raw telemetry, filling of telemetry gaps in the original processing, and improved time-tagging of the telemetry data was begun in May. Level 1 and 2 re-processing has been delayed due to additional patches needed for the processing software and due to personnel assignments to other problems, and is now expected to start in July 2000.

Next CUC Meeting and Telecons

Tananbaum asked that the CUC meet again in six months, and the members agreed that the current rate of change with Chandra required meetings on a 6 month interval. The CUC tentatively decided on a December 2000 meeting. The CUC also asked to have a monthly, one hour long telecon with the CXC Director to monitor progress and be informed of new developments.